diff --git a/Assets/FSR3 Upscaler Assets.asset b/Assets/FSR3 Upscaler Assets.asset new file mode 100644 index 0000000..b3d83b8 --- /dev/null +++ b/Assets/FSR3 Upscaler Assets.asset @@ -0,0 +1,26 @@ +%YAML 1.1 +%TAG !u! tag:unity3d.com,2011: +--- !u!114 &11400000 +MonoBehaviour: + m_ObjectHideFlags: 0 + m_CorrespondingSourceObject: {fileID: 0} + m_PrefabInstance: {fileID: 0} + m_PrefabAsset: {fileID: 0} + m_GameObject: {fileID: 0} + m_Enabled: 1 + m_EditorHideFlags: 0 + m_Script: {fileID: 11500000, guid: 2dbcc608a4754d049a14a0bcce2eb40b, type: 3} + m_Name: FSR3 Upscaler Assets + m_EditorClassIdentifier: + shaders: + prepareInputsPass: {fileID: 7200000, guid: 4f59e5b9179d74844ae06a30ae1e0629, type: 3} + lumaPyramidPass: {fileID: 7200000, guid: d253be05abcdc80428503d3e4cce3a36, type: 3} + shadingChangePyramidPass: {fileID: 7200000, guid: 251e663738905fa4d8817001682d802f, type: 3} + shadingChangePass: {fileID: 7200000, guid: 9a2bff2f97619ed4989d9b0577ba0641, type: 3} + prepareReactivityPass: {fileID: 7200000, guid: 20e44016ed34b0d4b8de499d1b566c69, type: 3} + lumaInstabilityPass: {fileID: 7200000, guid: a135306e6d1857e43a86ef20db2a47fe, type: 3} + accumulatePass: {fileID: 7200000, guid: c9b45f0ae7673694ba57a4aadfe212e9, type: 3} + sharpenPass: {fileID: 7200000, guid: 7aaf5cfff022de2499e9b0412f947f6c, type: 3} + autoGenReactivePass: {fileID: 7200000, guid: 5716b91fdaa4e9e439df6b96a796fe6e, type: 3} + tcrAutoGenPass: {fileID: 7200000, guid: 75cdc6ef23f08ed498d4da511923fcea, type: 3} + debugViewPass: {fileID: 7200000, guid: cb24a71d54164c54eb5e86839acd48c5, type: 3} diff --git a/Assets/Fsr3UpscalerAssets.asset.meta b/Assets/FSR3 Upscaler Assets.asset.meta similarity index 100% rename from Assets/Fsr3UpscalerAssets.asset.meta rename to Assets/FSR3 Upscaler Assets.asset.meta diff --git a/Assets/Fsr3UpscalerAssets.asset b/Assets/Fsr3UpscalerAssets.asset deleted file mode 100644 index f3f30ea..0000000 --- a/Assets/Fsr3UpscalerAssets.asset +++ /dev/null @@ -1,23 +0,0 @@ -%YAML 1.1 -%TAG !u! tag:unity3d.com,2011: ---- !u!114 &11400000 -MonoBehaviour: - m_ObjectHideFlags: 0 - m_CorrespondingSourceObject: {fileID: 0} - m_PrefabInstance: {fileID: 0} - m_PrefabAsset: {fileID: 0} - m_GameObject: {fileID: 0} - m_Enabled: 1 - m_EditorHideFlags: 0 - m_Script: {fileID: 11500000, guid: db26e15a33db6ab42a38daab0ba2712f, type: 3} - m_Name: Fsr3UpscalerAssets - m_EditorClassIdentifier: - shaders: - computeLuminancePyramidPass: {fileID: 7200000, guid: d253be05abcdc80428503d3e4cce3a36, type: 3} - reconstructPreviousDepthPass: {fileID: 7200000, guid: 4f59e5b9179d74844ae06a30ae1e0629, type: 3} - depthClipPass: {fileID: 7200000, guid: 20e44016ed34b0d4b8de499d1b566c69, type: 3} - lockPass: {fileID: 7200000, guid: a135306e6d1857e43a86ef20db2a47fe, type: 3} - accumulatePass: {fileID: 7200000, guid: c9b45f0ae7673694ba57a4aadfe212e9, type: 3} - sharpenPass: {fileID: 7200000, guid: 7aaf5cfff022de2499e9b0412f947f6c, type: 3} - autoGenReactivePass: {fileID: 7200000, guid: 5716b91fdaa4e9e439df6b96a796fe6e, type: 3} - tcrAutoGenPass: {fileID: 7200000, guid: 75cdc6ef23f08ed498d4da511923fcea, type: 3} diff --git a/Assets/Scenes/SampleScenePPV2.unity b/Assets/Scenes/SampleScenePPV2.unity index 600d5e0..578bb3b 100644 --- a/Assets/Scenes/SampleScenePPV2.unity +++ b/Assets/Scenes/SampleScenePPV2.unity @@ -228,10 +228,8 @@ GameObject: - component: {fileID: 963194228} - component: {fileID: 963194227} - component: {fileID: 963194226} - - component: {fileID: 963194234} - component: {fileID: 963194232} - component: {fileID: 963194233} - - component: {fileID: 963194229} m_Layer: 6 m_Name: Main Camera m_TagString: MainCamera @@ -305,40 +303,6 @@ Transform: m_Father: {fileID: 0} m_RootOrder: 0 m_LocalEulerAnglesHint: {x: 9.931, y: 109.145, z: 0} ---- !u!114 &963194229 -MonoBehaviour: - m_ObjectHideFlags: 0 - m_CorrespondingSourceObject: {fileID: 0} - m_PrefabInstance: {fileID: 0} - m_PrefabAsset: {fileID: 0} - m_GameObject: {fileID: 963194225} - m_Enabled: 1 - m_EditorHideFlags: 0 - m_Script: {fileID: 11500000, guid: 955cb66a9ecc20441a7e32934c9b4690, type: 3} - m_Name: - m_EditorClassIdentifier: - qualityMode: 2 - performSharpenPass: 1 - sharpness: 0.8 - enableFP16: 0 - enableAutoExposure: 1 - preExposure: 1 - exposure: {fileID: 0} - reactiveMask: {fileID: 0} - transparencyAndCompositionMask: {fileID: 0} - autoGenerateReactiveMask: 1 - generateReactiveParameters: - scale: 0.5 - cutoffThreshold: 0.2 - binaryValue: 0.9 - flags: 13 - autoGenerateTransparencyAndComposition: 0 - generateTransparencyAndCompositionParameters: - autoTcThreshold: 0.05 - autoTcScale: 1 - autoReactiveScale: 5 - autoReactiveMax: 0.9 - assets: {fileID: 11400000, guid: 4151befafd86e8740ab09463b4f1bdbb, type: 2} --- !u!114 &963194232 MonoBehaviour: m_ObjectHideFlags: 0 @@ -357,12 +321,35 @@ MonoBehaviour: m_Bits: 64 stopNaNPropagation: 0 finalBlitToCameraTarget: 0 - antialiasingMode: 0 + antialiasingMode: 4 temporalAntialiasing: jitterSpread: 0.75 sharpness: 0.25 stationaryBlending: 0.95 motionBlending: 0.85 + superResolution: + qualityMode: 2 + performSharpenPass: 1 + sharpness: 0.8 + enableFP16: 0 + exposureSource: 1 + preExposure: 1 + exposure: {fileID: 0} + enableDebugView: 0 + reactiveMask: {fileID: 0} + transparencyAndCompositionMask: {fileID: 0} + autoGenerateReactiveMask: 1 + generateReactiveParameters: + scale: 0.5 + cutoffThreshold: 0.2 + binaryValue: 0.9 + flags: 13 + autoGenerateTransparencyAndComposition: 0 + generateTransparencyAndCompositionParameters: + autoTcThreshold: 0.05 + autoTcScale: 1 + autoReactiveScale: 5 + autoReactiveMax: 0.9 subpixelMorphologicalAntialiasing: quality: 2 fastApproximateAntialiasing: @@ -397,6 +384,7 @@ MonoBehaviour: m_ShowCustomSorter: 0 breakBeforeColorGrading: 0 m_BeforeTransparentBundles: [] + m_BeforeUpscalingBundles: [] m_BeforeStackBundles: [] m_AfterStackBundles: [] --- !u!114 &963194233 @@ -416,18 +404,6 @@ MonoBehaviour: blendDistance: 0 weight: 1 priority: 1 ---- !u!114 &963194234 -MonoBehaviour: - m_ObjectHideFlags: 0 - m_CorrespondingSourceObject: {fileID: 0} - m_PrefabInstance: {fileID: 0} - m_PrefabAsset: {fileID: 0} - m_GameObject: {fileID: 963194225} - m_Enabled: 1 - m_EditorHideFlags: 0 - m_Script: {fileID: 11500000, guid: 56db3edb28006274aab71644ef3e4f35, type: 3} - m_Name: - m_EditorClassIdentifier: --- !u!1001 &1313173313 PrefabInstance: m_ObjectHideFlags: 0 diff --git a/Assets/Scripts/Core.meta b/Assets/Scripts/Core.meta deleted file mode 100644 index b102114..0000000 --- a/Assets/Scripts/Core.meta +++ /dev/null @@ -1,8 +0,0 @@ -fileFormatVersion: 2 -guid: 975411b519480574290d74d7e34d119d -folderAsset: yes -DefaultImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Scripts/Core/Fsr3ShaderIDs.cs b/Assets/Scripts/Core/Fsr3ShaderIDs.cs deleted file mode 100644 index 3a28843..0000000 --- a/Assets/Scripts/Core/Fsr3ShaderIDs.cs +++ /dev/null @@ -1,80 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -using UnityEngine; - -namespace FidelityFX -{ - internal static class Fsr3ShaderIDs - { - // Shader resource views, i.e. read-only bindings - internal static readonly int SrvInputColor = Shader.PropertyToID("r_input_color_jittered"); - internal static readonly int SrvOpaqueOnly = Shader.PropertyToID("r_input_opaque_only"); - internal static readonly int SrvInputMotionVectors = Shader.PropertyToID("r_input_motion_vectors"); - internal static readonly int SrvInputDepth = Shader.PropertyToID("r_input_depth"); - internal static readonly int SrvInputExposure = Shader.PropertyToID("r_input_exposure"); - internal static readonly int SrvAutoExposure = Shader.PropertyToID("r_auto_exposure"); - internal static readonly int SrvReactiveMask = Shader.PropertyToID("r_reactive_mask"); - internal static readonly int SrvTransparencyAndCompositionMask = Shader.PropertyToID("r_transparency_and_composition_mask"); - internal static readonly int SrvReconstructedPrevNearestDepth = Shader.PropertyToID("r_reconstructed_previous_nearest_depth"); - internal static readonly int SrvDilatedMotionVectors = Shader.PropertyToID("r_dilated_motion_vectors"); - internal static readonly int SrvPrevDilatedMotionVectors = Shader.PropertyToID("r_previous_dilated_motion_vectors"); - internal static readonly int SrvDilatedDepth = Shader.PropertyToID("r_dilated_depth"); - internal static readonly int SrvInternalUpscaled = Shader.PropertyToID("r_internal_upscaled_color"); - internal static readonly int SrvLockStatus = Shader.PropertyToID("r_lock_status"); - internal static readonly int SrvLockInputLuma = Shader.PropertyToID("r_lock_input_luma"); - internal static readonly int SrvPreparedInputColor = Shader.PropertyToID("r_prepared_input_color"); - internal static readonly int SrvLumaHistory = Shader.PropertyToID("r_luma_history"); - internal static readonly int SrvRcasInput = Shader.PropertyToID("r_rcas_input"); - internal static readonly int SrvLanczosLut = Shader.PropertyToID("r_lanczos_lut"); - internal static readonly int SrvSceneLuminanceMips = Shader.PropertyToID("r_imgMips"); - internal static readonly int SrvUpscaleMaximumBiasLut = Shader.PropertyToID("r_upsample_maximum_bias_lut"); - internal static readonly int SrvDilatedReactiveMasks = Shader.PropertyToID("r_dilated_reactive_masks"); - internal static readonly int SrvPrevColorPreAlpha = Shader.PropertyToID("r_input_prev_color_pre_alpha"); - internal static readonly int SrvPrevColorPostAlpha = Shader.PropertyToID("r_input_prev_color_post_alpha"); - - // Unordered access views, i.e. random read/write bindings - internal static readonly int UavReconstructedPrevNearestDepth = Shader.PropertyToID("rw_reconstructed_previous_nearest_depth"); - internal static readonly int UavDilatedMotionVectors = Shader.PropertyToID("rw_dilated_motion_vectors"); - internal static readonly int UavDilatedDepth = Shader.PropertyToID("rw_dilated_depth"); - internal static readonly int UavInternalUpscaled = Shader.PropertyToID("rw_internal_upscaled_color"); - internal static readonly int UavLockStatus = Shader.PropertyToID("rw_lock_status"); - internal static readonly int UavLockInputLuma = Shader.PropertyToID("rw_lock_input_luma"); - internal static readonly int UavNewLocks = Shader.PropertyToID("rw_new_locks"); - internal static readonly int UavPreparedInputColor = Shader.PropertyToID("rw_prepared_input_color"); - internal static readonly int UavLumaHistory = Shader.PropertyToID("rw_luma_history"); - internal static readonly int UavUpscaledOutput = Shader.PropertyToID("rw_upscaled_output"); - internal static readonly int UavExposureMipLumaChange = Shader.PropertyToID("rw_img_mip_shading_change"); - internal static readonly int UavExposureMip5 = Shader.PropertyToID("rw_img_mip_5"); - internal static readonly int UavDilatedReactiveMasks = Shader.PropertyToID("rw_dilated_reactive_masks"); - internal static readonly int UavAutoExposure = Shader.PropertyToID("rw_auto_exposure"); - internal static readonly int UavSpdAtomicCount = Shader.PropertyToID("rw_spd_global_atomic"); - internal static readonly int UavAutoReactive = Shader.PropertyToID("rw_output_autoreactive"); - internal static readonly int UavAutoComposition = Shader.PropertyToID("rw_output_autocomposition"); - internal static readonly int UavPrevColorPreAlpha = Shader.PropertyToID("rw_output_prev_color_pre_alpha"); - internal static readonly int UavPrevColorPostAlpha = Shader.PropertyToID("rw_output_prev_color_post_alpha"); - - // Constant buffer bindings - internal static readonly int CbFsr3Upscaler = Shader.PropertyToID("cbFSR3Upscaler"); - internal static readonly int CbSpd = Shader.PropertyToID("cbSPD"); - internal static readonly int CbRcas = Shader.PropertyToID("cbRCAS"); - internal static readonly int CbGenReactive = Shader.PropertyToID("cbGenerateReactive"); - } -} diff --git a/Assets/Scripts/Core/Fsr3Upscaler.cs b/Assets/Scripts/Core/Fsr3Upscaler.cs deleted file mode 100644 index c636c93..0000000 --- a/Assets/Scripts/Core/Fsr3Upscaler.cs +++ /dev/null @@ -1,333 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -using System; -using System.Runtime.InteropServices; -using UnityEngine; -using UnityEngine.Rendering; - -namespace FidelityFX -{ - /// - /// A collection of helper functions and data structures required by the FSR3 Upscaler process. - /// - public static class Fsr3Upscaler - { - /// - /// Creates a new FSR3 Upscaler context with standard parameters that are appropriate for the current platform. - /// - public static Fsr3UpscalerContext CreateContext(Vector2Int displaySize, Vector2Int maxRenderSize, Fsr3UpscalerShaders shaders, InitializationFlags flags = 0) - { - if (SystemInfo.usesReversedZBuffer) - flags |= InitializationFlags.EnableDepthInverted; - else - flags &= ~InitializationFlags.EnableDepthInverted; - -#if UNITY_EDITOR || DEVELOPMENT_BUILD - flags |= InitializationFlags.EnableDebugChecking; -#endif - - Debug.Log($"Setting up FSR3 Upscaler with render size: {maxRenderSize.x}x{maxRenderSize.y}, display size: {displaySize.x}x{displaySize.y}, flags: {flags}"); - - var contextDescription = new ContextDescription - { - Flags = flags, - DisplaySize = displaySize, - MaxRenderSize = maxRenderSize, - Shaders = shaders, - }; - - var context = new Fsr3UpscalerContext(); - context.Create(contextDescription); - return context; - } - - public static float GetUpscaleRatioFromQualityMode(QualityMode qualityMode) - { - switch (qualityMode) - { - case QualityMode.NativeAA: - return 1.0f; - case QualityMode.UltraQuality: - return 1.2f; - case QualityMode.Quality: - return 1.5f; - case QualityMode.Balanced: - return 1.7f; - case QualityMode.Performance: - return 2.0f; - case QualityMode.UltraPerformance: - return 3.0f; - default: - return 1.0f; - } - } - - public static void GetRenderResolutionFromQualityMode( - out int renderWidth, out int renderHeight, - int displayWidth, int displayHeight, QualityMode qualityMode) - { - float ratio = GetUpscaleRatioFromQualityMode(qualityMode); - renderWidth = Mathf.RoundToInt(displayWidth / ratio); - renderHeight = Mathf.RoundToInt(displayHeight / ratio); - } - - public static float GetMipmapBiasOffset(int renderWidth, int displayWidth) - { - return Mathf.Log((float)renderWidth / displayWidth, 2.0f) - 1.0f; - } - - public static int GetJitterPhaseCount(int renderWidth, int displayWidth) - { - const float basePhaseCount = 8.0f; - int jitterPhaseCount = (int)(basePhaseCount * Mathf.Pow((float)displayWidth / renderWidth, 2.0f)); - return jitterPhaseCount; - } - - public static void GetJitterOffset(out float outX, out float outY, int index, int phaseCount) - { - outX = Halton((index % phaseCount) + 1, 2) - 0.5f; - outY = Halton((index % phaseCount) + 1, 3) - 0.5f; - } - - // Calculate halton number for index and base. - private static float Halton(int index, int @base) - { - float f = 1.0f, result = 0.0f; - - for (int currentIndex = index; currentIndex > 0;) { - - f /= @base; - result += f * (currentIndex % @base); - currentIndex = (int)Mathf.Floor((float)currentIndex / @base); - } - - return result; - } - - public static float Lanczos2(float value) - { - return Mathf.Abs(value) < Mathf.Epsilon ? 1.0f : Mathf.Sin(Mathf.PI * value) / (Mathf.PI * value) * (Mathf.Sin(0.5f * Mathf.PI * value) / (0.5f * Mathf.PI * value)); - } - -#if !UNITY_2021_1_OR_NEWER - internal static void SetBufferData(this CommandBuffer commandBuffer, ComputeBuffer computeBuffer, Array data) - { - commandBuffer.SetComputeBufferData(computeBuffer, data); - } -#endif - - public enum QualityMode - { - NativeAA = 0, - UltraQuality = 1, - Quality = 2, - Balanced = 3, - Performance = 4, - UltraPerformance = 5, - } - - [Flags] - public enum InitializationFlags - { - EnableHighDynamicRange = 1 << 0, - EnableDisplayResolutionMotionVectors = 1 << 1, - EnableMotionVectorsJitterCancellation = 1 << 2, - EnableDepthInverted = 1 << 3, - EnableDepthInfinite = 1 << 4, - EnableAutoExposure = 1 << 5, - EnableDynamicResolution = 1 << 6, - EnableFP16Usage = 1 << 7, - EnableDebugChecking = 1 << 8, - } - - /// - /// A structure encapsulating the parameters required to initialize FidelityFX Super Resolution 3 upscaling. - /// - public struct ContextDescription - { - public InitializationFlags Flags; - public Vector2Int MaxRenderSize; - public Vector2Int DisplaySize; - public Fsr3UpscalerShaders Shaders; - } - - /// - /// A structure encapsulating the parameters for dispatching the various passes of FidelityFX Super Resolution 3. - /// - public class DispatchDescription - { - public ResourceView Color; - public ResourceView Depth; - public ResourceView MotionVectors; - public ResourceView Exposure; // optional - public ResourceView Reactive; // optional - public ResourceView TransparencyAndComposition; // optional - public ResourceView Output; - public Vector2 JitterOffset; - public Vector2 MotionVectorScale; - public Vector2Int RenderSize; - public Vector2Int InputResourceSize; - public bool EnableSharpening; - public float Sharpness; - public float FrameTimeDelta; // in seconds - public float PreExposure; - public bool Reset; - public float CameraNear; - public float CameraFar; - public float CameraFovAngleVertical; - public float ViewSpaceToMetersFactor; - - // EXPERIMENTAL reactive mask generation parameters - public bool EnableAutoReactive; - public ResourceView ColorOpaqueOnly; - public float AutoTcThreshold = 0.05f; - public float AutoTcScale = 1.0f; - public float AutoReactiveScale = 5.0f; - public float AutoReactiveMax = 0.9f; - } - - /// - /// A structure encapsulating the parameters for automatic generation of a reactive mask. - /// The default values for Scale, CutoffThreshold, BinaryValue and Flags were taken from the FSR3 demo project. - /// - public class GenerateReactiveDescription - { - public ResourceView ColorOpaqueOnly; - public ResourceView ColorPreUpscale; - public ResourceView OutReactive; - public Vector2Int RenderSize; - public float Scale = 0.5f; - public float CutoffThreshold = 0.2f; - public float BinaryValue = 0.9f; - public GenerateReactiveFlags Flags = GenerateReactiveFlags.ApplyTonemap | GenerateReactiveFlags.ApplyThreshold | GenerateReactiveFlags.UseComponentsMax; - } - - [Flags] - public enum GenerateReactiveFlags - { - ApplyTonemap = 1 << 0, - ApplyInverseTonemap = 1 << 1, - ApplyThreshold = 1 << 2, - UseComponentsMax = 1 << 3, - } - - [Serializable, StructLayout(LayoutKind.Sequential)] - internal struct UpscalerConstants - { - public Vector2Int renderSize; - public Vector2Int maxRenderSize; - public Vector2Int displaySize; - public Vector2Int inputColorResourceDimensions; - public Vector2Int lumaMipDimensions; - public int lumaMipLevelToUse; - public int frameIndex; - - public Vector4 deviceToViewDepth; - public Vector2 jitterOffset; - public Vector2 motionVectorScale; - public Vector2 downscaleFactor; - public Vector2 motionVectorJitterCancellation; - public float preExposure; - public float previousFramePreExposure; - public float tanHalfFOV; - public float jitterPhaseCount; - public float deltaTime; - public float dynamicResChangeFactor; - public float viewSpaceToMetersFactor; - - public int dummy; - } - - [Serializable, StructLayout(LayoutKind.Sequential)] - internal struct SpdConstants - { - public uint mips; - public uint numWorkGroups; - public uint workGroupOffsetX, workGroupOffsetY; - public uint renderSizeX, renderSizeY; - } - - [Serializable, StructLayout(LayoutKind.Sequential)] - internal struct GenerateReactiveConstants - { - public float scale; - public float threshold; - public float binaryValue; - public uint flags; - } - - [Serializable, StructLayout(LayoutKind.Sequential)] - internal struct GenerateReactiveConstants2 - { - public float autoTcThreshold; - public float autoTcScale; - public float autoReactiveScale; - public float autoReactiveMax; - } - - [Serializable, StructLayout(LayoutKind.Sequential)] - internal struct RcasConstants - { - public RcasConstants(uint sharpness, uint halfSharp) - { - this.sharpness = sharpness; - this.halfSharp = halfSharp; - dummy0 = dummy1 = 0; - } - - public readonly uint sharpness; - public readonly uint halfSharp; - public readonly uint dummy0; - public readonly uint dummy1; - } - } - - /// - /// An immutable structure wrapping all of the necessary information to bind a specific buffer or attachment of a render target to a compute shader. - /// - public readonly struct ResourceView - { - /// - /// This value is the equivalent of not setting any value at all; all struct fields will have their default values. - /// It does not refer to a valid texture, therefore any variable set to this value should be checked for IsValid and reassigned before being bound to a shader. - /// - public static readonly ResourceView Unassigned = new ResourceView(default); - - /// - /// This value contains a valid texture reference that can be bound to a shader, however it is just an empty placeholder texture. - /// Binding this to a shader can be seen as setting the texture variable inside the shader to null. - /// - public static readonly ResourceView None = new ResourceView(BuiltinRenderTextureType.None); - - public ResourceView(in RenderTargetIdentifier renderTarget, RenderTextureSubElement subElement = RenderTextureSubElement.Default, int mipLevel = 0) - { - RenderTarget = renderTarget; - SubElement = subElement; - MipLevel = mipLevel; - } - - public bool IsValid => !RenderTarget.Equals(default); - - public readonly RenderTargetIdentifier RenderTarget; - public readonly RenderTextureSubElement SubElement; - public readonly int MipLevel; - } -} diff --git a/Assets/Scripts/Core/Fsr3UpscalerAssets.cs b/Assets/Scripts/Core/Fsr3UpscalerAssets.cs deleted file mode 100644 index 3e4e24f..0000000 --- a/Assets/Scripts/Core/Fsr3UpscalerAssets.cs +++ /dev/null @@ -1,151 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -using UnityEngine; - -namespace FidelityFX -{ - /// - /// Scriptable object containing all shader resources required by FidelityFX Super Resolution 3 (FSR3) Upscaler. - /// These can be stored in an asset file and referenced from a scene or prefab, avoiding the need to load the shaders from a Resources folder. - /// - public class Fsr3UpscalerAssets : ScriptableObject - { - public Fsr3UpscalerShaders shaders; - -#if UNITY_EDITOR - private void Reset() - { - shaders = new Fsr3UpscalerShaders - { - computeLuminancePyramidPass = FindComputeShader("ffx_fsr3upscaler_compute_luminance_pyramid_pass"), - reconstructPreviousDepthPass = FindComputeShader("ffx_fsr3upscaler_reconstruct_previous_depth_pass"), - depthClipPass = FindComputeShader("ffx_fsr3upscaler_depth_clip_pass"), - lockPass = FindComputeShader("ffx_fsr3upscaler_lock_pass"), - accumulatePass = FindComputeShader("ffx_fsr3upscaler_accumulate_pass"), - sharpenPass = FindComputeShader("ffx_fsr3upscaler_rcas_pass"), - autoGenReactivePass = FindComputeShader("ffx_fsr3upscaler_autogen_reactive_pass"), - tcrAutoGenPass = FindComputeShader("ffx_fsr3upscaler_tcr_autogen_pass"), - }; - } - - private static ComputeShader FindComputeShader(string name) - { - string[] assetGuids = UnityEditor.AssetDatabase.FindAssets($"t:ComputeShader {name}"); - if (assetGuids == null || assetGuids.Length == 0) - return null; - - string assetPath = UnityEditor.AssetDatabase.GUIDToAssetPath(assetGuids[0]); - return UnityEditor.AssetDatabase.LoadAssetAtPath(assetPath); - } -#endif - } - - /// - /// All the compute shaders used by the FSR3 Upscaler. - /// - [System.Serializable] - public class Fsr3UpscalerShaders - { - /// - /// The compute shader used by the luminance pyramid computation pass. - /// - public ComputeShader computeLuminancePyramidPass; - - /// - /// The compute shader used by the previous depth reconstruction pass. - /// - public ComputeShader reconstructPreviousDepthPass; - - /// - /// The compute shader used by the depth clip pass. - /// - public ComputeShader depthClipPass; - - /// - /// The compute shader used by the lock pass. - /// - public ComputeShader lockPass; - - /// - /// The compute shader used by the accumulation pass. - /// - public ComputeShader accumulatePass; - - /// - /// The compute shader used by the RCAS sharpening pass. - /// - public ComputeShader sharpenPass; - - /// - /// The compute shader used to auto-generate a reactive mask. - /// - public ComputeShader autoGenReactivePass; - - /// - /// The compute shader used to auto-generate a transparency & composition mask. - /// - public ComputeShader tcrAutoGenPass; - - /// - /// Returns a copy of this class and its contents. - /// - public Fsr3UpscalerShaders Clone() - { - return (Fsr3UpscalerShaders)MemberwiseClone(); - } - - /// - /// Returns a copy of this class with clones of all its shaders. - /// This can be useful if you're running multiple FSR3 Upscaler instances with different shader configurations. - /// Be sure to clean up these clones through Dispose once you're done with them. - /// - public Fsr3UpscalerShaders DeepCopy() - { - return new Fsr3UpscalerShaders - { - computeLuminancePyramidPass = Object.Instantiate(computeLuminancePyramidPass), - reconstructPreviousDepthPass = Object.Instantiate(reconstructPreviousDepthPass), - depthClipPass = Object.Instantiate(depthClipPass), - lockPass = Object.Instantiate(lockPass), - accumulatePass = Object.Instantiate(accumulatePass), - sharpenPass = Object.Instantiate(sharpenPass), - autoGenReactivePass = Object.Instantiate(autoGenReactivePass), - tcrAutoGenPass = Object.Instantiate(tcrAutoGenPass), - }; - } - - /// - /// Destroy all the shaders within this instance. - /// Use this only on clones created through DeepCopy. - /// - public void Dispose() - { - Object.Destroy(computeLuminancePyramidPass); - Object.Destroy(reconstructPreviousDepthPass); - Object.Destroy(depthClipPass); - Object.Destroy(lockPass); - Object.Destroy(accumulatePass); - Object.Destroy(sharpenPass); - Object.Destroy(autoGenReactivePass); - Object.Destroy(tcrAutoGenPass); - } - } -} diff --git a/Assets/Scripts/Core/Fsr3UpscalerCallbacks.cs b/Assets/Scripts/Core/Fsr3UpscalerCallbacks.cs deleted file mode 100644 index 5b2c89c..0000000 --- a/Assets/Scripts/Core/Fsr3UpscalerCallbacks.cs +++ /dev/null @@ -1,81 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -using UnityEngine; - -namespace FidelityFX -{ - /// - /// A collection of callbacks required by the FSR3 Upscaler process. - /// This allows some customization by the game dev on how to integrate FSR3 upscaling into their own game setup. - /// - public interface IFsr3UpscalerCallbacks - { - /// - /// Apply a mipmap bias to in-game textures to prevent them from becoming blurry as the internal rendering resolution lowers. - /// This will need to be customized on a per-game basis, as there is no clear universal way to determine what are "in-game" textures. - /// The default implementation will simply apply a mipmap bias to all 2D textures, which will include things like UI textures and which might miss things like terrain texture arrays. - /// - /// Depending on how your game organizes its assets, you will want to create a filter that more specifically selects the textures that need to have this mipmap bias applied. - /// You may also want to store the bias offset value and apply it to any assets that are loaded in on demand. - /// - void ApplyMipmapBias(float biasOffset); - - void UndoMipmapBias(); - } - - /// - /// Default implementation of IFsr3UpscalerCallbacks using simple Resources calls. - /// These are fine for testing but a proper game will want to extend and override these methods. - /// - public class Fsr3UpscalerCallbacksBase: IFsr3UpscalerCallbacks - { - protected float CurrentBiasOffset = 0; - - public virtual void ApplyMipmapBias(float biasOffset) - { - if (float.IsNaN(biasOffset) || float.IsInfinity(biasOffset)) - return; - - CurrentBiasOffset += biasOffset; - - if (Mathf.Approximately(CurrentBiasOffset, 0f)) - { - CurrentBiasOffset = 0f; - } - - foreach (var texture in Resources.FindObjectsOfTypeAll()) - { - if (texture.mipmapCount <= 1) - continue; - - texture.mipMapBias += biasOffset; - } - } - - public virtual void UndoMipmapBias() - { - if (CurrentBiasOffset == 0f) - return; - - ApplyMipmapBias(-CurrentBiasOffset); - } - } -} diff --git a/Assets/Scripts/Core/Fsr3UpscalerContext.cs b/Assets/Scripts/Core/Fsr3UpscalerContext.cs deleted file mode 100644 index da02d9f..0000000 --- a/Assets/Scripts/Core/Fsr3UpscalerContext.cs +++ /dev/null @@ -1,610 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -using System; -using System.Runtime.InteropServices; -using UnityEngine; -using UnityEngine.Rendering; - -namespace FidelityFX -{ - /// - /// This class loosely matches the FfxFsr3UpscalerContext struct from the original FSR3 codebase. - /// It manages the various resources and compute passes required by the FSR3 Upscaler process. - /// Note that this class does not know anything about Unity render pipelines; all it knows is CommandBuffers and RenderTargetIdentifiers. - /// This should make it suitable for integration with any of the available Unity render pipelines. - /// - public class Fsr3UpscalerContext - { - private const int MaxQueuedFrames = 16; - - private Fsr3Upscaler.ContextDescription _contextDescription; - private CommandBuffer _commandBuffer; - - private Fsr3UpscalerPass _depthClipPass; - private Fsr3UpscalerPass _reconstructPreviousDepthPass; - private Fsr3UpscalerPass _lockPass; - private Fsr3UpscalerPass _accumulatePass; - private Fsr3UpscalerPass _sharpenPass; - private Fsr3UpscalerPass _computeLuminancePyramidPass; - private Fsr3UpscalerPass _generateReactivePass; - private Fsr3UpscalerPass _tcrAutogeneratePass; - - private readonly Fsr3UpscalerResources _resources = new Fsr3UpscalerResources(); - - private ComputeBuffer _upscalerConstantsBuffer; - private readonly Fsr3Upscaler.UpscalerConstants[] _upscalerConstantsArray = { new Fsr3Upscaler.UpscalerConstants() }; - private ref Fsr3Upscaler.UpscalerConstants UpscalerConsts => ref _upscalerConstantsArray[0]; - - private ComputeBuffer _spdConstantsBuffer; - private readonly Fsr3Upscaler.SpdConstants[] _spdConstantsArray = { new Fsr3Upscaler.SpdConstants() }; - private ref Fsr3Upscaler.SpdConstants SpdConsts => ref _spdConstantsArray[0]; - - private ComputeBuffer _rcasConstantsBuffer; - private readonly Fsr3Upscaler.RcasConstants[] _rcasConstantsArray = new Fsr3Upscaler.RcasConstants[1]; - private ref Fsr3Upscaler.RcasConstants RcasConsts => ref _rcasConstantsArray[0]; - - private ComputeBuffer _generateReactiveConstantsBuffer; - private readonly Fsr3Upscaler.GenerateReactiveConstants[] _generateReactiveConstantsArray = { new Fsr3Upscaler.GenerateReactiveConstants() }; - private ref Fsr3Upscaler.GenerateReactiveConstants GenReactiveConsts => ref _generateReactiveConstantsArray[0]; - - private ComputeBuffer _tcrAutogenerateConstantsBuffer; - private readonly Fsr3Upscaler.GenerateReactiveConstants2[] _tcrAutogenerateConstantsArray = { new Fsr3Upscaler.GenerateReactiveConstants2() }; - private ref Fsr3Upscaler.GenerateReactiveConstants2 TcrAutoGenConsts => ref _tcrAutogenerateConstantsArray[0]; - - private bool _firstExecution; - private Vector2 _previousJitterOffset; - private int _resourceFrameIndex; - - public void Create(Fsr3Upscaler.ContextDescription contextDescription) - { - _contextDescription = contextDescription; - _commandBuffer = new CommandBuffer { name = "FSR3 Upscaler" }; - - _upscalerConstantsBuffer = CreateConstantBuffer(); - _spdConstantsBuffer = CreateConstantBuffer(); - _rcasConstantsBuffer = CreateConstantBuffer(); - _generateReactiveConstantsBuffer = CreateConstantBuffer(); - _tcrAutogenerateConstantsBuffer = CreateConstantBuffer(); - - // Set defaults - _firstExecution = true; - _resourceFrameIndex = 0; - - UpscalerConsts.displaySize = _contextDescription.DisplaySize; - - _resources.Create(_contextDescription); - CreatePasses(); - } - - private void CreatePasses() - { - _computeLuminancePyramidPass = new Fsr3UpscalerComputeLuminancePyramidPass(_contextDescription, _resources, _upscalerConstantsBuffer, _spdConstantsBuffer); - _reconstructPreviousDepthPass = new Fsr3UpscalerReconstructPreviousDepthPass(_contextDescription, _resources, _upscalerConstantsBuffer); - _depthClipPass = new Fsr3UpscalerDepthClipPass(_contextDescription, _resources, _upscalerConstantsBuffer); - _lockPass = new Fsr3UpscalerLockPass(_contextDescription, _resources, _upscalerConstantsBuffer); - _accumulatePass = new Fsr3UpscalerAccumulatePass(_contextDescription, _resources, _upscalerConstantsBuffer); - _sharpenPass = new Fsr3UpscalerSharpenPass(_contextDescription, _resources, _upscalerConstantsBuffer, _rcasConstantsBuffer); - _generateReactivePass = new Fsr3UpscalerGenerateReactivePass(_contextDescription, _resources, _generateReactiveConstantsBuffer); - _tcrAutogeneratePass = new Fsr3UpscalerTcrAutogeneratePass(_contextDescription, _resources, _upscalerConstantsBuffer, _tcrAutogenerateConstantsBuffer); - } - - public void Destroy() - { - DestroyPass(ref _tcrAutogeneratePass); - DestroyPass(ref _generateReactivePass); - DestroyPass(ref _computeLuminancePyramidPass); - DestroyPass(ref _sharpenPass); - DestroyPass(ref _accumulatePass); - DestroyPass(ref _lockPass); - DestroyPass(ref _reconstructPreviousDepthPass); - DestroyPass(ref _depthClipPass); - - _resources.Destroy(); - - DestroyConstantBuffer(ref _tcrAutogenerateConstantsBuffer); - DestroyConstantBuffer(ref _generateReactiveConstantsBuffer); - DestroyConstantBuffer(ref _rcasConstantsBuffer); - DestroyConstantBuffer(ref _spdConstantsBuffer); - DestroyConstantBuffer(ref _upscalerConstantsBuffer); - - _commandBuffer.Dispose(); - _commandBuffer = null; - } - - public void Dispatch(Fsr3Upscaler.DispatchDescription dispatchParams) - { - _commandBuffer.Clear(); - Dispatch(dispatchParams, _commandBuffer); - Graphics.ExecuteCommandBuffer(_commandBuffer); - } - - public void Dispatch(Fsr3Upscaler.DispatchDescription dispatchParams, CommandBuffer commandBuffer) - { - if ((_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDebugChecking) != 0) - { - DebugCheckDispatch(dispatchParams); - } - - if (_firstExecution) - { - commandBuffer.SetRenderTarget(_resources.LockStatus[0]); - commandBuffer.ClearRenderTarget(false, true, Color.clear); - commandBuffer.SetRenderTarget(_resources.LockStatus[1]); - commandBuffer.ClearRenderTarget(false, true, Color.clear); - } - - int frameIndex = _resourceFrameIndex % 2; - bool resetAccumulation = dispatchParams.Reset || _firstExecution; - _firstExecution = false; - - // If auto exposure is enabled use the auto exposure SRV, otherwise what the app sends - if ((_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableAutoExposure) != 0) - dispatchParams.Exposure = new ResourceView(_resources.AutoExposure); - else if (!dispatchParams.Exposure.IsValid) - dispatchParams.Exposure = new ResourceView(_resources.DefaultExposure); - - if (dispatchParams.EnableAutoReactive) - { - // Create the auto-TCR resources only when we need them - if (_resources.AutoReactive == null) - _resources.CreateTcrAutogenResources(_contextDescription); - - if (resetAccumulation) - { - RenderTargetIdentifier opaqueOnly = dispatchParams.ColorOpaqueOnly.IsValid ? dispatchParams.ColorOpaqueOnly.RenderTarget : Fsr3ShaderIDs.SrvOpaqueOnly; - commandBuffer.Blit(_resources.PrevPreAlpha[frameIndex ^ 1], opaqueOnly); - } - } - else if (_resources.AutoReactive != null) - { - // Destroy the auto-TCR resources if we don't use the feature - _resources.DestroyTcrAutogenResources(); - } - - if (!dispatchParams.Reactive.IsValid) dispatchParams.Reactive = new ResourceView(_resources.DefaultReactive); - if (!dispatchParams.TransparencyAndComposition.IsValid) dispatchParams.TransparencyAndComposition = new ResourceView(_resources.DefaultReactive); - Fsr3UpscalerResources.CreateAliasableResources(commandBuffer, _contextDescription, dispatchParams); - - SetupConstants(dispatchParams, resetAccumulation); - - // Reactive mask bias - const int threadGroupWorkRegionDim = 8; - int dispatchSrcX = (UpscalerConsts.renderSize.x + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; - int dispatchSrcY = (UpscalerConsts.renderSize.y + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; - int dispatchDstX = (_contextDescription.DisplaySize.x + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; - int dispatchDstY = (_contextDescription.DisplaySize.y + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; - - // Clear reconstructed depth for max depth store - if (resetAccumulation) - { - commandBuffer.SetRenderTarget(_resources.LockStatus[frameIndex ^ 1]); - commandBuffer.ClearRenderTarget(false, true, Color.clear); - - commandBuffer.SetRenderTarget(_resources.InternalUpscaled[frameIndex ^ 1]); - commandBuffer.ClearRenderTarget(false, true, Color.clear); - - commandBuffer.SetRenderTarget(_resources.SceneLuminance); - commandBuffer.ClearRenderTarget(false, true, Color.clear); - - // Auto exposure always used to track luma changes in locking logic - commandBuffer.SetRenderTarget(_resources.AutoExposure); - commandBuffer.ClearRenderTarget(false, true, new Color(0f, 1e8f, 0f, 0f)); - - // Reset atomic counter to 0 - commandBuffer.SetRenderTarget(_resources.SpdAtomicCounter); - commandBuffer.ClearRenderTarget(false, true, Color.clear); - } - - // FSR3: need to clear here since we need the content of this surface for frame interpolation, so clearing in the lock pass is not an option - bool depthInverted = (_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDepthInverted) == Fsr3Upscaler.InitializationFlags.EnableDepthInverted; - commandBuffer.SetRenderTarget(Fsr3ShaderIDs.UavReconstructedPrevNearestDepth); - commandBuffer.ClearRenderTarget(false, true, depthInverted ? Color.clear : Color.white); - - // Auto exposure - SetupSpdConstants(dispatchParams, out var dispatchThreadGroupCount); - - // Initialize constant buffers data - commandBuffer.SetBufferData(_upscalerConstantsBuffer, _upscalerConstantsArray); - commandBuffer.SetBufferData(_spdConstantsBuffer, _spdConstantsArray); - - // Auto reactive - if (dispatchParams.EnableAutoReactive) - { - GenerateTransparencyCompositionReactive(dispatchParams, commandBuffer, frameIndex); - dispatchParams.Reactive = new ResourceView(_resources.AutoReactive); - dispatchParams.TransparencyAndComposition = new ResourceView(_resources.AutoComposition); - } - - // Compute luminance pyramid - _computeLuminancePyramidPass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchThreadGroupCount.x, dispatchThreadGroupCount.y); - - // Reconstruct previous depth - _reconstructPreviousDepthPass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchSrcX, dispatchSrcY); - - // Depth clip - _depthClipPass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchSrcX, dispatchSrcY); - - // Create locks - _lockPass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchSrcX, dispatchSrcY); - - // Accumulate - _accumulatePass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchDstX, dispatchDstY); - - if (dispatchParams.EnableSharpening) - { - // Compute the constants - SetupRcasConstants(dispatchParams); - commandBuffer.SetBufferData(_rcasConstantsBuffer, _rcasConstantsArray); - - // Dispatch RCAS - const int threadGroupWorkRegionDimRcas = 16; - int threadGroupsX = (Screen.width + threadGroupWorkRegionDimRcas - 1) / threadGroupWorkRegionDimRcas; - int threadGroupsY = (Screen.height + threadGroupWorkRegionDimRcas - 1) / threadGroupWorkRegionDimRcas; - _sharpenPass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, threadGroupsX, threadGroupsY); - } - - _resourceFrameIndex = (_resourceFrameIndex + 1) % MaxQueuedFrames; - - Fsr3UpscalerResources.DestroyAliasableResources(commandBuffer); - } - - public void GenerateReactiveMask(Fsr3Upscaler.GenerateReactiveDescription dispatchParams) - { - _commandBuffer.Clear(); - GenerateReactiveMask(dispatchParams, _commandBuffer); - Graphics.ExecuteCommandBuffer(_commandBuffer); - } - - public void GenerateReactiveMask(Fsr3Upscaler.GenerateReactiveDescription dispatchParams, CommandBuffer commandBuffer) - { - const int threadGroupWorkRegionDim = 8; - int dispatchSrcX = (dispatchParams.RenderSize.x + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; - int dispatchSrcY = (dispatchParams.RenderSize.y + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; - - GenReactiveConsts.scale = dispatchParams.Scale; - GenReactiveConsts.threshold = dispatchParams.CutoffThreshold; - GenReactiveConsts.binaryValue = dispatchParams.BinaryValue; - GenReactiveConsts.flags = (uint)dispatchParams.Flags; - commandBuffer.SetBufferData(_generateReactiveConstantsBuffer, _generateReactiveConstantsArray); - - ((Fsr3UpscalerGenerateReactivePass)_generateReactivePass).ScheduleDispatch(commandBuffer, dispatchParams, dispatchSrcX, dispatchSrcY); - } - - private void GenerateTransparencyCompositionReactive(Fsr3Upscaler.DispatchDescription dispatchParams, CommandBuffer commandBuffer, int frameIndex) - { - const int threadGroupWorkRegionDim = 8; - int dispatchSrcX = (dispatchParams.RenderSize.x + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; - int dispatchSrcY = (dispatchParams.RenderSize.y + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; - - TcrAutoGenConsts.autoTcThreshold = dispatchParams.AutoTcThreshold; - TcrAutoGenConsts.autoTcScale = dispatchParams.AutoTcScale; - TcrAutoGenConsts.autoReactiveScale = dispatchParams.AutoReactiveScale; - TcrAutoGenConsts.autoReactiveMax = dispatchParams.AutoReactiveMax; - commandBuffer.SetBufferData(_tcrAutogenerateConstantsBuffer, _tcrAutogenerateConstantsArray); - - _tcrAutogeneratePass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchSrcX, dispatchSrcY); - } - - private void SetupConstants(Fsr3Upscaler.DispatchDescription dispatchParams, bool resetAccumulation) - { - ref Fsr3Upscaler.UpscalerConstants constants = ref UpscalerConsts; - - constants.jitterOffset = dispatchParams.JitterOffset; - constants.renderSize = dispatchParams.RenderSize; - constants.maxRenderSize = _contextDescription.MaxRenderSize; - constants.inputColorResourceDimensions = dispatchParams.InputResourceSize; - - // Compute the horizontal FOV for the shader from the vertical one - float aspectRatio = (float)dispatchParams.RenderSize.x / dispatchParams.RenderSize.y; - float cameraAngleHorizontal = Mathf.Atan(Mathf.Tan(dispatchParams.CameraFovAngleVertical / 2.0f) * aspectRatio) * 2.0f; - constants.tanHalfFOV = Mathf.Tan(cameraAngleHorizontal * 0.5f); - constants.viewSpaceToMetersFactor = (dispatchParams.ViewSpaceToMetersFactor > 0.0f) ? dispatchParams.ViewSpaceToMetersFactor : 1.0f; - - // Compute params to enable device depth to view space depth computation in shader - constants.deviceToViewDepth = SetupDeviceDepthToViewSpaceDepthParams(dispatchParams); - - // To be updated if resource is larger than the actual image size - constants.downscaleFactor = new Vector2((float)constants.renderSize.x / _contextDescription.DisplaySize.x, (float)constants.renderSize.y / _contextDescription.DisplaySize.y); - constants.previousFramePreExposure = constants.preExposure; - constants.preExposure = (dispatchParams.PreExposure != 0) ? dispatchParams.PreExposure : 1.0f; - - // Motion vector data - Vector2Int motionVectorsTargetSize = (_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDisplayResolutionMotionVectors) != 0 ? constants.displaySize : constants.renderSize; - constants.motionVectorScale = dispatchParams.MotionVectorScale / motionVectorsTargetSize; - - // Compute jitter cancellation - if ((_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableMotionVectorsJitterCancellation) != 0) - { - constants.motionVectorJitterCancellation = (_previousJitterOffset - constants.jitterOffset) / motionVectorsTargetSize; - _previousJitterOffset = constants.jitterOffset; - } - - int jitterPhaseCount = Fsr3Upscaler.GetJitterPhaseCount(dispatchParams.RenderSize.x, _contextDescription.DisplaySize.x); - if (resetAccumulation || constants.jitterPhaseCount == 0) - { - constants.jitterPhaseCount = jitterPhaseCount; - } - else - { - int jitterPhaseCountDelta = (int)(jitterPhaseCount - constants.jitterPhaseCount); - if (jitterPhaseCountDelta > 0) - constants.jitterPhaseCount++; - else if (jitterPhaseCountDelta < 0) - constants.jitterPhaseCount--; - } - - // Convert delta time to seconds and clamp to [0, 1] - constants.deltaTime = Mathf.Clamp01(dispatchParams.FrameTimeDelta); - - if (resetAccumulation) - constants.frameIndex = 0; - else - constants.frameIndex++; - - // Shading change usage of the SPD mip levels - constants.lumaMipLevelToUse = Fsr3UpscalerPass.ShadingChangeMipLevel; - - float mipDiv = 2 << constants.lumaMipLevelToUse; - constants.lumaMipDimensions.x = (int)(constants.maxRenderSize.x / mipDiv); - constants.lumaMipDimensions.y = (int)(constants.maxRenderSize.y / mipDiv); - } - - private Vector4 SetupDeviceDepthToViewSpaceDepthParams(Fsr3Upscaler.DispatchDescription dispatchParams) - { - bool inverted = (_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDepthInverted) != 0; - bool infinite = (_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDepthInfinite) != 0; - - // make sure it has no impact if near and far plane values are swapped in dispatch params - // the flags "inverted" and "infinite" will decide what transform to use - float min = Mathf.Min(dispatchParams.CameraNear, dispatchParams.CameraFar); - float max = Mathf.Max(dispatchParams.CameraNear, dispatchParams.CameraFar); - - if (inverted) - { - (min, max) = (max, min); - } - - float q = max / (min - max); - float d = -1.0f; - - Vector4 matrixElemC = new Vector4(q, -1.0f - Mathf.Epsilon, q, 0.0f + Mathf.Epsilon); - Vector4 matrixElemE = new Vector4(q * min, -min - Mathf.Epsilon, q * min, max); - - // Revert x and y coords - float aspect = (float)dispatchParams.RenderSize.x / dispatchParams.RenderSize.y; - float cotHalfFovY = Mathf.Cos(0.5f * dispatchParams.CameraFovAngleVertical) / Mathf.Sin(0.5f * dispatchParams.CameraFovAngleVertical); - - int matrixIndex = (inverted ? 2 : 0) + (infinite ? 1 : 0); - return new Vector4( - d * matrixElemC[matrixIndex], - matrixElemE[matrixIndex], - aspect / cotHalfFovY, - 1.0f / cotHalfFovY); - } - - private void SetupRcasConstants(Fsr3Upscaler.DispatchDescription dispatchParams) - { - int sharpnessIndex = Mathf.RoundToInt(Mathf.Clamp01(dispatchParams.Sharpness) * (RcasConfigs.Length - 1)); - RcasConsts = RcasConfigs[sharpnessIndex]; - } - - private void SetupSpdConstants(Fsr3Upscaler.DispatchDescription dispatchParams, out Vector2Int dispatchThreadGroupCount) - { - RectInt rectInfo = new RectInt(0, 0, dispatchParams.RenderSize.x, dispatchParams.RenderSize.y); - SpdSetup(rectInfo, out dispatchThreadGroupCount, out var workGroupOffset, out var numWorkGroupsAndMips); - - // Downsample - ref Fsr3Upscaler.SpdConstants spdConstants = ref SpdConsts; - spdConstants.numWorkGroups = (uint)numWorkGroupsAndMips.x; - spdConstants.mips = (uint)numWorkGroupsAndMips.y; - spdConstants.workGroupOffsetX = (uint)workGroupOffset.x; - spdConstants.workGroupOffsetY = (uint)workGroupOffset.y; - spdConstants.renderSizeX = (uint)dispatchParams.RenderSize.x; - spdConstants.renderSizeY = (uint)dispatchParams.RenderSize.y; - } - - private static void SpdSetup(RectInt rectInfo, out Vector2Int dispatchThreadGroupCount, out Vector2Int workGroupOffset, out Vector2Int numWorkGroupsAndMips, int mips = -1) - { - workGroupOffset = new Vector2Int(rectInfo.x / 64, rectInfo.y / 64); - - int endIndexX = (rectInfo.x + rectInfo.width - 1) / 64; - int endIndexY = (rectInfo.y + rectInfo.height - 1) / 64; - - dispatchThreadGroupCount = new Vector2Int(endIndexX + 1 - workGroupOffset.x, endIndexY + 1 - workGroupOffset.y); - - numWorkGroupsAndMips = new Vector2Int(dispatchThreadGroupCount.x * dispatchThreadGroupCount.y, mips); - if (mips < 0) - { - float resolution = Math.Max(rectInfo.width, rectInfo.height); - numWorkGroupsAndMips.y = Math.Min(Mathf.FloorToInt(Mathf.Log(resolution, 2.0f)), 12); - } - } - - private void DebugCheckDispatch(Fsr3Upscaler.DispatchDescription dispatchParams) - { - if (!dispatchParams.Color.IsValid) - { - Debug.LogError("Color resource is null"); - } - - if (!dispatchParams.Depth.IsValid) - { - Debug.LogError("Depth resource is null"); - } - - if (!dispatchParams.MotionVectors.IsValid) - { - Debug.LogError("MotionVectors resource is null"); - } - - if (!dispatchParams.Output.IsValid) - { - Debug.LogError("Output resource is null"); - } - - if (dispatchParams.Exposure.IsValid && (_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableAutoExposure) != 0) - { - Debug.LogWarning("Exposure resource provided, however auto exposure flag is present"); - } - - if (Mathf.Abs(dispatchParams.JitterOffset.x) > 1.0f || Mathf.Abs(dispatchParams.JitterOffset.y) > 1.0f) - { - Debug.LogWarning("JitterOffset contains value outside of expected range [-1.0, 1.0]"); - } - - if (dispatchParams.MotionVectorScale.x > _contextDescription.MaxRenderSize.x || dispatchParams.MotionVectorScale.y > _contextDescription.MaxRenderSize.y) - { - Debug.LogWarning("MotionVectorScale contains scale value greater than MaxRenderSize"); - } - - if (dispatchParams.MotionVectorScale.x == 0.0f || dispatchParams.MotionVectorScale.y == 0.0f) - { - Debug.LogWarning("MotionVectorScale contains zero scale value"); - } - - if (dispatchParams.RenderSize.x > _contextDescription.MaxRenderSize.x || dispatchParams.RenderSize.y > _contextDescription.MaxRenderSize.y) - { - Debug.LogWarning("RenderSize is greater than context MaxRenderSize"); - } - - if (dispatchParams.RenderSize.x == 0 || dispatchParams.RenderSize.y == 0) - { - Debug.LogWarning("RenderSize contains zero dimension"); - } - - if (dispatchParams.FrameTimeDelta > 1.0f) - { - Debug.LogWarning("FrameTimeDelta is greater than 1.0f - this value should be seconds (~0.0166 for 60fps)"); - } - - if (dispatchParams.PreExposure == 0.0f) - { - Debug.LogError("PreExposure provided as 0.0f which is invalid"); - } - - bool infiniteDepth = (_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDepthInfinite) != 0; - bool inverseDepth = (_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDepthInverted) != 0; - - if (inverseDepth) - { - if (dispatchParams.CameraNear < dispatchParams.CameraFar) - { - Debug.LogWarning("EnableDepthInverted flag is present yet CameraNear is less than CameraFar"); - } - - if (infiniteDepth) - { - if (dispatchParams.CameraNear < float.MaxValue) - { - Debug.LogWarning("EnableDepthInfinite and EnableDepthInverted present, yet CameraNear != float.MaxValue"); - } - } - - if (dispatchParams.CameraFar < 0.075f) - { - Debug.LogWarning("EnableDepthInverted present, CameraFar value is very low which may result in depth separation artefacting"); - } - } - else - { - if (dispatchParams.CameraNear > dispatchParams.CameraFar) - { - Debug.LogWarning("CameraNear is greater than CameraFar in non-inverted-depth context"); - } - - if (infiniteDepth) - { - if (dispatchParams.CameraFar < float.MaxValue) - { - Debug.LogWarning("EnableDepthInfinite present, yet CameraFar != float.MaxValue"); - } - } - - if (dispatchParams.CameraNear < 0.075f) - { - Debug.LogWarning("CameraNear value is very low which may result in depth separation artefacting"); - } - } - - if (dispatchParams.CameraFovAngleVertical <= 0.0f) - { - Debug.LogError("CameraFovAngleVertical is 0.0f - this value should be > 0.0f"); - } - - if (dispatchParams.CameraFovAngleVertical > Mathf.PI) - { - Debug.LogError("CameraFovAngleVertical is greater than 180 degrees/PI"); - } - } - - /// - /// The FSR3 C++ codebase uses floats bitwise converted to ints to pass sharpness parameters to the RCAS shader. - /// This is not possible in C# without enabling unsafe code compilation, so to avoid that we instead use a table of precomputed values. - /// - private static readonly Fsr3Upscaler.RcasConstants[] RcasConfigs = new [] - { - new Fsr3Upscaler.RcasConstants(1048576000u, 872428544u), - new Fsr3Upscaler.RcasConstants(1049178080u, 877212745u), - new Fsr3Upscaler.RcasConstants(1049823372u, 882390168u), - new Fsr3Upscaler.RcasConstants(1050514979u, 887895276u), - new Fsr3Upscaler.RcasConstants(1051256227u, 893859143u), - new Fsr3Upscaler.RcasConstants(1052050675u, 900216232u), - new Fsr3Upscaler.RcasConstants(1052902144u, 907032080u), - new Fsr3Upscaler.RcasConstants(1053814727u, 914306687u), - new Fsr3Upscaler.RcasConstants(1054792807u, 922105590u), - new Fsr3Upscaler.RcasConstants(1055841087u, 930494326u), - new Fsr3Upscaler.RcasConstants(1056964608u, 939538432u), - new Fsr3Upscaler.RcasConstants(1057566688u, 944322633u), - new Fsr3Upscaler.RcasConstants(1058211980u, 949500056u), - new Fsr3Upscaler.RcasConstants(1058903587u, 955005164u), - new Fsr3Upscaler.RcasConstants(1059644835u, 960969031u), - new Fsr3Upscaler.RcasConstants(1060439283u, 967326120u), - new Fsr3Upscaler.RcasConstants(1061290752u, 974141968u), - new Fsr3Upscaler.RcasConstants(1062203335u, 981416575u), - new Fsr3Upscaler.RcasConstants(1063181415u, 989215478u), - new Fsr3Upscaler.RcasConstants(1064229695u, 997604214u), - new Fsr3Upscaler.RcasConstants(1065353216u, 1006648320), - }; - - private static ComputeBuffer CreateConstantBuffer() where TConstants: struct - { - return new ComputeBuffer(1, Marshal.SizeOf(), ComputeBufferType.Constant); - } - - private static void DestroyConstantBuffer(ref ComputeBuffer bufferRef) - { - if (bufferRef == null) - return; - - bufferRef.Release(); - bufferRef = null; - } - - private static void DestroyPass(ref Fsr3UpscalerPass pass) - { - if (pass == null) - return; - - pass.Dispose(); - pass = null; - } - } -} diff --git a/Assets/Scripts/Core/Fsr3UpscalerPass.cs b/Assets/Scripts/Core/Fsr3UpscalerPass.cs deleted file mode 100644 index e1dc225..0000000 --- a/Assets/Scripts/Core/Fsr3UpscalerPass.cs +++ /dev/null @@ -1,372 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -using System; -using System.Runtime.InteropServices; -using UnityEngine; -using UnityEngine.Rendering; - -namespace FidelityFX -{ - /// - /// Base class for all of the compute passes that make up the FSR3 Upscaler process. - /// This loosely matches the FfxPipelineState struct from the original FSR3 codebase, wrapped in an object-oriented blanket. - /// These classes are responsible for loading compute shaders, managing temporary resources, binding resources to shader kernels and dispatching said shaders. - /// - internal abstract class Fsr3UpscalerPass: IDisposable - { - internal const int ShadingChangeMipLevel = 4; // This matches the FFX_FSR3UPSCALER_SHADING_CHANGE_MIP_LEVEL define - - protected readonly Fsr3Upscaler.ContextDescription ContextDescription; - protected readonly Fsr3UpscalerResources Resources; - protected readonly ComputeBuffer Constants; - - protected ComputeShader ComputeShader; - protected int KernelIndex; - - protected Fsr3UpscalerPass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants) - { - ContextDescription = contextDescription; - Resources = resources; - Constants = constants; - } - - public virtual void Dispose() - { - } - - public abstract void ScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY); - - protected void InitComputeShader(string passName, ComputeShader shader) - { - InitComputeShader(passName, shader, ContextDescription.Flags); - } - - private void InitComputeShader(string passName, ComputeShader shader, Fsr3Upscaler.InitializationFlags flags) - { - if (shader == null) - { - throw new MissingReferenceException($"Shader for FSR3 Upscaler '{passName}' could not be loaded! Please ensure it is included in the project correctly."); - } - - ComputeShader = shader; - KernelIndex = ComputeShader.FindKernel("CS"); - - bool useLut = false; -#if UNITY_2022_1_OR_NEWER // This will also work in 2020.3.43+ and 2021.3.14+ - if (SystemInfo.computeSubGroupSize == 64) - { - useLut = true; - } -#endif - - // This matches the permutation rules from the CreatePipeline* functions - if ((flags & Fsr3Upscaler.InitializationFlags.EnableHighDynamicRange) != 0) ComputeShader.EnableKeyword("FFX_FSR3UPSCALER_OPTION_HDR_COLOR_INPUT"); - if ((flags & Fsr3Upscaler.InitializationFlags.EnableDisplayResolutionMotionVectors) == 0) ComputeShader.EnableKeyword("FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS"); - if ((flags & Fsr3Upscaler.InitializationFlags.EnableMotionVectorsJitterCancellation) != 0) ComputeShader.EnableKeyword("FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS"); - if ((flags & Fsr3Upscaler.InitializationFlags.EnableDepthInverted) != 0) ComputeShader.EnableKeyword("FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH"); - if (useLut) ComputeShader.EnableKeyword("FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE"); - if ((flags & Fsr3Upscaler.InitializationFlags.EnableFP16Usage) != 0) ComputeShader.EnableKeyword("FFX_HALF"); - - // Inform the shader which render pipeline we're currently using - var pipeline = GraphicsSettings.currentRenderPipeline; - if (pipeline != null && pipeline.GetType().Name.Contains("HDRenderPipeline")) - { - ComputeShader.EnableKeyword("UNITY_FSR3UPSCALER_HDRP"); - } - } - } - - internal class Fsr3UpscalerComputeLuminancePyramidPass : Fsr3UpscalerPass - { - private readonly ComputeBuffer _spdConstants; - - public Fsr3UpscalerComputeLuminancePyramidPass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants, ComputeBuffer spdConstants) - : base(contextDescription, resources, constants) - { - _spdConstants = spdConstants; - - InitComputeShader("compute_luminance_pyramid_pass", contextDescription.Shaders.computeLuminancePyramidPass); - } - - public override void ScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) - { - ref var color = ref dispatchParams.Color; - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputColor, color.RenderTarget, color.MipLevel, color.SubElement); - - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavSpdAtomicCount, Resources.SpdAtomicCounter); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavExposureMipLumaChange, Resources.SceneLuminance, ShadingChangeMipLevel); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavExposureMip5, Resources.SceneLuminance, 5); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavAutoExposure, Resources.AutoExposure); - - commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); - commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbSpd, _spdConstants, 0, Marshal.SizeOf()); - - commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); - } - } - - internal class Fsr3UpscalerReconstructPreviousDepthPass : Fsr3UpscalerPass - { - public Fsr3UpscalerReconstructPreviousDepthPass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants) - : base(contextDescription, resources, constants) - { - InitComputeShader("reconstruct_previous_depth_pass", contextDescription.Shaders.reconstructPreviousDepthPass); - } - - public override void ScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) - { - ref var color = ref dispatchParams.Color; - ref var depth = ref dispatchParams.Depth; - ref var motionVectors = ref dispatchParams.MotionVectors; - ref var exposure = ref dispatchParams.Exposure; - - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputColor, color.RenderTarget, color.MipLevel, color.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputDepth, depth.RenderTarget, depth.MipLevel, depth.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputMotionVectors, motionVectors.RenderTarget, motionVectors.MipLevel, motionVectors.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputExposure, exposure.RenderTarget, exposure.MipLevel, exposure.SubElement); - - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavDilatedMotionVectors, Resources.DilatedMotionVectors[frameIndex]); - - commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); - - commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); - } - } - - internal class Fsr3UpscalerDepthClipPass : Fsr3UpscalerPass - { - public Fsr3UpscalerDepthClipPass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants) - : base(contextDescription, resources, constants) - { - InitComputeShader("depth_clip_pass", contextDescription.Shaders.depthClipPass); - } - - public override void ScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) - { - ref var color = ref dispatchParams.Color; - ref var depth = ref dispatchParams.Depth; - ref var motionVectors = ref dispatchParams.MotionVectors; - ref var exposure = ref dispatchParams.Exposure; - ref var reactive = ref dispatchParams.Reactive; - ref var tac = ref dispatchParams.TransparencyAndComposition; - - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputColor, color.RenderTarget, color.MipLevel, color.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputDepth, depth.RenderTarget, depth.MipLevel, depth.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputMotionVectors, motionVectors.RenderTarget, motionVectors.MipLevel, motionVectors.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputExposure, exposure.RenderTarget, exposure.MipLevel, exposure.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvReactiveMask, reactive.RenderTarget, reactive.MipLevel, reactive.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvTransparencyAndCompositionMask, tac.RenderTarget, tac.MipLevel, tac.SubElement); - - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvReconstructedPrevNearestDepth, Fsr3ShaderIDs.UavReconstructedPrevNearestDepth); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvDilatedMotionVectors, Resources.DilatedMotionVectors[frameIndex]); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvDilatedDepth, Fsr3ShaderIDs.UavDilatedDepth); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvPrevDilatedMotionVectors, Resources.DilatedMotionVectors[frameIndex ^ 1]); - - commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); - - commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); - } - } - - internal class Fsr3UpscalerLockPass : Fsr3UpscalerPass - { - public Fsr3UpscalerLockPass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants) - : base(contextDescription, resources, constants) - { - InitComputeShader("lock_pass", contextDescription.Shaders.lockPass); - } - - public override void ScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) - { - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvLockInputLuma, Fsr3ShaderIDs.UavLockInputLuma); - commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); - - commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); - } - } - - internal class Fsr3UpscalerAccumulatePass : Fsr3UpscalerPass - { - private const string SharpeningKeyword = "FFX_FSR3UPSCALER_OPTION_APPLY_SHARPENING"; - -#if UNITY_2021_2_OR_NEWER - private readonly LocalKeyword _sharpeningKeyword; -#endif - - public Fsr3UpscalerAccumulatePass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants) - : base(contextDescription, resources, constants) - { - InitComputeShader("accumulate_pass", contextDescription.Shaders.accumulatePass); -#if UNITY_2021_2_OR_NEWER - _sharpeningKeyword = new LocalKeyword(ComputeShader, SharpeningKeyword); -#endif - } - - public override void ScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) - { -#if UNITY_2021_2_OR_NEWER - if (dispatchParams.EnableSharpening) - commandBuffer.EnableKeyword(ComputeShader, _sharpeningKeyword); - else - commandBuffer.DisableKeyword(ComputeShader, _sharpeningKeyword); -#else - if (dispatchParams.EnableSharpening) - commandBuffer.EnableShaderKeyword(SharpeningKeyword); - else - commandBuffer.DisableShaderKeyword(SharpeningKeyword); -#endif - - if ((ContextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDisplayResolutionMotionVectors) == 0) - { - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvDilatedMotionVectors, Resources.DilatedMotionVectors[frameIndex]); - } - else - { - ref var motionVectors = ref dispatchParams.MotionVectors; - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputMotionVectors, motionVectors.RenderTarget, motionVectors.MipLevel, motionVectors.SubElement); - } - - ref var exposure = ref dispatchParams.Exposure; - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputExposure, exposure.RenderTarget, exposure.MipLevel, exposure.SubElement); - - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvDilatedReactiveMasks, Fsr3ShaderIDs.UavDilatedReactiveMasks); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInternalUpscaled, Resources.InternalUpscaled[frameIndex ^ 1]); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvLockStatus, Resources.LockStatus[frameIndex ^ 1]); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvPreparedInputColor, Fsr3ShaderIDs.UavPreparedInputColor); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvLanczosLut, Resources.LanczosLut); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvUpscaleMaximumBiasLut, Resources.MaximumBiasLut); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvSceneLuminanceMips, Resources.SceneLuminance); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvAutoExposure, Resources.AutoExposure); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvLumaHistory, Resources.LumaHistory[frameIndex ^ 1]); - - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavInternalUpscaled, Resources.InternalUpscaled[frameIndex]); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavLockStatus, Resources.LockStatus[frameIndex]); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavLumaHistory, Resources.LumaHistory[frameIndex]); - - ref var output = ref dispatchParams.Output; - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavUpscaledOutput, output.RenderTarget, output.MipLevel, output.SubElement); - - commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); - - commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); - } - } - - internal class Fsr3UpscalerSharpenPass : Fsr3UpscalerPass - { - private readonly ComputeBuffer _rcasConstants; - - public Fsr3UpscalerSharpenPass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants, ComputeBuffer rcasConstants) - : base(contextDescription, resources, constants) - { - _rcasConstants = rcasConstants; - - InitComputeShader("rcas_pass", contextDescription.Shaders.sharpenPass); - } - - public override void ScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) - { - ref var exposure = ref dispatchParams.Exposure; - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputExposure, exposure.RenderTarget, exposure.MipLevel, exposure.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvRcasInput, Resources.InternalUpscaled[frameIndex]); - - ref var output = ref dispatchParams.Output; - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavUpscaledOutput, output.RenderTarget, output.MipLevel, output.SubElement); - - commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); - commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbRcas, _rcasConstants, 0, Marshal.SizeOf()); - - commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); - } - } - - internal class Fsr3UpscalerGenerateReactivePass : Fsr3UpscalerPass - { - private readonly ComputeBuffer _generateReactiveConstants; - - public Fsr3UpscalerGenerateReactivePass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer generateReactiveConstants) - : base(contextDescription, resources, null) - { - _generateReactiveConstants = generateReactiveConstants; - - InitComputeShader("autogen_reactive_pass", contextDescription.Shaders.autoGenReactivePass); - } - - public override void ScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) - { - } - - public void ScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.GenerateReactiveDescription dispatchParams, int dispatchX, int dispatchY) - { - ref var opaqueOnly = ref dispatchParams.ColorOpaqueOnly; - ref var color = ref dispatchParams.ColorPreUpscale; - ref var reactive = ref dispatchParams.OutReactive; - - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvOpaqueOnly, opaqueOnly.RenderTarget, opaqueOnly.MipLevel, opaqueOnly.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputColor, color.RenderTarget, color.MipLevel, color.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavAutoReactive, reactive.RenderTarget, reactive.MipLevel, reactive.SubElement); - - commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbGenReactive, _generateReactiveConstants, 0, Marshal.SizeOf()); - - commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); - } - } - - internal class Fsr3UpscalerTcrAutogeneratePass : Fsr3UpscalerPass - { - private readonly ComputeBuffer _tcrAutogenerateConstants; - - public Fsr3UpscalerTcrAutogeneratePass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants, ComputeBuffer tcrAutogenerateConstants) - : base(contextDescription, resources, constants) - { - _tcrAutogenerateConstants = tcrAutogenerateConstants; - - InitComputeShader("tcr_autogen_pass", contextDescription.Shaders.tcrAutoGenPass); - } - - public override void ScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) - { - ref var color = ref dispatchParams.Color; - ref var motionVectors = ref dispatchParams.MotionVectors; - ref var opaqueOnly = ref dispatchParams.ColorOpaqueOnly; - ref var reactive = ref dispatchParams.Reactive; - ref var tac = ref dispatchParams.TransparencyAndComposition; - - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvOpaqueOnly, opaqueOnly.RenderTarget, opaqueOnly.MipLevel, opaqueOnly.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputColor, color.RenderTarget, color.MipLevel, color.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputMotionVectors, motionVectors.RenderTarget, motionVectors.MipLevel, motionVectors.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvPrevColorPreAlpha, Resources.PrevPreAlpha[frameIndex ^ 1]); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvPrevColorPostAlpha, Resources.PrevPostAlpha[frameIndex ^ 1]); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvReactiveMask, reactive.RenderTarget, reactive.MipLevel, reactive.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvTransparencyAndCompositionMask, tac.RenderTarget, tac.MipLevel, tac.SubElement); - - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavAutoReactive, Resources.AutoReactive); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavAutoComposition, Resources.AutoComposition); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavPrevColorPreAlpha, Resources.PrevPreAlpha[frameIndex]); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavPrevColorPostAlpha, Resources.PrevPostAlpha[frameIndex]); - - commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); - commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbGenReactive, _tcrAutogenerateConstants, 0, Marshal.SizeOf()); - - commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); - } - } -} diff --git a/Assets/Scripts/Core/Fsr3UpscalerResources.cs b/Assets/Scripts/Core/Fsr3UpscalerResources.cs deleted file mode 100644 index 344a371..0000000 --- a/Assets/Scripts/Core/Fsr3UpscalerResources.cs +++ /dev/null @@ -1,258 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -using System; -using UnityEngine; -using UnityEngine.Experimental.Rendering; -using UnityEngine.Rendering; - -namespace FidelityFX -{ - /// - /// Helper class for bundling and managing persistent resources required by the FSR3 Upscaler process. - /// This includes lookup tables, default fallback resources and double-buffered resources that get swapped between frames. - /// - internal class Fsr3UpscalerResources - { - public Texture2D DefaultExposure; - public Texture2D DefaultReactive; - public Texture2D LanczosLut; - public Texture2D MaximumBiasLut; - public RenderTexture SpdAtomicCounter; - public RenderTexture AutoExposure; - public RenderTexture SceneLuminance; - public RenderTexture AutoReactive; - public RenderTexture AutoComposition; - public readonly RenderTexture[] DilatedMotionVectors = new RenderTexture[2]; - public readonly RenderTexture[] LockStatus = new RenderTexture[2]; - public readonly RenderTexture[] InternalUpscaled = new RenderTexture[2]; - public readonly RenderTexture[] LumaHistory = new RenderTexture[2]; - public readonly RenderTexture[] PrevPreAlpha = new RenderTexture[2]; - public readonly RenderTexture[] PrevPostAlpha = new RenderTexture[2]; - - public void Create(Fsr3Upscaler.ContextDescription contextDescription) - { - // Generate the data for the LUT - const int lanczos2LutWidth = 128; - float[] lanczos2Weights = new float[lanczos2LutWidth]; - for (int currentLanczosWidthIndex = 0; currentLanczosWidthIndex < lanczos2LutWidth; ++currentLanczosWidthIndex) - { - float x = 2.0f * currentLanczosWidthIndex / (lanczos2LutWidth - 1); - float y = Fsr3Upscaler.Lanczos2(x); - lanczos2Weights[currentLanczosWidthIndex] = y; - } - - float[] maximumBias = new float[MaximumBiasTextureWidth * MaximumBiasTextureHeight]; - for (int i = 0; i < maximumBias.Length; ++i) - { - maximumBias[i] = MaximumBias[i] / 2.0f; - } - - // Resource FSR3UPSCALER_LanczosLutData: FFX_RESOURCE_USAGE_READ_ONLY, FFX_SURFACE_FORMAT_R16_SNORM, FFX_RESOURCE_FLAGS_NONE - // R16_SNorm textures are not supported by Unity on most platforms, strangely enough. So instead we use R32_SFloat and upload pre-normalized float data. - LanczosLut = new Texture2D(lanczos2LutWidth, 1, GraphicsFormat.R32_SFloat, TextureCreationFlags.None) { name = "FSR3UPSCALER_LanczosLutData" }; - LanczosLut.SetPixelData(lanczos2Weights, 0); - LanczosLut.Apply(); - - // Resource FSR3UPSCALER_MaximumUpsampleBias: FFX_RESOURCE_USAGE_READ_ONLY, FFX_SURFACE_FORMAT_R16_SNORM, FFX_RESOURCE_FLAGS_NONE - MaximumBiasLut = new Texture2D(MaximumBiasTextureWidth, MaximumBiasTextureHeight, GraphicsFormat.R32_SFloat, TextureCreationFlags.None) { name = "FSR3UPSCALER_MaximumUpsampleBias" }; - MaximumBiasLut.SetPixelData(maximumBias, 0); - MaximumBiasLut.Apply(); - - // Resource FSR3UPSCALER_DefaultExposure: FFX_RESOURCE_USAGE_READ_ONLY, FFX_SURFACE_FORMAT_R32G32_FLOAT, FFX_RESOURCE_FLAGS_NONE - DefaultExposure = new Texture2D(1, 1, GraphicsFormat.R32G32_SFloat, TextureCreationFlags.None) { name = "FSR3UPSCALER_DefaultExposure" }; - DefaultExposure.SetPixel(0, 0, Color.clear); - DefaultExposure.Apply(); - - // Resource FSR3UPSCALER_DefaultReactivityMask: FFX_RESOURCE_USAGE_READ_ONLY, FFX_SURFACE_FORMAT_R8_UNORM, FFX_RESOURCE_FLAGS_NONE - DefaultReactive = new Texture2D(1, 1, GraphicsFormat.R8_UNorm, TextureCreationFlags.None) { name = "FSR3UPSCALER_DefaultReactivityMask" }; - DefaultReactive.SetPixel(0, 0, Color.clear); - DefaultReactive.Apply(); - - // Resource FSR3UPSCALER_SpdAtomicCounter: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R32_UINT, FFX_RESOURCE_FLAGS_ALIASABLE - // Despite what the original FSR3 codebase says, this resource really isn't aliasable. Resetting this counter to 0 every frame breaks auto-exposure on MacOS Metal. - SpdAtomicCounter = new RenderTexture(1, 1, 0, GraphicsFormat.R32_UInt) { name = "FSR3UPSCALER_SpdAtomicCounter", enableRandomWrite = true }; - SpdAtomicCounter.Create(); - - // Resource FSR3UPSCALER_AutoExposure: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R32G32_FLOAT, FFX_RESOURCE_FLAGS_NONE - AutoExposure = new RenderTexture(1, 1, 0, GraphicsFormat.R32G32_SFloat) { name = "FSR3UPSCALER_AutoExposure", enableRandomWrite = true }; - AutoExposure.Create(); - - // Resource FSR3UPSCALER_ExposureMips: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16_FLOAT, FFX_RESOURCE_FLAGS_ALIASABLE - // This is a rather special case: it's an aliasable resource, but because we require a mipmap chain and bind specific mip levels per shader, we can't easily use temporary RTs for this. - int w = contextDescription.MaxRenderSize.x / 2, h = contextDescription.MaxRenderSize.y / 2; - int mipCount = 1 + Mathf.FloorToInt(Mathf.Log(Math.Max(w, h), 2.0f)); - SceneLuminance = new RenderTexture(w, h, 0, GraphicsFormat.R16_SFloat, mipCount) { name = "FSR3UPSCALER_ExposureMips", enableRandomWrite = true, useMipMap = true, autoGenerateMips = false }; - SceneLuminance.Create(); - - // Resources FSR3UPSCALER_InternalDilatedVelocity1/2: FFX_RESOURCE_USAGE_RENDERTARGET | FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16G16_FLOAT, FFX_RESOURCE_FLAGS_NONE - CreateDoubleBufferedResource(DilatedMotionVectors, "FSR3UPSCALER_InternalDilatedVelocity", contextDescription.MaxRenderSize, GraphicsFormat.R16G16_SFloat); - - // Resources FSR3UPSCALER_LockStatus1/2: FFX_RESOURCE_USAGE_RENDERTARGET | FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16G16_FLOAT, FFX_RESOURCE_FLAGS_NONE - CreateDoubleBufferedResource(LockStatus, "FSR3UPSCALER_LockStatus", contextDescription.DisplaySize, GraphicsFormat.R16G16_SFloat); - - // Resources FSR3UPSCALER_InternalUpscaled1/2: FFX_RESOURCE_USAGE_RENDERTARGET | FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16G16B16A16_FLOAT, FFX_RESOURCE_FLAGS_NONE - CreateDoubleBufferedResource(InternalUpscaled, "FSR3UPSCALER_InternalUpscaled", contextDescription.DisplaySize, GraphicsFormat.R16G16B16A16_SFloat); - - // Resources FSR3UPSCALER_LumaHistory1/2: FFX_RESOURCE_USAGE_RENDERTARGET | FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R8G8B8A8_UNORM, FFX_RESOURCE_FLAGS_NONE - CreateDoubleBufferedResource(LumaHistory, "FSR3UPSCALER_LumaHistory", contextDescription.DisplaySize, GraphicsFormat.R8G8B8A8_UNorm); - } - - public void CreateTcrAutogenResources(Fsr3Upscaler.ContextDescription contextDescription) - { - // Resource FSR3UPSCALER_AutoReactive: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R8_UNORM, FFX_RESOURCE_FLAGS_NONE - AutoReactive = new RenderTexture(contextDescription.MaxRenderSize.x, contextDescription.MaxRenderSize.y, 0, GraphicsFormat.R8_UNorm) { name = "FSR3UPSCALER_AutoReactive", enableRandomWrite = true }; - AutoReactive.Create(); - - // Resource FSR3UPSCALER_AutoComposition: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R8_UNORM, FFX_RESOURCE_FLAGS_NONE - AutoComposition = new RenderTexture(contextDescription.MaxRenderSize.x, contextDescription.MaxRenderSize.y, 0, GraphicsFormat.R8_UNorm) { name = "FSR3UPSCALER_AutoComposition", enableRandomWrite = true }; - AutoComposition.Create(); - - // Resources FSR3UPSCALER_PrevPreAlpha0/1: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R11G11B10_FLOAT, FFX_RESOURCE_FLAGS_NONE - CreateDoubleBufferedResource(PrevPreAlpha, "FSR3UPSCALER_PrevPreAlpha", contextDescription.MaxRenderSize, GraphicsFormat.B10G11R11_UFloatPack32); - - // Resources FSR3UPSCALER_PrevPostAlpha0/1: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R11G11B10_FLOAT, FFX_RESOURCE_FLAGS_NONE - CreateDoubleBufferedResource(PrevPostAlpha, "FSR3UPSCALER_PrevPostAlpha", contextDescription.MaxRenderSize, GraphicsFormat.B10G11R11_UFloatPack32); - } - - // Set up shared aliasable resources, i.e. temporary render textures - // These do not need to persist between frames, but they do need to be available between passes - public static void CreateAliasableResources(CommandBuffer commandBuffer, Fsr3Upscaler.ContextDescription contextDescription, Fsr3Upscaler.DispatchDescription dispatchParams) - { - Vector2Int displaySize = contextDescription.DisplaySize; - Vector2Int maxRenderSize = contextDescription.MaxRenderSize; - - // FSR3UPSCALER_ReconstructedPrevNearestDepth: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R32_UINT, FFX_RESOURCE_FLAGS_ALIASABLE - commandBuffer.GetTemporaryRT(Fsr3ShaderIDs.UavReconstructedPrevNearestDepth, maxRenderSize.x, maxRenderSize.y, 0, default, GraphicsFormat.R32_UInt, 1, true); - - // FSR3UPSCALER_DilatedDepth: FFX_RESOURCE_USAGE_RENDERTARGET | FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R32_FLOAT, FFX_RESOURCE_FLAGS_ALIASABLE - commandBuffer.GetTemporaryRT(Fsr3ShaderIDs.UavDilatedDepth, maxRenderSize.x, maxRenderSize.y, 0, default, GraphicsFormat.R32_SFloat, 1, true); - - // FSR3UPSCALER_LockInputLuma: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16_FLOAT, FFX_RESOURCE_FLAGS_ALIASABLE - commandBuffer.GetTemporaryRT(Fsr3ShaderIDs.UavLockInputLuma, maxRenderSize.x, maxRenderSize.y, 0, default, GraphicsFormat.R16_SFloat, 1, true); - - // FSR3UPSCALER_DilatedReactiveMasks: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R8G8_UNORM, FFX_RESOURCE_FLAGS_ALIASABLE - commandBuffer.GetTemporaryRT(Fsr3ShaderIDs.UavDilatedReactiveMasks, maxRenderSize.x, maxRenderSize.y, 0, default, GraphicsFormat.R8G8_UNorm, 1, true); - - // FSR3UPSCALER_PreparedInputColor: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16G16B16A16_FLOAT, FFX_RESOURCE_FLAGS_ALIASABLE - commandBuffer.GetTemporaryRT(Fsr3ShaderIDs.UavPreparedInputColor, maxRenderSize.x, maxRenderSize.y, 0, default, GraphicsFormat.R16G16B16A16_SFloat, 1, true); - - // FSR3UPSCALER_NewLocks: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R8_UNORM, FFX_RESOURCE_FLAGS_ALIASABLE - commandBuffer.GetTemporaryRT(Fsr3ShaderIDs.UavNewLocks, displaySize.x, displaySize.y, 0, default, GraphicsFormat.R8_UNorm, 1, true); - } - - public static void DestroyAliasableResources(CommandBuffer commandBuffer) - { - // Release all of the aliasable resources used this frame - commandBuffer.ReleaseTemporaryRT(Fsr3ShaderIDs.UavReconstructedPrevNearestDepth); - commandBuffer.ReleaseTemporaryRT(Fsr3ShaderIDs.UavDilatedDepth); - commandBuffer.ReleaseTemporaryRT(Fsr3ShaderIDs.UavLockInputLuma); - commandBuffer.ReleaseTemporaryRT(Fsr3ShaderIDs.UavDilatedReactiveMasks); - commandBuffer.ReleaseTemporaryRT(Fsr3ShaderIDs.UavPreparedInputColor); - commandBuffer.ReleaseTemporaryRT(Fsr3ShaderIDs.UavNewLocks); - } - - private static void CreateDoubleBufferedResource(RenderTexture[] resource, string name, Vector2Int size, GraphicsFormat format) - { - for (int i = 0; i < 2; ++i) - { - resource[i] = new RenderTexture(size.x, size.y, 0, format) { name = name + (i + 1), enableRandomWrite = true }; - resource[i].Create(); - } - } - - public void Destroy() - { - DestroyTcrAutogenResources(); - - DestroyResource(LumaHistory); - DestroyResource(InternalUpscaled); - DestroyResource(LockStatus); - DestroyResource(DilatedMotionVectors); - DestroyResource(ref SceneLuminance); - DestroyResource(ref AutoExposure); - DestroyResource(ref DefaultReactive); - DestroyResource(ref DefaultExposure); - DestroyResource(ref MaximumBiasLut); - DestroyResource(ref LanczosLut); - } - - public void DestroyTcrAutogenResources() - { - DestroyResource(PrevPostAlpha); - DestroyResource(PrevPreAlpha); - DestroyResource(ref AutoComposition); - DestroyResource(ref AutoReactive); - } - - private static void DestroyResource(ref Texture2D resource) - { - if (resource == null) - return; - -#if UNITY_EDITOR - if (Application.isPlaying && !UnityEditor.EditorApplication.isPaused) - UnityEngine.Object.Destroy(resource); - else - UnityEngine.Object.DestroyImmediate(resource); -#else - UnityEngine.Object.Destroy(resource); -#endif - resource = null; - } - - private static void DestroyResource(ref RenderTexture resource) - { - if (resource == null) - return; - - resource.Release(); - resource = null; - } - - private static void DestroyResource(RenderTexture[] resource) - { - for (int i = 0; i < resource.Length; ++i) - DestroyResource(ref resource[i]); - } - - private const int MaximumBiasTextureWidth = 16; - private const int MaximumBiasTextureHeight = 16; - private static readonly float[] MaximumBias = - { - 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.876f, 1.809f, 1.772f, 1.753f, 1.748f, - 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.869f, 1.801f, 1.764f, 1.745f, 1.739f, - 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.976f, 1.841f, 1.774f, 1.737f, 1.716f, 1.71f, - 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.914f, 1.784f, 1.716f, 1.673f, 1.649f, 1.641f, - 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.793f, 1.676f, 1.604f, 1.562f, 1.54f, 1.533f, - 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.802f, 1.619f, 1.536f, 1.492f, 1.467f, 1.454f, 1.449f, - 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.812f, 1.575f, 1.496f, 1.456f, 1.432f, 1.416f, 1.408f, 1.405f, - 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.555f, 1.479f, 1.438f, 1.413f, 1.398f, 1.387f, 1.381f, 1.379f, - 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.812f, 1.555f, 1.474f, 1.43f, 1.404f, 1.387f, 1.376f, 1.368f, 1.363f, 1.362f, - 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.802f, 1.575f, 1.479f, 1.43f, 1.401f, 1.382f, 1.369f, 1.36f, 1.354f, 1.351f, 1.35f, - 2.0f, 2.0f, 1.976f, 1.914f, 1.793f, 1.619f, 1.496f, 1.438f, 1.404f, 1.382f, 1.367f, 1.357f, 1.349f, 1.344f, 1.341f, 1.34f, - 1.876f, 1.869f, 1.841f, 1.784f, 1.676f, 1.536f, 1.456f, 1.413f, 1.387f, 1.369f, 1.357f, 1.347f, 1.341f, 1.336f, 1.333f, 1.332f, - 1.809f, 1.801f, 1.774f, 1.716f, 1.604f, 1.492f, 1.432f, 1.398f, 1.376f, 1.36f, 1.349f, 1.341f, 1.335f, 1.33f, 1.328f, 1.327f, - 1.772f, 1.764f, 1.737f, 1.673f, 1.562f, 1.467f, 1.416f, 1.387f, 1.368f, 1.354f, 1.344f, 1.336f, 1.33f, 1.326f, 1.323f, 1.323f, - 1.753f, 1.745f, 1.716f, 1.649f, 1.54f, 1.454f, 1.408f, 1.381f, 1.363f, 1.351f, 1.341f, 1.333f, 1.328f, 1.323f, 1.321f, 1.32f, - 1.748f, 1.739f, 1.71f, 1.641f, 1.533f, 1.449f, 1.405f, 1.379f, 1.362f, 1.35f, 1.34f, 1.332f, 1.327f, 1.323f, 1.32f, 1.319f, - }; - } -} diff --git a/Assets/Scripts/Fsr3UpscalerImageEffect.cs b/Assets/Scripts/Fsr3UpscalerImageEffect.cs index a5591a5..bbde974 100644 --- a/Assets/Scripts/Fsr3UpscalerImageEffect.cs +++ b/Assets/Scripts/Fsr3UpscalerImageEffect.cs @@ -1,4 +1,4 @@ -// Copyright (c) 2023 Nico de Poel +// Copyright (c) 2024 Nico de Poel // // Permission is hereby granted, free of charge, to any person obtaining a copy // of this software and associated documentation files (the "Software"), to deal @@ -22,6 +22,7 @@ using System; using UnityEngine; using UnityEngine.Experimental.Rendering; using UnityEngine.Rendering; +using FidelityFX.FSR3; namespace FidelityFX { @@ -54,6 +55,10 @@ namespace FidelityFX [Tooltip("Optional 1x1 texture containing the exposure value for the current frame.")] public Texture exposure = null; + [Header("Debug")] + [Tooltip("Enable a debug view to analyze the upscaling process.")] + public bool enableDebugView = false; + [Header("Reactivity, Transparency & Composition")] [Tooltip("Optional texture to control the influence of the current frame on the reconstructed output. If unset, either an auto-generated or a default cleared reactive mask will be used.")] public Texture reactiveMask = null; @@ -326,12 +331,14 @@ namespace FidelityFX _dispatchDescription.MotionVectorScale.x = -scaledRenderSize.x; _dispatchDescription.MotionVectorScale.y = -scaledRenderSize.y; _dispatchDescription.RenderSize = scaledRenderSize; + _dispatchDescription.UpscaleSize = _displaySize; _dispatchDescription.FrameTimeDelta = Time.unscaledDeltaTime; _dispatchDescription.CameraNear = _renderCamera.nearClipPlane; _dispatchDescription.CameraFar = _renderCamera.farClipPlane; _dispatchDescription.CameraFovAngleVertical = _renderCamera.fieldOfView * Mathf.Deg2Rad; _dispatchDescription.ViewSpaceToMetersFactor = 1.0f; // 1 unit is 1 meter in Unity _dispatchDescription.Reset = _resetHistory; + _dispatchDescription.Flags = enableDebugView ? Fsr3Upscaler.DispatchFlags.DrawDebugView : 0; _resetHistory = false; // Set up the parameters for the optional experimental auto-TCR feature @@ -390,9 +397,6 @@ namespace FidelityFX _renderCamera.rect = _originalRect; _renderCamera.ResetProjectionMatrix(); - // Update the input resource descriptions - _dispatchDescription.InputResourceSize = new Vector2Int(src.width, src.height); - _dispatchCommandBuffer.Clear(); if (autoGenerateReactiveMask) diff --git a/Assets/Scripts/Fsr3UpscalerImageEffectHelper.cs b/Assets/Scripts/Fsr3UpscalerImageEffectHelper.cs index 785ab8c..87bc8c9 100644 --- a/Assets/Scripts/Fsr3UpscalerImageEffectHelper.cs +++ b/Assets/Scripts/Fsr3UpscalerImageEffectHelper.cs @@ -20,6 +20,7 @@ using System.Collections; using UnityEngine; +using FidelityFX.FSR3; namespace FidelityFX { diff --git a/Assets/Shaders.meta b/Assets/Shaders.meta deleted file mode 100644 index a1e9f1a..0000000 --- a/Assets/Shaders.meta +++ /dev/null @@ -1,8 +0,0 @@ -fileFormatVersion: 2 -guid: f1c65c7377a34fe41beb57484d87a08b -folderAsset: yes -DefaultImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3.meta b/Assets/Shaders/FSR3.meta deleted file mode 100644 index ca8a4cc..0000000 --- a/Assets/Shaders/FSR3.meta +++ /dev/null @@ -1,8 +0,0 @@ -fileFormatVersion: 2 -guid: 661ffc6dbd7389b4da99fb9f749745de -folderAsset: yes -DefaultImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/ffx_fsr3upscaler_accumulate_pass.compute b/Assets/Shaders/FSR3/ffx_fsr3upscaler_accumulate_pass.compute deleted file mode 100644 index 5cbfb80..0000000 --- a/Assets/Shaders/FSR3/ffx_fsr3upscaler_accumulate_pass.compute +++ /dev/null @@ -1,41 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -#pragma kernel CS - -#pragma multi_compile_local __ FFX_HALF -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_HDR_COLOR_INPUT -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_APPLY_SHARPENING - -#pragma multi_compile_local __ UNITY_FSR3UPSCALER_HDRP - -#include "ffx_fsr3upscaler_unity_common.cginc" - -// Ensure the correct value is defined for this keyword, as it is used to select one of multiple sampler functions -#ifdef FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE -#undef FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE -#define FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE 1 -#endif - -#include "shaders/ffx_fsr3upscaler_accumulate_pass.hlsl" diff --git a/Assets/Shaders/FSR3/ffx_fsr3upscaler_autogen_reactive_pass.compute b/Assets/Shaders/FSR3/ffx_fsr3upscaler_autogen_reactive_pass.compute deleted file mode 100644 index e13c001..0000000 --- a/Assets/Shaders/FSR3/ffx_fsr3upscaler_autogen_reactive_pass.compute +++ /dev/null @@ -1,32 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -#pragma kernel CS - -#pragma multi_compile_local __ FFX_HALF -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - -#pragma multi_compile_local __ UNITY_FSR3UPSCALER_HDRP - -#include "ffx_fsr3upscaler_unity_common.cginc" - -#include "shaders/ffx_fsr3upscaler_autogen_reactive_pass.hlsl" diff --git a/Assets/Shaders/FSR3/ffx_fsr3upscaler_compute_luminance_pyramid_pass.compute b/Assets/Shaders/FSR3/ffx_fsr3upscaler_compute_luminance_pyramid_pass.compute deleted file mode 100644 index d5903c0..0000000 --- a/Assets/Shaders/FSR3/ffx_fsr3upscaler_compute_luminance_pyramid_pass.compute +++ /dev/null @@ -1,42 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -#pragma kernel CS - -#pragma multi_compile_local __ FFX_HALF -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - -#pragma multi_compile_local __ UNITY_FSR3UPSCALER_HDRP - -#include "ffx_fsr3upscaler_unity_common.cginc" - -// Wave operations require shader model 6.0; this can only be enabled when using DXC on D3D12 -// These pragmas are commented out by default as Unity will sometimes ignore the #if's and try to enable these features anyway. -// Uncomment the below lines if you intend to try wave operations on DX12 with the DXC compiler. -//#if defined(UNITY_COMPILER_DXC) && defined(SHADER_API_D3D12) -//#pragma require WaveBasic // Required for WaveGetLaneIndex -//#pragma require WaveBallot // Required for WaveReadLaneAt -//#else -#define FFX_SPD_NO_WAVE_OPERATIONS -//#endif - -#include "shaders/ffx_fsr3upscaler_compute_luminance_pyramid_pass.hlsl" diff --git a/Assets/Shaders/FSR3/ffx_fsr3upscaler_depth_clip_pass.compute b/Assets/Shaders/FSR3/ffx_fsr3upscaler_depth_clip_pass.compute deleted file mode 100644 index 0ccd388..0000000 --- a/Assets/Shaders/FSR3/ffx_fsr3upscaler_depth_clip_pass.compute +++ /dev/null @@ -1,32 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -#pragma kernel CS - -#pragma multi_compile_local __ FFX_HALF -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - -#pragma multi_compile_local __ UNITY_FSR3UPSCALER_HDRP - -#include "ffx_fsr3upscaler_unity_common.cginc" - -#include "shaders/ffx_fsr3upscaler_depth_clip_pass.hlsl" diff --git a/Assets/Shaders/FSR3/ffx_fsr3upscaler_lock_pass.compute b/Assets/Shaders/FSR3/ffx_fsr3upscaler_lock_pass.compute deleted file mode 100644 index e38ad99..0000000 --- a/Assets/Shaders/FSR3/ffx_fsr3upscaler_lock_pass.compute +++ /dev/null @@ -1,30 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -#pragma kernel CS - -#pragma multi_compile_local __ FFX_HALF -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - -#include "ffx_fsr3upscaler_unity_common.cginc" - -#include "shaders/ffx_fsr3upscaler_lock_pass.hlsl" diff --git a/Assets/Shaders/FSR3/ffx_fsr3upscaler_rcas_pass.compute b/Assets/Shaders/FSR3/ffx_fsr3upscaler_rcas_pass.compute deleted file mode 100644 index be7bbb5..0000000 --- a/Assets/Shaders/FSR3/ffx_fsr3upscaler_rcas_pass.compute +++ /dev/null @@ -1,31 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -#pragma kernel CS - -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - -#pragma multi_compile_local __ UNITY_FSR3UPSCALER_HDRP - -#include "ffx_fsr3upscaler_unity_common.cginc" - -#include "shaders/ffx_fsr3upscaler_rcas_pass.hlsl" diff --git a/Assets/Shaders/FSR3/ffx_fsr3upscaler_reconstruct_previous_depth_pass.compute b/Assets/Shaders/FSR3/ffx_fsr3upscaler_reconstruct_previous_depth_pass.compute deleted file mode 100644 index ee2f276..0000000 --- a/Assets/Shaders/FSR3/ffx_fsr3upscaler_reconstruct_previous_depth_pass.compute +++ /dev/null @@ -1,33 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -#pragma kernel CS - -#pragma multi_compile_local __ FFX_HALF -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_HDR_COLOR_INPUT -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - -#pragma multi_compile_local __ UNITY_FSR3UPSCALER_HDRP - -#include "ffx_fsr3upscaler_unity_common.cginc" - -#include "shaders/ffx_fsr3upscaler_reconstruct_previous_depth_pass.hlsl" diff --git a/Assets/Shaders/FSR3/ffx_fsr3upscaler_tcr_autogen_pass.compute b/Assets/Shaders/FSR3/ffx_fsr3upscaler_tcr_autogen_pass.compute deleted file mode 100644 index 6338918..0000000 --- a/Assets/Shaders/FSR3/ffx_fsr3upscaler_tcr_autogen_pass.compute +++ /dev/null @@ -1,32 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -#pragma kernel CS - -#pragma multi_compile_local __ FFX_HALF -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - -#pragma multi_compile_local __ UNITY_FSR3UPSCALER_HDRP - -#include "ffx_fsr3upscaler_unity_common.cginc" - -#include "shaders/ffx_fsr3upscaler_tcr_autogen_pass.hlsl" diff --git a/Assets/Shaders/FSR3/ffx_fsr3upscaler_unity_common.cginc b/Assets/Shaders/FSR3/ffx_fsr3upscaler_unity_common.cginc deleted file mode 100644 index 758bb0c..0000000 --- a/Assets/Shaders/FSR3/ffx_fsr3upscaler_unity_common.cginc +++ /dev/null @@ -1,82 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -// Suppress a few warnings produced by FFX's HLSL code -#pragma warning(disable: 3078) // Loop control variable conflicts -#pragma warning(disable: 3203) // Signed/unsigned mismatch - -#define FFX_GPU // Compiling for GPU -#define FFX_HLSL // Compile for plain HLSL - -// Use the DXC shader compiler on modern graphics APIs to enable a few advanced features -// The DXC-related pragmas are disabled by default, as DXC doesn't support all platforms yet and will break on some platforms when enabled. -// Consider this to be an experimental feature. If you want to benefit from 16-bit floating point and wave operations, and don't care about supporting older graphics APIs, then it's worth a try. -//#if defined(SHADER_API_D3D12) || defined(SHADER_API_VULKAN) || defined(SHADER_API_METAL) -//#pragma use_dxc // Using DXC will currently break DX11 support since DX11 and DX12 share the same shader bytecode in Unity. -//#endif - -// Enable half precision data types on platforms that support it -//#if defined(UNITY_COMPILER_DXC) && defined(FFX_HALF) -//#pragma require Native16Bit -//#endif - -// Hack to work around the lack of texture atomics on Metal -#if defined(SHADER_API_METAL) -#define InterlockedAdd(dest, val, orig) { (orig) = (dest); (dest) += (val); } -#define InterlockedMin(dest, val) { (dest) = min((dest), (val)); } -#define InterlockedMax(dest, val) { (dest) = max((dest), (val)); } -#endif - -// Workaround for HDRP using texture arrays for its camera buffers on some platforms -// The below defines are copied from: Packages/com.unity.render-pipelines.high-definition/Runtime/ShaderLibrary/TextureXR.hlsl -#if defined(UNITY_FSR3UPSCALER_HDRP) - // Must be in sync with C# with property useTexArray in TextureXR.cs - #if ((defined(SHADER_API_D3D11) || defined(SHADER_API_D3D12)) && !defined(SHADER_API_XBOXONE) && !defined(SHADER_API_GAMECORE)) || defined(SHADER_API_PSSL) || defined(SHADER_API_VULKAN) - #define UNITY_TEXTURE2D_X_ARRAY_SUPPORTED - #endif - - // Control if TEXTURE2D_X macros will expand to texture arrays - #if defined(UNITY_TEXTURE2D_X_ARRAY_SUPPORTED) && !defined(DISABLE_TEXTURE2D_X_ARRAY) - #define USE_TEXTURE2D_X_AS_ARRAY - #endif - - // Early defines for single-pass instancing - #if defined(STEREO_INSTANCING_ON) && defined(UNITY_TEXTURE2D_X_ARRAY_SUPPORTED) - #define UNITY_STEREO_INSTANCING_ENABLED - #endif - - // Helper macros to handle XR single-pass with Texture2DArray - #if defined(USE_TEXTURE2D_X_AS_ARRAY) - - // Only single-pass stereo instancing used array indexing - #if defined(UNITY_STEREO_INSTANCING_ENABLED) - #define SLICE_ARRAY_INDEX unity_StereoEyeIndex - #else - #define SLICE_ARRAY_INDEX 0 - #endif - - // Declare and sample camera buffers as texture arrays - #define UNITY_FSR3_TEX2D(type) Texture2DArray - #define UNITY_FSR3_RWTEX2D(type) RWTexture2DArray - #define UNITY_FSR3_POS(pxPos) FfxUInt32x3(pxPos, SLICE_ARRAY_INDEX) - #define UNITY_FSR3_UV(uv) FfxFloat32x3(uv, SLICE_ARRAY_INDEX) - - #endif -#endif diff --git a/Assets/Shaders/FSR3/shaders.meta b/Assets/Shaders/FSR3/shaders.meta deleted file mode 100644 index 8a4ff2b..0000000 --- a/Assets/Shaders/FSR3/shaders.meta +++ /dev/null @@ -1,8 +0,0 @@ -fileFormatVersion: 2 -guid: 8364d4f86c613ec4d999d062f5f773b8 -folderAsset: yes -DefaultImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_accumulate_pass.hlsl b/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_accumulate_pass.hlsl deleted file mode 100644 index d2f1b32..0000000 --- a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_accumulate_pass.hlsl +++ /dev/null @@ -1,79 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#define FSR3UPSCALER_BIND_SRV_INPUT_EXPOSURE 0 -#define FSR3UPSCALER_BIND_SRV_DILATED_REACTIVE_MASKS 1 -#if FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS -#define FSR3UPSCALER_BIND_SRV_DILATED_MOTION_VECTORS 2 -#else -#define FSR3UPSCALER_BIND_SRV_INPUT_MOTION_VECTORS 2 -#endif -#define FSR3UPSCALER_BIND_SRV_INTERNAL_UPSCALED 3 -#define FSR3UPSCALER_BIND_SRV_LOCK_STATUS 4 -#define FSR3UPSCALER_BIND_SRV_PREPARED_INPUT_COLOR 5 -#define FSR3UPSCALER_BIND_SRV_LANCZOS_LUT 6 -#define FSR3UPSCALER_BIND_SRV_UPSCALE_MAXIMUM_BIAS_LUT 7 -#define FSR3UPSCALER_BIND_SRV_SCENE_LUMINANCE_MIPS 8 -#define FSR3UPSCALER_BIND_SRV_AUTO_EXPOSURE 9 -#define FSR3UPSCALER_BIND_SRV_LUMA_HISTORY 10 - -#define FSR3UPSCALER_BIND_UAV_INTERNAL_UPSCALED 0 -#define FSR3UPSCALER_BIND_UAV_LOCK_STATUS 1 -#define FSR3UPSCALER_BIND_UAV_UPSCALED_OUTPUT 2 -#define FSR3UPSCALER_BIND_UAV_NEW_LOCKS 3 -#define FSR3UPSCALER_BIND_UAV_LUMA_HISTORY 4 - -#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 - -#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" -#include "fsr3upscaler/ffx_fsr3upscaler_common.h" -#include "fsr3upscaler/ffx_fsr3upscaler_sample.h" -#include "fsr3upscaler/ffx_fsr3upscaler_upsample.h" -#include "fsr3upscaler/ffx_fsr3upscaler_postprocess_lock_status.h" -#include "fsr3upscaler/ffx_fsr3upscaler_reproject.h" -#include "fsr3upscaler/ffx_fsr3upscaler_accumulate.h" - -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 8 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 8 -#endif // FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#ifndef FFX_FSR3UPSCALER_NUM_THREADS -#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] -#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS - -FFX_PREFER_WAVE64 -FFX_FSR3UPSCALER_NUM_THREADS -FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT -void CS(uint2 uGroupId : SV_GroupID, uint2 uGroupThreadId : SV_GroupThreadID) -{ - const uint GroupRows = (uint(DisplaySize().y) + FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT - 1) / FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT; - uGroupId.y = GroupRows - uGroupId.y - 1; - - uint2 uDispatchThreadId = uGroupId * uint2(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT) + uGroupThreadId; - - Accumulate(uDispatchThreadId); -} diff --git a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_accumulate_pass.hlsl.meta b/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_accumulate_pass.hlsl.meta deleted file mode 100644 index 80f209e..0000000 --- a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_accumulate_pass.hlsl.meta +++ /dev/null @@ -1,7 +0,0 @@ -fileFormatVersion: 2 -guid: 507ab779c38eddb429cdcedf9c108d1b -ShaderIncludeImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_autogen_reactive_pass.hlsl b/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_autogen_reactive_pass.hlsl deleted file mode 100644 index 0d6e2eb..0000000 --- a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_autogen_reactive_pass.hlsl +++ /dev/null @@ -1,77 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#define FSR3UPSCALER_BIND_SRV_INPUT_OPAQUE_ONLY 0 -#define FSR3UPSCALER_BIND_SRV_INPUT_COLOR 1 - -#define FSR3UPSCALER_BIND_UAV_AUTOREACTIVE 0 - -#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 -#define FSR3UPSCALER_BIND_CB_REACTIVE 1 - -#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" -#include "fsr3upscaler/ffx_fsr3upscaler_common.h" - -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 8 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 8 -#endif // FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#ifndef FFX_FSR3UPSCALER_NUM_THREADS -#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] -#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS - -FFX_FSR3UPSCALER_NUM_THREADS -FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT -void CS(uint2 uGroupId : SV_GroupID, uint2 uGroupThreadId : SV_GroupThreadID) -{ - uint2 uDispatchThreadId = uGroupId * uint2(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT) + uGroupThreadId; - - float3 ColorPreAlpha = LoadOpaqueOnly( FFX_MIN16_I2(uDispatchThreadId) ).rgb; - float3 ColorPostAlpha = LoadInputColor(uDispatchThreadId).rgb; - - if (GenReactiveFlags() & FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_APPLY_TONEMAP) - { - ColorPreAlpha = Tonemap(ColorPreAlpha); - ColorPostAlpha = Tonemap(ColorPostAlpha); - } - - if (GenReactiveFlags() & FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_APPLY_INVERSETONEMAP) - { - ColorPreAlpha = InverseTonemap(ColorPreAlpha); - ColorPostAlpha = InverseTonemap(ColorPostAlpha); - } - - float out_reactive_value = 0.f; - float3 delta = abs(ColorPostAlpha - ColorPreAlpha); - - out_reactive_value = (GenReactiveFlags() & FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_USE_COMPONENTS_MAX) ? max(delta.x, max(delta.y, delta.z)) : length(delta); - out_reactive_value *= GenReactiveScale(); - - out_reactive_value = (GenReactiveFlags() & FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_APPLY_THRESHOLD) ? (out_reactive_value < GenReactiveThreshold() ? 0 : GenReactiveBinaryValue()) : out_reactive_value; - - rw_output_autoreactive[uDispatchThreadId] = out_reactive_value; -} diff --git a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_autogen_reactive_pass.hlsl.meta b/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_autogen_reactive_pass.hlsl.meta deleted file mode 100644 index c55f004..0000000 --- a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_autogen_reactive_pass.hlsl.meta +++ /dev/null @@ -1,7 +0,0 @@ -fileFormatVersion: 2 -guid: 52cdb7a7c30cb614984908593ed19082 -ShaderIncludeImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_compute_luminance_pyramid_pass.hlsl b/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_compute_luminance_pyramid_pass.hlsl deleted file mode 100644 index 93b7332..0000000 --- a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_compute_luminance_pyramid_pass.hlsl +++ /dev/null @@ -1,55 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#define FSR3UPSCALER_BIND_SRV_INPUT_COLOR 0 - -#define FSR3UPSCALER_BIND_UAV_SPD_GLOBAL_ATOMIC 0 -#define FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_LUMA_CHANGE 1 -#define FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_5 2 -#define FSR3UPSCALER_BIND_UAV_AUTO_EXPOSURE 3 - -#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 -#define FSR3UPSCALER_BIND_CB_SPD 1 - -#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" -#include "fsr3upscaler/ffx_fsr3upscaler_common.h" -#include "fsr3upscaler/ffx_fsr3upscaler_compute_luminance_pyramid.h" - -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 256 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 1 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#ifndef FFX_FSR3UPSCALER_NUM_THREADS -#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] -#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS - -FFX_FSR3UPSCALER_NUM_THREADS -FFX_FSR3UPSCALER_EMBED_CB2_ROOTSIG_CONTENT -void CS(uint3 WorkGroupId : SV_GroupID, uint LocalThreadIndex : SV_GroupIndex) -{ - ComputeAutoExposure(WorkGroupId, LocalThreadIndex); -} diff --git a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_compute_luminance_pyramid_pass.hlsl.meta b/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_compute_luminance_pyramid_pass.hlsl.meta deleted file mode 100644 index 508b43e..0000000 --- a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_compute_luminance_pyramid_pass.hlsl.meta +++ /dev/null @@ -1,7 +0,0 @@ -fileFormatVersion: 2 -guid: 2d149b52ba0f5bb468a94a71dbbcb66f -ShaderIncludeImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_depth_clip_pass.hlsl b/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_depth_clip_pass.hlsl deleted file mode 100644 index 70cc7ba..0000000 --- a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_depth_clip_pass.hlsl +++ /dev/null @@ -1,67 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#define FSR3UPSCALER_BIND_SRV_RECONSTRUCTED_PREV_NEAREST_DEPTH 0 -#define FSR3UPSCALER_BIND_SRV_DILATED_MOTION_VECTORS 1 -#define FSR3UPSCALER_BIND_SRV_DILATED_DEPTH 2 -#define FSR3UPSCALER_BIND_SRV_REACTIVE_MASK 3 -#define FSR3UPSCALER_BIND_SRV_TRANSPARENCY_AND_COMPOSITION_MASK 4 -#define FSR3UPSCALER_BIND_SRV_PREVIOUS_DILATED_MOTION_VECTORS 5 -#define FSR3UPSCALER_BIND_SRV_INPUT_MOTION_VECTORS 6 -#define FSR3UPSCALER_BIND_SRV_INPUT_COLOR 7 -#define FSR3UPSCALER_BIND_SRV_INPUT_DEPTH 8 -#define FSR3UPSCALER_BIND_SRV_INPUT_EXPOSURE 9 - -#define FSR3UPSCALER_BIND_UAV_DILATED_REACTIVE_MASKS 0 -#define FSR3UPSCALER_BIND_UAV_PREPARED_INPUT_COLOR 1 - -#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 - -#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" -#include "fsr3upscaler/ffx_fsr3upscaler_common.h" -#include "fsr3upscaler/ffx_fsr3upscaler_sample.h" -#include "fsr3upscaler/ffx_fsr3upscaler_depth_clip.h" - -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 8 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 8 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#ifndef FFX_FSR3UPSCALER_NUM_THREADS -#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] -#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS - -FFX_PREFER_WAVE64 -FFX_FSR3UPSCALER_NUM_THREADS -FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT -void CS( - int2 iGroupId : SV_GroupID, - int2 iDispatchThreadId : SV_DispatchThreadID, - int2 iGroupThreadId : SV_GroupThreadID, - int iGroupIndex : SV_GroupIndex) -{ - DepthClip(iDispatchThreadId); -} diff --git a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_depth_clip_pass.hlsl.meta b/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_depth_clip_pass.hlsl.meta deleted file mode 100644 index cde3a5e..0000000 --- a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_depth_clip_pass.hlsl.meta +++ /dev/null @@ -1,7 +0,0 @@ -fileFormatVersion: 2 -guid: da435b71cf57e2247b80ae0f0f86d1f8 -ShaderIncludeImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_lock_pass.hlsl b/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_lock_pass.hlsl deleted file mode 100644 index 26b28f0..0000000 --- a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_lock_pass.hlsl +++ /dev/null @@ -1,56 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#define FSR3UPSCALER_BIND_SRV_LOCK_INPUT_LUMA 0 - -#define FSR3UPSCALER_BIND_UAV_NEW_LOCKS 0 -#define FSR3UPSCALER_BIND_UAV_RECONSTRUCTED_PREV_NEAREST_DEPTH 1 - -#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 - -#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" -#include "fsr3upscaler/ffx_fsr3upscaler_common.h" -#include "fsr3upscaler/ffx_fsr3upscaler_sample.h" -#include "fsr3upscaler/ffx_fsr3upscaler_lock.h" - -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 8 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 8 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#ifndef FFX_FSR3UPSCALER_NUM_THREADS -#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] -#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS - -FFX_PREFER_WAVE64 -FFX_FSR3UPSCALER_NUM_THREADS -FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT -void CS(uint2 uGroupId : SV_GroupID, uint2 uGroupThreadId : SV_GroupThreadID) -{ - uint2 uDispatchThreadId = uGroupId * uint2(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT) + uGroupThreadId; - - ComputeLock(uDispatchThreadId); -} diff --git a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_lock_pass.hlsl.meta b/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_lock_pass.hlsl.meta deleted file mode 100644 index 45c99dc..0000000 --- a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_lock_pass.hlsl.meta +++ /dev/null @@ -1,7 +0,0 @@ -fileFormatVersion: 2 -guid: 98d2cbbda5e90dd4ebd1d70abbb63a09 -ShaderIncludeImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_rcas_pass.hlsl b/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_rcas_pass.hlsl deleted file mode 100644 index bebdeb3..0000000 --- a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_rcas_pass.hlsl +++ /dev/null @@ -1,53 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#define FSR3UPSCALER_BIND_SRV_INPUT_EXPOSURE 0 -#define FSR3UPSCALER_BIND_SRV_RCAS_INPUT 1 - -#define FSR3UPSCALER_BIND_UAV_UPSCALED_OUTPUT 0 - -#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 -#define FSR3UPSCALER_BIND_CB_RCAS 1 - -#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" -#include "fsr3upscaler/ffx_fsr3upscaler_common.h" -#include "fsr3upscaler/ffx_fsr3upscaler_rcas.h" - -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 64 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 1 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#ifndef FFX_FSR3UPSCALER_NUM_THREADS -#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] -#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS - -FFX_FSR3UPSCALER_NUM_THREADS -FFX_FSR3UPSCALER_EMBED_CB2_ROOTSIG_CONTENT -void CS(uint3 LocalThreadId : SV_GroupThreadID, uint3 WorkGroupId : SV_GroupID, uint3 Dtid : SV_DispatchThreadID) -{ - RCAS(LocalThreadId, WorkGroupId, Dtid); -} diff --git a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_rcas_pass.hlsl.meta b/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_rcas_pass.hlsl.meta deleted file mode 100644 index fb9bfe2..0000000 --- a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_rcas_pass.hlsl.meta +++ /dev/null @@ -1,7 +0,0 @@ -fileFormatVersion: 2 -guid: 9a15fc73170a9bc478801c8fa4d8d574 -ShaderIncludeImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_reconstruct_previous_depth_pass.hlsl b/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_reconstruct_previous_depth_pass.hlsl deleted file mode 100644 index f277fd1..0000000 --- a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_reconstruct_previous_depth_pass.hlsl +++ /dev/null @@ -1,64 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#define FSR3UPSCALER_BIND_SRV_INPUT_MOTION_VECTORS 0 -#define FSR3UPSCALER_BIND_SRV_INPUT_DEPTH 1 -#define FSR3UPSCALER_BIND_SRV_INPUT_COLOR 2 -#define FSR3UPSCALER_BIND_SRV_INPUT_EXPOSURE 3 - -#define FSR3UPSCALER_BIND_UAV_RECONSTRUCTED_PREV_NEAREST_DEPTH 0 -#define FSR3UPSCALER_BIND_UAV_DILATED_MOTION_VECTORS 1 -#define FSR3UPSCALER_BIND_UAV_DILATED_DEPTH 2 -#define FSR3UPSCALER_BIND_UAV_LOCK_INPUT_LUMA 3 - -#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 - -#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" -#include "fsr3upscaler/ffx_fsr3upscaler_common.h" -#include "fsr3upscaler/ffx_fsr3upscaler_sample.h" -#include "fsr3upscaler/ffx_fsr3upscaler_reconstruct_dilated_velocity_and_previous_depth.h" - -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 8 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 8 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#ifndef FFX_FSR3UPSCALER_NUM_THREADS -#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] -#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS - -FFX_PREFER_WAVE64 -FFX_FSR3UPSCALER_NUM_THREADS -FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT -void CS( - int2 iGroupId : SV_GroupID, - int2 iDispatchThreadId : SV_DispatchThreadID, - int2 iGroupThreadId : SV_GroupThreadID, - int iGroupIndex : SV_GroupIndex -) -{ - ReconstructAndDilate(iDispatchThreadId); -} diff --git a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_reconstruct_previous_depth_pass.hlsl.meta b/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_reconstruct_previous_depth_pass.hlsl.meta deleted file mode 100644 index 6489d6d..0000000 --- a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_reconstruct_previous_depth_pass.hlsl.meta +++ /dev/null @@ -1,7 +0,0 @@ -fileFormatVersion: 2 -guid: bafb3726a76b97a49bb343d8a4323754 -ShaderIncludeImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_tcr_autogen_pass.hlsl b/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_tcr_autogen_pass.hlsl deleted file mode 100644 index 6180885..0000000 --- a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_tcr_autogen_pass.hlsl +++ /dev/null @@ -1,90 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2022-2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -#define FSR3UPSCALER_BIND_SRV_INPUT_OPAQUE_ONLY 0 -#define FSR3UPSCALER_BIND_SRV_INPUT_COLOR 1 -#define FSR3UPSCALER_BIND_SRV_INPUT_MOTION_VECTORS 2 -#define FSR3UPSCALER_BIND_SRV_PREV_PRE_ALPHA_COLOR 3 -#define FSR3UPSCALER_BIND_SRV_PREV_POST_ALPHA_COLOR 4 -#define FSR3UPSCALER_BIND_SRV_REACTIVE_MASK 4 -#define FSR3UPSCALER_BIND_SRV_TRANSPARENCY_AND_COMPOSITION_MASK 5 - -#define FSR3UPSCALER_BIND_UAV_AUTOREACTIVE 0 -#define FSR3UPSCALER_BIND_UAV_AUTOCOMPOSITION 1 -#define FSR3UPSCALER_BIND_UAV_PREV_PRE_ALPHA_COLOR 2 -#define FSR3UPSCALER_BIND_UAV_PREV_POST_ALPHA_COLOR 3 - -#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 -#define FSR3UPSCALER_BIND_CB_AUTOREACTIVE 1 - -#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" -#include "fsr3upscaler/ffx_fsr3upscaler_common.h" -#include "fsr3upscaler/ffx_fsr3upscaler_tcr_autogen.h" - -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 8 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 8 -#endif // FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#ifndef FFX_FSR3UPSCALER_NUM_THREADS -#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] -#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS - -FFX_FSR3UPSCALER_NUM_THREADS -FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT -void CS(uint2 uGroupId : SV_GroupID, uint2 uGroupThreadId : SV_GroupThreadID) -{ - FFX_MIN16_I2 uDispatchThreadId = FFX_MIN16_I2(uGroupId * uint2(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT) + uGroupThreadId); - - // ToDo: take into account jitter (i.e. add delta of previous jitter and current jitter to previous UV - // fetch pre- and post-alpha color values - FFX_MIN16_F2 fUv = ( FFX_MIN16_F2(uDispatchThreadId) + FFX_MIN16_F2(0.5f, 0.5f) ) / FFX_MIN16_F2( RenderSize() ); - FFX_MIN16_F2 fPrevUV = fUv + FFX_MIN16_F2( LoadInputMotionVector(uDispatchThreadId) ); - FFX_MIN16_I2 iPrevIdx = FFX_MIN16_I2(fPrevUV * FFX_MIN16_F2(RenderSize()) - 0.5f); - - FFX_MIN16_F3 colorPreAlpha = FFX_MIN16_F3( LoadOpaqueOnly( uDispatchThreadId ) ); - FFX_MIN16_F3 colorPostAlpha = FFX_MIN16_F3( LoadInputColor( uDispatchThreadId ) ); - - FFX_MIN16_F2 outReactiveMask = 0; - - outReactiveMask.y = ComputeTransparencyAndComposition(uDispatchThreadId, iPrevIdx); - - if (outReactiveMask.y > 0.5f) - { - outReactiveMask.x = ComputeReactive(uDispatchThreadId, iPrevIdx); - outReactiveMask.x *= FFX_MIN16_F(fReactiveScale); - outReactiveMask.x = outReactiveMask.x < fReactiveMax ? outReactiveMask.x : FFX_MIN16_F( fReactiveMax ); - } - - outReactiveMask.y *= FFX_MIN16_F(fTcScale ); - - outReactiveMask.x = max( outReactiveMask.x, FFX_MIN16_F( LoadReactiveMask(uDispatchThreadId) ) ); - outReactiveMask.y = max( outReactiveMask.y, FFX_MIN16_F( LoadTransparencyAndCompositionMask(uDispatchThreadId) ) ); - - StoreAutoReactive(uDispatchThreadId, outReactiveMask); - - StorePrevPreAlpha(uDispatchThreadId, colorPreAlpha); - StorePrevPostAlpha(uDispatchThreadId, colorPostAlpha); -} diff --git a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_tcr_autogen_pass.hlsl.meta b/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_tcr_autogen_pass.hlsl.meta deleted file mode 100644 index 02c5f46..0000000 --- a/Assets/Shaders/FSR3/shaders/ffx_fsr3upscaler_tcr_autogen_pass.hlsl.meta +++ /dev/null @@ -1,7 +0,0 @@ -fileFormatVersion: 2 -guid: 712d171118b59fc4fb28d0d487060d42 -ShaderIncludeImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler.meta deleted file mode 100644 index 2626a2e..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler.meta +++ /dev/null @@ -1,8 +0,0 @@ -fileFormatVersion: 2 -guid: ae9c6d015ae76544f9e8117e79ea862b -folderAsset: yes -DefaultImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_common_types.h b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_common_types.h deleted file mode 100644 index f0b62ab..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_common_types.h +++ /dev/null @@ -1,616 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#ifndef FFX_COMMON_TYPES_H -#define FFX_COMMON_TYPES_H - -#if defined(FFX_CPU) -#define FFX_PARAMETER_IN -#define FFX_PARAMETER_OUT -#define FFX_PARAMETER_INOUT -#define FFX_PARAMETER_UNIFORM -#elif defined(FFX_HLSL) -#define FFX_PARAMETER_IN in -#define FFX_PARAMETER_OUT out -#define FFX_PARAMETER_INOUT inout -#define FFX_PARAMETER_UNIFORM uniform -#elif defined(FFX_GLSL) -#define FFX_PARAMETER_IN in -#define FFX_PARAMETER_OUT out -#define FFX_PARAMETER_INOUT inout -#define FFX_PARAMETER_UNIFORM const //[cacao_placeholder] until a better fit is found! -#endif // #if defined(FFX_CPU) - -#if defined(FFX_CPU) -/// A typedef for a boolean value. -/// -/// @ingroup CPUTypes -typedef bool FfxBoolean; - -/// A typedef for a unsigned 8bit integer. -/// -/// @ingroup CPUTypes -typedef uint8_t FfxUInt8; - -/// A typedef for a unsigned 16bit integer. -/// -/// @ingroup CPUTypes -typedef uint16_t FfxUInt16; - -/// A typedef for a unsigned 32bit integer. -/// -/// @ingroup CPUTypes -typedef uint32_t FfxUInt32; - -/// A typedef for a unsigned 64bit integer. -/// -/// @ingroup CPUTypes -typedef uint64_t FfxUInt64; - -/// A typedef for a signed 8bit integer. -/// -/// @ingroup CPUTypes -typedef int8_t FfxInt8; - -/// A typedef for a signed 16bit integer. -/// -/// @ingroup CPUTypes -typedef int16_t FfxInt16; - -/// A typedef for a signed 32bit integer. -/// -/// @ingroup CPUTypes -typedef int32_t FfxInt32; - -/// A typedef for a signed 64bit integer. -/// -/// @ingroup CPUTypes -typedef int64_t FfxInt64; - -/// A typedef for a floating point value. -/// -/// @ingroup CPUTypes -typedef float FfxFloat32; - -/// A typedef for a 2-dimensional floating point value. -/// -/// @ingroup CPUTypes -typedef float FfxFloat32x2[2]; - -/// A typedef for a 3-dimensional floating point value. -/// -/// @ingroup CPUTypes -typedef float FfxFloat32x3[3]; - -/// A typedef for a 4-dimensional floating point value. -/// -/// @ingroup CPUTypes -typedef float FfxFloat32x4[4]; - -/// A typedef for a 2-dimensional 32bit unsigned integer. -/// -/// @ingroup CPUTypes -typedef uint32_t FfxUInt32x2[2]; - -/// A typedef for a 3-dimensional 32bit unsigned integer. -/// -/// @ingroup CPUTypes -typedef uint32_t FfxUInt32x3[3]; - -/// A typedef for a 4-dimensional 32bit unsigned integer. -/// -/// @ingroup CPUTypes -typedef uint32_t FfxUInt32x4[4]; -#endif // #if defined(FFX_CPU) - -#if defined(FFX_HLSL) - -#define FfxFloat32Mat4 matrix -#define FfxFloat32Mat3 matrix - -/// A typedef for a boolean value. -/// -/// @ingroup HLSLTypes -typedef bool FfxBoolean; - -#if FFX_HLSL_SM>=62 - -/// @defgroup HLSL62Types HLSL 6.2 And Above Types -/// HLSL 6.2 and above type defines for all commonly used variables -/// -/// @ingroup HLSLTypes - -/// A typedef for a floating point value. -/// -/// @ingroup HLSL62Types -typedef float32_t FfxFloat32; - -/// A typedef for a 2-dimensional floating point value. -/// -/// @ingroup HLSL62Types -typedef float32_t2 FfxFloat32x2; - -/// A typedef for a 3-dimensional floating point value. -/// -/// @ingroup HLSL62Types -typedef float32_t3 FfxFloat32x3; - -/// A typedef for a 4-dimensional floating point value. -/// -/// @ingroup HLSL62Types -typedef float32_t4 FfxFloat32x4; - -/// A [cacao_placeholder] typedef for matrix type until confirmed. -typedef float4x4 FfxFloat32x4x4; -typedef float3x3 FfxFloat32x3x3; -typedef float2x2 FfxFloat32x2x2; - -/// A typedef for a unsigned 32bit integer. -/// -/// @ingroup HLSL62Types -typedef uint32_t FfxUInt32; - -/// A typedef for a 2-dimensional 32bit unsigned integer. -/// -/// @ingroup HLSL62Types -typedef uint32_t2 FfxUInt32x2; - -/// A typedef for a 3-dimensional 32bit unsigned integer. -/// -/// @ingroup HLSL62Types -typedef uint32_t3 FfxUInt32x3; - -/// A typedef for a 4-dimensional 32bit unsigned integer. -/// -/// @ingroup HLSL62Types -typedef uint32_t4 FfxUInt32x4; - -/// A typedef for a signed 32bit integer. -/// -/// @ingroup HLSL62Types -typedef int32_t FfxInt32; - -/// A typedef for a 2-dimensional signed 32bit integer. -/// -/// @ingroup HLSL62Types -typedef int32_t2 FfxInt32x2; - -/// A typedef for a 3-dimensional signed 32bit integer. -/// -/// @ingroup HLSL62Types -typedef int32_t3 FfxInt32x3; - -/// A typedef for a 4-dimensional signed 32bit integer. -/// -/// @ingroup HLSL62Types -typedef int32_t4 FfxInt32x4; - -#else // #if FFX_HLSL_SM>=62 - -/// @defgroup HLSLBaseTypes HLSL 6.1 And Below Types -/// HLSL 6.1 and below type defines for all commonly used variables -/// -/// @ingroup HLSLTypes - -#define FfxFloat32 float -#define FfxFloat32x2 float2 -#define FfxFloat32x3 float3 -#define FfxFloat32x4 float4 - -/// A [cacao_placeholder] typedef for matrix type until confirmed. -#define FfxFloat32x4x4 float4x4 -#define FfxFloat32x3x3 float3x3 -#define FfxFloat32x2x2 float2x2 - -/// A typedef for a unsigned 32bit integer. -/// -/// @ingroup GPU -typedef uint FfxUInt32; -typedef uint2 FfxUInt32x2; -typedef uint3 FfxUInt32x3; -typedef uint4 FfxUInt32x4; - -typedef int FfxInt32; -typedef int2 FfxInt32x2; -typedef int3 FfxInt32x3; -typedef int4 FfxInt32x4; - -#endif // #if FFX_HLSL_SM>=62 - -#if FFX_HALF - -#if FFX_HLSL_SM >= 62 - -typedef float16_t FfxFloat16; -typedef float16_t2 FfxFloat16x2; -typedef float16_t3 FfxFloat16x3; -typedef float16_t4 FfxFloat16x4; - -/// A typedef for an unsigned 16bit integer. -/// -/// @ingroup HLSLTypes -typedef uint16_t FfxUInt16; -typedef uint16_t2 FfxUInt16x2; -typedef uint16_t3 FfxUInt16x3; -typedef uint16_t4 FfxUInt16x4; - -/// A typedef for a signed 16bit integer. -/// -/// @ingroup HLSLTypes -typedef int16_t FfxInt16; -typedef int16_t2 FfxInt16x2; -typedef int16_t3 FfxInt16x3; -typedef int16_t4 FfxInt16x4; -#elif SHADER_API_PSSL -#pragma argument(realtypes) // Enable true 16-bit types - -typedef half FfxFloat16; -typedef half2 FfxFloat16x2; -typedef half3 FfxFloat16x3; -typedef half4 FfxFloat16x4; - -/// A typedef for an unsigned 16bit integer. -/// -/// @ingroup GPU -typedef ushort FfxUInt16; -typedef ushort2 FfxUInt16x2; -typedef ushort3 FfxUInt16x3; -typedef ushort4 FfxUInt16x4; - -/// A typedef for a signed 16bit integer. -/// -/// @ingroup GPU -typedef short FfxInt16; -typedef short2 FfxInt16x2; -typedef short3 FfxInt16x3; -typedef short4 FfxInt16x4; -#else // #if FFX_HLSL_SM>=62 -typedef min16float FfxFloat16; -typedef min16float2 FfxFloat16x2; -typedef min16float3 FfxFloat16x3; -typedef min16float4 FfxFloat16x4; - -/// A typedef for an unsigned 16bit integer. -/// -/// @ingroup HLSLTypes -typedef min16uint FfxUInt16; -typedef min16uint2 FfxUInt16x2; -typedef min16uint3 FfxUInt16x3; -typedef min16uint4 FfxUInt16x4; - -/// A typedef for a signed 16bit integer. -/// -/// @ingroup HLSLTypes -typedef min16int FfxInt16; -typedef min16int2 FfxInt16x2; -typedef min16int3 FfxInt16x3; -typedef min16int4 FfxInt16x4; -#endif // #if FFX_HLSL_SM>=62 - -#endif // FFX_HALF - -#endif // #if defined(FFX_HLSL) - -#if defined(FFX_GLSL) - -#define FfxFloat32Mat4 mat4 -#define FfxFloat32Mat3 mat3 - -/// A typedef for a boolean value. -/// -/// @ingroup GLSLTypes -#define FfxBoolean bool -#define FfxFloat32 float -#define FfxFloat32x2 vec2 -#define FfxFloat32x3 vec3 -#define FfxFloat32x4 vec4 -#define FfxUInt32 uint -#define FfxUInt32x2 uvec2 -#define FfxUInt32x3 uvec3 -#define FfxUInt32x4 uvec4 -#define FfxInt32 int -#define FfxInt32x2 ivec2 -#define FfxInt32x3 ivec3 -#define FfxInt32x4 ivec4 - -/// A [cacao_placeholder] typedef for matrix type until confirmed. -#define FfxFloat32x4x4 mat4 -#define FfxFloat32x3x3 mat3 -#define FfxFloat32x2x2 mat2 - -#if FFX_HALF -#define FfxFloat16 float16_t -#define FfxFloat16x2 f16vec2 -#define FfxFloat16x3 f16vec3 -#define FfxFloat16x4 f16vec4 -#define FfxUInt16 uint16_t -#define FfxUInt16x2 u16vec2 -#define FfxUInt16x3 u16vec3 -#define FfxUInt16x4 u16vec4 -#define FfxInt16 int16_t -#define FfxInt16x2 i16vec2 -#define FfxInt16x3 i16vec3 -#define FfxInt16x4 i16vec4 -#endif // FFX_HALF -#endif // #if defined(FFX_GLSL) - -// Global toggles: -// #define FFX_HALF (1) -// #define FFX_HLSL_SM (62) - -#if FFX_HALF && !defined(SHADER_API_PSSL) - -#if FFX_HLSL_SM >= 62 - -#define FFX_MIN16_SCALAR( TypeName, BaseComponentType ) typedef BaseComponentType##16_t TypeName; -#define FFX_MIN16_VECTOR( TypeName, BaseComponentType, COL ) typedef vector TypeName; -#define FFX_MIN16_MATRIX( TypeName, BaseComponentType, ROW, COL ) typedef matrix TypeName; - -#define FFX_16BIT_SCALAR( TypeName, BaseComponentType ) typedef BaseComponentType##16_t TypeName; -#define FFX_16BIT_VECTOR( TypeName, BaseComponentType, COL ) typedef vector TypeName; -#define FFX_16BIT_MATRIX( TypeName, BaseComponentType, ROW, COL ) typedef matrix TypeName; - -#else //FFX_HLSL_SM>=62 - -#define FFX_MIN16_SCALAR( TypeName, BaseComponentType ) typedef min16##BaseComponentType TypeName; -#define FFX_MIN16_VECTOR( TypeName, BaseComponentType, COL ) typedef vector TypeName; -#define FFX_MIN16_MATRIX( TypeName, BaseComponentType, ROW, COL ) typedef matrix TypeName; - -#define FFX_16BIT_SCALAR( TypeName, BaseComponentType ) FFX_MIN16_SCALAR( TypeName, BaseComponentType ); -#define FFX_16BIT_VECTOR( TypeName, BaseComponentType, COL ) FFX_MIN16_VECTOR( TypeName, BaseComponentType, COL ); -#define FFX_16BIT_MATRIX( TypeName, BaseComponentType, ROW, COL ) FFX_MIN16_MATRIX( TypeName, BaseComponentType, ROW, COL ); - -#endif //FFX_HLSL_SM>=62 - -#else //FFX_HALF - -#define FFX_MIN16_SCALAR( TypeName, BaseComponentType ) typedef BaseComponentType TypeName; -#define FFX_MIN16_VECTOR( TypeName, BaseComponentType, COL ) typedef vector TypeName; -#define FFX_MIN16_MATRIX( TypeName, BaseComponentType, ROW, COL ) typedef matrix TypeName; - -#define FFX_16BIT_SCALAR( TypeName, BaseComponentType ) typedef BaseComponentType TypeName; -#define FFX_16BIT_VECTOR( TypeName, BaseComponentType, COL ) typedef vector TypeName; -#define FFX_16BIT_MATRIX( TypeName, BaseComponentType, ROW, COL ) typedef matrix TypeName; - -#endif //FFX_HALF - -#if defined(FFX_GPU) -// Common typedefs: -#if defined(FFX_HLSL) && !defined(SHADER_API_PSSL) -FFX_MIN16_SCALAR( FFX_MIN16_F , float ); -FFX_MIN16_VECTOR( FFX_MIN16_F2, float, 2 ); -FFX_MIN16_VECTOR( FFX_MIN16_F3, float, 3 ); -FFX_MIN16_VECTOR( FFX_MIN16_F4, float, 4 ); - -FFX_MIN16_SCALAR( FFX_MIN16_I, int ); -FFX_MIN16_VECTOR( FFX_MIN16_I2, int, 2 ); -FFX_MIN16_VECTOR( FFX_MIN16_I3, int, 3 ); -FFX_MIN16_VECTOR( FFX_MIN16_I4, int, 4 ); - -FFX_MIN16_SCALAR( FFX_MIN16_U, uint ); -FFX_MIN16_VECTOR( FFX_MIN16_U2, uint, 2 ); -FFX_MIN16_VECTOR( FFX_MIN16_U3, uint, 3 ); -FFX_MIN16_VECTOR( FFX_MIN16_U4, uint, 4 ); - -FFX_16BIT_SCALAR( FFX_F16_t , float ); -FFX_16BIT_VECTOR( FFX_F16_t2, float, 2 ); -FFX_16BIT_VECTOR( FFX_F16_t3, float, 3 ); -FFX_16BIT_VECTOR( FFX_F16_t4, float, 4 ); - -FFX_16BIT_SCALAR( FFX_I16_t, int ); -FFX_16BIT_VECTOR( FFX_I16_t2, int, 2 ); -FFX_16BIT_VECTOR( FFX_I16_t3, int, 3 ); -FFX_16BIT_VECTOR( FFX_I16_t4, int, 4 ); - -FFX_16BIT_SCALAR( FFX_U16_t, uint ); -FFX_16BIT_VECTOR( FFX_U16_t2, uint, 2 ); -FFX_16BIT_VECTOR( FFX_U16_t3, uint, 3 ); -FFX_16BIT_VECTOR( FFX_U16_t4, uint, 4 ); - -#define TYPEDEF_MIN16_TYPES(Prefix) \ -typedef FFX_MIN16_F Prefix##_F; \ -typedef FFX_MIN16_F2 Prefix##_F2; \ -typedef FFX_MIN16_F3 Prefix##_F3; \ -typedef FFX_MIN16_F4 Prefix##_F4; \ -typedef FFX_MIN16_I Prefix##_I; \ -typedef FFX_MIN16_I2 Prefix##_I2; \ -typedef FFX_MIN16_I3 Prefix##_I3; \ -typedef FFX_MIN16_I4 Prefix##_I4; \ -typedef FFX_MIN16_U Prefix##_U; \ -typedef FFX_MIN16_U2 Prefix##_U2; \ -typedef FFX_MIN16_U3 Prefix##_U3; \ -typedef FFX_MIN16_U4 Prefix##_U4; - -#define TYPEDEF_16BIT_TYPES(Prefix) \ -typedef FFX_16BIT_F Prefix##_F; \ -typedef FFX_16BIT_F2 Prefix##_F2; \ -typedef FFX_16BIT_F3 Prefix##_F3; \ -typedef FFX_16BIT_F4 Prefix##_F4; \ -typedef FFX_16BIT_I Prefix##_I; \ -typedef FFX_16BIT_I2 Prefix##_I2; \ -typedef FFX_16BIT_I3 Prefix##_I3; \ -typedef FFX_16BIT_I4 Prefix##_I4; \ -typedef FFX_16BIT_U Prefix##_U; \ -typedef FFX_16BIT_U2 Prefix##_U2; \ -typedef FFX_16BIT_U3 Prefix##_U3; \ -typedef FFX_16BIT_U4 Prefix##_U4; - -#define TYPEDEF_FULL_PRECISION_TYPES(Prefix) \ -typedef FfxFloat32 Prefix##_F; \ -typedef FfxFloat32x2 Prefix##_F2; \ -typedef FfxFloat32x3 Prefix##_F3; \ -typedef FfxFloat32x4 Prefix##_F4; \ -typedef FfxInt32 Prefix##_I; \ -typedef FfxInt32x2 Prefix##_I2; \ -typedef FfxInt32x3 Prefix##_I3; \ -typedef FfxInt32x4 Prefix##_I4; \ -typedef FfxUInt32 Prefix##_U; \ -typedef FfxUInt32x2 Prefix##_U2; \ -typedef FfxUInt32x3 Prefix##_U3; \ -typedef FfxUInt32x4 Prefix##_U4; -#endif // #if defined(FFX_HLSL) - -#if defined(SHADER_API_PSSL) - -#define unorm -#define globallycoherent - -#if FFX_HALF - -#define FFX_MIN16_F half -#define FFX_MIN16_F2 half2 -#define FFX_MIN16_F3 half3 -#define FFX_MIN16_F4 half4 - -#define FFX_MIN16_I short -#define FFX_MIN16_I2 short2 -#define FFX_MIN16_I3 short3 -#define FFX_MIN16_I4 short4 - -#define FFX_MIN16_U ushort -#define FFX_MIN16_U2 ushort2 -#define FFX_MIN16_U3 ushort3 -#define FFX_MIN16_U4 ushort4 - -#define FFX_16BIT_F half -#define FFX_16BIT_F2 half2 -#define FFX_16BIT_F3 half3 -#define FFX_16BIT_F4 half4 - -#define FFX_16BIT_I short -#define FFX_16BIT_I2 short2 -#define FFX_16BIT_I3 short3 -#define FFX_16BIT_I4 short4 - -#define FFX_16BIT_U ushort -#define FFX_16BIT_U2 ushort2 -#define FFX_16BIT_U3 ushort3 -#define FFX_16BIT_U4 ushort4 - -#else // FFX_HALF - -#define FFX_MIN16_F float -#define FFX_MIN16_F2 float2 -#define FFX_MIN16_F3 float3 -#define FFX_MIN16_F4 float4 - -#define FFX_MIN16_I int -#define FFX_MIN16_I2 int2 -#define FFX_MIN16_I3 int3 -#define FFX_MIN16_I4 int4 - -#define FFX_MIN16_U uint -#define FFX_MIN16_U2 uint2 -#define FFX_MIN16_U3 uint3 -#define FFX_MIN16_U4 uint4 - -#define FFX_16BIT_F float -#define FFX_16BIT_F2 float2 -#define FFX_16BIT_F3 float3 -#define FFX_16BIT_F4 float4 - -#define FFX_16BIT_I int -#define FFX_16BIT_I2 int2 -#define FFX_16BIT_I3 int3 -#define FFX_16BIT_I4 int4 - -#define FFX_16BIT_U uint -#define FFX_16BIT_U2 uint2 -#define FFX_16BIT_U3 uint3 -#define FFX_16BIT_U4 uint4 - -#endif // FFX_HALF - -#endif // #if defined(SHADER_API_PSSL) - -#if defined(FFX_GLSL) - -#if FFX_HALF - -#define FFX_MIN16_F float16_t -#define FFX_MIN16_F2 f16vec2 -#define FFX_MIN16_F3 f16vec3 -#define FFX_MIN16_F4 f16vec4 - -#define FFX_MIN16_I int16_t -#define FFX_MIN16_I2 i16vec2 -#define FFX_MIN16_I3 i16vec3 -#define FFX_MIN16_I4 i16vec4 - -#define FFX_MIN16_U uint16_t -#define FFX_MIN16_U2 u16vec2 -#define FFX_MIN16_U3 u16vec3 -#define FFX_MIN16_U4 u16vec4 - -#define FFX_16BIT_F float16_t -#define FFX_16BIT_F2 f16vec2 -#define FFX_16BIT_F3 f16vec3 -#define FFX_16BIT_F4 f16vec4 - -#define FFX_16BIT_I int16_t -#define FFX_16BIT_I2 i16vec2 -#define FFX_16BIT_I3 i16vec3 -#define FFX_16BIT_I4 i16vec4 - -#define FFX_16BIT_U uint16_t -#define FFX_16BIT_U2 u16vec2 -#define FFX_16BIT_U3 u16vec3 -#define FFX_16BIT_U4 u16vec4 - -#else // FFX_HALF - -#define FFX_MIN16_F float -#define FFX_MIN16_F2 vec2 -#define FFX_MIN16_F3 vec3 -#define FFX_MIN16_F4 vec4 - -#define FFX_MIN16_I int -#define FFX_MIN16_I2 ivec2 -#define FFX_MIN16_I3 ivec3 -#define FFX_MIN16_I4 ivec4 - -#define FFX_MIN16_U uint -#define FFX_MIN16_U2 uvec2 -#define FFX_MIN16_U3 uvec3 -#define FFX_MIN16_U4 uvec4 - -#define FFX_16BIT_F float -#define FFX_16BIT_F2 vec2 -#define FFX_16BIT_F3 vec3 -#define FFX_16BIT_F4 vec4 - -#define FFX_16BIT_I int -#define FFX_16BIT_I2 ivec2 -#define FFX_16BIT_I3 ivec3 -#define FFX_16BIT_I4 ivec4 - -#define FFX_16BIT_U uint -#define FFX_16BIT_U2 uvec2 -#define FFX_16BIT_U3 uvec3 -#define FFX_16BIT_U4 uvec4 - -#endif // FFX_HALF - -#endif // #if defined(FFX_GLSL) - -#endif // #if defined(FFX_GPU) -#endif // #ifndef FFX_COMMON_TYPES_H diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_common_types.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_common_types.h.meta deleted file mode 100644 index bbd7f6a..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_common_types.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 7974b728d5c1b6d4a8a8e3965d03f96d -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core.h b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core.h deleted file mode 100644 index 02f6b3f..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core.h +++ /dev/null @@ -1,80 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -/// @defgroup FfxGPU GPU -/// The FidelityFX SDK GPU References -/// -/// @ingroup ffxSDK - -/// @defgroup FfxHLSL HLSL References -/// FidelityFX SDK HLSL GPU References -/// -/// @ingroup FfxGPU - -/// @defgroup FfxGLSL GLSL References -/// FidelityFX SDK GLSL GPU References -/// -/// @ingroup FfxGPU - -/// @defgroup FfxGPUEffects FidelityFX GPU References -/// FidelityFX Effect GPU Reference Documentation -/// -/// @ingroup FfxGPU - -/// @defgroup GPUCore GPU Core -/// GPU defines and functions -/// -/// @ingroup FfxGPU - -#if !defined(FFX_CORE_H) -#define FFX_CORE_H - -#ifdef __hlsl_dx_compiler -#pragma dxc diagnostic push -#pragma dxc diagnostic ignored "-Wambig-lit-shift" -#endif //__hlsl_dx_compiler - -#include "ffx_common_types.h" - -#if defined(FFX_CPU) - #include "ffx_core_cpu.h" -#endif // #if defined(FFX_CPU) - -#if defined(FFX_GLSL) && defined(FFX_GPU) - #include "ffx_core_glsl.h" -#endif // #if defined(FFX_GLSL) && defined(FFX_GPU) - -#if defined(FFX_HLSL) && defined(FFX_GPU) - #include "ffx_core_hlsl.h" -#endif // #if defined(FFX_HLSL) && defined(FFX_GPU) - -#if defined(FFX_GPU) - #include "ffx_core_gpu_common.h" - #include "ffx_core_gpu_common_half.h" - #include "ffx_core_portability.h" -#endif // #if defined(FFX_GPU) - -#ifdef __hlsl_dx_compiler -#pragma dxc diagnostic pop -#endif //__hlsl_dx_compiler - -#endif // #if !defined(FFX_CORE_H) diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core.h.meta deleted file mode 100644 index 8533462..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: b91c5f52b89ff554dacb51045a802ed8 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_cpu.h b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_cpu.h deleted file mode 100644 index 865258d..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_cpu.h +++ /dev/null @@ -1,338 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -/// A define for a true value in a boolean expression. -/// -/// @ingroup CPUTypes -#define FFX_TRUE (1) - -/// A define for a false value in a boolean expression. -/// -/// @ingroup CPUTypes -#define FFX_FALSE (0) - -#if !defined(FFX_STATIC) -/// A define to abstract declaration of static variables and functions. -/// -/// @ingroup CPUTypes -#define FFX_STATIC static -#endif // #if !defined(FFX_STATIC) - -/// @defgroup CPUCore CPU Core -/// Core CPU-side defines and functions -/// -/// @ingroup ffxHost - -#ifdef __clang__ -#pragma clang diagnostic ignored "-Wunused-variable" -#endif - -/// Interpret the bit layout of an IEEE-754 floating point value as an unsigned integer. -/// -/// @param [in] x A 32bit floating value. -/// -/// @returns -/// An unsigned 32bit integer value containing the bit pattern of x. -/// -/// @ingroup CPUCore -FFX_STATIC FfxUInt32 ffxAsUInt32(FfxFloat32 x) -{ - union - { - FfxFloat32 f; - FfxUInt32 u; - } bits; - - bits.f = x; - return bits.u; -} - -FFX_STATIC FfxFloat32 ffxDot2(FfxFloat32x2 a, FfxFloat32x2 b) -{ - return a[0] * b[0] + a[1] * b[1]; -} - -FFX_STATIC FfxFloat32 ffxDot3(FfxFloat32x3 a, FfxFloat32x3 b) -{ - return a[0] * b[0] + a[1] * b[1] + a[2] * b[2]; -} - -FFX_STATIC FfxFloat32 ffxDot4(FfxFloat32x4 a, FfxFloat32x4 b) -{ - return a[0] * b[0] + a[1] * b[1] + a[2] * b[2] + a[3] * b[3]; -} - -/// Compute the linear interopation between two values. -/// -/// Implemented by calling the GLSL mix instrinsic function. Implements the -/// following math: -/// -/// (1 - t) * x + t * y -/// -/// @param [in] x The first value to lerp between. -/// @param [in] y The second value to lerp between. -/// @param [in] t The value to determine how much of x and how much of y. -/// -/// @returns -/// A linearly interpolated value between x and y according to t. -/// -/// @ingroup CPUCore -FFX_STATIC FfxFloat32 ffxLerp(FfxFloat32 x, FfxFloat32 y, FfxFloat32 t) -{ - return y * t + (-x * t + x); -} - -/// Compute the reciprocal of a value. -/// -/// @param [in] x The value to compute the reciprocal for. -/// -/// @returns -/// The reciprocal value of x. -/// -/// @ingroup CPUCore -FFX_STATIC FfxFloat32 ffxReciprocal(FfxFloat32 x) -{ - return 1.0f / x; -} - -/// Compute the square root of a value. -/// -/// @param [in] x The first value to compute the min of. -/// -/// @returns -/// The the square root of x. -/// -/// @ingroup CPUCore -FFX_STATIC FfxFloat32 ffxSqrt(FfxFloat32 x) -{ - return sqrt(x); -} - -FFX_STATIC FfxUInt32 AShrSU1(FfxUInt32 a, FfxUInt32 b) -{ - return FfxUInt32(FfxInt32(a) >> FfxInt32(b)); -} - -/// Compute the factional part of a decimal value. -/// -/// This function calculates x - floor(x). -/// -/// @param [in] x The value to compute the fractional part from. -/// -/// @returns -/// The fractional part of x. -/// -/// @ingroup CPUCore -FFX_STATIC FfxFloat32 ffxFract(FfxFloat32 x) -{ - return x - floor(x); -} - -/// Compute the reciprocal square root of a value. -/// -/// @param [in] x The value to compute the reciprocal for. -/// -/// @returns -/// The reciprocal square root value of x. -/// -/// @ingroup CPUCore -FFX_STATIC FfxFloat32 rsqrt(FfxFloat32 x) -{ - return ffxReciprocal(ffxSqrt(x)); -} - -FFX_STATIC FfxFloat32 ffxMin(FfxFloat32 x, FfxFloat32 y) -{ - return x < y ? x : y; -} - -FFX_STATIC FfxUInt32 ffxMin(FfxUInt32 x, FfxUInt32 y) -{ - return x < y ? x : y; -} - -FFX_STATIC FfxFloat32 ffxMax(FfxFloat32 x, FfxFloat32 y) -{ - return x > y ? x : y; -} - -FFX_STATIC FfxUInt32 ffxMax(FfxUInt32 x, FfxUInt32 y) -{ - return x > y ? x : y; -} - -/// Clamp a value to a [0..1] range. -/// -/// @param [in] x The value to clamp to [0..1] range. -/// -/// @returns -/// The clamped version of x. -/// -/// @ingroup CPUCore -FFX_STATIC FfxFloat32 ffxSaturate(FfxFloat32 x) -{ - return ffxMin(1.0f, ffxMax(0.0f, x)); -} - -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// - -FFX_STATIC void opAAddOneF3(FfxFloat32x3 d, FfxFloat32x3 a, FfxFloat32 b) -{ - d[0] = a[0] + b; - d[1] = a[1] + b; - d[2] = a[2] + b; - return; -} - -FFX_STATIC void opACpyF3(FfxFloat32x3 d, FfxFloat32x3 a) -{ - d[0] = a[0]; - d[1] = a[1]; - d[2] = a[2]; - return; -} - -FFX_STATIC void opAMulF3(FfxFloat32x3 d, FfxFloat32x3 a, FfxFloat32x3 b) -{ - d[0] = a[0] * b[0]; - d[1] = a[1] * b[1]; - d[2] = a[2] * b[2]; - return; -} - -FFX_STATIC void opAMulOneF3(FfxFloat32x3 d, FfxFloat32x3 a, FfxFloat32 b) -{ - d[0] = a[0] * b; - d[1] = a[1] * b; - d[2] = a[2] * b; - return; -} - -FFX_STATIC void opARcpF3(FfxFloat32x3 d, FfxFloat32x3 a) -{ - d[0] = ffxReciprocal(a[0]); - d[1] = ffxReciprocal(a[1]); - d[2] = ffxReciprocal(a[2]); - return; -} - -/// Convert FfxFloat32 to half (in lower 16-bits of output). -/// -/// This function implements the same fast technique that is documented here: ftp://ftp.fox-toolkit.org/pub/fasthalffloatconversion.pdf -/// -/// The function supports denormals. -/// -/// Some conversion rules are to make computations possibly "safer" on the GPU, -/// -INF & -NaN -> -65504 -/// +INF & +NaN -> +65504 -/// -/// @param [in] f The 32bit floating point value to convert. -/// -/// @returns -/// The closest 16bit floating point value to f. -/// -/// @ingroup CPUCore -FFX_STATIC FfxUInt32 f32tof16(FfxFloat32 f) -{ - static FfxUInt16 base[512] = { - 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, - 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, - 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, - 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, - 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, - 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0001, 0x0002, 0x0004, 0x0008, 0x0010, 0x0020, 0x0040, 0x0080, 0x0100, 0x0200, 0x0400, - 0x0800, 0x0c00, 0x1000, 0x1400, 0x1800, 0x1c00, 0x2000, 0x2400, 0x2800, 0x2c00, 0x3000, 0x3400, 0x3800, 0x3c00, 0x4000, 0x4400, 0x4800, 0x4c00, 0x5000, - 0x5400, 0x5800, 0x5c00, 0x6000, 0x6400, 0x6800, 0x6c00, 0x7000, 0x7400, 0x7800, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, - 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, - 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, - 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, - 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, - 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, - 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, - 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, - 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, - 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, - 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, - 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8001, 0x8002, - 0x8004, 0x8008, 0x8010, 0x8020, 0x8040, 0x8080, 0x8100, 0x8200, 0x8400, 0x8800, 0x8c00, 0x9000, 0x9400, 0x9800, 0x9c00, 0xa000, 0xa400, 0xa800, 0xac00, - 0xb000, 0xb400, 0xb800, 0xbc00, 0xc000, 0xc400, 0xc800, 0xcc00, 0xd000, 0xd400, 0xd800, 0xdc00, 0xe000, 0xe400, 0xe800, 0xec00, 0xf000, 0xf400, 0xf800, - 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, - 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, - 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, - 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, - 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, - 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff - }; - - static FfxUInt8 shift[512] = { - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x17, 0x16, 0x15, 0x14, 0x13, 0x12, 0x11, 0x10, 0x0f, 0x0e, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, - 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x17, 0x16, 0x15, 0x14, 0x13, 0x12, 0x11, 0x10, 0x0f, 0x0e, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, - 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18 - }; - - union - { - FfxFloat32 f; - FfxUInt32 u; - } bits; - - bits.f = f; - FfxUInt32 u = bits.u; - FfxUInt32 i = u >> 23; - return (FfxUInt32)(base[i]) + ((u & 0x7fffff) >> shift[i]); -} - -/// Pack 2x32-bit floating point values in a single 32bit value. -/// -/// This function first converts each component of value into their nearest 16-bit floating -/// point representation, and then stores the X and Y components in the lower and upper 16 bits of the -/// 32bit unsigned integer respectively. -/// -/// @param [in] x A 2-dimensional floating point value to convert and pack. -/// -/// @returns -/// A packed 32bit value containing 2 16bit floating point values. -/// -/// @ingroup CPUCore -FFX_STATIC FfxUInt32 packHalf2x16(FfxFloat32x2 x) -{ - return f32tof16(x[0]) + (f32tof16(x[1]) << 16); -} diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_cpu.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_cpu.h.meta deleted file mode 100644 index f6508bc..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_cpu.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 4c88c0b7a4dec1e479272449c19ca981 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common.h b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common.h deleted file mode 100644 index 2f687df..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common.h +++ /dev/null @@ -1,2784 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -/// A define for a true value in a boolean expression. -/// -/// @ingroup GPUCore -#define FFX_TRUE (true) - -/// A define for a false value in a boolean expression. -/// -/// @ingroup GPUCore -#define FFX_FALSE (false) - -/// A define value for positive infinity. -/// -/// @ingroup GPUCore -#define FFX_POSITIVE_INFINITY_FLOAT ffxAsFloat(0x7f800000u) - -/// A define value for negative infinity. -/// -/// @ingroup GPUCore -#define FFX_NEGATIVE_INFINITY_FLOAT ffxAsFloat(0xff800000u) - -/// A define value for PI. -/// -/// @ingroup GPUCore -#define FFX_PI (3.14159) - -FFX_STATIC const FfxFloat32 FFX_FP16_MIN = 6.10e-05f; -FFX_STATIC const FfxFloat32 FFX_FP16_MAX = 65504.0f; -FFX_STATIC const FfxFloat32 FFX_TONEMAP_EPSILON = 1.0f / FFX_FP16_MAX; - -/// Compute the reciprocal of value. -/// -/// @param [in] value The value to compute the reciprocal of. -/// -/// @returns -/// The 1 / value. -/// -/// @ingroup GPUCore -FfxFloat32 ffxReciprocal(FfxFloat32 value) -{ - return rcp(value); -} - -/// Compute the reciprocal of value. -/// -/// @param [in] value The value to compute the reciprocal of. -/// -/// @returns -/// The 1 / value. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxReciprocal(FfxFloat32x2 value) -{ - return rcp(value); -} - -/// Compute the reciprocal of value. -/// -/// @param [in] value The value to compute the reciprocal of. -/// -/// @returns -/// The 1 / value. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxReciprocal(FfxFloat32x3 value) -{ - return rcp(value); -} - -/// Compute the reciprocal of value. -/// -/// @param [in] value The value to compute the reciprocal of. -/// -/// @returns -/// The 1 / value. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxReciprocal(FfxFloat32x4 value) -{ - return rcp(value); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat32 ffxMin(FfxFloat32 x, FfxFloat32 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxMin(FfxFloat32x2 x, FfxFloat32x2 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxMin(FfxFloat32x3 x, FfxFloat32x3 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxMin(FfxFloat32x4 x, FfxFloat32x4 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt32 ffxMin(FfxInt32 x, FfxInt32 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt32x2 ffxMin(FfxInt32x2 x, FfxInt32x2 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt32x3 ffxMin(FfxInt32x3 x, FfxInt32x3 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt32x4 ffxMin(FfxInt32x4 x, FfxInt32x4 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt32 ffxMin(FfxUInt32 x, FfxUInt32 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxMin(FfxUInt32x2 x, FfxUInt32x2 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt32x3 ffxMin(FfxUInt32x3 x, FfxUInt32x3 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt32x4 ffxMin(FfxUInt32x4 x, FfxUInt32x4 y) -{ - return min(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat32 ffxMax(FfxFloat32 x, FfxFloat32 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxMax(FfxFloat32x2 x, FfxFloat32x2 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxMax(FfxFloat32x3 x, FfxFloat32x3 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxMax(FfxFloat32x4 x, FfxFloat32x4 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt32 ffxMax(FfxInt32 x, FfxInt32 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt32x2 ffxMax(FfxInt32x2 x, FfxInt32x2 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt32x3 ffxMax(FfxInt32x3 x, FfxInt32x3 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt32x4 ffxMax(FfxInt32x4 x, FfxInt32x4 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt32 ffxMax(FfxUInt32 x, FfxUInt32 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxMax(FfxUInt32x2 x, FfxUInt32x2 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt32x3 ffxMax(FfxUInt32x3 x, FfxUInt32x3 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt32x4 ffxMax(FfxUInt32x4 x, FfxUInt32x4 y) -{ - return max(x, y); -} - -/// Compute the value of the first parameter raised to the power of the second. -/// -/// @param [in] x The value to raise to the power y. -/// @param [in] y The power to which to raise x. -/// -/// @returns -/// The value of the first parameter raised to the power of the second. -/// -/// @ingroup GPUCore -FfxFloat32 ffxPow(FfxFloat32 x, FfxFloat32 y) -{ - return pow(x, y); -} - -/// Compute the value of the first parameter raised to the power of the second. -/// -/// @param [in] x The value to raise to the power y. -/// @param [in] y The power to which to raise x. -/// -/// @returns -/// The value of the first parameter raised to the power of the second. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxPow(FfxFloat32x2 x, FfxFloat32x2 y) -{ - return pow(x, y); -} - -/// Compute the value of the first parameter raised to the power of the second. -/// -/// @param [in] x The value to raise to the power y. -/// @param [in] y The power to which to raise x. -/// -/// @returns -/// The value of the first parameter raised to the power of the second. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxPow(FfxFloat32x3 x, FfxFloat32x3 y) -{ - return pow(x, y); -} - -/// Compute the value of the first parameter raised to the power of the second. -/// -/// @param [in] x The value to raise to the power y. -/// @param [in] y The power to which to raise x. -/// -/// @returns -/// The value of the first parameter raised to the power of the second. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxPow(FfxFloat32x4 x, FfxFloat32x4 y) -{ - return pow(x, y); -} - -/// Compute the square root of a value. -/// -/// @param [in] x The first value to compute the min of. -/// -/// @returns -/// The the square root of x. -/// -/// @ingroup GPUCore -FfxFloat32 ffxSqrt(FfxFloat32 x) -{ - return sqrt(x); -} - -/// Compute the square root of a value. -/// -/// @param [in] x The first value to compute the min of. -/// -/// @returns -/// The the square root of x. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxSqrt(FfxFloat32x2 x) -{ - return sqrt(x); -} - -/// Compute the square root of a value. -/// -/// @param [in] x The first value to compute the min of. -/// -/// @returns -/// The the square root of x. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxSqrt(FfxFloat32x3 x) -{ - return sqrt(x); -} - -/// Compute the square root of a value. -/// -/// @param [in] x The first value to compute the min of. -/// -/// @returns -/// The the square root of x. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxSqrt(FfxFloat32x4 x) -{ - return sqrt(x); -} - -/// Copy the sign bit from 's' to positive 'd'. -/// -/// @param [in] d The value to copy the sign bit into. -/// @param [in] s The value to copy the sign bit from. -/// -/// @returns -/// The value of d with the sign bit from s. -/// -/// @ingroup GPUCore -FfxFloat32 ffxCopySignBit(FfxFloat32 d, FfxFloat32 s) -{ - return ffxAsFloat(ffxAsUInt32(d) | (ffxAsUInt32(s) & FfxUInt32(0x80000000u))); -} - -/// Copy the sign bit from 's' to positive 'd'. -/// -/// @param [in] d The value to copy the sign bit into. -/// @param [in] s The value to copy the sign bit from. -/// -/// @returns -/// The value of d with the sign bit from s. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxCopySignBit(FfxFloat32x2 d, FfxFloat32x2 s) -{ - return ffxAsFloat(ffxAsUInt32(d) | (ffxAsUInt32(s) & ffxBroadcast2(0x80000000u))); -} - -/// Copy the sign bit from 's' to positive 'd'. -/// -/// @param [in] d The value to copy the sign bit into. -/// @param [in] s The value to copy the sign bit from. -/// -/// @returns -/// The value of d with the sign bit from s. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxCopySignBit(FfxFloat32x3 d, FfxFloat32x3 s) -{ - return ffxAsFloat(ffxAsUInt32(d) | (ffxAsUInt32(s) & ffxBroadcast3(0x80000000u))); -} - -/// Copy the sign bit from 's' to positive 'd'. -/// -/// @param [in] d The value to copy the sign bit into. -/// @param [in] s The value to copy the sign bit from. -/// -/// @returns -/// The value of d with the sign bit from s. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxCopySignBit(FfxFloat32x4 d, FfxFloat32x4 s) -{ - return ffxAsFloat(ffxAsUInt32(d) | (ffxAsUInt32(s) & ffxBroadcast4(0x80000000u))); -} - -/// A single operation to return the following: -/// m = NaN := 0 -/// m >= 0 := 0 -/// m < 0 := 1 -/// -/// Uses the following useful floating point logic, -/// saturate(+a*(-INF)==-INF) := 0 -/// saturate( 0*(-INF)== NaN) := 0 -/// saturate(-a*(-INF)==+INF) := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against 0. -/// -/// @returns -/// 1.0 when the value is negative, or 0.0 when the value is 0 or position. -/// -/// @ingroup GPUCore -FfxFloat32 ffxIsSigned(FfxFloat32 m) -{ - return ffxSaturate(m * FfxFloat32(FFX_NEGATIVE_INFINITY_FLOAT)); -} - -/// A single operation to return the following: -/// m = NaN := 0 -/// m >= 0 := 0 -/// m < 0 := 1 -/// -/// Uses the following useful floating point logic, -/// saturate(+a*(-INF)==-INF) := 0 -/// saturate( 0*(-INF)== NaN) := 0 -/// saturate(-a*(-INF)==+INF) := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against 0. -/// -/// @returns -/// 1.0 when the value is negative, or 0.0 when the value is 0 or position. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxIsSigned(FfxFloat32x2 m) -{ - return ffxSaturate(m * ffxBroadcast2(FFX_NEGATIVE_INFINITY_FLOAT)); -} - -/// A single operation to return the following: -/// m = NaN := 0 -/// m >= 0 := 0 -/// m < 0 := 1 -/// -/// Uses the following useful floating point logic, -/// saturate(+a*(-INF)==-INF) := 0 -/// saturate( 0*(-INF)== NaN) := 0 -/// saturate(-a*(-INF)==+INF) := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against 0. -/// -/// @returns -/// 1.0 when the value is negative, or 0.0 when the value is 0 or position. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxIsSigned(FfxFloat32x3 m) -{ - return ffxSaturate(m * ffxBroadcast3(FFX_NEGATIVE_INFINITY_FLOAT)); -} - -/// A single operation to return the following: -/// m = NaN := 0 -/// m >= 0 := 0 -/// m < 0 := 1 -/// -/// Uses the following useful floating point logic, -/// saturate(+a*(-INF)==-INF) := 0 -/// saturate( 0*(-INF)== NaN) := 0 -/// saturate(-a*(-INF)==+INF) := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against for have the sign set. -/// -/// @returns -/// 1.0 when the value is negative, or 0.0 when the value is 0 or positive. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxIsSigned(FfxFloat32x4 m) -{ - return ffxSaturate(m * ffxBroadcast4(FFX_NEGATIVE_INFINITY_FLOAT)); -} - -/// A single operation to return the following: -/// m = NaN := 1 -/// m > 0 := 0 -/// m <= 0 := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against zero. -/// -/// @returns -/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. -/// -/// @ingroup GPUCore -FfxFloat32 ffxIsGreaterThanZero(FfxFloat32 m) -{ - return ffxSaturate(m * FfxFloat32(FFX_POSITIVE_INFINITY_FLOAT)); -} - -/// A single operation to return the following: -/// m = NaN := 1 -/// m > 0 := 0 -/// m <= 0 := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against zero. -/// -/// @returns -/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxIsGreaterThanZero(FfxFloat32x2 m) -{ - return ffxSaturate(m * ffxBroadcast2(FFX_POSITIVE_INFINITY_FLOAT)); -} - -/// A single operation to return the following: -/// m = NaN := 1 -/// m > 0 := 0 -/// m <= 0 := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against zero. -/// -/// @returns -/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxIsGreaterThanZero(FfxFloat32x3 m) -{ - return ffxSaturate(m * ffxBroadcast3(FFX_POSITIVE_INFINITY_FLOAT)); -} - -/// A single operation to return the following: -/// m = NaN := 1 -/// m > 0 := 0 -/// m <= 0 := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against zero. -/// -/// @returns -/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxIsGreaterThanZero(FfxFloat32x4 m) -{ - return ffxSaturate(m * ffxBroadcast4(FFX_POSITIVE_INFINITY_FLOAT)); -} - -/// Convert a 32bit floating point value to sortable integer. -/// -/// - If sign bit=0, flip the sign bit (positives). -/// - If sign bit=1, flip all bits (negatives). -/// -/// The function has the side effects that: -/// - Larger integers are more positive values. -/// - Float zero is mapped to center of integers (so clear to integer zero is a nice default for atomic max usage). -/// -/// @param [in] value The floating point value to make sortable. -/// -/// @returns -/// The sortable integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxFloatToSortableInteger(FfxUInt32 value) -{ - return value ^ ((AShrSU1(value, FfxUInt32(31))) | FfxUInt32(0x80000000)); -} - -/// Convert a sortable integer to a 32bit floating point value. -/// -/// The function has the side effects that: -/// - If sign bit=1, flip the sign bit (positives). -/// - If sign bit=0, flip all bits (negatives). -/// -/// @param [in] value The floating point value to make sortable. -/// -/// @returns -/// The sortable integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxSortableIntegerToFloat(FfxUInt32 value) -{ - return value ^ ((~AShrSU1(value, FfxUInt32(31))) | FfxUInt32(0x80000000)); -} - -/// Calculate a low-quality approximation for the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the square root for. -/// -/// @returns -/// An approximation of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximateSqrt(FfxFloat32 value) -{ - return ffxAsFloat((ffxAsUInt32(value) >> FfxUInt32(1)) + FfxUInt32(0x1fbc4639)); -} - -/// Calculate a low-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximateReciprocal(FfxFloat32 value) -{ - return ffxAsFloat(FfxUInt32(0x7ef07ebb) - ffxAsUInt32(value)); -} - -/// Calculate a medium-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to medium quality. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximateReciprocalMedium(FfxFloat32 value) -{ - FfxFloat32 b = ffxAsFloat(FfxUInt32(0x7ef19fff) - ffxAsUInt32(value)); - return b * (-b * value + FfxFloat32(2.0)); -} - -/// Calculate a low-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the reciprocal square root for. -/// -/// @returns -/// An approximation of the reciprocal square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximateReciprocalSquareRoot(FfxFloat32 value) -{ - return ffxAsFloat(FfxUInt32(0x5f347d74) - (ffxAsUInt32(value) >> FfxUInt32(1))); -} - -/// Calculate a low-quality approximation for the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the square root for. -/// -/// @returns -/// An approximation of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximateSqrt(FfxFloat32x2 value) -{ - return ffxAsFloat((ffxAsUInt32(value) >> ffxBroadcast2(1u)) + ffxBroadcast2(0x1fbc4639u)); -} - -/// Calculate a low-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximateReciprocal(FfxFloat32x2 value) -{ - return ffxAsFloat(ffxBroadcast2(0x7ef07ebbu) - ffxAsUInt32(value)); -} - -/// Calculate a medium-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to medium quality. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximateReciprocalMedium(FfxFloat32x2 value) -{ - FfxFloat32x2 b = ffxAsFloat(ffxBroadcast2(0x7ef19fffu) - ffxAsUInt32(value)); - return b * (-b * value + ffxBroadcast2(2.0f)); -} - -/// Calculate a low-quality approximation for the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the square root for. -/// -/// @returns -/// An approximation of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximateReciprocalSquareRoot(FfxFloat32x2 value) -{ - return ffxAsFloat(ffxBroadcast2(0x5f347d74u) - (ffxAsUInt32(value) >> ffxBroadcast2(1u))); -} - -/// Calculate a low-quality approximation for the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the square root for. -/// -/// @returns -/// An approximation of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximateSqrt(FfxFloat32x3 value) -{ - return ffxAsFloat((ffxAsUInt32(value) >> ffxBroadcast3(1u)) + ffxBroadcast3(0x1fbc4639u)); -} - -/// Calculate a low-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximateReciprocal(FfxFloat32x3 value) -{ - return ffxAsFloat(ffxBroadcast3(0x7ef07ebbu) - ffxAsUInt32(value)); -} - -/// Calculate a medium-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to medium quality. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximateReciprocalMedium(FfxFloat32x3 value) -{ - FfxFloat32x3 b = ffxAsFloat(ffxBroadcast3(0x7ef19fffu) - ffxAsUInt32(value)); - return b * (-b * value + ffxBroadcast3(2.0f)); -} - -/// Calculate a low-quality approximation for the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the square root for. -/// -/// @returns -/// An approximation of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximateReciprocalSquareRoot(FfxFloat32x3 value) -{ - return ffxAsFloat(ffxBroadcast3(0x5f347d74u) - (ffxAsUInt32(value) >> ffxBroadcast3(1u))); -} - -/// Calculate a low-quality approximation for the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the square root for. -/// -/// @returns -/// An approximation of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximateSqrt(FfxFloat32x4 value) -{ - return ffxAsFloat((ffxAsUInt32(value) >> ffxBroadcast4(1u)) + ffxBroadcast4(0x1fbc4639u)); -} - -/// Calculate a low-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximateReciprocal(FfxFloat32x4 value) -{ - return ffxAsFloat(ffxBroadcast4(0x7ef07ebbu) - ffxAsUInt32(value)); -} - -/// Calculate a medium-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to medium quality. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximateReciprocalMedium(FfxFloat32x4 value) -{ - FfxFloat32x4 b = ffxAsFloat(ffxBroadcast4(0x7ef19fffu) - ffxAsUInt32(value)); - return b * (-b * value + ffxBroadcast4(2.0f)); -} - -/// Calculate a low-quality approximation for the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the square root for. -/// -/// @returns -/// An approximation of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximateReciprocalSquareRoot(FfxFloat32x4 value) -{ - return ffxAsFloat(ffxBroadcast4(0x5f347d74u) - (ffxAsUInt32(value) >> ffxBroadcast4(1u))); -} - -/// Calculate dot product of 'a' and 'b'. -/// -/// @param [in] a First vector input. -/// @param [in] b Second vector input. -/// -/// @returns -/// The value of a dot b. -/// -/// @ingroup GPUCore -FfxFloat32 ffxDot2(FfxFloat32x2 a, FfxFloat32x2 b) -{ - return dot(a, b); -} - -/// Calculate dot product of 'a' and 'b'. -/// -/// @param [in] a First vector input. -/// @param [in] b Second vector input. -/// -/// @returns -/// The value of a dot b. -/// -/// @ingroup GPUCore -FfxFloat32 ffxDot3(FfxFloat32x3 a, FfxFloat32x3 b) -{ - return dot(a, b); -} - -/// Calculate dot product of 'a' and 'b'. -/// -/// @param [in] a First vector input. -/// @param [in] b Second vector input. -/// -/// @returns -/// The value of a dot b. -/// -/// @ingroup GPUCore -FfxFloat32 ffxDot4(FfxFloat32x4 a, FfxFloat32x4 b) -{ - return dot(a, b); -} - - -/// Compute an approximate conversion from PQ to Gamma2 space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between PQ and Gamma2. -/// -/// @returns -/// The value a converted into Gamma2. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximatePQToGamma2Medium(FfxFloat32 a) -{ - return a * a * a * a; -} - -/// Compute an approximate conversion from PQ to linear space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between PQ and linear. -/// -/// @returns -/// The value a converted into linear. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximatePQToLinear(FfxFloat32 a) -{ - return a * a * a * a * a * a * a * a; -} - -/// Compute an approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximateGamma2ToPQ(FfxFloat32 a) -{ - return ffxAsFloat((ffxAsUInt32(a) >> FfxUInt32(2)) + FfxUInt32(0x2F9A4E46)); -} - -/// Compute a more accurate approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximateGamma2ToPQMedium(FfxFloat32 a) -{ - FfxFloat32 b = ffxAsFloat((ffxAsUInt32(a) >> FfxUInt32(2)) + FfxUInt32(0x2F9A4E46)); - FfxFloat32 b4 = b * b * b * b; - return b - b * (b4 - a) / (FfxFloat32(4.0) * b4); -} - -/// Compute a high accuracy approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximateGamma2ToPQHigh(FfxFloat32 a) -{ - return ffxSqrt(ffxSqrt(a)); -} - -/// Compute an approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximateLinearToPQ(FfxFloat32 a) -{ - return ffxAsFloat((ffxAsUInt32(a) >> FfxUInt32(3)) + FfxUInt32(0x378D8723)); -} - -/// Compute a more accurate approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximateLinearToPQMedium(FfxFloat32 a) -{ - FfxFloat32 b = ffxAsFloat((ffxAsUInt32(a) >> FfxUInt32(3)) + FfxUInt32(0x378D8723)); - FfxFloat32 b8 = b * b * b * b * b * b * b * b; - return b - b * (b8 - a) / (FfxFloat32(8.0) * b8); -} - -/// Compute a very accurate approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximateLinearToPQHigh(FfxFloat32 a) -{ - return ffxSqrt(ffxSqrt(ffxSqrt(a))); -} - -/// Compute an approximate conversion from PQ to Gamma2 space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between PQ and Gamma2. -/// -/// @returns -/// The value a converted into Gamma2. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximatePQToGamma2Medium(FfxFloat32x2 a) -{ - return a * a * a * a; -} - -/// Compute an approximate conversion from PQ to linear space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between PQ and linear. -/// -/// @returns -/// The value a converted into linear. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximatePQToLinear(FfxFloat32x2 a) -{ - return a * a * a * a * a * a * a * a; -} - -/// Compute an approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximateGamma2ToPQ(FfxFloat32x2 a) -{ - return ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast2(2u)) + ffxBroadcast2(0x2F9A4E46u)); -} - -/// Compute a more accurate approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximateGamma2ToPQMedium(FfxFloat32x2 a) -{ - FfxFloat32x2 b = ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast2(2u)) + ffxBroadcast2(0x2F9A4E46u)); - FfxFloat32x2 b4 = b * b * b * b; - return b - b * (b4 - a) / (FfxFloat32(4.0) * b4); -} - -/// Compute a high accuracy approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximateGamma2ToPQHigh(FfxFloat32x2 a) -{ - return ffxSqrt(ffxSqrt(a)); -} - -/// Compute an approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximateLinearToPQ(FfxFloat32x2 a) -{ - return ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast2(3u)) + ffxBroadcast2(0x378D8723u)); -} - -/// Compute a more accurate approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximateLinearToPQMedium(FfxFloat32x2 a) -{ - FfxFloat32x2 b = ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast2(3u)) + ffxBroadcast2(0x378D8723u)); - FfxFloat32x2 b8 = b * b * b * b * b * b * b * b; - return b - b * (b8 - a) / (FfxFloat32(8.0) * b8); -} - -/// Compute a very accurate approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximateLinearToPQHigh(FfxFloat32x2 a) -{ - return ffxSqrt(ffxSqrt(ffxSqrt(a))); -} - -/// Compute an approximate conversion from PQ to Gamma2 space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between PQ and Gamma2. -/// -/// @returns -/// The value a converted into Gamma2. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximatePQToGamma2Medium(FfxFloat32x3 a) -{ - return a * a * a * a; -} - -/// Compute an approximate conversion from PQ to linear space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between PQ and linear. -/// -/// @returns -/// The value a converted into linear. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximatePQToLinear(FfxFloat32x3 a) -{ - return a * a * a * a * a * a * a * a; -} - -/// Compute an approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximateGamma2ToPQ(FfxFloat32x3 a) -{ - return ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast3(2u)) + ffxBroadcast3(0x2F9A4E46u)); -} - -/// Compute a more accurate approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximateGamma2ToPQMedium(FfxFloat32x3 a) -{ - FfxFloat32x3 b = ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast3(2u)) + ffxBroadcast3(0x2F9A4E46u)); - FfxFloat32x3 b4 = b * b * b * b; - return b - b * (b4 - a) / (FfxFloat32(4.0) * b4); -} - -/// Compute a high accuracy approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximateGamma2ToPQHigh(FfxFloat32x3 a) -{ - return ffxSqrt(ffxSqrt(a)); -} - -/// Compute an approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximateLinearToPQ(FfxFloat32x3 a) -{ - return ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast3(3u)) + ffxBroadcast3(0x378D8723u)); -} - -/// Compute a more accurate approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximateLinearToPQMedium(FfxFloat32x3 a) -{ - FfxFloat32x3 b = ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast3(3u)) + ffxBroadcast3(0x378D8723u)); - FfxFloat32x3 b8 = b * b * b * b * b * b * b * b; - return b - b * (b8 - a) / (FfxFloat32(8.0) * b8); -} - -/// Compute a very accurate approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximateLinearToPQHigh(FfxFloat32x3 a) -{ - return ffxSqrt(ffxSqrt(ffxSqrt(a))); -} - -/// Compute an approximate conversion from PQ to Gamma2 space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between PQ and Gamma2. -/// -/// @returns -/// The value a converted into Gamma2. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximatePQToGamma2Medium(FfxFloat32x4 a) -{ - return a * a * a * a; -} - -/// Compute an approximate conversion from PQ to linear space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between PQ and linear. -/// -/// @returns -/// The value a converted into linear. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximatePQToLinear(FfxFloat32x4 a) -{ - return a * a * a * a * a * a * a * a; -} - -/// Compute an approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximateGamma2ToPQ(FfxFloat32x4 a) -{ - return ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast4(2u)) + ffxBroadcast4(0x2F9A4E46u)); -} - -/// Compute a more accurate approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximateGamma2ToPQMedium(FfxFloat32x4 a) -{ - FfxFloat32x4 b = ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast4(2u)) + ffxBroadcast4(0x2F9A4E46u)); - FfxFloat32x4 b4 = b * b * b * b * b * b * b * b; - return b - b * (b4 - a) / (FfxFloat32(4.0) * b4); -} - -/// Compute a high accuracy approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximateGamma2ToPQHigh(FfxFloat32x4 a) -{ - return ffxSqrt(ffxSqrt(a)); -} - -/// Compute an approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximateLinearToPQ(FfxFloat32x4 a) -{ - return ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast4(3u)) + ffxBroadcast4(0x378D8723u)); -} - -/// Compute a more accurate approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximateLinearToPQMedium(FfxFloat32x4 a) -{ - FfxFloat32x4 b = ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast4(3u)) + ffxBroadcast4(0x378D8723u)); - FfxFloat32x4 b8 = b * b * b * b * b * b * b * b; - return b - b * (b8 - a) / (FfxFloat32(8.0) * b8); -} - -/// Compute a very accurate approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximateLinearToPQHigh(FfxFloat32x4 a) -{ - return ffxSqrt(ffxSqrt(ffxSqrt(a))); -} - -// An approximation of sine. -// -// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range -// is {-1/4 to 1/4} representing {-1 to 1}. -// -// @param [in] value The value to calculate approximate sine for. -// -// @returns -// The approximate sine of value. -FfxFloat32 ffxParabolicSin(FfxFloat32 value) -{ - return value * abs(value) - value; -} - -// An approximation of sine. -// -// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range -// is {-1/4 to 1/4} representing {-1 to 1}. -// -// @param [in] value The value to calculate approximate sine for. -// -// @returns -// The approximate sine of value. -FfxFloat32x2 ffxParabolicSin(FfxFloat32x2 x) -{ - return x * abs(x) - x; -} - -// An approximation of cosine. -// -// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range -// is {-1/4 to 1/4} representing {-1 to 1}. -// -// @param [in] value The value to calculate approximate cosine for. -// -// @returns -// The approximate cosine of value. -FfxFloat32 ffxParabolicCos(FfxFloat32 x) -{ - x = ffxFract(x * FfxFloat32(0.5) + FfxFloat32(0.75)); - x = x * FfxFloat32(2.0) - FfxFloat32(1.0); - return ffxParabolicSin(x); -} - -// An approximation of cosine. -// -// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range -// is {-1/4 to 1/4} representing {-1 to 1}. -// -// @param [in] value The value to calculate approximate cosine for. -// -// @returns -// The approximate cosine of value. -FfxFloat32x2 ffxParabolicCos(FfxFloat32x2 x) -{ - x = ffxFract(x * ffxBroadcast2(0.5f) + ffxBroadcast2(0.75f)); - x = x * ffxBroadcast2(2.0f) - ffxBroadcast2(1.0f); - return ffxParabolicSin(x); -} - -// An approximation of both sine and cosine. -// -// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range -// is {-1/4 to 1/4} representing {-1 to 1}. -// -// @param [in] value The value to calculate approximate cosine for. -// -// @returns -// A FfxFloat32x2 containing approximations of both sine and cosine of value. -FfxFloat32x2 ffxParabolicSinCos(FfxFloat32 x) -{ - FfxFloat32 y = ffxFract(x * FfxFloat32(0.5) + FfxFloat32(0.75)); - y = y * FfxFloat32(2.0) - FfxFloat32(1.0); - return ffxParabolicSin(FfxFloat32x2(x, y)); -} - -/// Conditional free logic AND operation using values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxUInt32 ffxZeroOneAnd(FfxUInt32 x, FfxUInt32 y) -{ - return min(x, y); -} - -/// Conditional free logic AND operation using two values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxZeroOneAnd(FfxUInt32x2 x, FfxUInt32x2 y) -{ - return min(x, y); -} - -/// Conditional free logic AND operation using two values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxUInt32x3 ffxZeroOneAnd(FfxUInt32x3 x, FfxUInt32x3 y) -{ - return min(x, y); -} - -/// Conditional free logic AND operation using two values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxUInt32x4 ffxZeroOneAnd(FfxUInt32x4 x, FfxUInt32x4 y) -{ - return min(x, y); -} - -/// Conditional free logic NOT operation using two values. -/// -/// @param [in] x The first value to be fed into the NOT operator. -/// -/// @returns -/// Result of the NOT operation. -/// -/// @ingroup GPUCore -FfxUInt32 ffxZeroOneAnd(FfxUInt32 x) -{ - return x ^ FfxUInt32(1); -} - -/// Conditional free logic NOT operation using two values. -/// -/// @param [in] x The first value to be fed into the NOT operator. -/// -/// @returns -/// Result of the NOT operation. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxZeroOneAnd(FfxUInt32x2 x) -{ - return x ^ ffxBroadcast2(1u); -} - -/// Conditional free logic NOT operation using two values. -/// -/// @param [in] x The first value to be fed into the NOT operator. -/// -/// @returns -/// Result of the NOT operation. -/// -/// @ingroup GPUCore -FfxUInt32x3 ffxZeroOneAnd(FfxUInt32x3 x) -{ - return x ^ ffxBroadcast3(1u); -} - -/// Conditional free logic NOT operation using two values. -/// -/// @param [in] x The first value to be fed into the NOT operator. -/// -/// @returns -/// Result of the NOT operation. -/// -/// @ingroup GPUCore -FfxUInt32x4 ffxZeroOneAnd(FfxUInt32x4 x) -{ - return x ^ ffxBroadcast4(1u); -} - -/// Conditional free logic OR operation using two values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxUInt32 ffxZeroOneOr(FfxUInt32 x, FfxUInt32 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxZeroOneOr(FfxUInt32x2 x, FfxUInt32x2 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxUInt32x3 ffxZeroOneOr(FfxUInt32x3 x, FfxUInt32x3 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxUInt32x4 ffxZeroOneOr(FfxUInt32x4 x, FfxUInt32x4 y) -{ - return max(x, y); -} - -/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxUInt32 ffxZeroOneAndToU1(FfxFloat32 x) -{ - return FfxUInt32(FfxFloat32(1.0) - x); -} - -/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxZeroOneAndToU2(FfxFloat32x2 x) -{ - return FfxUInt32x2(ffxBroadcast2(1.0) - x); -} - -/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxUInt32x3 ffxZeroOneAndToU3(FfxFloat32x3 x) -{ - return FfxUInt32x3(ffxBroadcast3(1.0) - x); -} - -/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxUInt32x4 ffxZeroOneAndToU4(FfxFloat32x4 x) -{ - return FfxUInt32x4(ffxBroadcast4(1.0) - x); -} - -/// Conditional free logic AND operation using two values followed by a NOT operation -/// using the resulting value and a third value. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// @param [in] z The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat32 ffxZeroOneAndOr(FfxFloat32 x, FfxFloat32 y, FfxFloat32 z) -{ - return ffxSaturate(x * y + z); -} - -/// Conditional free logic AND operation using two values followed by a NOT operation -/// using the resulting value and a third value. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// @param [in] z The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxZeroOneAndOr(FfxFloat32x2 x, FfxFloat32x2 y, FfxFloat32x2 z) -{ - return ffxSaturate(x * y + z); -} - -/// Conditional free logic AND operation using two values followed by a NOT operation -/// using the resulting value and a third value. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// @param [in] z The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxZeroOneAndOr(FfxFloat32x3 x, FfxFloat32x3 y, FfxFloat32x3 z) -{ - return ffxSaturate(x * y + z); -} - -/// Conditional free logic AND operation using two values followed by a NOT operation -/// using the resulting value and a third value. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// @param [in] z The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxZeroOneAndOr(FfxFloat32x4 x, FfxFloat32x4 y, FfxFloat32x4 z) -{ - return ffxSaturate(x * y + z); -} - -/// Given a value, returns 1.0 if greater than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the greater than zero comparison. -/// -/// @ingroup GPUCore -FfxFloat32 ffxZeroOneIsGreaterThanZero(FfxFloat32 x) -{ - return ffxSaturate(x * FfxFloat32(FFX_POSITIVE_INFINITY_FLOAT)); -} - -/// Given a value, returns 1.0 if greater than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the greater than zero comparison. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxZeroOneIsGreaterThanZero(FfxFloat32x2 x) -{ - return ffxSaturate(x * ffxBroadcast2(FFX_POSITIVE_INFINITY_FLOAT)); -} - -/// Given a value, returns 1.0 if greater than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the greater than zero comparison. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxZeroOneIsGreaterThanZero(FfxFloat32x3 x) -{ - return ffxSaturate(x * ffxBroadcast3(FFX_POSITIVE_INFINITY_FLOAT)); -} - -/// Given a value, returns 1.0 if greater than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the greater than zero comparison. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxZeroOneIsGreaterThanZero(FfxFloat32x4 x) -{ - return ffxSaturate(x * ffxBroadcast4(FFX_POSITIVE_INFINITY_FLOAT)); -} - -/// Conditional free logic signed NOT operation using two FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat32 ffxZeroOneAnd(FfxFloat32 x) -{ - return FfxFloat32(1.0) - x; -} - -/// Conditional free logic signed NOT operation using two FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxZeroOneAnd(FfxFloat32x2 x) -{ - return ffxBroadcast2(1.0) - x; -} - -/// Conditional free logic signed NOT operation using two FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxZeroOneAnd(FfxFloat32x3 x) -{ - return ffxBroadcast3(1.0) - x; -} - -/// Conditional free logic signed NOT operation using two FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxZeroOneAnd(FfxFloat32x4 x) -{ - return ffxBroadcast4(1.0) - x; -} - -/// Conditional free logic OR operation using two FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxFloat32 ffxZeroOneOr(FfxFloat32 x, FfxFloat32 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxZeroOneOr(FfxFloat32x2 x, FfxFloat32x2 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxZeroOneOr(FfxFloat32x3 x, FfxFloat32x3 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxZeroOneOr(FfxFloat32x4 x, FfxFloat32x4 y) -{ - return max(x, y); -} - -/// Choose between two FfxFloat32 values if the first paramter is greater than zero. -/// -/// @param [in] x The value to compare against zero. -/// @param [in] y The value to return if the comparision is greater than zero. -/// @param [in] z The value to return if the comparision is less than or equal to zero. -/// -/// @returns -/// The selected value. -/// -/// @ingroup GPUCore -FfxFloat32 ffxZeroOneSelect(FfxFloat32 x, FfxFloat32 y, FfxFloat32 z) -{ - FfxFloat32 r = (-x) * z + z; - return x * y + r; -} - -/// Choose between two FfxFloat32 values if the first paramter is greater than zero. -/// -/// @param [in] x The value to compare against zero. -/// @param [in] y The value to return if the comparision is greater than zero. -/// @param [in] z The value to return if the comparision is less than or equal to zero. -/// -/// @returns -/// The selected value. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxZeroOneSelect(FfxFloat32x2 x, FfxFloat32x2 y, FfxFloat32x2 z) -{ - FfxFloat32x2 r = (-x) * z + z; - return x * y + r; -} - -/// Choose between two FfxFloat32 values if the first paramter is greater than zero. -/// -/// @param [in] x The value to compare against zero. -/// @param [in] y The value to return if the comparision is greater than zero. -/// @param [in] z The value to return if the comparision is less than or equal to zero. -/// -/// @returns -/// The selected value. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxZeroOneSelect(FfxFloat32x3 x, FfxFloat32x3 y, FfxFloat32x3 z) -{ - FfxFloat32x3 r = (-x) * z + z; - return x * y + r; -} - -/// Choose between two FfxFloat32 values if the first paramter is greater than zero. -/// -/// @param [in] x The value to compare against zero. -/// @param [in] y The value to return if the comparision is greater than zero. -/// @param [in] z The value to return if the comparision is less than or equal to zero. -/// -/// @returns -/// The selected value. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxZeroOneSelect(FfxFloat32x4 x, FfxFloat32x4 y, FfxFloat32x4 z) -{ - FfxFloat32x4 r = (-x) * z + z; - return x * y + r; -} - -/// Given a value, returns 1.0 if less than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the sign value. -/// -/// @ingroup GPUCore -FfxFloat32 ffxZeroOneIsSigned(FfxFloat32 x) -{ - return ffxSaturate(x * FfxFloat32(FFX_NEGATIVE_INFINITY_FLOAT)); -} - -/// Given a value, returns 1.0 if less than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the sign value. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxZeroOneIsSigned(FfxFloat32x2 x) -{ - return ffxSaturate(x * ffxBroadcast2(FFX_NEGATIVE_INFINITY_FLOAT)); -} - -/// Given a value, returns 1.0 if less than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the sign value. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxZeroOneIsSigned(FfxFloat32x3 x) -{ - return ffxSaturate(x * ffxBroadcast3(FFX_NEGATIVE_INFINITY_FLOAT)); -} - -/// Given a value, returns 1.0 if less than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the sign value. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxZeroOneIsSigned(FfxFloat32x4 x) -{ - return ffxSaturate(x * ffxBroadcast4(FFX_NEGATIVE_INFINITY_FLOAT)); -} - -/// Compute a Rec.709 color space. -/// -/// Rec.709 is used for some HDTVs. -/// -/// Both Rec.709 and sRGB have a linear segment which as spec'ed would intersect the curved segment 2 times. -/// (a.) For 8-bit sRGB, steps {0 to 10.3} are in the linear region (4% of the encoding range). -/// (b.) For 8-bit 709, steps {0 to 20.7} are in the linear region (8% of the encoding range). -/// -/// @param [in] color The color to convert to Rec. 709. -/// -/// @returns -/// The color in linear space. -/// -/// @ingroup GPUCore -FfxFloat32 ffxRec709FromLinear(FfxFloat32 color) -{ - FfxFloat32x3 j = FfxFloat32x3(0.018 * 4.5, 4.5, 0.45); - FfxFloat32x2 k = FfxFloat32x2(1.099, -0.099); - return clamp(j.x, color * j.y, pow(color, j.z) * k.x + k.y); -} - -/// Compute a Rec.709 color space. -/// -/// Rec.709 is used for some HDTVs. -/// -/// Both Rec.709 and sRGB have a linear segment which as spec'ed would intersect the curved segment 2 times. -/// (a.) For 8-bit sRGB, steps {0 to 10.3} are in the linear region (4% of the encoding range). -/// (b.) For 8-bit 709, steps {0 to 20.7} are in the linear region (8% of the encoding range). -/// -/// @param [in] color The color to convert to Rec. 709. -/// -/// @returns -/// The color in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxRec709FromLinear(FfxFloat32x2 color) -{ - FfxFloat32x3 j = FfxFloat32x3(0.018 * 4.5, 4.5, 0.45); - FfxFloat32x2 k = FfxFloat32x2(1.099, -0.099); - return clamp(j.xx, color * j.yy, pow(color, j.zz) * k.xx + k.yy); -} - -/// Compute a Rec.709 color space. -/// -/// Rec.709 is used for some HDTVs. -/// -/// Both Rec.709 and sRGB have a linear segment which as spec'ed would intersect the curved segment 2 times. -/// (a.) For 8-bit sRGB, steps {0 to 10.3} are in the linear region (4% of the encoding range). -/// (b.) For 8-bit 709, steps {0 to 20.7} are in the linear region (8% of the encoding range). -/// -/// @param [in] color The color to convert to Rec. 709. -/// -/// @returns -/// The color in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxRec709FromLinear(FfxFloat32x3 color) -{ - FfxFloat32x3 j = FfxFloat32x3(0.018 * 4.5, 4.5, 0.45); - FfxFloat32x2 k = FfxFloat32x2(1.099, -0.099); - return clamp(j.xxx, color * j.yyy, pow(color, j.zzz) * k.xxx + k.yyy); -} - -/// Compute a gamma value from a linear value. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// Note: 'rcpX' is '1/x', where the 'x' is what would be used in ffxLinearFromGamma. -/// -/// @param [in] value The value to convert to gamma space from linear. -/// @param [in] power The reciprocal of power value used for the gamma curve. -/// -/// @returns -/// A value in gamma space. -/// -/// @ingroup GPUCore -FfxFloat32 ffxGammaFromLinear(FfxFloat32 value, FfxFloat32 power) -{ - return pow(value, FfxFloat32(power)); -} - -/// Compute a gamma value from a linear value. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// Note: 'rcpX' is '1/x', where the 'x' is what would be used in ffxLinearFromGamma. -/// -/// @param [in] value The value to convert to gamma space from linear. -/// @param [in] power The reciprocal of power value used for the gamma curve. -/// -/// @returns -/// A value in gamma space. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxGammaFromLinear(FfxFloat32x2 value, FfxFloat32 power) -{ - return pow(value, ffxBroadcast2(power)); -} - -/// Compute a gamma value from a linear value. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// Note: 'rcpX' is '1/x', where the 'x' is what would be used in ffxLinearFromGamma. -/// -/// @param [in] value The value to convert to gamma space from linear. -/// @param [in] power The reciprocal of power value used for the gamma curve. -/// -/// @returns -/// A value in gamma space. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxGammaFromLinear(FfxFloat32x3 value, FfxFloat32 power) -{ - return pow(value, ffxBroadcast3(power)); -} - -/// Compute a PQ value from a linear value. -/// -/// @param [in] value The value to convert to PQ from linear. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32 ffxPQToLinear(FfxFloat32 value) -{ - FfxFloat32 p = pow(value, FfxFloat32(0.159302)); - return pow((FfxFloat32(0.835938) + FfxFloat32(18.8516) * p) / (FfxFloat32(1.0) + FfxFloat32(18.6875) * p), FfxFloat32(78.8438)); -} - -/// Compute a PQ value from a linear value. -/// -/// @param [in] value The value to convert to PQ from linear. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxPQToLinear(FfxFloat32x2 value) -{ - FfxFloat32x2 p = pow(value, ffxBroadcast2(0.159302)); - return pow((ffxBroadcast2(0.835938) + ffxBroadcast2(18.8516) * p) / (ffxBroadcast2(1.0) + ffxBroadcast2(18.6875) * p), ffxBroadcast2(78.8438)); -} - -/// Compute a PQ value from a linear value. -/// -/// @param [in] value The value to convert to PQ from linear. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxPQToLinear(FfxFloat32x3 value) -{ - FfxFloat32x3 p = pow(value, ffxBroadcast3(0.159302)); - return pow((ffxBroadcast3(0.835938) + ffxBroadcast3(18.8516) * p) / (ffxBroadcast3(1.0) + ffxBroadcast3(18.6875) * p), ffxBroadcast3(78.8438)); -} - -/// Compute a linear value from a SRGB value. -/// -/// @param [in] value The value to convert to linear from SRGB. -/// -/// @returns -/// A value in SRGB space. -/// -/// @ingroup GPUCore -FfxFloat32 ffxSrgbToLinear(FfxFloat32 value) -{ - FfxFloat32x3 j = FfxFloat32x3(0.0031308 * 12.92, 12.92, 1.0 / 2.4); - FfxFloat32x2 k = FfxFloat32x2(1.055, -0.055); - return clamp(j.x, value * j.y, pow(value, j.z) * k.x + k.y); -} - -/// Compute a linear value from a SRGB value. -/// -/// @param [in] value The value to convert to linear from SRGB. -/// -/// @returns -/// A value in SRGB space. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxSrgbToLinear(FfxFloat32x2 value) -{ - FfxFloat32x3 j = FfxFloat32x3(0.0031308 * 12.92, 12.92, 1.0 / 2.4); - FfxFloat32x2 k = FfxFloat32x2(1.055, -0.055); - return clamp(j.xx, value * j.yy, pow(value, j.zz) * k.xx + k.yy); -} - -/// Compute a linear value from a SRGB value. -/// -/// @param [in] value The value to convert to linear from SRGB. -/// -/// @returns -/// A value in SRGB space. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxSrgbToLinear(FfxFloat32x3 value) -{ - FfxFloat32x3 j = FfxFloat32x3(0.0031308 * 12.92, 12.92, 1.0 / 2.4); - FfxFloat32x2 k = FfxFloat32x2(1.055, -0.055); - return clamp(j.xxx, value * j.yyy, pow(value, j.zzz) * k.xxx + k.yyy); -} - -/// Compute a linear value from a REC.709 value. -/// -/// @param [in] color The value to convert to linear from REC.709. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32 ffxLinearFromRec709(FfxFloat32 color) -{ - FfxFloat32x3 j = FfxFloat32x3(0.081 / 4.5, 1.0 / 4.5, 1.0 / 0.45); - FfxFloat32x2 k = FfxFloat32x2(1.0 / 1.099, 0.099 / 1.099); - return ffxZeroOneSelect(ffxZeroOneIsSigned(color - j.x), color * j.y, pow(color * k.x + k.y, j.z)); -} - -/// Compute a linear value from a REC.709 value. -/// -/// @param [in] color The value to convert to linear from REC.709. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxLinearFromRec709(FfxFloat32x2 color) -{ - FfxFloat32x3 j = FfxFloat32x3(0.081 / 4.5, 1.0 / 4.5, 1.0 / 0.45); - FfxFloat32x2 k = FfxFloat32x2(1.0 / 1.099, 0.099 / 1.099); - return ffxZeroOneSelect(ffxZeroOneIsSigned(color - j.xx), color * j.yy, pow(color * k.xx + k.yy, j.zz)); -} - -/// Compute a linear value from a REC.709 value. -/// -/// @param [in] color The value to convert to linear from REC.709. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxLinearFromRec709(FfxFloat32x3 color) -{ - FfxFloat32x3 j = FfxFloat32x3(0.081 / 4.5, 1.0 / 4.5, 1.0 / 0.45); - FfxFloat32x2 k = FfxFloat32x2(1.0 / 1.099, 0.099 / 1.099); - return ffxZeroOneSelect(ffxZeroOneIsSigned(color - j.xxx), color * j.yyy, pow(color * k.xxx + k.yyy, j.zzz)); -} - -/// Compute a linear value from a value in a gamma space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] color The value to convert to linear in gamma space. -/// @param [in] power The power value used for the gamma curve. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32 ffxLinearFromGamma(FfxFloat32 color, FfxFloat32 power) -{ - return pow(color, FfxFloat32(power)); -} - -/// Compute a linear value from a value in a gamma space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] color The value to convert to linear in gamma space. -/// @param [in] power The power value used for the gamma curve. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxLinearFromGamma(FfxFloat32x2 color, FfxFloat32 power) -{ - return pow(color, ffxBroadcast2(power)); -} - -/// Compute a linear value from a value in a gamma space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] color The value to convert to linear in gamma space. -/// @param [in] power The power value used for the gamma curve. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxLinearFromGamma(FfxFloat32x3 color, FfxFloat32 power) -{ - return pow(color, ffxBroadcast3(power)); -} - -/// Compute a linear value from a value in a PQ space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] value The value to convert to linear in PQ space. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32 ffxLinearFromPQ(FfxFloat32 value) -{ - FfxFloat32 p = pow(value, FfxFloat32(0.0126833)); - return pow(ffxSaturate(p - FfxFloat32(0.835938)) / (FfxFloat32(18.8516) - FfxFloat32(18.6875) * p), FfxFloat32(6.27739)); -} - -/// Compute a linear value from a value in a PQ space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] value The value to convert to linear in PQ space. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxLinearFromPQ(FfxFloat32x2 value) -{ - FfxFloat32x2 p = pow(value, ffxBroadcast2(0.0126833)); - return pow(ffxSaturate(p - ffxBroadcast2(0.835938)) / (ffxBroadcast2(18.8516) - ffxBroadcast2(18.6875) * p), ffxBroadcast2(6.27739)); -} - -/// Compute a linear value from a value in a PQ space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] value The value to convert to linear in PQ space. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxLinearFromPQ(FfxFloat32x3 value) -{ - FfxFloat32x3 p = pow(value, ffxBroadcast3(0.0126833)); - return pow(ffxSaturate(p - ffxBroadcast3(0.835938)) / (ffxBroadcast3(18.8516) - ffxBroadcast3(18.6875) * p), ffxBroadcast3(6.27739)); -} - -/// Compute a linear value from a value in a SRGB space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] value The value to convert to linear in SRGB space. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32 ffxLinearFromSrgb(FfxFloat32 value) -{ - FfxFloat32x3 j = FfxFloat32x3(0.04045 / 12.92, 1.0 / 12.92, 2.4); - FfxFloat32x2 k = FfxFloat32x2(1.0 / 1.055, 0.055 / 1.055); - return ffxZeroOneSelect(ffxZeroOneIsSigned(value - j.x), value * j.y, pow(value * k.x + k.y, j.z)); -} - -/// Compute a linear value from a value in a SRGB space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] value The value to convert to linear in SRGB space. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxLinearFromSrgb(FfxFloat32x2 value) -{ - FfxFloat32x3 j = FfxFloat32x3(0.04045 / 12.92, 1.0 / 12.92, 2.4); - FfxFloat32x2 k = FfxFloat32x2(1.0 / 1.055, 0.055 / 1.055); - return ffxZeroOneSelect(ffxZeroOneIsSigned(value - j.xx), value * j.yy, pow(value * k.xx + k.yy, j.zz)); -} - -/// Compute a linear value from a value in a SRGB space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] value The value to convert to linear in SRGB space. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxLinearFromSrgb(FfxFloat32x3 value) -{ - FfxFloat32x3 j = FfxFloat32x3(0.04045 / 12.92, 1.0 / 12.92, 2.4); - FfxFloat32x2 k = FfxFloat32x2(1.0 / 1.055, 0.055 / 1.055); - return ffxZeroOneSelect(ffxZeroOneIsSigned(value - j.xxx), value * j.yyy, pow(value * k.xxx + k.yyy, j.zzz)); -} - -/// A remapping of 64x1 to 8x8 imposing rotated 2x2 pixel quads in quad linear. -/// -/// 543210 -/// ====== -/// ..xxx. -/// yy...y -/// -/// @param [in] a The input 1D coordinates to remap. -/// -/// @returns -/// The remapped 2D coordinates. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxRemapForQuad(FfxUInt32 a) -{ - return FfxUInt32x2(bitfieldExtract(a, 1u, 3u), bitfieldInsertMask(bitfieldExtract(a, 3u, 3u), a, 1u)); -} - -/// A helper function performing a remap 64x1 to 8x8 remapping which is necessary for 2D wave reductions. -/// -/// The 64-wide lane indices to 8x8 remapping is performed as follows: -/// -/// 00 01 08 09 10 11 18 19 -/// 02 03 0a 0b 12 13 1a 1b -/// 04 05 0c 0d 14 15 1c 1d -/// 06 07 0e 0f 16 17 1e 1f -/// 20 21 28 29 30 31 38 39 -/// 22 23 2a 2b 32 33 3a 3b -/// 24 25 2c 2d 34 35 3c 3d -/// 26 27 2e 2f 36 37 3e 3f -/// -/// @param [in] a The input 1D coordinate to remap. -/// -/// @returns -/// The remapped 2D coordinates. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxRemapForWaveReduction(FfxUInt32 a) -{ - return FfxUInt32x2(bitfieldInsertMask(bitfieldExtract(a, 2u, 3u), a, 1u), bitfieldInsertMask(bitfieldExtract(a, 3u, 3u), bitfieldExtract(a, 1u, 2u), 2u)); -} diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common.h.meta deleted file mode 100644 index 14292a9..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 05b921699d1374a429e32afca13137e2 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common_half.h b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common_half.h deleted file mode 100644 index 4c73daf..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common_half.h +++ /dev/null @@ -1,2979 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#if FFX_HALF -#if FFX_HLSL_SM >= 62 -/// A define value for 16bit positive infinity. -/// -/// @ingroup GPUCore -#define FFX_POSITIVE_INFINITY_HALF FFX_TO_FLOAT16((uint16_t)0x7c00u) - -/// A define value for 16bit negative infinity. -/// -/// @ingroup GPUCore -#define FFX_NEGATIVE_INFINITY_HALF FFX_TO_FLOAT16((uint16_t)0xfc00u) -#else -/// A define value for 16bit positive infinity. -/// -/// @ingroup GPUCore -#define FFX_POSITIVE_INFINITY_HALF FFX_TO_FLOAT16(0x7c00u) - -/// A define value for 16bit negative infinity. -/// -/// @ingroup GPUCore -#define FFX_NEGATIVE_INFINITY_HALF FFX_TO_FLOAT16(0xfc00u) -#endif // #if FFX_HLSL_SM>=62 - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat16 ffxMin(FfxFloat16 x, FfxFloat16 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxMin(FfxFloat16x2 x, FfxFloat16x2 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxMin(FfxFloat16x3 x, FfxFloat16x3 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxMin(FfxFloat16x4 x, FfxFloat16x4 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt16 ffxMin(FfxInt16 x, FfxInt16 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt16x2 ffxMin(FfxInt16x2 x, FfxInt16x2 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt16x3 ffxMin(FfxInt16x3 x, FfxInt16x3 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt16x4 ffxMin(FfxInt16x4 x, FfxInt16x4 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt16 ffxMin(FfxUInt16 x, FfxUInt16 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxMin(FfxUInt16x2 x, FfxUInt16x2 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt16x3 ffxMin(FfxUInt16x3 x, FfxUInt16x3 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt16x4 ffxMin(FfxUInt16x4 x, FfxUInt16x4 y) -{ - return min(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat16 ffxMax(FfxFloat16 x, FfxFloat16 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxMax(FfxFloat16x2 x, FfxFloat16x2 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxMax(FfxFloat16x3 x, FfxFloat16x3 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxMax(FfxFloat16x4 x, FfxFloat16x4 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt16 ffxMax(FfxInt16 x, FfxInt16 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt16x2 ffxMax(FfxInt16x2 x, FfxInt16x2 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt16x3 ffxMax(FfxInt16x3 x, FfxInt16x3 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt16x4 ffxMax(FfxInt16x4 x, FfxInt16x4 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt16 ffxMax(FfxUInt16 x, FfxUInt16 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxMax(FfxUInt16x2 x, FfxUInt16x2 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt16x3 ffxMax(FfxUInt16x3 x, FfxUInt16x3 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt16x4 ffxMax(FfxUInt16x4 x, FfxUInt16x4 y) -{ - return max(x, y); -} - -/// Compute the value of the first parameter raised to the power of the second. -/// -/// @param [in] x The value to raise to the power y. -/// @param [in] y The power to which to raise x. -/// -/// @returns -/// The value of the first parameter raised to the power of the second. -/// -/// @ingroup GPUCore -FfxFloat16 ffxPow(FfxFloat16 x, FfxFloat16 y) -{ - return pow(x, y); -} - -/// Compute the value of the first parameter raised to the power of the second. -/// -/// @param [in] x The value to raise to the power y. -/// @param [in] y The power to which to raise x. -/// -/// @returns -/// The value of the first parameter raised to the power of the second. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPow(FfxFloat16x2 x, FfxFloat16x2 y) -{ - return pow(x, y); -} - -/// Compute the value of the first parameter raised to the power of the second. -/// -/// @param [in] x The value to raise to the power y. -/// @param [in] y The power to which to raise x. -/// -/// @returns -/// The value of the first parameter raised to the power of the second. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxPow(FfxFloat16x3 x, FfxFloat16x3 y) -{ - return pow(x, y); -} - -/// Compute the value of the first parameter raised to the power of the second. -/// -/// @param [in] x The value to raise to the power y. -/// @param [in] y The power to which to raise x. -/// -/// @returns -/// The value of the first parameter raised to the power of the second. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxPow(FfxFloat16x4 x, FfxFloat16x4 y) -{ - return pow(x, y); -} - -/// Compute the square root of a value. -/// -/// @param [in] x The first value to compute the min of. -/// -/// @returns -/// The the square root of x. -/// -/// @ingroup GPUCore -FfxFloat16 ffxSqrt(FfxFloat16 x) -{ - return sqrt(x); -} - -/// Compute the square root of a value. -/// -/// @param [in] x The first value to compute the min of. -/// -/// @returns -/// The the square root of x. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxSqrt(FfxFloat16x2 x) -{ - return sqrt(x); -} - -/// Compute the square root of a value. -/// -/// @param [in] x The first value to compute the min of. -/// -/// @returns -/// The the square root of x. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxSqrt(FfxFloat16x3 x) -{ - return sqrt(x); -} - -/// Compute the square root of a value. -/// -/// @param [in] x The first value to compute the min of. -/// -/// @returns -/// The the square root of x. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxSqrt(FfxFloat16x4 x) -{ - return sqrt(x); -} - -/// Copy the sign bit from 's' to positive 'd'. -/// -/// @param [in] d The value to copy the sign bit into. -/// @param [in] s The value to copy the sign bit from. -/// -/// @returns -/// The value of d with the sign bit from s. -/// -/// @ingroup GPUCore -FfxFloat16 ffxCopySignBitHalf(FfxFloat16 d, FfxFloat16 s) -{ - return FFX_TO_FLOAT16(FFX_TO_UINT16(d) | (FFX_TO_UINT16(s) & FFX_BROADCAST_UINT16(0x8000u))); -} - -/// Copy the sign bit from 's' to positive 'd'. -/// -/// @param [in] d The value to copy the sign bit into. -/// @param [in] s The value to copy the sign bit from. -/// -/// @returns -/// The value of d with the sign bit from s. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxCopySignBitHalf(FfxFloat16x2 d, FfxFloat16x2 s) -{ - return FFX_TO_FLOAT16X2(FFX_TO_UINT16X2(d) | (FFX_TO_UINT16X2(s) & FFX_BROADCAST_UINT16X2(0x8000u))); -} - -/// Copy the sign bit from 's' to positive 'd'. -/// -/// @param [in] d The value to copy the sign bit into. -/// @param [in] s The value to copy the sign bit from. -/// -/// @returns -/// The value of d with the sign bit from s. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxCopySignBitHalf(FfxFloat16x3 d, FfxFloat16x3 s) -{ - return FFX_TO_FLOAT16X3(FFX_TO_UINT16X3(d) | (FFX_TO_UINT16X3(s) & FFX_BROADCAST_UINT16X3(0x8000u))); -} - -/// Copy the sign bit from 's' to positive 'd'. -/// -/// @param [in] d The value to copy the sign bit into. -/// @param [in] s The value to copy the sign bit from. -/// -/// @returns -/// The value of d with the sign bit from s. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxCopySignBitHalf(FfxFloat16x4 d, FfxFloat16x4 s) -{ - return FFX_TO_FLOAT16X4(FFX_TO_UINT16X4(d) | (FFX_TO_UINT16X4(s) & FFX_BROADCAST_UINT16X4(0x8000u))); -} - -/// A single operation to return the following: -/// m = NaN := 0 -/// m >= 0 := 0 -/// m < 0 := 1 -/// -/// Uses the following useful floating point logic, -/// saturate(+a*(-INF)==-INF) := 0 -/// saturate( 0*(-INF)== NaN) := 0 -/// saturate(-a*(-INF)==+INF) := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against 0. -/// -/// @returns -/// 1.0 when the value is negative, or 0.0 when the value is 0 or position. -/// -/// @ingroup GPUCore -FfxFloat16 ffxIsSignedHalf(FfxFloat16 m) -{ - return FfxFloat16(ffxSaturate(m * FFX_BROADCAST_FLOAT16(FFX_NEGATIVE_INFINITY_HALF))); -} - -/// A single operation to return the following: -/// m = NaN := 0 -/// m >= 0 := 0 -/// m < 0 := 1 -/// -/// Uses the following useful floating point logic, -/// saturate(+a*(-INF)==-INF) := 0 -/// saturate( 0*(-INF)== NaN) := 0 -/// saturate(-a*(-INF)==+INF) := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against 0. -/// -/// @returns -/// 1.0 when the value is negative, or 0.0 when the value is 0 or position. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxIsSignedHalf(FfxFloat16x2 m) -{ - return FfxFloat16x2(ffxSaturate(m * FFX_BROADCAST_FLOAT16X2(FFX_NEGATIVE_INFINITY_HALF))); -} - -/// A single operation to return the following: -/// m = NaN := 0 -/// m >= 0 := 0 -/// m < 0 := 1 -/// -/// Uses the following useful floating point logic, -/// saturate(+a*(-INF)==-INF) := 0 -/// saturate( 0*(-INF)== NaN) := 0 -/// saturate(-a*(-INF)==+INF) := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against 0. -/// -/// @returns -/// 1.0 when the value is negative, or 0.0 when the value is 0 or position. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxIsSignedHalf(FfxFloat16x3 m) -{ - return FfxFloat16x3(ffxSaturate(m * FFX_BROADCAST_FLOAT16X3(FFX_NEGATIVE_INFINITY_HALF))); -} - -/// A single operation to return the following: -/// m = NaN := 0 -/// m >= 0 := 0 -/// m < 0 := 1 -/// -/// Uses the following useful floating point logic, -/// saturate(+a*(-INF)==-INF) := 0 -/// saturate( 0*(-INF)== NaN) := 0 -/// saturate(-a*(-INF)==+INF) := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against 0. -/// -/// @returns -/// 1.0 when the value is negative, or 0.0 when the value is 0 or position. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxIsSignedHalf(FfxFloat16x4 m) -{ - return FfxFloat16x4(ffxSaturate(m * FFX_BROADCAST_FLOAT16X4(FFX_NEGATIVE_INFINITY_HALF))); -} - -/// A single operation to return the following: -/// m = NaN := 1 -/// m > 0 := 0 -/// m <= 0 := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against zero. -/// -/// @returns -/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. -/// -/// @ingroup GPUCore -FfxFloat16 ffxIsGreaterThanZeroHalf(FfxFloat16 m) -{ - return FfxFloat16(ffxSaturate(m * FFX_BROADCAST_FLOAT16(FFX_POSITIVE_INFINITY_HALF))); -} - -/// A single operation to return the following: -/// m = NaN := 1 -/// m > 0 := 0 -/// m <= 0 := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against zero. -/// -/// @returns -/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxIsGreaterThanZeroHalf(FfxFloat16x2 m) -{ - return FfxFloat16x2(ffxSaturate(m * FFX_BROADCAST_FLOAT16X2(FFX_POSITIVE_INFINITY_HALF))); -} - -/// A single operation to return the following: -/// m = NaN := 1 -/// m > 0 := 0 -/// m <= 0 := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against zero. -/// -/// @returns -/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxIsGreaterThanZeroHalf(FfxFloat16x3 m) -{ - return FfxFloat16x3(ffxSaturate(m * FFX_BROADCAST_FLOAT16X3(FFX_POSITIVE_INFINITY_HALF))); -} - -/// A single operation to return the following: -/// m = NaN := 1 -/// m > 0 := 0 -/// m <= 0 := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against zero. -/// -/// @returns -/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxIsGreaterThanZeroHalf(FfxFloat16x4 m) -{ - return FfxFloat16x4(ffxSaturate(m * FFX_BROADCAST_FLOAT16X4(FFX_POSITIVE_INFINITY_HALF))); -} - -/// Convert a 16bit floating point value to sortable integer. -/// -/// - If sign bit=0, flip the sign bit (positives). -/// - If sign bit=1, flip all bits (negatives). -/// -/// The function has the side effects that: -/// - Larger integers are more positive values. -/// - Float zero is mapped to center of integers (so clear to integer zero is a nice default for atomic max usage). -/// -/// @param [in] x The floating point value to make sortable. -/// -/// @returns -/// The sortable integer value. -/// -/// @ingroup GPUCore -FfxUInt16 ffxFloatToSortableIntegerHalf(FfxUInt16 x) -{ - return x ^ ((ffxBitShiftRightHalf(x, FFX_BROADCAST_UINT16(15))) | FFX_BROADCAST_UINT16(0x8000)); -} - -/// Convert a sortable integer to a 16bit floating point value. -/// -/// The function has the side effects that: -/// - If sign bit=1, flip the sign bit (positives). -/// - If sign bit=0, flip all bits (negatives). -/// -/// @param [in] x The sortable integer value to make floating point. -/// -/// @returns -/// The floating point value. -/// -/// @ingroup GPUCore -FfxUInt16 ffxSortableIntegerToFloatHalf(FfxUInt16 x) -{ - return x ^ ((~ffxBitShiftRightHalf(x, FFX_BROADCAST_UINT16(15))) | FFX_BROADCAST_UINT16(0x8000)); -} - -/// Convert a pair of 16bit floating point values to a pair of sortable integers. -/// -/// - If sign bit=0, flip the sign bit (positives). -/// - If sign bit=1, flip all bits (negatives). -/// -/// The function has the side effects that: -/// - Larger integers are more positive values. -/// - Float zero is mapped to center of integers (so clear to integer zero is a nice default for atomic max usage). -/// -/// @param [in] x The floating point values to make sortable. -/// -/// @returns -/// The sortable integer values. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxFloatToSortableIntegerHalf(FfxUInt16x2 x) -{ - return x ^ ((ffxBitShiftRightHalf(x, FFX_BROADCAST_UINT16X2(15))) | FFX_BROADCAST_UINT16X2(0x8000)); -} - -/// Convert a pair of sortable integers to a pair of 16bit floating point values. -/// -/// The function has the side effects that: -/// - If sign bit=1, flip the sign bit (positives). -/// - If sign bit=0, flip all bits (negatives). -/// -/// @param [in] x The sortable integer values to make floating point. -/// -/// @returns -/// The floating point values. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxSortableIntegerToFloatHalf(FfxUInt16x2 x) -{ - return x ^ ((~ffxBitShiftRightHalf(x, FFX_BROADCAST_UINT16X2(15))) | FFX_BROADCAST_UINT16X2(0x8000)); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// [Zero] Y0 [Zero] X0 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesZeroY0ZeroX0(FfxUInt32x2 i) -{ - return ((i.x) & 0xffu) | ((i.y << 16) & 0xff0000u); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// [Zero] Y1 [Zero] X1 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesZeroY1ZeroX1(FfxUInt32x2 i) -{ - return ((i.x >> 8) & 0xffu) | ((i.y << 8) & 0xff0000u); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// [Zero] Y2 [Zero] X2 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesZeroY2ZeroX2(FfxUInt32x2 i) -{ - return ((i.x >> 16) & 0xffu) | ((i.y) & 0xff0000u); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// [Zero] Y3 [Zero] X3 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesZeroY3ZeroX3(FfxUInt32x2 i) -{ - return ((i.x >> 24) & 0xffu) | ((i.y >> 8) & 0xff0000u); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// Y3 Y2 Y1 X0 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesY3Y2Y1X0(FfxUInt32x2 i) -{ - return ((i.x) & 0x000000ffu) | (i.y & 0xffffff00u); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// Y3 Y2 Y1 X2 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesY3Y2Y1X2(FfxUInt32x2 i) -{ - return ((i.x >> 16) & 0x000000ffu) | (i.y & 0xffffff00u); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// Y3 Y2 X0 Y0 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesY3Y2X0Y0(FfxUInt32x2 i) -{ - return ((i.x << 8) & 0x0000ff00u) | (i.y & 0xffff00ffu); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// Y3 Y2 X2 Y0 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesY3Y2X2Y0(FfxUInt32x2 i) -{ - return ((i.x >> 8) & 0x0000ff00u) | (i.y & 0xffff00ffu); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// Y3 X0 Y1 Y0 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesY3X0Y1Y0(FfxUInt32x2 i) -{ - return ((i.x << 16) & 0x00ff0000u) | (i.y & 0xff00ffffu); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// Y3 X2 Y1 Y0 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesY3X2Y1Y0(FfxUInt32x2 i) -{ - return ((i.x) & 0x00ff0000u) | (i.y & 0xff00ffffu); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// X0 Y2 Y1 Y0 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesX0Y2Y1Y0(FfxUInt32x2 i) -{ - return ((i.x << 24) & 0xff000000u) | (i.y & 0x00ffffffu); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// X2 Y2 Y1 Y0 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesX2Y2Y1Y0(FfxUInt32x2 i) -{ - return ((i.x << 8) & 0xff000000u) | (i.y & 0x00ffffffu); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// Y2 X2 Y0 X0 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesY2X2Y0X0(FfxUInt32x2 i) -{ - return ((i.x) & 0x00ff00ffu) | ((i.y << 8) & 0xff00ff00u); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// Y2 Y0 X2 X0 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesY2Y0X2X0(FfxUInt32x2 i) -{ - return (((i.x) & 0xffu) | ((i.x >> 8) & 0xff00u) | ((i.y << 16) & 0xff0000u) | ((i.y << 8) & 0xff000000u)); -} - -/// Takes two Float16x2 values x and y, normalizes them and builds a single Uint16x2 value in the format {{x0,y0},{x1,y1}}. -/// -/// @param [in] x The first float16x2 value to pack. -/// @param [in] y The second float16x2 value to pack. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxPackX0Y0X1Y1UnsignedToUint16x2(FfxFloat16x2 x, FfxFloat16x2 y) -{ - x *= FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0); - y *= FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0); - return FFX_UINT32_TO_UINT16X2(ffxPackBytesY2X2Y0X0(FfxUInt32x2(FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(x)), FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(y))))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[0:7], -/// d.y[0:7] into r.y[0:7], i.x[8:15] into r.x[8:15], r.y[8:15] and i.y[0:15] into r.x[16:31], r.y[16:31] using 3 ops. -/// -/// r=ffxPermuteUByte0Float16x2ToUint2(d,i) -/// Where 'k0' is an SGPR with {1.0/32768.0} packed into the lower 16-bits -/// Where 'k1' is an SGPR with 0x???? -/// Where 'k2' is an SGPR with 0x???? -/// V_PK_FMA_F16 i,i,k0.x,0 -/// V_PERM_B32 r.x,i,i,k1 -/// V_PERM_B32 r.y,i,i,k2 -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteUByte0Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0))); - return FfxUInt32x2(ffxPackBytesY3Y2Y1X0(FfxUInt32x2(d.x, b)), ffxPackBytesY3Y2Y1X2(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[8:15], -/// d.y[0:7] into r.y[8:15], i.x[0:7] into r.x[0:7], r.y[0:7] and i.y[0:15] into r.x[16:31], r.y[16:31] using 3 ops. -/// -/// r=ffxPermuteUByte1Float16x2ToUint2(d,i) -/// Where 'k0' is an SGPR with {1.0/32768.0} packed into the lower 16-bits -/// Where 'k1' is an SGPR with 0x???? -/// Where 'k2' is an SGPR with 0x???? -/// V_PK_FMA_F16 i,i,k0.x,0 -/// V_PERM_B32 r.x,i,i,k1 -/// V_PERM_B32 r.y,i,i,k2 -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteUByte1Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0))); - return FfxUInt32x2(ffxPackBytesY3Y2X0Y0(FfxUInt32x2(d.x, b)), ffxPackBytesY3Y2X2Y0(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[16:23], -/// d.y[0:7] into r.y[16:23], i.x[0:15] into r.x[0:15], r.y[0:15] and i.y[8:15] into r.x[24:31], r.y[24:31] using 3 ops. -/// -/// r=ffxPermuteUByte2Float16x2ToUint2(d,i) -/// Where 'k0' is an SGPR with {1.0/32768.0} packed into the lower 16-bits -/// Where 'k1' is an SGPR with 0x???? -/// Where 'k2' is an SGPR with 0x???? -/// V_PK_FMA_F16 i,i,k0.x,0 -/// V_PERM_B32 r.x,i,i,k1 -/// V_PERM_B32 r.y,i,i,k2 -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteUByte2Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0))); - return FfxUInt32x2(ffxPackBytesY3X0Y1Y0(FfxUInt32x2(d.x, b)), ffxPackBytesY3X2Y1Y0(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[24:31], -/// d.y[0:7] into r.y[24:31], i.x[0:15] into r.x[0:15], r.y[0:15] and i.y[0:7] into r.x[16:23], r.y[16:23] using 3 ops. -/// -/// r=ffxPermuteUByte3Float16x2ToUint2(d,i) -/// Where 'k0' is an SGPR with {1.0/32768.0} packed into the lower 16-bits -/// Where 'k1' is an SGPR with 0x???? -/// Where 'k2' is an SGPR with 0x???? -/// V_PK_FMA_F16 i,i,k0.x,0 -/// V_PERM_B32 r.x,i,i,k1 -/// V_PERM_B32 r.y,i,i,k2 -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteUByte3Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0))); - return FfxUInt32x2(ffxPackBytesX0Y2Y1Y0(FfxUInt32x2(d.x, b)), ffxPackBytesX2Y2Y1Y0(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[0:7] into r.x[0:7] and i.y[0:7] into r.y[0:7] using 2 ops. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteUByte0Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY0ZeroX0(i))) * FFX_BROADCAST_FLOAT16X2(32768.0); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[8:15] into r.x[0:7] and i.y[8:15] into r.y[0:7] using 2 ops. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteUByte1Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY1ZeroX1(i))) * FFX_BROADCAST_FLOAT16X2(32768.0); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[16:23] into r.x[0:7] and i.y[16:23] into r.y[0:7] using 2 ops. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteUByte2Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY2ZeroX2(i))) * FFX_BROADCAST_FLOAT16X2(32768.0); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[24:31] into r.x[0:7] and i.y[24:31] into r.y[0:7] using 2 ops. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteUByte3Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY3ZeroX3(i))) * FFX_BROADCAST_FLOAT16X2(32768.0); -} - -/// Takes two Float16x2 values x and y, normalizes them and builds a single Uint16x2 value in the format {{x0,y0},{x1,y1}}. -/// -/// @param [in] x The first float16x2 value to pack. -/// @param [in] y The second float16x2 value to pack. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxPackX0Y0X1Y1SignedToUint16x2(FfxFloat16x2 x, FfxFloat16x2 y) -{ - x = x * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0); - y = y * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0); - return FFX_UINT32_TO_UINT16X2(ffxPackBytesY2X2Y0X0(FfxUInt32x2(FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(x)), FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(y))))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[0:7], -/// d.y[0:7] into r.y[0:7], i.x[8:15] into r.x[8:15], r.y[8:15] and i.y[0:15] into r.x[16:31], r.y[16:31] using 3 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteSByte0Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))); - return FfxUInt32x2(ffxPackBytesY3Y2Y1X0(FfxUInt32x2(d.x, b)), ffxPackBytesY3Y2Y1X2(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[8:15], -/// d.y[0:7] into r.y[8:15], i.x[0:7] into r.x[0:7], r.y[0:7] and i.y[0:15] into r.x[16:31], r.y[16:31] using 3 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteSByte1Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))); - return FfxUInt32x2(ffxPackBytesY3Y2X0Y0(FfxUInt32x2(d.x, b)), ffxPackBytesY3Y2X2Y0(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[16:23], -/// d.y[0:7] into r.y[16:23], i.x[0:15] into r.x[0:15], r.y[0:15] and i.y[8:15] into r.x[24:31], r.y[24:31] using 3 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteSByte2Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))); - return FfxUInt32x2(ffxPackBytesY3X0Y1Y0(FfxUInt32x2(d.x, b)), ffxPackBytesY3X2Y1Y0(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[24:31], -/// d.y[0:7] into r.y[24:31], i.x[0:15] into r.x[0:15], r.y[0:15] and i.y[0:7] into r.x[16:23], r.y[16:23] using 3 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteSByte3Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))); - return FfxUInt32x2(ffxPackBytesX0Y2Y1Y0(FfxUInt32x2(d.x, b)), ffxPackBytesX2Y2Y1Y0(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[0:7], -/// d.y[0:7] into r.y[0:7], i.x[8:15] into r.x[8:15], r.y[8:15] and i.y[0:15] into r.x[16:31], r.y[16:31] using 3 ops. -/// -/// Zero-based flips the MSB bit of the byte (making 128 "exact zero" actually zero). -/// This is useful if there is a desire for cleared values to decode as zero. -/// -/// Handles signed byte values. -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteZeroBasedSByte0Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))) ^ 0x00800080u; - return FfxUInt32x2(ffxPackBytesY3Y2Y1X0(FfxUInt32x2(d.x, b)), ffxPackBytesY3Y2Y1X2(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[8:15], -/// d.y[0:7] into r.y[8:15], i.x[0:7] into r.x[0:7], r.y[0:7] and i.y[0:15] into r.x[16:31], r.y[16:31] using 3 ops. -/// -/// Zero-based flips the MSB bit of the byte (making 128 "exact zero" actually zero). -/// This is useful if there is a desire for cleared values to decode as zero. -/// -/// Handles signed byte values. -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteZeroBasedSByte1Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))) ^ 0x00800080u; - return FfxUInt32x2(ffxPackBytesY3Y2X0Y0(FfxUInt32x2(d.x, b)), ffxPackBytesY3Y2X2Y0(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[16:23], -/// d.y[0:7] into r.y[16:23], i.x[0:15] into r.x[0:15], r.y[0:15] and i.y[8:15] into r.x[24:31], r.y[24:31] using 3 ops. -/// -/// Zero-based flips the MSB bit of the byte (making 128 "exact zero" actually zero). -/// This is useful if there is a desire for cleared values to decode as zero. -/// -/// Handles signed byte values. -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteZeroBasedSByte2Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))) ^ 0x00800080u; - return FfxUInt32x2(ffxPackBytesY3X0Y1Y0(FfxUInt32x2(d.x, b)), ffxPackBytesY3X2Y1Y0(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[24:31], -/// d.y[0:7] into r.y[24:31], i.x[0:15] into r.x[0:15], r.y[0:15] and i.y[0:7] into r.x[16:23], r.y[16:23] using 3 ops. -/// -/// Zero-based flips the MSB bit of the byte (making 128 "exact zero" actually zero). -/// This is useful if there is a desire for cleared values to decode as zero. -/// -/// Handles signed byte values. -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteZeroBasedSByte3Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))) ^ 0x00800080u; - return FfxUInt32x2(ffxPackBytesX0Y2Y1Y0(FfxUInt32x2(d.x, b)), ffxPackBytesX2Y2Y1Y0(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[0:7] into r.x[0:7] and i.y[0:7] into r.y[0:7] using 2 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteSByte0Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY0ZeroX0(i))) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[8:15] into r.x[0:7] and i.y[8:15] into r.y[0:7] using 2 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteSByte1Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY1ZeroX1(i))) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[16:23] into r.x[0:7] and i.y[16:23] into r.y[0:7] using 2 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteSByte2Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY2ZeroX2(i))) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[24:31] into r.x[0:7] and i.y[24:31] into r.y[0:7] using 2 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteSByte3Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY3ZeroX3(i))) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[0:7] into r.x[0:7] and i.y[0:7] into r.y[0:7] using 2 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteZeroBasedSByte0Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY0ZeroX0(i) ^ 0x00800080u)) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[8:15] into r.x[0:7] and i.y[8:15] into r.y[0:7] using 2 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteZeroBasedSByte1Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY1ZeroX1(i) ^ 0x00800080u)) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[16:23] into r.x[0:7] and i.y[16:23] into r.y[0:7] using 2 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteZeroBasedSByte2Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY2ZeroX2(i) ^ 0x00800080u)) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[24:31] into r.x[0:7] and i.y[24:31] into r.y[0:7] using 2 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteZeroBasedSByte3Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY3ZeroX3(i) ^ 0x00800080u)) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); -} - -/// Calculate a half-precision low-quality approximation for the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the square root for. -/// -/// @returns -/// An approximation of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat16 ffxApproximateSqrtHalf(FfxFloat16 a) -{ - return FFX_TO_FLOAT16((FFX_TO_UINT16(a) >> FFX_BROADCAST_UINT16(1)) + FFX_BROADCAST_UINT16(0x1de2)); -} - -/// Calculate a half-precision low-quality approximation for the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the square root for. -/// -/// @returns -/// An approximation of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxApproximateSqrtHalf(FfxFloat16x2 a) -{ - return FFX_TO_FLOAT16X2((FFX_TO_UINT16X2(a) >> FFX_BROADCAST_UINT16X2(1)) + FFX_BROADCAST_UINT16X2(0x1de2)); -} - -/// Calculate a half-precision low-quality approximation for the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the square root for. -/// -/// @returns -/// An approximation of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxApproximateSqrtHalf(FfxFloat16x3 a) -{ - return FFX_TO_FLOAT16X3((FFX_TO_UINT16X3(a) >> FFX_BROADCAST_UINT16X3(1)) + FFX_BROADCAST_UINT16X3(0x1de2)); -} - -/// Calculate a half-precision low-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat16 ffxApproximateReciprocalHalf(FfxFloat16 a) -{ - return FFX_TO_FLOAT16(FFX_BROADCAST_UINT16(0x7784) - FFX_TO_UINT16(a)); -} - -/// Calculate a half-precision low-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxApproximateReciprocalHalf(FfxFloat16x2 a) -{ - return FFX_TO_FLOAT16X2(FFX_BROADCAST_UINT16X2(0x7784) - FFX_TO_UINT16X2(a)); -} - -/// Calculate a half-precision low-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxApproximateReciprocalHalf(FfxFloat16x3 a) -{ - return FFX_TO_FLOAT16X3(FFX_BROADCAST_UINT16X3(0x7784) - FFX_TO_UINT16X3(a)); -} - -/// Calculate a half-precision low-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxApproximateReciprocalHalf(FfxFloat16x4 a) -{ - return FFX_TO_FLOAT16X4(FFX_BROADCAST_UINT16X4(0x7784) - FFX_TO_UINT16X4(a)); -} - -/// Calculate a half-precision medium-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to medium quality. -/// -/// @ingroup GPUCore -FfxFloat16 ffxApproximateReciprocalMediumHalf(FfxFloat16 a) -{ - FfxFloat16 b = FFX_TO_FLOAT16(FFX_BROADCAST_UINT16(0x778d) - FFX_TO_UINT16(a)); - return b * (-b * a + FFX_BROADCAST_FLOAT16(2.0)); -} - -/// Calculate a half-precision medium-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to medium quality. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxApproximateReciprocalMediumHalf(FfxFloat16x2 a) -{ - FfxFloat16x2 b = FFX_TO_FLOAT16X2(FFX_BROADCAST_UINT16X2(0x778d) - FFX_TO_UINT16X2(a)); - return b * (-b * a + FFX_BROADCAST_FLOAT16X2(2.0)); -} - -/// Calculate a half-precision medium-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to medium quality. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxApproximateReciprocalMediumHalf(FfxFloat16x3 a) -{ - FfxFloat16x3 b = FFX_TO_FLOAT16X3(FFX_BROADCAST_UINT16X3(0x778d) - FFX_TO_UINT16X3(a)); - return b * (-b * a + FFX_BROADCAST_FLOAT16X3(2.0)); -} - -/// Calculate a half-precision medium-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to medium quality. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxApproximateReciprocalMediumHalf(FfxFloat16x4 a) -{ - FfxFloat16x4 b = FFX_TO_FLOAT16X4(FFX_BROADCAST_UINT16X4(0x778d) - FFX_TO_UINT16X4(a)); - return b * (-b * a + FFX_BROADCAST_FLOAT16X4(2.0)); -} - -/// Calculate a half-precision low-quality approximation for the reciprocal of the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal of the square root for. -/// -/// @returns -/// An approximation of the reciprocal of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat16 ffxApproximateReciprocalSquareRootHalf(FfxFloat16 a) -{ - return FFX_TO_FLOAT16(FFX_BROADCAST_UINT16(0x59a3) - (FFX_TO_UINT16(a) >> FFX_BROADCAST_UINT16(1))); -} - -/// Calculate a half-precision low-quality approximation for the reciprocal of the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal of the square root for. -/// -/// @returns -/// An approximation of the reciprocal of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxApproximateReciprocalSquareRootHalf(FfxFloat16x2 a) -{ - return FFX_TO_FLOAT16X2(FFX_BROADCAST_UINT16X2(0x59a3) - (FFX_TO_UINT16X2(a) >> FFX_BROADCAST_UINT16X2(1))); -} - -/// Calculate a half-precision low-quality approximation for the reciprocal of the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal of the square root for. -/// -/// @returns -/// An approximation of the reciprocal of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxApproximateReciprocalSquareRootHalf(FfxFloat16x3 a) -{ - return FFX_TO_FLOAT16X3(FFX_BROADCAST_UINT16X3(0x59a3) - (FFX_TO_UINT16X3(a) >> FFX_BROADCAST_UINT16X3(1))); -} - -/// Calculate a half-precision low-quality approximation for the reciprocal of the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal of the square root for. -/// -/// @returns -/// An approximation of the reciprocal of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxApproximateReciprocalSquareRootHalf(FfxFloat16x4 a) -{ - return FFX_TO_FLOAT16X4(FFX_BROADCAST_UINT16X4(0x59a3) - (FFX_TO_UINT16X4(a) >> FFX_BROADCAST_UINT16X4(1))); -} - -/// An approximation of sine. -/// -/// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range -/// is {-1/4 to 1/4} representing {-1 to 1}. -/// -/// @param [in] x The value to calculate approximate sine for. -/// -/// @returns -/// The approximate sine of value. -FfxFloat16 ffxParabolicSinHalf(FfxFloat16 x) -{ - return x * abs(x) - x; -} - -/// An approximation of sine. -/// -/// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range -/// is {-1/4 to 1/4} representing {-1 to 1}. -/// -/// @param [in] x The value to calculate approximate sine for. -/// -/// @returns -/// The approximate sine of value. -FfxFloat16x2 ffxParabolicSinHalf(FfxFloat16x2 x) -{ - return x * abs(x) - x; -} - -/// An approximation of cosine. -/// -/// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range -/// is {-1/4 to 1/4} representing {-1 to 1}. -/// -/// @param [in] x The value to calculate approximate cosine for. -/// -/// @returns -/// The approximate cosine of value. -FfxFloat16 ffxParabolicCosHalf(FfxFloat16 x) -{ - x = ffxFract(x * FFX_BROADCAST_FLOAT16(0.5) + FFX_BROADCAST_FLOAT16(0.75)); - x = x * FFX_BROADCAST_FLOAT16(2.0) - FFX_BROADCAST_FLOAT16(1.0); - return ffxParabolicSinHalf(x); -} - -/// An approximation of cosine. -/// -/// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range -/// is {-1/4 to 1/4} representing {-1 to 1}. -/// -/// @param [in] x The value to calculate approximate cosine for. -/// -/// @returns -/// The approximate cosine of value. -FfxFloat16x2 ffxParabolicCosHalf(FfxFloat16x2 x) -{ - x = ffxFract(x * FFX_BROADCAST_FLOAT16X2(0.5) + FFX_BROADCAST_FLOAT16X2(0.75)); - x = x * FFX_BROADCAST_FLOAT16X2(2.0) - FFX_BROADCAST_FLOAT16X2(1.0); - return ffxParabolicSinHalf(x); -} - -/// An approximation of both sine and cosine. -/// -/// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range -/// is {-1/4 to 1/4} representing {-1 to 1}. -/// -/// @param [in] x The value to calculate approximate cosine for. -/// -/// @returns -/// A FfxFloat32x2 containing approximations of both sine and cosine of value. -FfxFloat16x2 ffxParabolicSinCosHalf(FfxFloat16 x) -{ - FfxFloat16 y = ffxFract(x * FFX_BROADCAST_FLOAT16(0.5) + FFX_BROADCAST_FLOAT16(0.75)); - y = y * FFX_BROADCAST_FLOAT16(2.0) - FFX_BROADCAST_FLOAT16(1.0); - return ffxParabolicSinHalf(FfxFloat16x2(x, y)); -} - -/// Conditional free logic AND operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxUInt16 ffxZeroOneAndHalf(FfxUInt16 x, FfxUInt16 y) -{ - return min(x, y); -} - -/// Conditional free logic AND operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxZeroOneAndHalf(FfxUInt16x2 x, FfxUInt16x2 y) -{ - return min(x, y); -} - -/// Conditional free logic AND operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxUInt16x3 ffxZeroOneAndHalf(FfxUInt16x3 x, FfxUInt16x3 y) -{ - return min(x, y); -} - -/// Conditional free logic AND operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxUInt16x4 ffxZeroOneAndHalf(FfxUInt16x4 x, FfxUInt16x4 y) -{ - return min(x, y); -} - -/// Conditional free logic NOT operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the NOT operator. -/// @param [in] y The second value to be fed into the NOT operator. -/// -/// @returns -/// Result of the NOT operation. -/// -/// @ingroup GPUCore -FfxUInt16 ffxZeroOneNotHalf(FfxUInt16 x) -{ - return x ^ FFX_BROADCAST_UINT16(1); -} - -/// Conditional free logic NOT operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the NOT operator. -/// @param [in] y The second value to be fed into the NOT operator. -/// -/// @returns -/// Result of the NOT operation. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxZeroOneNotHalf(FfxUInt16x2 x) -{ - return x ^ FFX_BROADCAST_UINT16X2(1); -} - -/// Conditional free logic NOT operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the NOT operator. -/// @param [in] y The second value to be fed into the NOT operator. -/// -/// @returns -/// Result of the NOT operation. -/// -/// @ingroup GPUCore -FfxUInt16x3 ffxZeroOneNotHalf(FfxUInt16x3 x) -{ - return x ^ FFX_BROADCAST_UINT16X3(1); -} - -/// Conditional free logic NOT operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the NOT operator. -/// @param [in] y The second value to be fed into the NOT operator. -/// -/// @returns -/// Result of the NOT operation. -/// -/// @ingroup GPUCore -FfxUInt16x4 ffxZeroOneNotHalf(FfxUInt16x4 x) -{ - return x ^ FFX_BROADCAST_UINT16X4(1); -} - -/// Conditional free logic OR operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxUInt16 ffxZeroOneOrHalf(FfxUInt16 x, FfxUInt16 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxZeroOneOrHalf(FfxUInt16x2 x, FfxUInt16x2 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxUInt16x3 ffxZeroOneOrHalf(FfxUInt16x3 x, FfxUInt16x3 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxUInt16x4 ffxZeroOneOrHalf(FfxUInt16x4 x, FfxUInt16x4 y) -{ - return max(x, y); -} - -/// Convert a half-precision FfxFloat32 value between 0.0f and 1.0f to a half-precision Uint. -/// -/// @param [in] x The value to converted to a Uint. -/// -/// @returns -/// The converted Uint value. -/// -/// @ingroup GPUCore -FfxUInt16 ffxZeroOneFloat16ToUint16(FfxFloat16 x) -{ - return FFX_TO_UINT16(x * FFX_TO_FLOAT16(FFX_TO_UINT16(1))); -} - -/// Convert a half-precision FfxFloat32 value between 0.0f and 1.0f to a half-precision Uint. -/// -/// @param [in] x The value to converted to a Uint. -/// -/// @returns -/// The converted Uint value. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxZeroOneFloat16x2ToUint16x2(FfxFloat16x2 x) -{ - return FFX_TO_UINT16X2(x * FFX_TO_FLOAT16X2(FfxUInt16x2(1, 1))); -} - -/// Convert a half-precision FfxFloat32 value between 0.0f and 1.0f to a half-precision Uint. -/// -/// @param [in] x The value to converted to a Uint. -/// -/// @returns -/// The converted Uint value. -/// -/// @ingroup GPUCore -FfxUInt16x3 ffxZeroOneFloat16x3ToUint16x3(FfxFloat16x3 x) -{ - return FFX_TO_UINT16X3(x * FFX_TO_FLOAT16X3(FfxUInt16x3(1, 1, 1))); -} - -/// Convert a half-precision FfxFloat32 value between 0.0f and 1.0f to a half-precision Uint. -/// -/// @param [in] x The value to converted to a Uint. -/// -/// @returns -/// The converted Uint value. -/// -/// @ingroup GPUCore -FfxUInt16x4 ffxZeroOneFloat16x4ToUint16x4(FfxFloat16x4 x) -{ - return FFX_TO_UINT16X4(x * FFX_TO_FLOAT16X4(FfxUInt16x4(1, 1, 1, 1))); -} - -/// Convert a half-precision FfxUInt32 value between 0 and 1 to a half-precision FfxFloat32. -/// -/// @param [in] x The value to converted to a half-precision FfxFloat32. -/// -/// @returns -/// The converted half-precision FfxFloat32 value. -/// -/// @ingroup GPUCore -FfxFloat16 ffxZeroOneUint16ToFloat16(FfxUInt16 x) -{ - return FFX_TO_FLOAT16(x * FFX_TO_UINT16(FFX_TO_FLOAT16(1.0))); -} - -/// Convert a half-precision FfxUInt32 value between 0 and 1 to a half-precision FfxFloat32. -/// -/// @param [in] x The value to converted to a half-precision FfxFloat32. -/// -/// @returns -/// The converted half-precision FfxFloat32 value. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxZeroOneUint16x2ToFloat16x2(FfxUInt16x2 x) -{ - return FFX_TO_FLOAT16X2(x * FFX_TO_UINT16X2(FfxUInt16x2(FFX_TO_FLOAT16(1.0), FFX_TO_FLOAT16(1.0)))); -} - -/// Convert a half-precision FfxUInt32 value between 0 and 1 to a half-precision FfxFloat32. -/// -/// @param [in] x The value to converted to a half-precision FfxFloat32. -/// -/// @returns -/// The converted half-precision FfxFloat32 value. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxZeroOneUint16x3ToFloat16x3(FfxUInt16x3 x) -{ - return FFX_TO_FLOAT16X3(x * FFX_TO_UINT16X3(FfxUInt16x3(FFX_TO_FLOAT16(1.0), FFX_TO_FLOAT16(1.0), FFX_TO_FLOAT16(1.0)))); -} - -/// Convert a half-precision FfxUInt32 value between 0 and 1 to a half-precision FfxFloat32. -/// -/// @param [in] x The value to converted to a half-precision FfxFloat32. -/// -/// @returns -/// The converted half-precision FfxFloat32 value. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxZeroOneUint16x4ToFloat16x4(FfxUInt16x4 x) -{ - return FFX_TO_FLOAT16X4(x * FFX_TO_UINT16X4(FfxUInt16x4(FFX_TO_FLOAT16(1.0), FFX_TO_FLOAT16(1.0), FFX_TO_FLOAT16(1.0), FFX_TO_FLOAT16(1.0)))); -} - -/// Conditional free logic AND operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxFloat16 ffxZeroOneAndHalf(FfxFloat16 x, FfxFloat16 y) -{ - return min(x, y); -} - -/// Conditional free logic AND operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxZeroOneAndHalf(FfxFloat16x2 x, FfxFloat16x2 y) -{ - return min(x, y); -} - -/// Conditional free logic AND operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxZeroOneAndHalf(FfxFloat16x3 x, FfxFloat16x3 y) -{ - return min(x, y); -} - -/// Conditional free logic AND operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxZeroOneAndHalf(FfxFloat16x4 x, FfxFloat16x4 y) -{ - return min(x, y); -} - -/// Conditional free logic AND NOT operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND NOT operator. -/// @param [in] y The second value to be fed into the AND NOT operator. -/// -/// @returns -/// Result of the AND NOT operation. -/// -/// @ingroup GPUCore -FfxFloat16 ffxSignedZeroOneAndOrHalf(FfxFloat16 x, FfxFloat16 y) -{ - return (-x) * y + FFX_BROADCAST_FLOAT16(1.0); -} - -/// Conditional free logic AND NOT operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND NOT operator. -/// @param [in] y The second value to be fed into the AND NOT operator. -/// -/// @returns -/// Result of the AND NOT operation. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxSignedZeroOneAndOrHalf(FfxFloat16x2 x, FfxFloat16x2 y) -{ - return (-x) * y + FFX_BROADCAST_FLOAT16X2(1.0); -} - -/// Conditional free logic AND NOT operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND NOT operator. -/// @param [in] y The second value to be fed into the AND NOT operator. -/// -/// @returns -/// Result of the AND NOT operation. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxSignedZeroOneAndOrHalf(FfxFloat16x3 x, FfxFloat16x3 y) -{ - return (-x) * y + FFX_BROADCAST_FLOAT16X3(1.0); -} - -/// Conditional free logic AND NOT operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND NOT operator. -/// @param [in] y The second value to be fed into the AND NOT operator. -/// -/// @returns -/// Result of the AND NOT operation. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxSignedZeroOneAndOrHalf(FfxFloat16x4 x, FfxFloat16x4 y) -{ - return (-x) * y + FFX_BROADCAST_FLOAT16X4(1.0); -} - -/// Conditional free logic AND operation using two half-precision values followed by -/// a NOT operation using the resulting value and a third half-precision value. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// @param [in] z The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat16 ffxZeroOneAndOrHalf(FfxFloat16 x, FfxFloat16 y, FfxFloat16 z) -{ - return FfxFloat16(ffxSaturate(x * y + z)); -} - -/// Conditional free logic AND operation using two half-precision values followed by -/// a NOT operation using the resulting value and a third half-precision value. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// @param [in] z The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxZeroOneAndOrHalf(FfxFloat16x2 x, FfxFloat16x2 y, FfxFloat16x2 z) -{ - return FfxFloat16x2(ffxSaturate(x * y + z)); -} - -/// Conditional free logic AND operation using two half-precision values followed by -/// a NOT operation using the resulting value and a third half-precision value. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// @param [in] z The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxZeroOneAndOrHalf(FfxFloat16x3 x, FfxFloat16x3 y, FfxFloat16x3 z) -{ - return FfxFloat16x3(ffxSaturate(x * y + z)); -} - -/// Conditional free logic AND operation using two half-precision values followed by -/// a NOT operation using the resulting value and a third half-precision value. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// @param [in] z The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxZeroOneAndOrHalf(FfxFloat16x4 x, FfxFloat16x4 y, FfxFloat16x4 z) -{ - return FfxFloat16x4(ffxSaturate(x * y + z)); -} - -/// Given a half-precision value, returns 1.0 if greater than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the greater than zero comparison. -/// -/// @ingroup GPUCore -FfxFloat16 ffxZeroOneIsGreaterThanZeroHalf(FfxFloat16 x) -{ - return FfxFloat16(ffxSaturate(x * FFX_BROADCAST_FLOAT16(FFX_POSITIVE_INFINITY_HALF))); -} - -/// Given a half-precision value, returns 1.0 if greater than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the greater than zero comparison. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxZeroOneIsGreaterThanZeroHalf(FfxFloat16x2 x) -{ - return FfxFloat16x2(ffxSaturate(x * FFX_BROADCAST_FLOAT16X2(FFX_POSITIVE_INFINITY_HALF))); -} - -/// Given a half-precision value, returns 1.0 if greater than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the greater than zero comparison. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxZeroOneIsGreaterThanZeroHalf(FfxFloat16x3 x) -{ - return FfxFloat16x3(ffxSaturate(x * FFX_BROADCAST_FLOAT16X3(FFX_POSITIVE_INFINITY_HALF))); -} - -/// Given a half-precision value, returns 1.0 if greater than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the greater than zero comparison. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxZeroOneIsGreaterThanZeroHalf(FfxFloat16x4 x) -{ - return FfxFloat16x4(ffxSaturate(x * FFX_BROADCAST_FLOAT16X4(FFX_POSITIVE_INFINITY_HALF))); -} - -/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat16 ffxZeroOneNotHalf(FfxFloat16 x) -{ - return FFX_BROADCAST_FLOAT16(1.0) - x; -} - -/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxZeroOneNotHalf(FfxFloat16x2 x) -{ - return FFX_BROADCAST_FLOAT16X2(1.0) - x; -} - -/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxZeroOneNotHalf(FfxFloat16x3 x) -{ - return FFX_BROADCAST_FLOAT16X3(1.0) - x; -} - -/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxZeroOneNotHalf(FfxFloat16x4 x) -{ - return FFX_BROADCAST_FLOAT16X4(1.0) - x; -} - -/// Conditional free logic OR operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxFloat16 ffxZeroOneOrHalf(FfxFloat16 x, FfxFloat16 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxZeroOneOrHalf(FfxFloat16x2 x, FfxFloat16x2 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxZeroOneOrHalf(FfxFloat16x3 x, FfxFloat16x3 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxZeroOneOrHalf(FfxFloat16x4 x, FfxFloat16x4 y) -{ - return max(x, y); -} - -/// Choose between two half-precision FfxFloat32 values if the first paramter is greater than zero. -/// -/// @param [in] x The value to compare against zero. -/// @param [in] y The value to return if the comparision is greater than zero. -/// @param [in] z The value to return if the comparision is less than or equal to zero. -/// -/// @returns -/// The selected value. -/// -/// @ingroup GPUCore -FfxFloat16 ffxZeroOneSelectHalf(FfxFloat16 x, FfxFloat16 y, FfxFloat16 z) -{ - FfxFloat16 r = (-x) * z + z; - return x * y + r; -} - -/// Choose between two half-precision FfxFloat32 values if the first paramter is greater than zero. -/// -/// @param [in] x The value to compare against zero. -/// @param [in] y The value to return if the comparision is greater than zero. -/// @param [in] z The value to return if the comparision is less than or equal to zero. -/// -/// @returns -/// The selected value. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxZeroOneSelectHalf(FfxFloat16x2 x, FfxFloat16x2 y, FfxFloat16x2 z) -{ - FfxFloat16x2 r = (-x) * z + z; - return x * y + r; -} - -/// Choose between two half-precision FfxFloat32 values if the first paramter is greater than zero. -/// -/// @param [in] x The value to compare against zero. -/// @param [in] y The value to return if the comparision is greater than zero. -/// @param [in] z The value to return if the comparision is less than or equal to zero. -/// -/// @returns -/// The selected value. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxZeroOneSelectHalf(FfxFloat16x3 x, FfxFloat16x3 y, FfxFloat16x3 z) -{ - FfxFloat16x3 r = (-x) * z + z; - return x * y + r; -} - -/// Choose between two half-precision FfxFloat32 values if the first paramter is greater than zero. -/// -/// @param [in] x The value to compare against zero. -/// @param [in] y The value to return if the comparision is greater than zero. -/// @param [in] z The value to return if the comparision is less than or equal to zero. -/// -/// @returns -/// The selected value. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxZeroOneSelectHalf(FfxFloat16x4 x, FfxFloat16x4 y, FfxFloat16x4 z) -{ - FfxFloat16x4 r = (-x) * z + z; - return x * y + r; -} - -/// Given a half-precision value, returns 1.0 if less than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the sign value. -/// -/// @ingroup GPUCore -FfxFloat16 ffxZeroOneIsSignedHalf(FfxFloat16 x) -{ - return FfxFloat16(ffxSaturate(x * FFX_BROADCAST_FLOAT16(FFX_NEGATIVE_INFINITY_HALF))); -} - -/// Given a half-precision value, returns 1.0 if less than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the sign value. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxZeroOneIsSignedHalf(FfxFloat16x2 x) -{ - return FfxFloat16x2(ffxSaturate(x * FFX_BROADCAST_FLOAT16X2(FFX_NEGATIVE_INFINITY_HALF))); -} - -/// Given a half-precision value, returns 1.0 if less than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the sign value. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxZeroOneIsSignedHalf(FfxFloat16x3 x) -{ - return FfxFloat16x3(ffxSaturate(x * FFX_BROADCAST_FLOAT16X3(FFX_NEGATIVE_INFINITY_HALF))); -} - -/// Given a half-precision value, returns 1.0 if less than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the sign value. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxZeroOneIsSignedHalf(FfxFloat16x4 x) -{ - return FfxFloat16x4(ffxSaturate(x * FFX_BROADCAST_FLOAT16X4(FFX_NEGATIVE_INFINITY_HALF))); -} - -/// Compute a Rec.709 color space. -/// -/// Rec.709 is used for some HDTVs. -/// -/// Both Rec.709 and sRGB have a linear segment which as spec'ed would intersect the curved segment 2 times. -/// (a.) For 8-bit sRGB, steps {0 to 10.3} are in the linear region (4% of the encoding range). -/// (b.) For 8-bit 709, steps {0 to 20.7} are in the linear region (8% of the encoding range). -/// -/// @param [in] c The color to convert to Rec. 709. -/// -/// @returns -/// The color in Rec.709 space. -/// -/// @ingroup GPUCore -FfxFloat16 ffxRec709FromLinearHalf(FfxFloat16 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.018 * 4.5, 4.5, 0.45); - FfxFloat16x2 k = FfxFloat16x2(1.099, -0.099); - return clamp(j.x, c * j.y, pow(c, j.z) * k.x + k.y); -} - -/// Compute a Rec.709 color space. -/// -/// Rec.709 is used for some HDTVs. -/// -/// Both Rec.709 and sRGB have a linear segment which as spec'ed would intersect the curved segment 2 times. -/// (a.) For 8-bit sRGB, steps {0 to 10.3} are in the linear region (4% of the encoding range). -/// (b.) For 8-bit 709, steps {0 to 20.7} are in the linear region (8% of the encoding range). -/// -/// @param [in] c The color to convert to Rec. 709. -/// -/// @returns -/// The color in Rec.709 space. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxRec709FromLinearHalf(FfxFloat16x2 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.018 * 4.5, 4.5, 0.45); - FfxFloat16x2 k = FfxFloat16x2(1.099, -0.099); - return clamp(j.xx, c * j.yy, pow(c, j.zz) * k.xx + k.yy); -} - -/// Compute a Rec.709 color space. -/// -/// Rec.709 is used for some HDTVs. -/// -/// Both Rec.709 and sRGB have a linear segment which as spec'ed would intersect the curved segment 2 times. -/// (a.) For 8-bit sRGB, steps {0 to 10.3} are in the linear region (4% of the encoding range). -/// (b.) For 8-bit 709, steps {0 to 20.7} are in the linear region (8% of the encoding range). -/// -/// @param [in] c The color to convert to Rec. 709. -/// -/// @returns -/// The color in Rec.709 space. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxRec709FromLinearHalf(FfxFloat16x3 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.018 * 4.5, 4.5, 0.45); - FfxFloat16x2 k = FfxFloat16x2(1.099, -0.099); - return clamp(j.xxx, c * j.yyy, pow(c, j.zzz) * k.xxx + k.yyy); -} - -/// Compute a gamma value from a linear value. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// Note: 'rcpX' is '1/x', where the 'x' is what would be used in ffxLinearFromGammaHalf. -/// -/// @param [in] c The value to convert to gamma space from linear. -/// @param [in] rcpX The reciprocal of power value used for the gamma curve. -/// -/// @returns -/// A value in gamma space. -/// -/// @ingroup GPUCore -FfxFloat16 ffxGammaFromLinearHalf(FfxFloat16 c, FfxFloat16 rcpX) -{ - return pow(c, FFX_BROADCAST_FLOAT16(rcpX)); -} - -/// Compute a gamma value from a linear value. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// Note: 'rcpX' is '1/x', where the 'x' is what would be used in ffxLinearFromGammaHalf. -/// -/// @param [in] c The value to convert to gamma space from linear. -/// @param [in] rcpX The reciprocal of power value used for the gamma curve. -/// -/// @returns -/// A value in gamma space. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxGammaFromLinearHalf(FfxFloat16x2 c, FfxFloat16 rcpX) -{ - return pow(c, FFX_BROADCAST_FLOAT16X2(rcpX)); -} - -/// Compute a gamma value from a linear value. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// Note: 'rcpX' is '1/x', where the 'x' is what would be used in ffxLinearFromGammaHalf. -/// -/// @param [in] c The value to convert to gamma space from linear. -/// @param [in] rcpX The reciprocal of power value used for the gamma curve. -/// -/// @returns -/// A value in gamma space. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxGammaFromLinearHalf(FfxFloat16x3 c, FfxFloat16 rcpX) -{ - return pow(c, FFX_BROADCAST_FLOAT16X3(rcpX)); -} - -/// Compute an SRGB value from a linear value. -/// -/// @param [in] c The value to convert to SRGB from linear. -/// -/// @returns -/// A value in SRGB space. -/// -/// @ingroup GPUCore -FfxFloat16 ffxSrgbFromLinearHalf(FfxFloat16 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.0031308 * 12.92, 12.92, 1.0 / 2.4); - FfxFloat16x2 k = FfxFloat16x2(1.055, -0.055); - return clamp(j.x, c * j.y, pow(c, j.z) * k.x + k.y); -} - -/// Compute an SRGB value from a linear value. -/// -/// @param [in] c The value to convert to SRGB from linear. -/// -/// @returns -/// A value in SRGB space. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxSrgbFromLinearHalf(FfxFloat16x2 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.0031308 * 12.92, 12.92, 1.0 / 2.4); - FfxFloat16x2 k = FfxFloat16x2(1.055, -0.055); - return clamp(j.xx, c * j.yy, pow(c, j.zz) * k.xx + k.yy); -} - -/// Compute an SRGB value from a linear value. -/// -/// @param [in] c The value to convert to SRGB from linear. -/// -/// @returns -/// A value in SRGB space. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxSrgbFromLinearHalf(FfxFloat16x3 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.0031308 * 12.92, 12.92, 1.0 / 2.4); - FfxFloat16x2 k = FfxFloat16x2(1.055, -0.055); - return clamp(j.xxx, c * j.yyy, pow(c, j.zzz) * k.xxx + k.yyy); -} - -/// Compute the square root of a value. -/// -/// @param [in] c The value to compute the square root for. -/// -/// @returns -/// A square root of the input value. -/// -/// @ingroup GPUCore -FfxFloat16 ffxSquareRootHalf(FfxFloat16 c) -{ - return sqrt(c); -} - -/// Compute the square root of a value. -/// -/// @param [in] c The value to compute the square root for. -/// -/// @returns -/// A square root of the input value. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxSquareRootHalf(FfxFloat16x2 c) -{ - return sqrt(c); -} - -/// Compute the square root of a value. -/// -/// @param [in] c The value to compute the square root for. -/// -/// @returns -/// A square root of the input value. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxSquareRootHalf(FfxFloat16x3 c) -{ - return sqrt(c); -} - -/// Compute the cube root of a value. -/// -/// @param [in] c The value to compute the cube root for. -/// -/// @returns -/// A cube root of the input value. -/// -/// @ingroup GPUCore -FfxFloat16 ffxCubeRootHalf(FfxFloat16 c) -{ - return pow(c, FFX_BROADCAST_FLOAT16(1.0 / 3.0)); -} - -/// Compute the cube root of a value. -/// -/// @param [in] c The value to compute the cube root for. -/// -/// @returns -/// A cube root of the input value. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxCubeRootHalf(FfxFloat16x2 c) -{ - return pow(c, FFX_BROADCAST_FLOAT16X2(1.0 / 3.0)); -} - -/// Compute the cube root of a value. -/// -/// @param [in] c The value to compute the cube root for. -/// -/// @returns -/// A cube root of the input value. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxCubeRootHalf(FfxFloat16x3 c) -{ - return pow(c, FFX_BROADCAST_FLOAT16X3(1.0 / 3.0)); -} - -/// Compute a linear value from a REC.709 value. -/// -/// @param [in] c The value to convert to linear from REC.709. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat16 ffxLinearFromRec709Half(FfxFloat16 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.081 / 4.5, 1.0 / 4.5, 1.0 / 0.45); - FfxFloat16x2 k = FfxFloat16x2(1.0 / 1.099, 0.099 / 1.099); - return ffxZeroOneSelectHalf(ffxZeroOneIsSignedHalf(c - j.x), c * j.y, pow(c * k.x + k.y, j.z)); -} - -/// Compute a linear value from a REC.709 value. -/// -/// @param [in] c The value to convert to linear from REC.709. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxLinearFromRec709Half(FfxFloat16x2 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.081 / 4.5, 1.0 / 4.5, 1.0 / 0.45); - FfxFloat16x2 k = FfxFloat16x2(1.0 / 1.099, 0.099 / 1.099); - return ffxZeroOneSelectHalf(ffxZeroOneIsSignedHalf(c - j.xx), c * j.yy, pow(c * k.xx + k.yy, j.zz)); -} - -/// Compute a linear value from a REC.709 value. -/// -/// @param [in] c The value to convert to linear from REC.709. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxLinearFromRec709Half(FfxFloat16x3 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.081 / 4.5, 1.0 / 4.5, 1.0 / 0.45); - FfxFloat16x2 k = FfxFloat16x2(1.0 / 1.099, 0.099 / 1.099); - return ffxZeroOneSelectHalf(ffxZeroOneIsSignedHalf(c - j.xxx), c * j.yyy, pow(c * k.xxx + k.yyy, j.zzz)); -} - -/// Compute a linear value from a value in a gamma space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] c The value to convert to linear in gamma space. -/// @param [in] x The power value used for the gamma curve. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat16 ffxLinearFromGammaHalf(FfxFloat16 c, FfxFloat16 x) -{ - return pow(c, FFX_BROADCAST_FLOAT16(x)); -} - -/// Compute a linear value from a value in a gamma space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] c The value to convert to linear in gamma space. -/// @param [in] x The power value used for the gamma curve. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxLinearFromGammaHalf(FfxFloat16x2 c, FfxFloat16 x) -{ - return pow(c, FFX_BROADCAST_FLOAT16X2(x)); -} - -/// Compute a linear value from a value in a gamma space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] c The value to convert to linear in gamma space. -/// @param [in] x The power value used for the gamma curve. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxLinearFromGammaHalf(FfxFloat16x3 c, FfxFloat16 x) -{ - return pow(c, FFX_BROADCAST_FLOAT16X3(x)); -} - -/// Compute a linear value from a value in a SRGB space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] c The value to convert to linear in SRGB space. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat16 ffxLinearFromSrgbHalf(FfxFloat16 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.04045 / 12.92, 1.0 / 12.92, 2.4); - FfxFloat16x2 k = FfxFloat16x2(1.0 / 1.055, 0.055 / 1.055); - return ffxZeroOneSelectHalf(ffxZeroOneIsSignedHalf(c - j.x), c * j.y, pow(c * k.x + k.y, j.z)); -} - -/// Compute a linear value from a value in a SRGB space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] c The value to convert to linear in SRGB space. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxLinearFromSrgbHalf(FfxFloat16x2 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.04045 / 12.92, 1.0 / 12.92, 2.4); - FfxFloat16x2 k = FfxFloat16x2(1.0 / 1.055, 0.055 / 1.055); - return ffxZeroOneSelectHalf(ffxZeroOneIsSignedHalf(c - j.xx), c * j.yy, pow(c * k.xx + k.yy, j.zz)); -} - -/// Compute a linear value from a value in a SRGB space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] c The value to convert to linear in SRGB space. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxLinearFromSrgbHalf(FfxFloat16x3 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.04045 / 12.92, 1.0 / 12.92, 2.4); - FfxFloat16x2 k = FfxFloat16x2(1.0 / 1.055, 0.055 / 1.055); - return ffxZeroOneSelectHalf(ffxZeroOneIsSignedHalf(c - j.xxx), c * j.yyy, pow(c * k.xxx + k.yyy, j.zzz)); -} - -/// A remapping of 64x1 to 8x8 imposing rotated 2x2 pixel quads in quad linear. -/// -/// 543210 -/// ====== -/// ..xxx. -/// yy...y -/// -/// @param [in] a The input 1D coordinates to remap. -/// -/// @returns -/// The remapped 2D coordinates. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxRemapForQuadHalf(FfxUInt32 a) -{ - return FfxUInt16x2(bitfieldExtract(a, 1u, 3u), bitfieldInsertMask(bitfieldExtract(a, 3u, 3u), a, 1u)); -} - -/// A helper function performing a remap 64x1 to 8x8 remapping which is necessary for 2D wave reductions. -/// -/// The 64-wide lane indices to 8x8 remapping is performed as follows: -/// -/// 00 01 08 09 10 11 18 19 -/// 02 03 0a 0b 12 13 1a 1b -/// 04 05 0c 0d 14 15 1c 1d -/// 06 07 0e 0f 16 17 1e 1f -/// 20 21 28 29 30 31 38 39 -/// 22 23 2a 2b 32 33 3a 3b -/// 24 25 2c 2d 34 35 3c 3d -/// 26 27 2e 2f 36 37 3e 3f -/// -/// @param [in] a The input 1D coordinate to remap. -/// -/// @returns -/// The remapped 2D coordinates. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxRemapForWaveReductionHalf(FfxUInt32 a) -{ - return FfxUInt16x2(bitfieldInsertMask(bitfieldExtract(a, 2u, 3u), a, 1u), bitfieldInsertMask(bitfieldExtract(a, 3u, 3u), bitfieldExtract(a, 1u, 2u), 2u)); -} - -#endif // FFX_HALF diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common_half.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common_half.h.meta deleted file mode 100644 index 234aa90..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common_half.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 1bdb323791a91a5438ee8e1e63187840 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_hlsl.h b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_hlsl.h deleted file mode 100644 index 337eb06..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_hlsl.h +++ /dev/null @@ -1,1651 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -/// @defgroup HLSLCore HLSL Core -/// HLSL core defines and functions -/// -/// @ingroup FfxHLSL - -#define DECLARE_SRV_REGISTER(regIndex) t##regIndex -#define DECLARE_UAV_REGISTER(regIndex) u##regIndex -#define DECLARE_CB_REGISTER(regIndex) b##regIndex -#define FFX_DECLARE_SRV(regIndex) register(DECLARE_SRV_REGISTER(regIndex)) -#define FFX_DECLARE_UAV(regIndex) register(DECLARE_UAV_REGISTER(regIndex)) -#define FFX_DECLARE_CB(regIndex) register(DECLARE_CB_REGISTER(regIndex)) - -/// A define for abstracting shared memory between shading languages. -/// -/// @ingroup HLSLCore -#define FFX_GROUPSHARED groupshared - -/// A define for abstracting compute memory barriers between shading languages. -/// -/// @ingroup HLSLCore -#define FFX_GROUP_MEMORY_BARRIER GroupMemoryBarrierWithGroupSync - -/// A define for abstracting compute atomic additions between shading languages. -/// -/// @ingroup HLSLCore -#define FFX_ATOMIC_ADD(x, y) InterlockedAdd(x, y) - -/// A define added to accept static markup on functions to aid CPU/GPU portability of code. -/// -/// @ingroup HLSLCore -#define FFX_STATIC static - -/// A define for abstracting loop unrolling between shading languages. -/// -/// @ingroup HLSLCore -#define FFX_UNROLL [unroll] - -/// A define for abstracting a 'greater than' comparison operator between two types. -/// -/// @ingroup HLSLCore -#define FFX_GREATER_THAN(x, y) x > y - -/// A define for abstracting a 'greater than or equal' comparison operator between two types. -/// -/// @ingroup HLSLCore -#define FFX_GREATER_THAN_EQUAL(x, y) x >= y - -/// A define for abstracting a 'less than' comparison operator between two types. -/// -/// @ingroup HLSLCore -#define FFX_LESS_THAN(x, y) x < y - -/// A define for abstracting a 'less than or equal' comparison operator between two types. -/// -/// @ingroup HLSLCore -#define FFX_LESS_THAN_EQUAL(x, y) x <= y - -/// A define for abstracting an 'equal' comparison operator between two types. -/// -/// @ingroup HLSLCore -#define FFX_EQUAL(x, y) x == y - -/// A define for abstracting a 'not equal' comparison operator between two types. -/// -/// @ingroup HLSLCore -#define FFX_NOT_EQUAL(x, y) x != y - -/// A define for abstracting matrix multiply operations between shading languages. -/// -/// @ingroup HLSLCore -#define FFX_MATRIX_MULTIPLY(a, b) mul(a, b) - -/// A define for abstracting vector transformations between shading languages. -/// -/// @ingroup HLSLCore -#define FFX_TRANSFORM_VECTOR(a, b) mul(a, b) - -/// A define for abstracting modulo operations between shading languages. -/// -/// @ingroup HLSLCore -#define FFX_MODULO(a, b) (fmod(a, b)) - -/// Broadcast a scalar value to a 1-dimensional floating point vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_FLOAT32(x) FfxFloat32(x) - -/// Broadcast a scalar value to a 2-dimensional floating point vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_FLOAT32X2(x) FfxFloat32(x) - -/// Broadcast a scalar value to a 3-dimensional floating point vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_FLOAT32X3(x) FfxFloat32(x) - -/// Broadcast a scalar value to a 4-dimensional floating point vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_FLOAT32X4(x) FfxFloat32(x) - -/// Broadcast a scalar value to a 1-dimensional unsigned integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_UINT32(x) FfxUInt32(x) - -/// Broadcast a scalar value to a 2-dimensional unsigned integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_UINT32X2(x) FfxUInt32(x) - -/// Broadcast a scalar value to a 4-dimensional unsigned integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_UINT32X3(x) FfxUInt32(x) - -/// Broadcast a scalar value to a 4-dimensional unsigned integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_UINT32X4(x) FfxUInt32(x) - -/// Broadcast a scalar value to a 1-dimensional signed integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_INT32(x) FfxInt32(x) - -/// Broadcast a scalar value to a 2-dimensional signed integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_INT32X2(x) FfxInt32(x) - -/// Broadcast a scalar value to a 3-dimensional signed integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_INT32X3(x) FfxInt32(x) - -/// Broadcast a scalar value to a 4-dimensional signed integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_INT32X4(x) FfxInt32(x) - -/// Broadcast a scalar value to a 1-dimensional half-precision floating point vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_FLOAT16(a) FFX_MIN16_F(a) - -/// Broadcast a scalar value to a 2-dimensional half-precision floating point vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_FLOAT16X2(a) FFX_MIN16_F(a) - -/// Broadcast a scalar value to a 3-dimensional half-precision floating point vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_FLOAT16X3(a) FFX_MIN16_F(a) - -/// Broadcast a scalar value to a 4-dimensional half-precision floating point vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_FLOAT16X4(a) FFX_MIN16_F(a) - -/// Broadcast a scalar value to a 1-dimensional half-precision unsigned integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_UINT16(a) FFX_MIN16_U(a) - -/// Broadcast a scalar value to a 2-dimensional half-precision unsigned integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_UINT16X2(a) FFX_MIN16_U(a) - -/// Broadcast a scalar value to a 3-dimensional half-precision unsigned integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_UINT16X3(a) FFX_MIN16_U(a) - -/// Broadcast a scalar value to a 4-dimensional half-precision unsigned integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_UINT16X4(a) FFX_MIN16_U(a) - -/// Broadcast a scalar value to a 1-dimensional half-precision signed integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_INT16(a) FFX_MIN16_I(a) - -/// Broadcast a scalar value to a 2-dimensional half-precision signed integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_INT16X2(a) FFX_MIN16_I(a) - -/// Broadcast a scalar value to a 3-dimensional half-precision signed integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_INT16X3(a) FFX_MIN16_I(a) - -/// Broadcast a scalar value to a 4-dimensional half-precision signed integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_INT16X4(a) FFX_MIN16_I(a) - -/// Pack 2x32-bit floating point values in a single 32bit value. -/// -/// This function first converts each component of value into their nearest 16-bit floating -/// point representation, and then stores the X and Y components in the lower and upper 16 bits of the -/// 32bit unsigned integer respectively. -/// -/// @param [in] value A 2-dimensional floating point value to convert and pack. -/// -/// @returns -/// A packed 32bit value containing 2 16bit floating point values. -/// -/// @ingroup HLSLCore -FfxUInt32 packHalf2x16(FfxFloat32x2 value) -{ - return f32tof16(value.x) | (f32tof16(value.y) << 16); -} - -/// Broadcast a scalar value to a 2-dimensional floating point vector. -/// -/// @param [in] value The value to to broadcast. -/// -/// @returns -/// A 2-dimensional floating point vector with value in each component. -/// -/// @ingroup HLSLCore -FfxFloat32x2 ffxBroadcast2(FfxFloat32 value) -{ - return FfxFloat32x2(value, value); -} - -/// Broadcast a scalar value to a 3-dimensional floating point vector. -/// -/// @param [in] value The value to to broadcast. -/// -/// @returns -/// A 3-dimensional floating point vector with value in each component. -/// -/// @ingroup HLSLCore -FfxFloat32x3 ffxBroadcast3(FfxFloat32 value) -{ - return FfxFloat32x3(value, value, value); -} - -/// Broadcast a scalar value to a 4-dimensional floating point vector. -/// -/// @param [in] value The value to to broadcast. -/// -/// @returns -/// A 4-dimensional floating point vector with value in each component. -/// -/// @ingroup HLSLCore -FfxFloat32x4 ffxBroadcast4(FfxFloat32 value) -{ - return FfxFloat32x4(value, value, value, value); -} - -/// Broadcast a scalar value to a 2-dimensional signed integer vector. -/// -/// @param [in] value The value to to broadcast. -/// -/// @returns -/// A 2-dimensional signed integer vector with value in each component. -/// -/// @ingroup HLSLCore -FfxInt32x2 ffxBroadcast2(FfxInt32 value) -{ - return FfxInt32x2(value, value); -} - -/// Broadcast a scalar value to a 3-dimensional signed integer vector. -/// -/// @param [in] value The value to to broadcast. -/// -/// @returns -/// A 3-dimensional signed integer vector with value in each component. -/// -/// @ingroup HLSLCore -FfxUInt32x3 ffxBroadcast3(FfxInt32 value) -{ - return FfxUInt32x3(value, value, value); -} - -/// Broadcast a scalar value to a 4-dimensional signed integer vector. -/// -/// @param [in] value The value to to broadcast. -/// -/// @returns -/// A 4-dimensional signed integer vector with value in each component. -/// -/// @ingroup HLSLCore -FfxInt32x4 ffxBroadcast4(FfxInt32 value) -{ - return FfxInt32x4(value, value, value, value); -} - -/// Broadcast a scalar value to a 2-dimensional unsigned integer vector. -/// -/// @param [in] value The value to to broadcast. -/// -/// @returns -/// A 2-dimensional unsigned integer vector with value in each component. -/// -/// @ingroup HLSLCore -FfxUInt32x2 ffxBroadcast2(FfxUInt32 value) -{ - return FfxUInt32x2(value, value); -} - -/// Broadcast a scalar value to a 3-dimensional unsigned integer vector. -/// -/// @param [in] value The value to to broadcast. -/// -/// @returns -/// A 3-dimensional unsigned integer vector with value in each component. -/// -/// @ingroup HLSLCore -FfxUInt32x3 ffxBroadcast3(FfxUInt32 value) -{ - return FfxUInt32x3(value, value, value); -} - -/// Broadcast a scalar value to a 4-dimensional unsigned integer vector. -/// -/// @param [in] value The value to to broadcast. -/// -/// @returns -/// A 4-dimensional unsigned integer vector with value in each component. -/// -/// @ingroup HLSLCore -FfxUInt32x4 ffxBroadcast4(FfxUInt32 value) -{ - return FfxUInt32x4(value, value, value, value); -} - -FfxUInt32 bitfieldExtract(FfxUInt32 src, FfxUInt32 off, FfxUInt32 bits) -{ - FfxUInt32 mask = (1u << bits) - 1; - return (src >> off) & mask; -} - -FfxUInt32 bitfieldInsert(FfxUInt32 src, FfxUInt32 ins, FfxUInt32 mask) -{ - return (ins & mask) | (src & (~mask)); -} - -FfxUInt32 bitfieldInsertMask(FfxUInt32 src, FfxUInt32 ins, FfxUInt32 bits) -{ - FfxUInt32 mask = (1u << bits) - 1; - return (ins & mask) | (src & (~mask)); -} - -/// Interprets the bit pattern of x as an unsigned integer. -/// -/// @param [in] x The input value. -/// -/// @returns -/// The input interpreted as an unsigned integer. -/// -/// @ingroup HLSLCore -FfxUInt32 ffxAsUInt32(FfxFloat32 x) -{ - return asuint(x); -} - -/// Interprets the bit pattern of x as an unsigned integer. -/// -/// @param [in] x The input value. -/// -/// @returns -/// The input interpreted as an unsigned integer. -/// -/// @ingroup HLSLCore -FfxUInt32x2 ffxAsUInt32(FfxFloat32x2 x) -{ - return asuint(x); -} - -/// Interprets the bit pattern of x as an unsigned integer. -/// -/// @param [in] x The input value. -/// -/// @returns -/// The input interpreted as an unsigned integer. -/// -/// @ingroup HLSLCore -FfxUInt32x3 ffxAsUInt32(FfxFloat32x3 x) -{ - return asuint(x); -} - -/// Interprets the bit pattern of x as an unsigned integer. -/// -/// @param [in] x The input value. -/// -/// @returns -/// The input interpreted as an unsigned integer. -/// -/// @ingroup HLSLCore -FfxUInt32x4 ffxAsUInt32(FfxFloat32x4 x) -{ - return asuint(x); -} - -/// Interprets the bit pattern of x as a floating-point number. -/// -/// @param [in] x The input value. -/// -/// @returns -/// The input interpreted as a floating-point number. -/// -/// @ingroup HLSLCore -FfxFloat32 ffxAsFloat(FfxUInt32 x) -{ - return asfloat(x); -} - -/// Interprets the bit pattern of x as a floating-point number. -/// -/// @param [in] x The input value. -/// -/// @returns -/// The input interpreted as a floating-point number. -/// -/// @ingroup HLSLCore -FfxFloat32x2 ffxAsFloat(FfxUInt32x2 x) -{ - return asfloat(x); -} - -/// Interprets the bit pattern of x as a floating-point number. -/// -/// @param [in] x The input value. -/// -/// @returns -/// The input interpreted as a floating-point number. -/// -/// @ingroup HLSLCore -FfxFloat32x3 ffxAsFloat(FfxUInt32x3 x) -{ - return asfloat(x); -} - -/// Interprets the bit pattern of x as a floating-point number. -/// -/// @param [in] x The input value. -/// -/// @returns -/// The input interpreted as a floating-point number. -/// -/// @ingroup HLSLCore -FfxFloat32x4 ffxAsFloat(FfxUInt32x4 x) -{ - return asfloat(x); -} - -/// Compute the linear interopation between two values. -/// -/// Implemented by calling the HLSL mix instrinsic function. Implements the -/// following math: -/// -/// (1 - t) * x + t * y -/// -/// @param [in] x The first value to lerp between. -/// @param [in] y The second value to lerp between. -/// @param [in] t The value to determine how much of x and how much of y. -/// -/// @returns -/// A linearly interpolated value between x and y according to t. -/// -/// @ingroup HLSLCore -FfxFloat32 ffxLerp(FfxFloat32 x, FfxFloat32 y, FfxFloat32 t) -{ - return lerp(x, y, t); -} - -/// Compute the linear interopation between two values. -/// -/// Implemented by calling the HLSL mix instrinsic function. Implements the -/// following math: -/// -/// (1 - t) * x + t * y -/// -/// @param [in] x The first value to lerp between. -/// @param [in] y The second value to lerp between. -/// @param [in] t The value to determine how much of x and how much of y. -/// -/// @returns -/// A linearly interpolated value between x and y according to t. -/// -/// @ingroup HLSLCore -FfxFloat32x2 ffxLerp(FfxFloat32x2 x, FfxFloat32x2 y, FfxFloat32 t) -{ - return lerp(x, y, t); -} - -/// Compute the linear interopation between two values. -/// -/// Implemented by calling the HLSL mix instrinsic function. Implements the -/// following math: -/// -/// (1 - t) * x + t * y -/// -/// @param [in] x The first value to lerp between. -/// @param [in] y The second value to lerp between. -/// @param [in] t The value to determine how much of x and how much of y. -/// -/// @returns -/// A linearly interpolated value between x and y according to t. -/// -/// @ingroup HLSLCore -FfxFloat32x2 ffxLerp(FfxFloat32x2 x, FfxFloat32x2 y, FfxFloat32x2 t) -{ - return lerp(x, y, t); -} - -/// Compute the linear interopation between two values. -/// -/// Implemented by calling the HLSL mix instrinsic function. Implements the -/// following math: -/// -/// (1 - t) * x + t * y -/// -/// @param [in] x The first value to lerp between. -/// @param [in] y The second value to lerp between. -/// @param [in] t The value to determine how much of x and how much of y. -/// -/// @returns -/// A linearly interpolated value between x and y according to t. -/// -/// @ingroup HLSLCore -FfxFloat32x3 ffxLerp(FfxFloat32x3 x, FfxFloat32x3 y, FfxFloat32 t) -{ - return lerp(x, y, t); -} - -/// Compute the linear interopation between two values. -/// -/// Implemented by calling the HLSL mix instrinsic function. Implements the -/// following math: -/// -/// (1 - t) * x + t * y -/// -/// @param [in] x The first value to lerp between. -/// @param [in] y The second value to lerp between. -/// @param [in] t The value to determine how much of x and how much of y. -/// -/// @returns -/// A linearly interpolated value between x and y according to t. -/// -/// @ingroup HLSLCore -FfxFloat32x3 ffxLerp(FfxFloat32x3 x, FfxFloat32x3 y, FfxFloat32x3 t) -{ - return lerp(x, y, t); -} - -/// Compute the linear interopation between two values. -/// -/// Implemented by calling the HLSL mix instrinsic function. Implements the -/// following math: -/// -/// (1 - t) * x + t * y -/// -/// @param [in] x The first value to lerp between. -/// @param [in] y The second value to lerp between. -/// @param [in] t The value to determine how much of x and how much of y. -/// -/// @returns -/// A linearly interpolated value between x and y according to t. -/// -/// @ingroup HLSLCore -FfxFloat32x4 ffxLerp(FfxFloat32x4 x, FfxFloat32x4 y, FfxFloat32 t) -{ - return lerp(x, y, t); -} - -/// Compute the linear interopation between two values. -/// -/// Implemented by calling the HLSL mix instrinsic function. Implements the -/// following math: -/// -/// (1 - t) * x + t * y -/// -/// @param [in] x The first value to lerp between. -/// @param [in] y The second value to lerp between. -/// @param [in] t The value to determine how much of x and how much of y. -/// -/// @returns -/// A linearly interpolated value between x and y according to t. -/// -/// @ingroup HLSLCore -FfxFloat32x4 ffxLerp(FfxFloat32x4 x, FfxFloat32x4 y, FfxFloat32x4 t) -{ - return lerp(x, y, t); -} - -/// Clamp a value to a [0..1] range. -/// -/// @param [in] x The value to clamp to [0..1] range. -/// -/// @returns -/// The clamped version of x. -/// -/// @ingroup HLSLCore -FfxFloat32 ffxSaturate(FfxFloat32 x) -{ - return saturate(x); -} - -/// Clamp a value to a [0..1] range. -/// -/// @param [in] x The value to clamp to [0..1] range. -/// -/// @returns -/// The clamped version of x. -/// -/// @ingroup HLSLCore -FfxFloat32x2 ffxSaturate(FfxFloat32x2 x) -{ - return saturate(x); -} - -/// Clamp a value to a [0..1] range. -/// -/// @param [in] x The value to clamp to [0..1] range. -/// -/// @returns -/// The clamped version of x. -/// -/// @ingroup HLSLCore -FfxFloat32x3 ffxSaturate(FfxFloat32x3 x) -{ - return saturate(x); -} - -/// Clamp a value to a [0..1] range. -/// -/// @param [in] x The value to clamp to [0..1] range. -/// -/// @returns -/// The clamped version of x. -/// -/// @ingroup HLSLCore -FfxFloat32x4 ffxSaturate(FfxFloat32x4 x) -{ - return saturate(x); -} - -/// Compute the factional part of a decimal value. -/// -/// This function calculates x - floor(x). Where floor is the intrinsic HLSL function. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. It is -/// worth further noting that this function is intentionally distinct from the HLSL frac intrinsic -/// function. -/// -/// @param [in] x The value to compute the fractional part from. -/// -/// @returns -/// The fractional part of x. -/// -/// @ingroup HLSLCore -FfxFloat32 ffxFract(FfxFloat32 x) -{ - return x - floor(x); -} - -/// Compute the factional part of a decimal value. -/// -/// This function calculates x - floor(x). Where floor is the intrinsic HLSL function. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. It is -/// worth further noting that this function is intentionally distinct from the HLSL frac intrinsic -/// function. -/// -/// @param [in] x The value to compute the fractional part from. -/// -/// @returns -/// The fractional part of x. -/// -/// @ingroup HLSLCore -FfxFloat32x2 ffxFract(FfxFloat32x2 x) -{ - return x - floor(x); -} - -/// Compute the factional part of a decimal value. -/// -/// This function calculates x - floor(x). Where floor is the intrinsic HLSL function. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. It is -/// worth further noting that this function is intentionally distinct from the HLSL frac intrinsic -/// function. -/// -/// @param [in] x The value to compute the fractional part from. -/// -/// @returns -/// The fractional part of x. -/// -/// @ingroup HLSLCore -FfxFloat32x3 ffxFract(FfxFloat32x3 x) -{ - return x - floor(x); -} - -/// Compute the factional part of a decimal value. -/// -/// This function calculates x - floor(x). Where floor is the intrinsic HLSL function. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. It is -/// worth further noting that this function is intentionally distinct from the HLSL frac intrinsic -/// function. -/// -/// @param [in] x The value to compute the fractional part from. -/// -/// @returns -/// The fractional part of x. -/// -/// @ingroup HLSLCore -FfxFloat32x4 ffxFract(FfxFloat32x4 x) -{ - return x - floor(x); -} - -/// Compute the maximum of three values. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the max calculation. -/// @param [in] y The second value to include in the max calcuation. -/// @param [in] z The third value to include in the max calcuation. -/// -/// @returns -/// The maximum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32 ffxMax3(FfxFloat32 x, FfxFloat32 y, FfxFloat32 z) -{ - return max(x, max(y, z)); -} - -/// Compute the maximum of three values. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the max calculation. -/// @param [in] y The second value to include in the max calcuation. -/// @param [in] z The third value to include in the max calcuation. -/// -/// @returns -/// The maximum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32x2 ffxMax3(FfxFloat32x2 x, FfxFloat32x2 y, FfxFloat32x2 z) -{ - return max(x, max(y, z)); -} - -/// Compute the maximum of three values. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the max calculation. -/// @param [in] y The second value to include in the max calcuation. -/// @param [in] z The third value to include in the max calcuation. -/// -/// @returns -/// The maximum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32x3 ffxMax3(FfxFloat32x3 x, FfxFloat32x3 y, FfxFloat32x3 z) -{ - return max(x, max(y, z)); -} - -/// Compute the maximum of three values. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the max calculation. -/// @param [in] y The second value to include in the max calcuation. -/// @param [in] z The third value to include in the max calcuation. -/// -/// @returns -/// The maximum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32x4 ffxMax3(FfxFloat32x4 x, FfxFloat32x4 y, FfxFloat32x4 z) -{ - return max(x, max(y, z)); -} - -/// Compute the maximum of three values. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the max calculation. -/// @param [in] y The second value to include in the max calcuation. -/// @param [in] z The third value to include in the max calcuation. -/// -/// @returns -/// The maximum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxUInt32 ffxMax3(FfxUInt32 x, FfxUInt32 y, FfxUInt32 z) -{ - return max(x, max(y, z)); -} - -/// Compute the maximum of three values. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the max calculation. -/// @param [in] y The second value to include in the max calcuation. -/// @param [in] z The third value to include in the max calcuation. -/// -/// @returns -/// The maximum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxUInt32x2 ffxMax3(FfxUInt32x2 x, FfxUInt32x2 y, FfxUInt32x2 z) -{ - return max(x, max(y, z)); -} - -/// Compute the maximum of three values. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the max calculation. -/// @param [in] y The second value to include in the max calcuation. -/// @param [in] z The third value to include in the max calcuation. -/// -/// @returns -/// The maximum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxUInt32x3 ffxMax3(FfxUInt32x3 x, FfxUInt32x3 y, FfxUInt32x3 z) -{ - return max(x, max(y, z)); -} - -/// Compute the maximum of three values. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the max calculation. -/// @param [in] y The second value to include in the max calcuation. -/// @param [in] z The third value to include in the max calcuation. -/// -/// @returns -/// The maximum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxUInt32x4 ffxMax3(FfxUInt32x4 x, FfxUInt32x4 y, FfxUInt32x4 z) -{ - return max(x, max(y, z)); -} - -/// Compute the median of three values. -/// -/// NOTE: This function should compile down to a single V_MED3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the median calculation. -/// @param [in] y The second value to include in the median calcuation. -/// @param [in] z The third value to include in the median calcuation. -/// -/// @returns -/// The median value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32 ffxMed3(FfxFloat32 x, FfxFloat32 y, FfxFloat32 z) -{ - return max(min(x, y), min(max(x, y), z)); -} - -/// Compute the median of three values. -/// -/// NOTE: This function should compile down to a single V_MED3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the median calculation. -/// @param [in] y The second value to include in the median calcuation. -/// @param [in] z The third value to include in the median calcuation. -/// -/// @returns -/// The median value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32x2 ffxMed3(FfxFloat32x2 x, FfxFloat32x2 y, FfxFloat32x2 z) -{ - return max(min(x, y), min(max(x, y), z)); -} - -/// Compute the median of three values. -/// -/// NOTE: This function should compile down to a single V_MED3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the median calculation. -/// @param [in] y The second value to include in the median calcuation. -/// @param [in] z The third value to include in the median calcuation. -/// -/// @returns -/// The median value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32x3 ffxMed3(FfxFloat32x3 x, FfxFloat32x3 y, FfxFloat32x3 z) -{ - return max(min(x, y), min(max(x, y), z)); -} - -/// Compute the median of three values. -/// -/// NOTE: This function should compile down to a single V_MED3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the median calculation. -/// @param [in] y The second value to include in the median calcuation. -/// @param [in] z The third value to include in the median calcuation. -/// -/// @returns -/// The median value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32x4 ffxMed3(FfxFloat32x4 x, FfxFloat32x4 y, FfxFloat32x4 z) -{ - return max(min(x, y), min(max(x, y), z)); -} - -/// Compute the median of three values. -/// -/// NOTE: This function should compile down to a single V_MED3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the median calculation. -/// @param [in] y The second value to include in the median calcuation. -/// @param [in] z The third value to include in the median calcuation. -/// -/// @returns -/// The median value of x, y, and z. -/// -/// @ingroup HLSL -FfxInt32 ffxMed3(FfxInt32 x, FfxInt32 y, FfxInt32 z) -{ - return max(min(x, y), min(max(x, y), z)); - // return min(max(min(y, z), x), max(y, z)); - // return max(max(x, y), z) == x ? max(y, z) : (max(max(x, y), z) == y ? max(x, z) : max(x, y)); -} - -/// Compute the median of three values. -/// -/// NOTE: This function should compile down to a single V_MED3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the median calculation. -/// @param [in] y The second value to include in the median calcuation. -/// @param [in] z The third value to include in the median calcuation. -/// -/// @returns -/// The median value of x, y, and z. -/// -/// @ingroup HLSL -FfxInt32x2 ffxMed3(FfxInt32x2 x, FfxInt32x2 y, FfxInt32x2 z) -{ - return max(min(x, y), min(max(x, y), z)); - // return min(max(min(y, z), x), max(y, z)); - // return max(max(x, y), z) == x ? max(y, z) : (max(max(x, y), z) == y ? max(x, z) : max(x, y)); -} - -/// Compute the median of three values. -/// -/// NOTE: This function should compile down to a single V_MED3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the median calculation. -/// @param [in] y The second value to include in the median calcuation. -/// @param [in] z The third value to include in the median calcuation. -/// -/// @returns -/// The median value of x, y, and z. -/// -/// @ingroup HLSL -FfxInt32x3 ffxMed3(FfxInt32x3 x, FfxInt32x3 y, FfxInt32x3 z) -{ - return max(min(x, y), min(max(x, y), z)); -} - -/// Compute the median of three values. -/// -/// NOTE: This function should compile down to a single V_MED3_I32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the median calculation. -/// @param [in] y The second value to include in the median calcuation. -/// @param [in] z The third value to include in the median calcuation. -/// -/// @returns -/// The median value of x, y, and z. -/// -/// @ingroup HLSL -FfxInt32x4 ffxMed3(FfxInt32x4 x, FfxInt32x4 y, FfxInt32x4 z) -{ - return max(min(x, y), min(max(x, y), z)); -} - -/// Compute the minimum of three values. -/// -/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the min calculation. -/// @param [in] y The second value to include in the min calcuation. -/// @param [in] z The third value to include in the min calcuation. -/// -/// @returns -/// The minimum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32 ffxMin3(FfxFloat32 x, FfxFloat32 y, FfxFloat32 z) -{ - return min(x, min(y, z)); -} - -/// Compute the minimum of three values. -/// -/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the min calculation. -/// @param [in] y The second value to include in the min calcuation. -/// @param [in] z The third value to include in the min calcuation. -/// -/// @returns -/// The minimum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32x2 ffxMin3(FfxFloat32x2 x, FfxFloat32x2 y, FfxFloat32x2 z) -{ - return min(x, min(y, z)); -} - -/// Compute the minimum of three values. -/// -/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the min calculation. -/// @param [in] y The second value to include in the min calcuation. -/// @param [in] z The third value to include in the min calcuation. -/// -/// @returns -/// The minimum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32x3 ffxMin3(FfxFloat32x3 x, FfxFloat32x3 y, FfxFloat32x3 z) -{ - return min(x, min(y, z)); -} - -/// Compute the minimum of three values. -/// -/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the min calculation. -/// @param [in] y The second value to include in the min calcuation. -/// @param [in] z The third value to include in the min calcuation. -/// -/// @returns -/// The minimum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32x4 ffxMin3(FfxFloat32x4 x, FfxFloat32x4 y, FfxFloat32x4 z) -{ - return min(x, min(y, z)); -} - -/// Compute the minimum of three values. -/// -/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the min calculation. -/// @param [in] y The second value to include in the min calcuation. -/// @param [in] z The third value to include in the min calcuation. -/// -/// @returns -/// The minimum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxUInt32 ffxMin3(FfxUInt32 x, FfxUInt32 y, FfxUInt32 z) -{ - return min(x, min(y, z)); -} - -/// Compute the minimum of three values. -/// -/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the min calculation. -/// @param [in] y The second value to include in the min calcuation. -/// @param [in] z The third value to include in the min calcuation. -/// -/// @returns -/// The minimum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxUInt32x2 ffxMin3(FfxUInt32x2 x, FfxUInt32x2 y, FfxUInt32x2 z) -{ - return min(x, min(y, z)); -} - -/// Compute the minimum of three values. -/// -/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the min calculation. -/// @param [in] y The second value to include in the min calculation. -/// @param [in] z The third value to include in the min calculation. -/// -/// @returns -/// The minimum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxUInt32x3 ffxMin3(FfxUInt32x3 x, FfxUInt32x3 y, FfxUInt32x3 z) -{ - return min(x, min(y, z)); -} - -/// Compute the minimum of three values. -/// -/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the min calculation. -/// @param [in] y The second value to include in the min calcuation. -/// @param [in] z The third value to include in the min calcuation. -/// -/// @returns -/// The minimum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxUInt32x4 ffxMin3(FfxUInt32x4 x, FfxUInt32x4 y, FfxUInt32x4 z) -{ - return min(x, min(y, z)); -} - - -FfxUInt32 AShrSU1(FfxUInt32 a, FfxUInt32 b) -{ - return FfxUInt32(FfxInt32(a) >> FfxInt32(b)); -} - -FfxUInt32 ffxPackF32(FfxFloat32x2 v){ - FfxUInt32x2 p = FfxUInt32x2(f32tof16(FfxFloat32x2(v).x), f32tof16(FfxFloat32x2(v).y)); - return p.x | (p.y << 16); -} - -FfxFloat32x2 ffxUnpackF32(FfxUInt32 a){ - return f16tof32(FfxUInt32x2(a & 0xFFFF, a >> 16)); -} - -//============================================================================================================================== -// HLSL HALF -//============================================================================================================================== -//============================================================================================================================== -// Need to use manual unpack to get optimal execution (don't use packed types in buffers directly). -// Unpack requires this pattern: https://gpuopen.com/first-steps-implementing-fp16/ -FFX_MIN16_F2 ffxUint32ToFloat16x2(FfxUInt32 x) -{ - FfxFloat32x2 t = f16tof32(FfxUInt32x2(x & 0xFFFF, x >> 16)); - return FFX_MIN16_F2(t); -} -FFX_MIN16_F4 ffxUint32x2ToFloat16x4(FfxUInt32x2 x) -{ - return FFX_MIN16_F4(ffxUint32ToFloat16x2(x.x), ffxUint32ToFloat16x2(x.y)); -} -FFX_MIN16_U2 ffxUint32ToUint16x2(FfxUInt32 x) -{ - FfxUInt32x2 t = FfxUInt32x2(x & 0xFFFF, x >> 16); - return FFX_MIN16_U2(t); -} -FFX_MIN16_U4 ffxUint32x2ToUint16x4(FfxUInt32x2 x) -{ - return FFX_MIN16_U4(ffxUint32ToUint16x2(x.x), ffxUint32ToUint16x2(x.y)); -} - -/// @brief Inverts the value while avoiding division by zero. If the value is zero, zero is returned. -/// @param v Value to invert. -/// @return If v = 0 returns 0. If v != 0 returns 1/v. -FfxFloat32 ffxInvertSafe(FfxFloat32 v){ - FfxFloat32 s = sign(v); - FfxFloat32 s2 = s*s; - return s2/(v + s2 - 1.0); -} - -/// @brief Inverts the value while avoiding division by zero. If the value is zero, zero is returned. -/// @param v Value to invert. -/// @return If v = 0 returns 0. If v != 0 returns 1/v. -FfxFloat32x2 ffxInvertSafe(FfxFloat32x2 v){ - FfxFloat32x2 s = sign(v); - FfxFloat32x2 s2 = s*s; - return s2/(v + s2 - FfxFloat32x2(1.0, 1.0)); -} - -/// @brief Inverts the value while avoiding division by zero. If the value is zero, zero is returned. -/// @param v Value to invert. -/// @return If v = 0 returns 0. If v != 0 returns 1/v. -FfxFloat32x3 ffxInvertSafe(FfxFloat32x3 v){ - FfxFloat32x3 s = sign(v); - FfxFloat32x3 s2 = s*s; - return s2/(v + s2 - FfxFloat32x3(1.0, 1.0, 1.0)); -} - -/// @brief Inverts the value while avoiding division by zero. If the value is zero, zero is returned. -/// @param v Value to invert. -/// @return If v = 0 returns 0. If v != 0 returns 1/v. -FfxFloat32x4 ffxInvertSafe(FfxFloat32x4 v){ - FfxFloat32x4 s = sign(v); - FfxFloat32x4 s2 = s*s; - return s2/(v + s2 - FfxFloat32x4(1.0, 1.0, 1.0, 1.0)); -} - -#define FFX_UINT32_TO_FLOAT16X2(x) ffxUint32ToFloat16x2(FfxUInt32(x)) -#if FFX_HALF - -#define FFX_UINT32X2_TO_FLOAT16X4(x) ffxUint32x2ToFloat16x4(FfxUInt32x2(x)) -#define FFX_UINT32_TO_UINT16X2(x) ffxUint32ToUint16x2(FfxUInt32(x)) -#define FFX_UINT32X2_TO_UINT16X4(x) ffxUint32x2ToUint16x4(FfxUInt32x2(x)) - -FfxUInt32 ffxPackF16(FfxFloat16x2 v){ - FfxUInt32x2 p = FfxUInt32x2(f32tof16(FfxFloat32x2(v).x), f32tof16(FfxFloat32x2(v).y)); - return p.x | (p.y << 16); -} - -FfxFloat16x2 ffxUnpackF16(FfxUInt32 a){ - return FfxFloat16x2(f16tof32(FfxUInt32x2(a & 0xFFFF, a >> 16))); -} - -//------------------------------------------------------------------------------------------------------------------------------ -FfxUInt32 FFX_MIN16_F2ToUint32(FFX_MIN16_F2 x) -{ - return f32tof16(x.x) + (f32tof16(x.y) << 16); -} -FfxUInt32x2 FFX_MIN16_F4ToUint32x2(FFX_MIN16_F4 x) -{ - return FfxUInt32x2(FFX_MIN16_F2ToUint32(x.xy), FFX_MIN16_F2ToUint32(x.zw)); -} -FfxUInt32 FFX_MIN16_U2ToUint32(FFX_MIN16_U2 x) -{ - return FfxUInt32(x.x) + (FfxUInt32(x.y) << 16); -} -FfxUInt32x2 FFX_MIN16_U4ToUint32x2(FFX_MIN16_U4 x) -{ - return FfxUInt32x2(FFX_MIN16_U2ToUint32(x.xy), FFX_MIN16_U2ToUint32(x.zw)); -} -#define FFX_FLOAT16X2_TO_UINT32(x) FFX_MIN16_F2ToUint32(FFX_MIN16_F2(x)) -#define FFX_FLOAT16X4_TO_UINT32X2(x) FFX_MIN16_F4ToUint32x2(FFX_MIN16_F4(x)) -#define FFX_UINT16X2_TO_UINT32(x) FFX_MIN16_U2ToUint32(FFX_MIN16_U2(x)) -#define FFX_UINT16X4_TO_UINT32X2(x) FFX_MIN16_U4ToUint32x2(FFX_MIN16_U4(x)) - -#if (FFX_HLSL_SM >= 62) && !defined(FFX_NO_16_BIT_CAST) -#define FFX_TO_UINT16(x) asuint16(x) -#define FFX_TO_UINT16X2(x) asuint16(x) -#define FFX_TO_UINT16X3(x) asuint16(x) -#define FFX_TO_UINT16X4(x) asuint16(x) -#else -#define FFX_TO_UINT16(a) FFX_MIN16_U(f32tof16(FfxFloat32(a))) -#define FFX_TO_UINT16X2(a) FFX_MIN16_U2(FFX_TO_UINT16((a).x), FFX_TO_UINT16((a).y)) -#define FFX_TO_UINT16X3(a) FFX_MIN16_U3(FFX_TO_UINT16((a).x), FFX_TO_UINT16((a).y), FFX_TO_UINT16((a).z)) -#define FFX_TO_UINT16X4(a) FFX_MIN16_U4(FFX_TO_UINT16((a).x), FFX_TO_UINT16((a).y), FFX_TO_UINT16((a).z), FFX_TO_UINT16((a).w)) -#endif // #if (FFX_HLSL_SM>=62) && !defined(FFX_NO_16_BIT_CAST) - -#if (FFX_HLSL_SM >= 62) && !defined(FFX_NO_16_BIT_CAST) -#define FFX_TO_FLOAT16(x) asfloat16(x) -#define FFX_TO_FLOAT16X2(x) asfloat16(x) -#define FFX_TO_FLOAT16X3(x) asfloat16(x) -#define FFX_TO_FLOAT16X4(x) asfloat16(x) -#else -#define FFX_TO_FLOAT16(a) FFX_MIN16_F(f16tof32(FfxUInt32(a))) -#define FFX_TO_FLOAT16X2(a) FFX_MIN16_F2(FFX_TO_FLOAT16((a).x), FFX_TO_FLOAT16((a).y)) -#define FFX_TO_FLOAT16X3(a) FFX_MIN16_F3(FFX_TO_FLOAT16((a).x), FFX_TO_FLOAT16((a).y), FFX_TO_FLOAT16((a).z)) -#define FFX_TO_FLOAT16X4(a) FFX_MIN16_F4(FFX_TO_FLOAT16((a).x), FFX_TO_FLOAT16((a).y), FFX_TO_FLOAT16((a).z), FFX_TO_FLOAT16((a).w)) -#endif // #if (FFX_HLSL_SM>=62) && !defined(FFX_NO_16_BIT_CAST) - -//============================================================================================================================== -#define FFX_BROADCAST_FLOAT16(a) FFX_MIN16_F(a) -#define FFX_BROADCAST_FLOAT16X2(a) FFX_MIN16_F(a) -#define FFX_BROADCAST_FLOAT16X3(a) FFX_MIN16_F(a) -#define FFX_BROADCAST_FLOAT16X4(a) FFX_MIN16_F(a) - -//------------------------------------------------------------------------------------------------------------------------------ -#define FFX_BROADCAST_INT16(a) FFX_MIN16_I(a) -#define FFX_BROADCAST_INT16X2(a) FFX_MIN16_I(a) -#define FFX_BROADCAST_INT16X3(a) FFX_MIN16_I(a) -#define FFX_BROADCAST_INT16X4(a) FFX_MIN16_I(a) - -//------------------------------------------------------------------------------------------------------------------------------ -#define FFX_BROADCAST_UINT16(a) FFX_MIN16_U(a) -#define FFX_BROADCAST_UINT16X2(a) FFX_MIN16_U(a) -#define FFX_BROADCAST_UINT16X3(a) FFX_MIN16_U(a) -#define FFX_BROADCAST_UINT16X4(a) FFX_MIN16_U(a) - -//============================================================================================================================== -FFX_MIN16_U ffxAbsHalf(FFX_MIN16_U a) -{ - return FFX_MIN16_U(abs(FFX_MIN16_I(a))); -} -FFX_MIN16_U2 ffxAbsHalf(FFX_MIN16_U2 a) -{ - return FFX_MIN16_U2(abs(FFX_MIN16_I2(a))); -} -FFX_MIN16_U3 ffxAbsHalf(FFX_MIN16_U3 a) -{ - return FFX_MIN16_U3(abs(FFX_MIN16_I3(a))); -} -FFX_MIN16_U4 ffxAbsHalf(FFX_MIN16_U4 a) -{ - return FFX_MIN16_U4(abs(FFX_MIN16_I4(a))); -} -//------------------------------------------------------------------------------------------------------------------------------ -FFX_MIN16_F ffxClampHalf(FFX_MIN16_F x, FFX_MIN16_F n, FFX_MIN16_F m) -{ - return max(n, min(x, m)); -} -FFX_MIN16_F2 ffxClampHalf(FFX_MIN16_F2 x, FFX_MIN16_F2 n, FFX_MIN16_F2 m) -{ - return max(n, min(x, m)); -} -FFX_MIN16_F3 ffxClampHalf(FFX_MIN16_F3 x, FFX_MIN16_F3 n, FFX_MIN16_F3 m) -{ - return max(n, min(x, m)); -} -FFX_MIN16_F4 ffxClampHalf(FFX_MIN16_F4 x, FFX_MIN16_F4 n, FFX_MIN16_F4 m) -{ - return max(n, min(x, m)); -} -//------------------------------------------------------------------------------------------------------------------------------ -// V_FRACT_F16 (note DX frac() is different). -FFX_MIN16_F ffxFract(FFX_MIN16_F x) -{ - return x - floor(x); -} -FFX_MIN16_F2 ffxFract(FFX_MIN16_F2 x) -{ - return x - floor(x); -} -FFX_MIN16_F3 ffxFract(FFX_MIN16_F3 x) -{ - return x - floor(x); -} -FFX_MIN16_F4 ffxFract(FFX_MIN16_F4 x) -{ - return x - floor(x); -} -//------------------------------------------------------------------------------------------------------------------------------ -FFX_MIN16_F ffxLerp(FFX_MIN16_F x, FFX_MIN16_F y, FFX_MIN16_F a) -{ - return lerp(x, y, a); -} -FFX_MIN16_F2 ffxLerp(FFX_MIN16_F2 x, FFX_MIN16_F2 y, FFX_MIN16_F a) -{ - return lerp(x, y, a); -} -FFX_MIN16_F2 ffxLerp(FFX_MIN16_F2 x, FFX_MIN16_F2 y, FFX_MIN16_F2 a) -{ - return lerp(x, y, a); -} -FFX_MIN16_F3 ffxLerp(FFX_MIN16_F3 x, FFX_MIN16_F3 y, FFX_MIN16_F a) -{ - return lerp(x, y, a); -} -FFX_MIN16_F3 ffxLerp(FFX_MIN16_F3 x, FFX_MIN16_F3 y, FFX_MIN16_F3 a) -{ - return lerp(x, y, a); -} -FFX_MIN16_F4 ffxLerp(FFX_MIN16_F4 x, FFX_MIN16_F4 y, FFX_MIN16_F a) -{ - return lerp(x, y, a); -} -FFX_MIN16_F4 ffxLerp(FFX_MIN16_F4 x, FFX_MIN16_F4 y, FFX_MIN16_F4 a) -{ - return lerp(x, y, a); -} -//------------------------------------------------------------------------------------------------------------------------------ -FFX_MIN16_F ffxMax3Half(FFX_MIN16_F x, FFX_MIN16_F y, FFX_MIN16_F z) -{ - return max(x, max(y, z)); -} -FFX_MIN16_F2 ffxMax3Half(FFX_MIN16_F2 x, FFX_MIN16_F2 y, FFX_MIN16_F2 z) -{ - return max(x, max(y, z)); -} -FFX_MIN16_F3 ffxMax3Half(FFX_MIN16_F3 x, FFX_MIN16_F3 y, FFX_MIN16_F3 z) -{ - return max(x, max(y, z)); -} -FFX_MIN16_F4 ffxMax3Half(FFX_MIN16_F4 x, FFX_MIN16_F4 y, FFX_MIN16_F4 z) -{ - return max(x, max(y, z)); -} -//------------------------------------------------------------------------------------------------------------------------------ -FFX_MIN16_F ffxMin3Half(FFX_MIN16_F x, FFX_MIN16_F y, FFX_MIN16_F z) -{ - return min(x, min(y, z)); -} -FFX_MIN16_F2 ffxMin3Half(FFX_MIN16_F2 x, FFX_MIN16_F2 y, FFX_MIN16_F2 z) -{ - return min(x, min(y, z)); -} -FFX_MIN16_F3 ffxMin3Half(FFX_MIN16_F3 x, FFX_MIN16_F3 y, FFX_MIN16_F3 z) -{ - return min(x, min(y, z)); -} -FFX_MIN16_F4 ffxMin3Half(FFX_MIN16_F4 x, FFX_MIN16_F4 y, FFX_MIN16_F4 z) -{ - return min(x, min(y, z)); -} -//------------------------------------------------------------------------------------------------------------------------------ -FFX_MIN16_F ffxMed3Half(FFX_MIN16_F x, FFX_MIN16_F y, FFX_MIN16_F z) -{ - return max(min(x, y), min(max(x, y), z)); -} -FFX_MIN16_F2 ffxMed3Half(FFX_MIN16_F2 x, FFX_MIN16_F2 y, FFX_MIN16_F2 z) -{ - return max(min(x, y), min(max(x, y), z)); -} -FFX_MIN16_F3 ffxMed3Half(FFX_MIN16_F3 x, FFX_MIN16_F3 y, FFX_MIN16_F3 z) -{ - return max(min(x, y), min(max(x, y), z)); -} -FFX_MIN16_F4 ffxMed3Half(FFX_MIN16_F4 x, FFX_MIN16_F4 y, FFX_MIN16_F4 z) -{ - return max(min(x, y), min(max(x, y), z)); -} -//------------------------------------------------------------------------------------------------------------------------------ -FFX_MIN16_I ffxMed3Half(FFX_MIN16_I x, FFX_MIN16_I y, FFX_MIN16_I z) -{ - return max(min(x, y), min(max(x, y), z)); -} -FFX_MIN16_I2 ffxMed3Half(FFX_MIN16_I2 x, FFX_MIN16_I2 y, FFX_MIN16_I2 z) -{ - return max(min(x, y), min(max(x, y), z)); -} -FFX_MIN16_I3 ffxMed3Half(FFX_MIN16_I3 x, FFX_MIN16_I3 y, FFX_MIN16_I3 z) -{ - return max(min(x, y), min(max(x, y), z)); -} -FFX_MIN16_I4 ffxMed3Half(FFX_MIN16_I4 x, FFX_MIN16_I4 y, FFX_MIN16_I4 z) -{ - return max(min(x, y), min(max(x, y), z)); -} -//------------------------------------------------------------------------------------------------------------------------------ -FFX_MIN16_F ffxReciprocalHalf(FFX_MIN16_F x) -{ - return rcp(x); -} -FFX_MIN16_F2 ffxReciprocalHalf(FFX_MIN16_F2 x) -{ - return rcp(x); -} -FFX_MIN16_F3 ffxReciprocalHalf(FFX_MIN16_F3 x) -{ - return rcp(x); -} -FFX_MIN16_F4 ffxReciprocalHalf(FFX_MIN16_F4 x) -{ - return rcp(x); -} -//------------------------------------------------------------------------------------------------------------------------------ -FFX_MIN16_F ffxReciprocalSquareRootHalf(FFX_MIN16_F x) -{ - return rsqrt(x); -} -FFX_MIN16_F2 ffxReciprocalSquareRootHalf(FFX_MIN16_F2 x) -{ - return rsqrt(x); -} -FFX_MIN16_F3 ffxReciprocalSquareRootHalf(FFX_MIN16_F3 x) -{ - return rsqrt(x); -} -FFX_MIN16_F4 ffxReciprocalSquareRootHalf(FFX_MIN16_F4 x) -{ - return rsqrt(x); -} -//------------------------------------------------------------------------------------------------------------------------------ -FFX_MIN16_F ffxSaturate(FFX_MIN16_F x) -{ - return saturate(x); -} -FFX_MIN16_F2 ffxSaturate(FFX_MIN16_F2 x) -{ - return saturate(x); -} -FFX_MIN16_F3 ffxSaturate(FFX_MIN16_F3 x) -{ - return saturate(x); -} -FFX_MIN16_F4 ffxSaturate(FFX_MIN16_F4 x) -{ - return saturate(x); -} -//------------------------------------------------------------------------------------------------------------------------------ -FFX_MIN16_U ffxBitShiftRightHalf(FFX_MIN16_U a, FFX_MIN16_U b) -{ - return FFX_MIN16_U(FFX_MIN16_I(a) >> FFX_MIN16_I(b)); -} -FFX_MIN16_U2 ffxBitShiftRightHalf(FFX_MIN16_U2 a, FFX_MIN16_U2 b) -{ - return FFX_MIN16_U2(FFX_MIN16_I2(a) >> FFX_MIN16_I2(b)); -} -FFX_MIN16_U3 ffxBitShiftRightHalf(FFX_MIN16_U3 a, FFX_MIN16_U3 b) -{ - return FFX_MIN16_U3(FFX_MIN16_I3(a) >> FFX_MIN16_I3(b)); -} -FFX_MIN16_U4 ffxBitShiftRightHalf(FFX_MIN16_U4 a, FFX_MIN16_U4 b) -{ - return FFX_MIN16_U4(FFX_MIN16_I4(a) >> FFX_MIN16_I4(b)); -} -#endif // FFX_HALF - -//============================================================================================================================== -// HLSL WAVE -//============================================================================================================================== -#if defined(FFX_WAVE) -// Where 'x' must be a compile time literal. -FfxFloat32 AWaveXorF1(FfxFloat32 v, FfxUInt32 x) -{ - return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); -} -FfxFloat32x2 AWaveXorF2(FfxFloat32x2 v, FfxUInt32 x) -{ - return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); -} -FfxFloat32x3 AWaveXorF3(FfxFloat32x3 v, FfxUInt32 x) -{ - return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); -} -FfxFloat32x4 AWaveXorF4(FfxFloat32x4 v, FfxUInt32 x) -{ - return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); -} -FfxUInt32 AWaveXorU1(FfxUInt32 v, FfxUInt32 x) -{ - return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); -} -FfxUInt32x2 AWaveXorU1(FfxUInt32x2 v, FfxUInt32 x) -{ - return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); -} -FfxUInt32x3 AWaveXorU1(FfxUInt32x3 v, FfxUInt32 x) -{ - return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); -} -FfxUInt32x4 AWaveXorU1(FfxUInt32x4 v, FfxUInt32 x) -{ - return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); -} -FfxBoolean AWaveIsFirstLane() -{ - return WaveIsFirstLane(); -} -FfxUInt32 AWaveLaneIndex() -{ - return WaveGetLaneIndex(); -} -FfxBoolean AWaveReadAtLaneIndexB1(FfxBoolean v, FfxUInt32 x) -{ - return WaveReadLaneAt(v, x); -} -FfxUInt32 AWavePrefixCountBits(FfxBoolean v) -{ - return WavePrefixCountBits(v); -} -FfxUInt32 AWaveActiveCountBits(FfxBoolean v) -{ - return WaveActiveCountBits(v); -} -FfxUInt32 AWaveReadLaneFirstU1(FfxUInt32 v) -{ - return WaveReadLaneFirst(v); -} -FfxUInt32 WaveOr(FfxUInt32 a) -{ - return WaveActiveBitOr(a); -} -FfxFloat32 WaveMin(FfxFloat32 a) -{ - return WaveActiveMin(a); -} -FfxFloat32 WaveMax(FfxFloat32 a) -{ - return WaveActiveMax(a); -} -FfxUInt32 WaveLaneCount() -{ - return WaveGetLaneCount(); -} -FfxBoolean WaveAllTrue(FfxBoolean v) -{ - return WaveActiveAllTrue(v); -} -FfxFloat32 QuadReadX(FfxFloat32 v) -{ - return QuadReadAcrossX(v); -} -FfxFloat32x2 QuadReadX(FfxFloat32x2 v) -{ - return QuadReadAcrossX(v); -} -FfxFloat32 QuadReadY(FfxFloat32 v) -{ - return QuadReadAcrossY(v); -} -FfxFloat32x2 QuadReadY(FfxFloat32x2 v) -{ - return QuadReadAcrossY(v); -} - -#if FFX_HALF -FfxFloat16x2 ffxWaveXorFloat16x2(FfxFloat16x2 v, FfxUInt32 x) -{ - return FFX_UINT32_TO_FLOAT16X2(WaveReadLaneAt(FFX_FLOAT16X2_TO_UINT32(v), WaveGetLaneIndex() ^ x)); -} -FfxFloat16x4 ffxWaveXorFloat16x4(FfxFloat16x4 v, FfxUInt32 x) -{ - return FFX_UINT32X2_TO_FLOAT16X4(WaveReadLaneAt(FFX_FLOAT16X4_TO_UINT32X2(v), WaveGetLaneIndex() ^ x)); -} -FfxUInt16x2 ffxWaveXorUint16x2(FfxUInt16x2 v, FfxUInt32 x) -{ - return FFX_UINT32_TO_UINT16X2(WaveReadLaneAt(FFX_UINT16X2_TO_UINT32(v), WaveGetLaneIndex() ^ x)); -} -FfxUInt16x4 ffxWaveXorUint16x4(FfxUInt16x4 v, FfxUInt32 x) -{ - return FFX_UINT32X2_TO_UINT16X4(WaveReadLaneAt(FFX_UINT16X4_TO_UINT32X2(v), WaveGetLaneIndex() ^ x)); -} -#endif // FFX_HALF -#endif // #if defined(FFX_WAVE) diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_hlsl.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_hlsl.h.meta deleted file mode 100644 index 2aa0691..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_hlsl.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 299f67e8b7e1d1a48a577bf8b328ac92 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_portability.h b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_portability.h deleted file mode 100644 index 84a62d6..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_portability.h +++ /dev/null @@ -1,51 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -FfxFloat32x3 opAAddOneF3(FfxFloat32x3 d, FfxFloat32x3 a, FfxFloat32 b) -{ - d = a + ffxBroadcast3(b); - return d; -} - -FfxFloat32x3 opACpyF3(FfxFloat32x3 d, FfxFloat32x3 a) -{ - d = a; - return d; -} - -FfxFloat32x3 opAMulF3(FfxFloat32x3 d, FfxFloat32x3 a, FfxFloat32x3 b) -{ - d = a * b; - return d; -} - -FfxFloat32x3 opAMulOneF3(FfxFloat32x3 d, FfxFloat32x3 a, FfxFloat32 b) -{ - d = a * ffxBroadcast3(b); - return d; -} - -FfxFloat32x3 opARcpF3(FfxFloat32x3 d, FfxFloat32x3 a) -{ - d = rcp(a); - return d; -} diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_portability.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_portability.h.meta deleted file mode 100644 index b333bf0..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_portability.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 8d2ace0bd52e0e1438e08ddaccd3ba24 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_accumulate.h b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_accumulate.h deleted file mode 100644 index c425de7..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_accumulate.h +++ /dev/null @@ -1,288 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#ifndef FFX_FSR3UPSCALER_ACCUMULATE_H -#define FFX_FSR3UPSCALER_ACCUMULATE_H - -FfxFloat32 GetPxHrVelocity(FfxFloat32x2 fMotionVector) -{ - return length(fMotionVector * DisplaySize()); -} -#if FFX_HALF -FFX_MIN16_F GetPxHrVelocity(FFX_MIN16_F2 fMotionVector) -{ - return length(fMotionVector * FFX_MIN16_F2(DisplaySize())); -} -#endif - -void Accumulate(const AccumulationPassCommonParams params, FFX_PARAMETER_INOUT FfxFloat32x3 fHistoryColor, FfxFloat32x3 fAccumulation, FFX_PARAMETER_IN FfxFloat32x4 fUpsampledColorAndWeight) -{ - // Aviod invalid values when accumulation and upsampled weight is 0 - fAccumulation = ffxMax(FSR3UPSCALER_EPSILON.xxx, fAccumulation + fUpsampledColorAndWeight.www); - -#if FFX_FSR3UPSCALER_OPTION_HDR_COLOR_INPUT - //YCoCg -> RGB -> Tonemap -> YCoCg (Use RGB tonemapper to avoid color desaturation) - fUpsampledColorAndWeight.xyz = RGBToYCoCg(Tonemap(YCoCgToRGB(fUpsampledColorAndWeight.xyz))); - fHistoryColor = RGBToYCoCg(Tonemap(YCoCgToRGB(fHistoryColor))); -#endif - - const FfxFloat32x3 fAlpha = fUpsampledColorAndWeight.www / fAccumulation; - fHistoryColor = ffxLerp(fHistoryColor, fUpsampledColorAndWeight.xyz, fAlpha); - - fHistoryColor = YCoCgToRGB(fHistoryColor); - -#if FFX_FSR3UPSCALER_OPTION_HDR_COLOR_INPUT - fHistoryColor = InverseTonemap(fHistoryColor); -#endif -} - -void RectifyHistory( - const AccumulationPassCommonParams params, - RectificationBox clippingBox, - FFX_PARAMETER_INOUT FfxFloat32x3 fHistoryColor, - FFX_PARAMETER_INOUT FfxFloat32x3 fAccumulation, - FfxFloat32 fLockContributionThisFrame, - FfxFloat32 fTemporalReactiveFactor, - FfxFloat32 fLumaInstabilityFactor) -{ - const FfxFloat32 fVecolityFactor = ffxSaturate(params.fHrVelocity / 20.0f); - const FfxFloat32 fBoxScaleT = ffxMax(params.fDepthClipFactor, ffxMax(params.fAccumulationMask, fVecolityFactor)); - const FfxFloat32 fBoxScale = ffxLerp(3.0f, 1.0f, fBoxScaleT); - - const FfxFloat32x3 fScaledBoxVec = clippingBox.boxVec * fBoxScale; - const FfxFloat32x3 boxMin = clippingBox.boxCenter - fScaledBoxVec; - const FfxFloat32x3 boxMax = clippingBox.boxCenter + fScaledBoxVec; - - if (any(FFX_GREATER_THAN(boxMin, fHistoryColor)) || any(FFX_GREATER_THAN(fHistoryColor, boxMax))) { - - const FfxFloat32x3 fClampedHistoryColor = clamp(fHistoryColor, boxMin, boxMax); - - FfxFloat32x3 fHistoryContribution = ffxMax(fLumaInstabilityFactor, fLockContributionThisFrame).xxx; - - const FfxFloat32 fReactiveFactor = params.fDilatedReactiveFactor; - const FfxFloat32 fReactiveContribution = 1.0f - ffxPow(fReactiveFactor, 1.0f / 2.0f); - fHistoryContribution *= fReactiveContribution; - - // Scale history color using rectification info, also using accumulation mask to avoid potential invalid color protection - fHistoryColor = ffxLerp(fClampedHistoryColor, fHistoryColor, ffxSaturate(fHistoryContribution)); - - // Scale accumulation using rectification info - const FfxFloat32x3 fAccumulationMin = ffxMin(fAccumulation, FFX_BROADCAST_FLOAT32X3(0.1f)); - fAccumulation = ffxLerp(fAccumulationMin, fAccumulation, ffxSaturate(fHistoryContribution)); - } -} - -void WriteUpscaledOutput(FfxInt32x2 iPxHrPos, FfxFloat32x3 fUpscaledColor) -{ - StoreUpscaledOutput(iPxHrPos, fUpscaledColor); -} - -void FinalizeLockStatus(const AccumulationPassCommonParams params, FfxFloat32x2 fLockStatus, FfxFloat32 fUpsampledWeight) -{ - // we expect similar motion for next frame - // kill lock if that location is outside screen, avoid locks to be clamped to screen borders - FfxFloat32x2 fEstimatedUvNextFrame = params.fHrUv - params.fMotionVector; - if (IsUvInside(fEstimatedUvNextFrame) == false) { - KillLock(fLockStatus); - } - else { - // Decrease lock lifetime - const FfxFloat32 fLifetimeDecreaseLanczosMax = FfxFloat32(JitterSequenceLength()) * FfxFloat32(fAverageLanczosWeightPerFrame); - const FfxFloat32 fLifetimeDecrease = FfxFloat32(fUpsampledWeight / fLifetimeDecreaseLanczosMax); - fLockStatus[LOCK_LIFETIME_REMAINING] = ffxMax(FfxFloat32(0), fLockStatus[LOCK_LIFETIME_REMAINING] - fLifetimeDecrease); - } - - StoreLockStatus(params.iPxHrPos, fLockStatus); -} - - -FfxFloat32x3 ComputeBaseAccumulationWeight(const AccumulationPassCommonParams params, FfxFloat32 fThisFrameReactiveFactor, FfxBoolean bInMotionLastFrame, FfxFloat32 fUpsampledWeight, LockState lockState) -{ - // Always assume max accumulation was reached - FfxFloat32 fBaseAccumulation = fMaxAccumulationLanczosWeight * FfxFloat32(params.bIsExistingSample) * (1.0f - fThisFrameReactiveFactor) * (1.0f - params.fDepthClipFactor); - - fBaseAccumulation = ffxMin(fBaseAccumulation, ffxLerp(fBaseAccumulation, fUpsampledWeight * 10.0f, ffxMax(FfxFloat32(bInMotionLastFrame), ffxSaturate(params.fHrVelocity * FfxFloat32(10))))); - - fBaseAccumulation = ffxMin(fBaseAccumulation, ffxLerp(fBaseAccumulation, fUpsampledWeight, ffxSaturate(params.fHrVelocity / FfxFloat32(20)))); - - return fBaseAccumulation.xxx; -} - -FfxFloat32 ComputeLumaInstabilityFactor(const AccumulationPassCommonParams params, RectificationBox clippingBox, FfxFloat32 fThisFrameReactiveFactor, FfxFloat32 fLuminanceDiff) -{ - const FfxFloat32 fUnormThreshold = 1.0f / 255.0f; - const FfxInt32 N_MINUS_1 = 0; - const FfxInt32 N_MINUS_2 = 1; - const FfxInt32 N_MINUS_3 = 2; - const FfxInt32 N_MINUS_4 = 3; - - FfxFloat32 fCurrentFrameLuma = clippingBox.boxCenter.x; - -#if FFX_FSR3UPSCALER_OPTION_HDR_COLOR_INPUT - fCurrentFrameLuma = fCurrentFrameLuma / (1.0f + ffxMax(0.0f, fCurrentFrameLuma)); -#endif - - fCurrentFrameLuma = round(fCurrentFrameLuma * 255.0f) / 255.0f; - - const FfxBoolean bSampleLumaHistory = (ffxMax(ffxMax(params.fDepthClipFactor, params.fAccumulationMask), fLuminanceDiff) < 0.1f) && (params.bIsNewSample == false); - FfxFloat32x4 fCurrentFrameLumaHistory = bSampleLumaHistory ? SampleLumaHistory(params.fReprojectedHrUv) : FFX_BROADCAST_FLOAT32X4(0.0f); - - FfxFloat32 fLumaInstability = 0.0f; - FfxFloat32 fDiffs0 = (fCurrentFrameLuma - fCurrentFrameLumaHistory[N_MINUS_1]); - - FfxFloat32 fMin = abs(fDiffs0); - - if (fMin >= fUnormThreshold) { - for (int i = N_MINUS_2; i <= N_MINUS_4; i++) { - FfxFloat32 fDiffs1 = (fCurrentFrameLuma - fCurrentFrameLumaHistory[i]); - - if (sign(fDiffs0) == sign(fDiffs1)) { - - // Scale difference to protect historically similar values - const FfxFloat32 fMinBias = 1.0f; - fMin = ffxMin(fMin, abs(fDiffs1) * fMinBias); - } - } - - const FfxFloat32 fBoxSize = clippingBox.boxVec.x; - const FfxFloat32 fBoxSizeFactor = ffxPow(ffxSaturate(fBoxSize / 0.1f), 6.0f); - - fLumaInstability = FfxFloat32(fMin != abs(fDiffs0)) * fBoxSizeFactor; - fLumaInstability = FfxFloat32(fLumaInstability > fUnormThreshold); - - fLumaInstability *= 1.0f - ffxMax(params.fAccumulationMask, ffxPow(fThisFrameReactiveFactor, 1.0f / 6.0f)); - } - - //shift history - fCurrentFrameLumaHistory[N_MINUS_4] = fCurrentFrameLumaHistory[N_MINUS_3]; - fCurrentFrameLumaHistory[N_MINUS_3] = fCurrentFrameLumaHistory[N_MINUS_2]; - fCurrentFrameLumaHistory[N_MINUS_2] = fCurrentFrameLumaHistory[N_MINUS_1]; - fCurrentFrameLumaHistory[N_MINUS_1] = fCurrentFrameLuma; - - StoreLumaHistory(params.iPxHrPos, fCurrentFrameLumaHistory); - - return fLumaInstability * FfxFloat32(fCurrentFrameLumaHistory[N_MINUS_4] != 0); -} - -FfxFloat32 ComputeTemporalReactiveFactor(const AccumulationPassCommonParams params, FfxFloat32 fTemporalReactiveFactor) -{ - FfxFloat32 fNewFactor = ffxMin(0.99f, fTemporalReactiveFactor); - - fNewFactor = ffxMax(fNewFactor, ffxLerp(fNewFactor, 0.4f, ffxSaturate(params.fHrVelocity))); - - fNewFactor = ffxMax(fNewFactor * fNewFactor, ffxMax(params.fDepthClipFactor * 0.1f, params.fDilatedReactiveFactor)); - - // Force reactive factor for new samples - fNewFactor = params.bIsNewSample ? 1.0f : fNewFactor; - - if (ffxSaturate(params.fHrVelocity * 10.0f) >= 1.0f) { - fNewFactor = ffxMax(FSR3UPSCALER_EPSILON, fNewFactor) * -1.0f; - } - - return fNewFactor; -} - -AccumulationPassCommonParams InitParams(FfxInt32x2 iPxHrPos) -{ - AccumulationPassCommonParams params; - - params.iPxHrPos = iPxHrPos; - const FfxFloat32x2 fHrUv = (iPxHrPos + 0.5f) / DisplaySize(); - params.fHrUv = fHrUv; - - const FfxFloat32x2 fLrUvJittered = fHrUv + Jitter() / RenderSize(); - params.fLrUv_HwSampler = ClampUv(fLrUvJittered, RenderSize(), MaxRenderSize()); - - params.fMotionVector = GetMotionVector(iPxHrPos, fHrUv); - params.fHrVelocity = GetPxHrVelocity(params.fMotionVector); - - ComputeReprojectedUVs(params, params.fReprojectedHrUv, params.bIsExistingSample); - - params.fDepthClipFactor = ffxSaturate(SampleDepthClip(params.fLrUv_HwSampler)); - - const FfxFloat32x2 fDilatedReactiveMasks = SampleDilatedReactiveMasks(params.fLrUv_HwSampler); - params.fDilatedReactiveFactor = fDilatedReactiveMasks.x; - params.fAccumulationMask = fDilatedReactiveMasks.y; - params.bIsResetFrame = (0 == FrameIndex()); - - params.bIsNewSample = (params.bIsExistingSample == false || params.bIsResetFrame); - - return params; -} - -void Accumulate(FfxInt32x2 iPxHrPos) -{ - const AccumulationPassCommonParams params = InitParams(iPxHrPos); - - FfxFloat32x3 fHistoryColor = FfxFloat32x3(0, 0, 0); - FfxFloat32x2 fLockStatus; - InitializeNewLockSample(fLockStatus); - - FfxFloat32 fTemporalReactiveFactor = 0.0f; - FfxBoolean bInMotionLastFrame = FFX_FALSE; - LockState lockState = { FFX_FALSE , FFX_FALSE }; - if (params.bIsExistingSample && !params.bIsResetFrame) { - ReprojectHistoryColor(params, fHistoryColor, fTemporalReactiveFactor, bInMotionLastFrame); - lockState = ReprojectHistoryLockStatus(params, fLockStatus); - } - - FfxFloat32 fThisFrameReactiveFactor = ffxMax(params.fDilatedReactiveFactor, fTemporalReactiveFactor); - - FfxFloat32 fLuminanceDiff = 0.0f; - FfxFloat32 fLockContributionThisFrame = 0.0f; - UpdateLockStatus(params, fThisFrameReactiveFactor, lockState, fLockStatus, fLockContributionThisFrame, fLuminanceDiff); - - // Load upsampled input color - RectificationBox clippingBox; - FfxFloat32x4 fUpsampledColorAndWeight = ComputeUpsampledColorAndWeight(params, clippingBox, fThisFrameReactiveFactor); - - const FfxFloat32 fLumaInstabilityFactor = ComputeLumaInstabilityFactor(params, clippingBox, fThisFrameReactiveFactor, fLuminanceDiff); - - - FfxFloat32x3 fAccumulation = ComputeBaseAccumulationWeight(params, fThisFrameReactiveFactor, bInMotionLastFrame, fUpsampledColorAndWeight.w, lockState); - - if (params.bIsNewSample) { - fHistoryColor = YCoCgToRGB(fUpsampledColorAndWeight.xyz); - } - else { - RectifyHistory(params, clippingBox, fHistoryColor, fAccumulation, fLockContributionThisFrame, fThisFrameReactiveFactor, fLumaInstabilityFactor); - - Accumulate(params, fHistoryColor, fAccumulation, fUpsampledColorAndWeight); - } - - fHistoryColor = UnprepareRgb(fHistoryColor, Exposure()); - - FinalizeLockStatus(params, fLockStatus, fUpsampledColorAndWeight.w); - - // Get new temporal reactive factor - fTemporalReactiveFactor = ComputeTemporalReactiveFactor(params, fThisFrameReactiveFactor); - - StoreInternalColorAndWeight(iPxHrPos, FfxFloat32x4(fHistoryColor, fTemporalReactiveFactor)); - - // Output final color when RCAS is disabled -#if FFX_FSR3UPSCALER_OPTION_APPLY_SHARPENING == 0 - WriteUpscaledOutput(iPxHrPos, fHistoryColor); -#endif - StoreNewLocks(iPxHrPos, 0); -} - -#endif // FFX_FSR3UPSCALER_ACCUMULATE_H diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_accumulate.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_accumulate.h.meta deleted file mode 100644 index 79fa7a3..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_accumulate.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 3fc2f7a2c8c31324a949e1761bf599cc -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h deleted file mode 100644 index 13b317a..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h +++ /dev/null @@ -1,928 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#include "ffx_fsr3upscaler_resources.h" - -#if defined(FFX_GPU) -#ifdef __hlsl_dx_compiler -#pragma dxc diagnostic push -#pragma dxc diagnostic ignored "-Wambig-lit-shift" -#endif //__hlsl_dx_compiler -#include "ffx_core.h" -#ifdef __hlsl_dx_compiler -#pragma dxc diagnostic pop -#endif //__hlsl_dx_compiler -#endif // #if defined(FFX_GPU) - -#if defined(FFX_GPU) -#ifndef FFX_PREFER_WAVE64 -#define FFX_PREFER_WAVE64 -#endif // FFX_PREFER_WAVE64 - -#if defined(FFX_GPU) -#pragma warning(disable: 3205) // conversion from larger type to smaller -#endif // #if defined(FFX_GPU) - -#define DECLARE_SRV_REGISTER(regIndex) t##regIndex -#define DECLARE_UAV_REGISTER(regIndex) u##regIndex -#define DECLARE_CB_REGISTER(regIndex) b##regIndex -#define FFX_FSR3UPSCALER_DECLARE_SRV(regIndex) register(DECLARE_SRV_REGISTER(regIndex)) -#define FFX_FSR3UPSCALER_DECLARE_UAV(regIndex) register(DECLARE_UAV_REGISTER(regIndex)) -#define FFX_FSR3UPSCALER_DECLARE_CB(regIndex) register(DECLARE_CB_REGISTER(regIndex)) - -#if defined(FSR3UPSCALER_BIND_CB_FSR3UPSCALER) - cbuffer cbFSR3Upscaler : FFX_FSR3UPSCALER_DECLARE_CB(FSR3UPSCALER_BIND_CB_FSR3UPSCALER) - { - FfxInt32x2 iRenderSize; - FfxInt32x2 iMaxRenderSize; - FfxInt32x2 iDisplaySize; - FfxInt32x2 iInputColorResourceDimensions; - FfxInt32x2 iLumaMipDimensions; - FfxInt32 iLumaMipLevelToUse; - FfxInt32 iFrameIndex; - - FfxFloat32x4 fDeviceToViewDepth; - FfxFloat32x2 fJitter; - FfxFloat32x2 fMotionVectorScale; - FfxFloat32x2 fDownscaleFactor; - FfxFloat32x2 fMotionVectorJitterCancellation; - FfxFloat32 fPreExposure; - FfxFloat32 fPreviousFramePreExposure; - FfxFloat32 fTanHalfFOV; - FfxFloat32 fJitterSequenceLength; - FfxFloat32 fDeltaTime; - FfxFloat32 fDynamicResChangeFactor; - FfxFloat32 fViewSpaceToMetersFactor; - - FfxInt32 iDummy; - }; - -#define FFX_FSR3UPSCALER_CONSTANT_BUFFER_1_SIZE (sizeof(cbFSR3Upscaler) / 4) // Number of 32-bit values. This must be kept in sync with the cbFSR3Upscaler size. - -/* Define getter functions in the order they are defined in the CB! */ -FfxInt32x2 RenderSize() -{ - return iRenderSize; -} - -FfxInt32x2 MaxRenderSize() -{ - return iMaxRenderSize; -} - -FfxInt32x2 DisplaySize() -{ - return iDisplaySize; -} - -FfxInt32x2 InputColorResourceDimensions() -{ - return iInputColorResourceDimensions; -} - -FfxInt32x2 LumaMipDimensions() -{ - return iLumaMipDimensions; -} - -FfxInt32 LumaMipLevelToUse() -{ - return iLumaMipLevelToUse; -} - -FfxInt32 FrameIndex() -{ - return iFrameIndex; -} - -FfxFloat32x2 Jitter() -{ - return fJitter; -} - -FfxFloat32x4 DeviceToViewSpaceTransformFactors() -{ - return fDeviceToViewDepth; -} - -FfxFloat32x2 MotionVectorScale() -{ - return fMotionVectorScale; -} - -FfxFloat32x2 DownscaleFactor() -{ - return fDownscaleFactor; -} - -FfxFloat32x2 MotionVectorJitterCancellation() -{ - return fMotionVectorJitterCancellation; -} - -FfxFloat32 PreExposure() -{ - return fPreExposure; -} - -FfxFloat32 PreviousFramePreExposure() -{ - return fPreviousFramePreExposure; -} - -FfxFloat32 TanHalfFoV() -{ - return fTanHalfFOV; -} - -FfxFloat32 JitterSequenceLength() -{ - return fJitterSequenceLength; -} - -FfxFloat32 DeltaTime() -{ - return fDeltaTime; -} - -FfxFloat32 DynamicResChangeFactor() -{ - return fDynamicResChangeFactor; -} - -FfxFloat32 ViewSpaceToMetersFactor() -{ - return fViewSpaceToMetersFactor; -} -#endif // #if defined(FSR3UPSCALER_BIND_CB_FSR3UPSCALER) - -#define FFX_FSR3UPSCALER_ROOTSIG_STRINGIFY(p) FFX_FSR3UPSCALER_ROOTSIG_STR(p) -#define FFX_FSR3UPSCALER_ROOTSIG_STR(p) #p -#define FFX_FSR3UPSCALER_ROOTSIG [RootSignature( "DescriptorTable(UAV(u0, numDescriptors = " FFX_FSR3UPSCALER_ROOTSIG_STRINGIFY(FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_COUNT) ")), " \ - "DescriptorTable(SRV(t0, numDescriptors = " FFX_FSR3UPSCALER_ROOTSIG_STRINGIFY(FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_COUNT) ")), " \ - "RootConstants(num32BitConstants=" FFX_FSR3UPSCALER_ROOTSIG_STRINGIFY(FFX_FSR3UPSCALER_CONSTANT_BUFFER_1_SIZE) ", b0), " \ - "StaticSampler(s0, filter = FILTER_MIN_MAG_MIP_POINT, " \ - "addressU = TEXTURE_ADDRESS_CLAMP, " \ - "addressV = TEXTURE_ADDRESS_CLAMP, " \ - "addressW = TEXTURE_ADDRESS_CLAMP, " \ - "comparisonFunc = COMPARISON_NEVER, " \ - "borderColor = STATIC_BORDER_COLOR_TRANSPARENT_BLACK), " \ - "StaticSampler(s1, filter = FILTER_MIN_MAG_MIP_LINEAR, " \ - "addressU = TEXTURE_ADDRESS_CLAMP, " \ - "addressV = TEXTURE_ADDRESS_CLAMP, " \ - "addressW = TEXTURE_ADDRESS_CLAMP, " \ - "comparisonFunc = COMPARISON_NEVER, " \ - "borderColor = STATIC_BORDER_COLOR_TRANSPARENT_BLACK)" )] - -#define FFX_FSR3UPSCALER_CONSTANT_BUFFER_2_SIZE 6 // Number of 32-bit values. This must be kept in sync with max( cbRCAS , cbSPD) size. - -#define FFX_FSR3UPSCALER_CB2_ROOTSIG [RootSignature( "DescriptorTable(UAV(u0, numDescriptors = " FFX_FSR3UPSCALER_ROOTSIG_STRINGIFY(FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_COUNT) ")), " \ - "DescriptorTable(SRV(t0, numDescriptors = " FFX_FSR3UPSCALER_ROOTSIG_STRINGIFY(FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_COUNT) ")), " \ - "RootConstants(num32BitConstants=" FFX_FSR3UPSCALER_ROOTSIG_STRINGIFY(FFX_FSR3UPSCALER_CONSTANT_BUFFER_1_SIZE) ", b0), " \ - "RootConstants(num32BitConstants=" FFX_FSR3UPSCALER_ROOTSIG_STRINGIFY(FFX_FSR3UPSCALER_CONSTANT_BUFFER_2_SIZE) ", b1), " \ - "StaticSampler(s0, filter = FILTER_MIN_MAG_MIP_POINT, " \ - "addressU = TEXTURE_ADDRESS_CLAMP, " \ - "addressV = TEXTURE_ADDRESS_CLAMP, " \ - "addressW = TEXTURE_ADDRESS_CLAMP, " \ - "comparisonFunc = COMPARISON_NEVER, " \ - "borderColor = STATIC_BORDER_COLOR_TRANSPARENT_BLACK), " \ - "StaticSampler(s1, filter = FILTER_MIN_MAG_MIP_LINEAR, " \ - "addressU = TEXTURE_ADDRESS_CLAMP, " \ - "addressV = TEXTURE_ADDRESS_CLAMP, " \ - "addressW = TEXTURE_ADDRESS_CLAMP, " \ - "comparisonFunc = COMPARISON_NEVER, " \ - "borderColor = STATIC_BORDER_COLOR_TRANSPARENT_BLACK)" )] -#if defined(FFX_FSR3UPSCALER_EMBED_ROOTSIG) -#define FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT FFX_FSR3UPSCALER_ROOTSIG -#define FFX_FSR3UPSCALER_EMBED_CB2_ROOTSIG_CONTENT FFX_FSR3UPSCALER_CB2_ROOTSIG -#else -#define FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT -#define FFX_FSR3UPSCALER_EMBED_CB2_ROOTSIG_CONTENT -#endif // #if FFX_FSR3UPSCALER_EMBED_ROOTSIG - -#if defined(FSR3UPSCALER_BIND_CB_AUTOREACTIVE) -cbuffer cbGenerateReactive : FFX_FSR3UPSCALER_DECLARE_CB(FSR3UPSCALER_BIND_CB_AUTOREACTIVE) -{ - FfxFloat32 fTcThreshold; // 0.1 is a good starting value, lower will result in more TC pixels - FfxFloat32 fTcScale; - FfxFloat32 fReactiveScale; - FfxFloat32 fReactiveMax; -}; - -FfxFloat32 TcThreshold() -{ - return fTcThreshold; -} - -FfxFloat32 TcScale() -{ - return fTcScale; -} - -FfxFloat32 ReactiveScale() -{ - return fReactiveScale; -} - -FfxFloat32 ReactiveMax() -{ - return fReactiveMax; -} -#endif // #if defined(FSR3UPSCALER_BIND_CB_AUTOREACTIVE) - -#if defined(FSR3UPSCALER_BIND_CB_RCAS) -cbuffer cbRCAS : FFX_FSR3UPSCALER_DECLARE_CB(FSR3UPSCALER_BIND_CB_RCAS) -{ - FfxUInt32x4 rcasConfig; -}; - -FfxUInt32x4 RCASConfig() -{ - return rcasConfig; -} -#endif // #if defined(FSR3UPSCALER_BIND_CB_RCAS) - - -#if defined(FSR3UPSCALER_BIND_CB_REACTIVE) -cbuffer cbGenerateReactive : FFX_FSR3UPSCALER_DECLARE_CB(FSR3UPSCALER_BIND_CB_REACTIVE) -{ - FfxFloat32 gen_reactive_scale; - FfxFloat32 gen_reactive_threshold; - FfxFloat32 gen_reactive_binaryValue; - FfxUInt32 gen_reactive_flags; -}; - -FfxFloat32 GenReactiveScale() -{ - return gen_reactive_scale; -} - -FfxFloat32 GenReactiveThreshold() -{ - return gen_reactive_threshold; -} - -FfxFloat32 GenReactiveBinaryValue() -{ - return gen_reactive_binaryValue; -} - -FfxUInt32 GenReactiveFlags() -{ - return gen_reactive_flags; -} -#endif // #if defined(FSR3UPSCALER_BIND_CB_REACTIVE) - -#if defined(FSR3UPSCALER_BIND_CB_SPD) -cbuffer cbSPD : FFX_FSR3UPSCALER_DECLARE_CB(FSR3UPSCALER_BIND_CB_SPD) { - - FfxUInt32 mips; - FfxUInt32 numWorkGroups; - FfxUInt32x2 workGroupOffset; - FfxUInt32x2 renderSize; -}; - -FfxUInt32 MipCount() -{ - return mips; -} - -FfxUInt32 NumWorkGroups() -{ - return numWorkGroups; -} - -FfxUInt32x2 WorkGroupOffset() -{ - return workGroupOffset; -} - -FfxUInt32x2 SPD_RenderSize() -{ - return renderSize; -} -#endif // #if defined(FSR3UPSCALER_BIND_CB_SPD) - -// Declare and sample camera buffers as regular textures, unless overridden -#if !defined(UNITY_FSR3_TEX2D) -#define UNITY_FSR3_TEX2D(type) Texture2D -#endif -#if !defined(UNITY_FSR3_RWTEX2D) -#define UNITY_FSR3_RWTEX2D(type) RWTexture2D -#endif -#if !defined(UNITY_FSR3_POS) -#define UNITY_FSR3_POS(pxPos) (pxPos) -#endif -#if !defined(UNITY_FSR3_UV) -#define UNITY_FSR3_UV(uv) (uv) -#endif - -SamplerState s_PointClamp : register(s0); -SamplerState s_LinearClamp : register(s1); - - // SRVs - #if defined FSR3UPSCALER_BIND_SRV_INPUT_COLOR - UNITY_FSR3_TEX2D(FfxFloat32x4) r_input_color_jittered : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_INPUT_COLOR); - #endif - #if defined FSR3UPSCALER_BIND_SRV_INPUT_OPAQUE_ONLY - UNITY_FSR3_TEX2D(FfxFloat32x4) r_input_opaque_only : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_INPUT_OPAQUE_ONLY); - #endif - #if defined FSR3UPSCALER_BIND_SRV_INPUT_MOTION_VECTORS - UNITY_FSR3_TEX2D(FfxFloat32x4) r_input_motion_vectors : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_INPUT_MOTION_VECTORS); - #endif - #if defined FSR3UPSCALER_BIND_SRV_INPUT_DEPTH - UNITY_FSR3_TEX2D(FfxFloat32) r_input_depth : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_INPUT_DEPTH); - #endif - #if defined FSR3UPSCALER_BIND_SRV_INPUT_EXPOSURE - Texture2D r_input_exposure : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_INPUT_EXPOSURE); - #endif - #if defined FSR3UPSCALER_BIND_SRV_AUTO_EXPOSURE - Texture2D r_auto_exposure : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_AUTO_EXPOSURE); - #endif - #if defined FSR3UPSCALER_BIND_SRV_REACTIVE_MASK - UNITY_FSR3_TEX2D(FfxFloat32) r_reactive_mask : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_REACTIVE_MASK); - #endif - #if defined FSR3UPSCALER_BIND_SRV_TRANSPARENCY_AND_COMPOSITION_MASK - UNITY_FSR3_TEX2D(FfxFloat32) r_transparency_and_composition_mask : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_TRANSPARENCY_AND_COMPOSITION_MASK); - #endif - #if defined FSR3UPSCALER_BIND_SRV_RECONSTRUCTED_PREV_NEAREST_DEPTH - Texture2D r_reconstructed_previous_nearest_depth : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_RECONSTRUCTED_PREV_NEAREST_DEPTH); - #endif - #if defined FSR3UPSCALER_BIND_SRV_DILATED_MOTION_VECTORS - Texture2D r_dilated_motion_vectors : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_DILATED_MOTION_VECTORS); - #endif - #if defined FSR3UPSCALER_BIND_SRV_PREVIOUS_DILATED_MOTION_VECTORS - Texture2D r_previous_dilated_motion_vectors : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_PREVIOUS_DILATED_MOTION_VECTORS); - #endif - #if defined FSR3UPSCALER_BIND_SRV_DILATED_DEPTH - Texture2D r_dilated_depth : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_DILATED_DEPTH); - #endif - #if defined FSR3UPSCALER_BIND_SRV_INTERNAL_UPSCALED - Texture2D r_internal_upscaled_color : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_INTERNAL_UPSCALED); - #endif - #if defined FSR3UPSCALER_BIND_SRV_LOCK_STATUS - Texture2D r_lock_status : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_LOCK_STATUS); - #endif - #if defined FSR3UPSCALER_BIND_SRV_LOCK_INPUT_LUMA - Texture2D r_lock_input_luma : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_LOCK_INPUT_LUMA); - #endif - #if defined FSR3UPSCALER_BIND_SRV_NEW_LOCKS - Texture2D r_new_locks : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_NEW_LOCKS); - #endif - #if defined FSR3UPSCALER_BIND_SRV_PREPARED_INPUT_COLOR - Texture2D r_prepared_input_color : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_PREPARED_INPUT_COLOR); - #endif - #if defined FSR3UPSCALER_BIND_SRV_LUMA_HISTORY - Texture2D r_luma_history : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_LUMA_HISTORY); - #endif - #if defined FSR3UPSCALER_BIND_SRV_RCAS_INPUT - Texture2D r_rcas_input : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_RCAS_INPUT); - #endif - #if defined FSR3UPSCALER_BIND_SRV_LANCZOS_LUT - Texture2D r_lanczos_lut : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_LANCZOS_LUT); - #endif - #if defined FSR3UPSCALER_BIND_SRV_SCENE_LUMINANCE_MIPS - Texture2D r_imgMips : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_SCENE_LUMINANCE_MIPS); - #endif - #if defined FSR3UPSCALER_BIND_SRV_UPSCALE_MAXIMUM_BIAS_LUT - Texture2D r_upsample_maximum_bias_lut : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_UPSCALE_MAXIMUM_BIAS_LUT); - #endif - #if defined FSR3UPSCALER_BIND_SRV_DILATED_REACTIVE_MASKS - Texture2D r_dilated_reactive_masks : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_DILATED_REACTIVE_MASKS); - #endif - - #if defined FSR3UPSCALER_BIND_SRV_PREV_PRE_ALPHA_COLOR - Texture2D r_input_prev_color_pre_alpha : FFX_FSR3UPSCALER_DECLARE_SRV(FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_PREV_PRE_ALPHA_COLOR); - #endif - #if defined FSR3UPSCALER_BIND_SRV_PREV_POST_ALPHA_COLOR - Texture2D r_input_prev_color_post_alpha : FFX_FSR3UPSCALER_DECLARE_SRV(FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_PREV_POST_ALPHA_COLOR); - #endif - - // UAV declarations - #if defined FSR3UPSCALER_BIND_UAV_RECONSTRUCTED_PREV_NEAREST_DEPTH - RWTexture2D rw_reconstructed_previous_nearest_depth : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_RECONSTRUCTED_PREV_NEAREST_DEPTH); - #endif - #if defined FSR3UPSCALER_BIND_UAV_DILATED_MOTION_VECTORS - RWTexture2D rw_dilated_motion_vectors : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_DILATED_MOTION_VECTORS); - #endif - #if defined FSR3UPSCALER_BIND_UAV_DILATED_DEPTH - RWTexture2D rw_dilated_depth : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_DILATED_DEPTH); - #endif - #if defined FSR3UPSCALER_BIND_UAV_INTERNAL_UPSCALED - RWTexture2D rw_internal_upscaled_color : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_INTERNAL_UPSCALED); - #endif - #if defined FSR3UPSCALER_BIND_UAV_LOCK_STATUS - RWTexture2D rw_lock_status : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_LOCK_STATUS); - #endif - #if defined FSR3UPSCALER_BIND_UAV_LOCK_INPUT_LUMA - RWTexture2D rw_lock_input_luma : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_LOCK_INPUT_LUMA); - #endif - #if defined FSR3UPSCALER_BIND_UAV_NEW_LOCKS - RWTexture2D rw_new_locks : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_NEW_LOCKS); - #endif - #if defined FSR3UPSCALER_BIND_UAV_PREPARED_INPUT_COLOR - RWTexture2D rw_prepared_input_color : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_PREPARED_INPUT_COLOR); - #endif - #if defined FSR3UPSCALER_BIND_UAV_LUMA_HISTORY - RWTexture2D rw_luma_history : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_LUMA_HISTORY); - #endif - #if defined FSR3UPSCALER_BIND_UAV_UPSCALED_OUTPUT - UNITY_FSR3_RWTEX2D(FfxFloat32x4) rw_upscaled_output : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_UPSCALED_OUTPUT); - #endif - #if defined FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_LUMA_CHANGE - globallycoherent RWTexture2D rw_img_mip_shading_change : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_LUMA_CHANGE); - #endif - #if defined FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_5 - globallycoherent RWTexture2D rw_img_mip_5 : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_5); - #endif - #if defined FSR3UPSCALER_BIND_UAV_DILATED_REACTIVE_MASKS - RWTexture2D rw_dilated_reactive_masks : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_DILATED_REACTIVE_MASKS); - #endif - #if defined FSR3UPSCALER_BIND_UAV_EXPOSURE - RWTexture2D rw_exposure : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_EXPOSURE); - #endif - #if defined FSR3UPSCALER_BIND_UAV_AUTO_EXPOSURE - RWTexture2D rw_auto_exposure : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_AUTO_EXPOSURE); - #endif - #if defined FSR3UPSCALER_BIND_UAV_SPD_GLOBAL_ATOMIC - globallycoherent RWTexture2D rw_spd_global_atomic : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_SPD_GLOBAL_ATOMIC); - #endif - - #if defined FSR3UPSCALER_BIND_UAV_AUTOREACTIVE - RWTexture2D rw_output_autoreactive : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_AUTOREACTIVE); - #endif - #if defined FSR3UPSCALER_BIND_UAV_AUTOCOMPOSITION - RWTexture2D rw_output_autocomposition : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_AUTOCOMPOSITION); - #endif - #if defined FSR3UPSCALER_BIND_UAV_PREV_PRE_ALPHA_COLOR - RWTexture2D rw_output_prev_color_pre_alpha : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_PREV_PRE_ALPHA_COLOR); - #endif - #if defined FSR3UPSCALER_BIND_UAV_PREV_POST_ALPHA_COLOR - RWTexture2D rw_output_prev_color_post_alpha : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_PREV_POST_ALPHA_COLOR); - #endif - -#if defined(FSR3UPSCALER_BIND_SRV_SCENE_LUMINANCE_MIPS) -FfxFloat32 LoadMipLuma(FfxUInt32x2 iPxPos, FfxUInt32 mipLevel) -{ - return r_imgMips.mips[mipLevel][iPxPos]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_SCENE_LUMINANCE_MIPS) -FfxFloat32 SampleMipLuma(FfxFloat32x2 fUV, FfxUInt32 mipLevel) -{ - return r_imgMips.SampleLevel(s_LinearClamp, fUV, mipLevel); -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_INPUT_DEPTH) -FfxFloat32 LoadInputDepth(FfxUInt32x2 iPxPos) -{ - return r_input_depth[UNITY_FSR3_POS(iPxPos)]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_INPUT_DEPTH) -FfxFloat32 SampleInputDepth(FfxFloat32x2 fUV) -{ - return r_input_depth.SampleLevel(s_LinearClamp, UNITY_FSR3_UV(fUV), 0).x; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_REACTIVE_MASK) -FfxFloat32 LoadReactiveMask(FfxUInt32x2 iPxPos) -{ - return r_reactive_mask[UNITY_FSR3_POS(iPxPos)]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_TRANSPARENCY_AND_COMPOSITION_MASK) -FfxFloat32 LoadTransparencyAndCompositionMask(FfxUInt32x2 iPxPos) -{ - return r_transparency_and_composition_mask[UNITY_FSR3_POS(iPxPos)]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_INPUT_COLOR) -FfxFloat32x3 LoadInputColor(FfxUInt32x2 iPxPos) -{ - return r_input_color_jittered[UNITY_FSR3_POS(iPxPos)].rgb; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_INPUT_COLOR) -FfxFloat32x3 SampleInputColor(FfxFloat32x2 fUV) -{ - return r_input_color_jittered.SampleLevel(s_LinearClamp, UNITY_FSR3_UV(fUV), 0).rgb; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_PREPARED_INPUT_COLOR) -FfxFloat32x3 LoadPreparedInputColor(FfxUInt32x2 iPxPos) -{ - return r_prepared_input_color[iPxPos].xyz; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_INPUT_MOTION_VECTORS) -FfxFloat32x2 LoadInputMotionVector(FfxUInt32x2 iPxDilatedMotionVectorPos) -{ - FfxFloat32x2 fSrcMotionVector = r_input_motion_vectors[UNITY_FSR3_POS(iPxDilatedMotionVectorPos)].xy; - - FfxFloat32x2 fUvMotionVector = fSrcMotionVector * MotionVectorScale(); - -#if FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS - fUvMotionVector -= MotionVectorJitterCancellation(); -#endif - - return fUvMotionVector; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_INTERNAL_UPSCALED) -FfxFloat32x4 LoadHistory(FfxUInt32x2 iPxHistory) -{ - return r_internal_upscaled_color[iPxHistory]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_LUMA_HISTORY) -void StoreLumaHistory(FfxUInt32x2 iPxPos, FfxFloat32x4 fLumaHistory) -{ - rw_luma_history[iPxPos] = fLumaHistory; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_LUMA_HISTORY) -FfxFloat32x4 SampleLumaHistory(FfxFloat32x2 fUV) -{ - return r_luma_history.SampleLevel(s_LinearClamp, fUV, 0); -} -#endif - -FfxFloat32x4 LoadRCAS_Input(FfxInt32x2 iPxPos) -{ -#if defined(FSR3UPSCALER_BIND_SRV_RCAS_INPUT) - return r_rcas_input[iPxPos]; -#else - return 0.0; -#endif -} - -#if defined(FSR3UPSCALER_BIND_UAV_INTERNAL_UPSCALED) -void StoreReprojectedHistory(FfxUInt32x2 iPxHistory, FfxFloat32x4 fHistory) -{ - rw_internal_upscaled_color[iPxHistory] = fHistory; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_INTERNAL_UPSCALED) -void StoreInternalColorAndWeight(FfxUInt32x2 iPxPos, FfxFloat32x4 fColorAndWeight) -{ - rw_internal_upscaled_color[iPxPos] = fColorAndWeight; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_UPSCALED_OUTPUT) -void StoreUpscaledOutput(FfxUInt32x2 iPxPos, FfxFloat32x3 fColor) -{ - rw_upscaled_output[UNITY_FSR3_POS(iPxPos)] = FfxFloat32x4(fColor, 1.f); -} -#endif - -//LOCK_LIFETIME_REMAINING == 0 -//Should make LockInitialLifetime() return a const 1.0f later -#if defined(FSR3UPSCALER_BIND_SRV_LOCK_STATUS) -FfxFloat32x2 LoadLockStatus(FfxUInt32x2 iPxPos) -{ - return r_lock_status[iPxPos]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_LOCK_STATUS) -void StoreLockStatus(FfxUInt32x2 iPxPos, FfxFloat32x2 fLockStatus) -{ - rw_lock_status[iPxPos] = fLockStatus; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_LOCK_INPUT_LUMA) -FfxFloat32 LoadLockInputLuma(FfxUInt32x2 iPxPos) -{ - return r_lock_input_luma[iPxPos]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_LOCK_INPUT_LUMA) -void StoreLockInputLuma(FfxUInt32x2 iPxPos, FfxFloat32 fLuma) -{ - rw_lock_input_luma[iPxPos] = fLuma; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_NEW_LOCKS) -FfxFloat32 LoadNewLocks(FfxUInt32x2 iPxPos) -{ - return r_new_locks[iPxPos]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_NEW_LOCKS) -FfxFloat32 LoadRwNewLocks(FfxUInt32x2 iPxPos) -{ - return rw_new_locks[iPxPos]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_NEW_LOCKS) -void StoreNewLocks(FfxUInt32x2 iPxPos, FfxFloat32 newLock) -{ - rw_new_locks[iPxPos] = newLock; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_PREPARED_INPUT_COLOR) -void StorePreparedInputColor(FFX_PARAMETER_IN FfxUInt32x2 iPxPos, FFX_PARAMETER_IN FfxFloat32x4 fTonemapped) -{ - rw_prepared_input_color[iPxPos] = fTonemapped; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_PREPARED_INPUT_COLOR) -FfxFloat32 SampleDepthClip(FfxFloat32x2 fUV) -{ - return r_prepared_input_color.SampleLevel(s_LinearClamp, fUV, 0).w; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_LOCK_STATUS) -FfxFloat32x2 SampleLockStatus(FfxFloat32x2 fUV) -{ - FfxFloat32x2 fLockStatus = r_lock_status.SampleLevel(s_LinearClamp, fUV, 0); - return fLockStatus; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_RECONSTRUCTED_PREV_NEAREST_DEPTH) -FfxFloat32 LoadReconstructedPrevDepth(FfxUInt32x2 iPxPos) -{ - return asfloat(r_reconstructed_previous_nearest_depth[iPxPos]); -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_RECONSTRUCTED_PREV_NEAREST_DEPTH) -void StoreReconstructedDepth(FfxUInt32x2 iPxSample, FfxFloat32 fDepth) -{ - FfxUInt32 uDepth = asuint(fDepth); - - #if FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - InterlockedMax(rw_reconstructed_previous_nearest_depth[iPxSample], uDepth); - #else - InterlockedMin(rw_reconstructed_previous_nearest_depth[iPxSample], uDepth); // min for standard, max for inverted depth - #endif -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_RECONSTRUCTED_PREV_NEAREST_DEPTH) -void SetReconstructedDepth(FfxUInt32x2 iPxSample, const FfxUInt32 uValue) -{ - rw_reconstructed_previous_nearest_depth[iPxSample] = uValue; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_DILATED_DEPTH) -void StoreDilatedDepth(FFX_PARAMETER_IN FfxUInt32x2 iPxPos, FFX_PARAMETER_IN FfxFloat32 fDepth) -{ - rw_dilated_depth[iPxPos] = fDepth; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_DILATED_MOTION_VECTORS) -void StoreDilatedMotionVector(FFX_PARAMETER_IN FfxUInt32x2 iPxPos, FFX_PARAMETER_IN FfxFloat32x2 fMotionVector) -{ - rw_dilated_motion_vectors[iPxPos] = fMotionVector; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_DILATED_MOTION_VECTORS) -FfxFloat32x2 LoadDilatedMotionVector(FfxUInt32x2 iPxInput) -{ - return r_dilated_motion_vectors[iPxInput].xy; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_PREVIOUS_DILATED_MOTION_VECTORS) -FfxFloat32x2 LoadPreviousDilatedMotionVector(FfxUInt32x2 iPxInput) -{ - return r_previous_dilated_motion_vectors[iPxInput].xy; -} - -FfxFloat32x2 SamplePreviousDilatedMotionVector(FfxFloat32x2 uv) -{ - return r_previous_dilated_motion_vectors.SampleLevel(s_LinearClamp, uv, 0).xy; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_DILATED_DEPTH) -FfxFloat32 LoadDilatedDepth(FfxUInt32x2 iPxInput) -{ - return r_dilated_depth[iPxInput]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_INPUT_EXPOSURE) -FfxFloat32 Exposure() -{ - FfxFloat32 exposure = r_input_exposure[FfxUInt32x2(0, 0)].x; - - if (exposure == 0.0f) { - exposure = 1.0f; - } - - return exposure; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_AUTO_EXPOSURE) -FfxFloat32 AutoExposure() -{ - FfxFloat32 exposure = r_auto_exposure[FfxUInt32x2(0, 0)].x; - - if (exposure == 0.0f) { - exposure = 1.0f; - } - - return exposure; -} -#endif - -FfxFloat32 SampleLanczos2Weight(FfxFloat32 x) -{ -#if defined(FSR3UPSCALER_BIND_SRV_LANCZOS_LUT) - return r_lanczos_lut.SampleLevel(s_LinearClamp, FfxFloat32x2(x / 2, 0.5f), 0); -#else - return 0.f; -#endif -} - -#if defined(FSR3UPSCALER_BIND_SRV_UPSCALE_MAXIMUM_BIAS_LUT) -FfxFloat32 SampleUpsampleMaximumBias(FfxFloat32x2 uv) -{ - // Stored as a SNORM, so make sure to multiply by 2 to retrieve the actual expected range. - return FfxFloat32(2.0) * r_upsample_maximum_bias_lut.SampleLevel(s_LinearClamp, abs(uv) * 2.0, 0); -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_DILATED_REACTIVE_MASKS) -FfxFloat32x2 SampleDilatedReactiveMasks(FfxFloat32x2 fUV) -{ - return r_dilated_reactive_masks.SampleLevel(s_LinearClamp, fUV, 0); -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_DILATED_REACTIVE_MASKS) -FfxFloat32x2 LoadDilatedReactiveMasks(FFX_PARAMETER_IN FfxUInt32x2 iPxPos) -{ - return r_dilated_reactive_masks[iPxPos]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_DILATED_REACTIVE_MASKS) -void StoreDilatedReactiveMasks(FFX_PARAMETER_IN FfxUInt32x2 iPxPos, FFX_PARAMETER_IN FfxFloat32x2 fDilatedReactiveMasks) -{ - rw_dilated_reactive_masks[iPxPos] = fDilatedReactiveMasks; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_INPUT_OPAQUE_ONLY) -FfxFloat32x3 LoadOpaqueOnly(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos) -{ - return r_input_opaque_only[UNITY_FSR3_POS(iPxPos)].xyz; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_PREV_PRE_ALPHA_COLOR) -FfxFloat32x3 LoadPrevPreAlpha(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos) -{ - return r_input_prev_color_pre_alpha[iPxPos]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_PREV_POST_ALPHA_COLOR) -FfxFloat32x3 LoadPrevPostAlpha(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos) -{ - return r_input_prev_color_post_alpha[iPxPos]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_AUTOREACTIVE) -#if defined(FSR3UPSCALER_BIND_UAV_AUTOCOMPOSITION) -void StoreAutoReactive(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos, FFX_PARAMETER_IN FFX_MIN16_F2 fReactive) -{ - rw_output_autoreactive[iPxPos] = fReactive.x; - - rw_output_autocomposition[iPxPos] = fReactive.y; -} -#endif -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_PREV_PRE_ALPHA_COLOR) -void StorePrevPreAlpha(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos, FFX_PARAMETER_IN FFX_MIN16_F3 color) -{ - rw_output_prev_color_pre_alpha[iPxPos] = color; - -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_PREV_POST_ALPHA_COLOR) -void StorePrevPostAlpha(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos, FFX_PARAMETER_IN FFX_MIN16_F3 color) -{ - rw_output_prev_color_post_alpha[iPxPos] = color; -} -#endif - -FfxFloat32x2 SPD_LoadExposureBuffer() -{ -#if defined FSR3UPSCALER_BIND_UAV_AUTO_EXPOSURE - return rw_auto_exposure[FfxInt32x2(0, 0)]; -#else - return FfxFloat32x2(0.f, 0.f); -#endif // #if defined FSR3UPSCALER_BIND_UAV_AUTO_EXPOSURE -} - -void SPD_SetExposureBuffer(FfxFloat32x2 value) -{ -#if defined FSR3UPSCALER_BIND_UAV_AUTO_EXPOSURE - rw_auto_exposure[FfxInt32x2(0, 0)] = value; -#endif // #if defined FSR3UPSCALER_BIND_UAV_AUTO_EXPOSURE -} - -FfxFloat32x4 SPD_LoadMipmap5(FfxInt32x2 iPxPos) -{ -#if defined FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_5 - return FfxFloat32x4(rw_img_mip_5[iPxPos], 0, 0, 0); -#else - return FfxFloat32x4(0.f, 0.f, 0.f, 0.f); -#endif // #if defined FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_5 -} - -void SPD_SetMipmap(FfxInt32x2 iPxPos, FfxUInt32 slice, FfxFloat32 value) -{ - switch (slice) - { - case FFX_FSR3UPSCALER_SHADING_CHANGE_MIP_LEVEL: -#if defined FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_LUMA_CHANGE - rw_img_mip_shading_change[iPxPos] = value; -#endif // #if defined FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_LUMA_CHANGE - break; - case 5: -#if defined FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_5 - rw_img_mip_5[iPxPos] = value; -#endif // #if defined FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_5 - break; - default: - - // avoid flattened side effect -#if defined(FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_LUMA_CHANGE) - rw_img_mip_shading_change[iPxPos] = rw_img_mip_shading_change[iPxPos]; -#elif defined(FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_5) - rw_img_mip_5[iPxPos] = rw_img_mip_5[iPxPos]; -#endif // #if defined FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_5 - break; - } -} - -void SPD_IncreaseAtomicCounter(inout FfxUInt32 spdCounter) -{ -#if defined FSR3UPSCALER_BIND_UAV_SPD_GLOBAL_ATOMIC - InterlockedAdd(rw_spd_global_atomic[FfxInt32x2(0, 0)], 1, spdCounter); -#endif // #if defined FSR3UPSCALER_BIND_UAV_SPD_GLOBAL_ATOMIC -} - -void SPD_ResetAtomicCounter() -{ -#if defined FSR3UPSCALER_BIND_UAV_SPD_GLOBAL_ATOMIC - rw_spd_global_atomic[FfxInt32x2(0, 0)] = 0; -#endif // #if defined FSR3UPSCALER_BIND_UAV_SPD_GLOBAL_ATOMIC -} - -#endif // #if defined(FFX_GPU) diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h.meta deleted file mode 100644 index e78e6a1..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: ba849fdeb042e7f458c81408414db834 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_common.h b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_common.h deleted file mode 100644 index 1f78a29..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_common.h +++ /dev/null @@ -1,566 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#if !defined(FFX_FSR3UPSCALER_COMMON_H) -#define FFX_FSR3UPSCALER_COMMON_H - -#if defined(FFX_CPU) || defined(FFX_GPU) -//Locks -#define LOCK_LIFETIME_REMAINING 0 -#define LOCK_TEMPORAL_LUMA 1 -#endif // #if defined(FFX_CPU) || defined(FFX_GPU) - -#if defined(FFX_GPU) -FFX_STATIC const FfxFloat32 FSR3UPSCALER_FP16_MIN = 6.10e-05f; -FFX_STATIC const FfxFloat32 FSR3UPSCALER_FP16_MAX = 65504.0f; -FFX_STATIC const FfxFloat32 FSR3UPSCALER_EPSILON = 1e-03f; -FFX_STATIC const FfxFloat32 FSR3UPSCALER_TONEMAP_EPSILON = 1.0f / FSR3UPSCALER_FP16_MAX; -FFX_STATIC const FfxFloat32 FSR3UPSCALER_FLT_MAX = 3.402823466e+38f; -FFX_STATIC const FfxFloat32 FSR3UPSCALER_FLT_MIN = 1.175494351e-38f; - -// treat vector truncation warnings as errors -#pragma warning(error: 3206) - -// suppress warnings -#pragma warning(disable: 3205) // conversion from larger type to smaller -#pragma warning(disable: 3571) // in ffxPow(f, e), f could be negative - -// Reconstructed depth usage -FFX_STATIC const FfxFloat32 fReconstructedDepthBilinearWeightThreshold = 0.01f; - -// Accumulation -FFX_STATIC const FfxFloat32 fUpsampleLanczosWeightScale = 1.0f / 12.0f; -FFX_STATIC const FfxFloat32 fMaxAccumulationLanczosWeight = 1.0f; -FFX_STATIC const FfxFloat32 fAverageLanczosWeightPerFrame = 0.74f * fUpsampleLanczosWeightScale; // Average lanczos weight for jitter accumulated samples -FFX_STATIC const FfxFloat32 fAccumulationMaxOnMotion = 3.0f * fUpsampleLanczosWeightScale; - -// Auto exposure -FFX_STATIC const FfxFloat32 resetAutoExposureAverageSmoothing = 1e8f; - -struct AccumulationPassCommonParams -{ - FfxInt32x2 iPxHrPos; - FfxFloat32x2 fHrUv; - FfxFloat32x2 fLrUv_HwSampler; - FfxFloat32x2 fMotionVector; - FfxFloat32x2 fReprojectedHrUv; - FfxFloat32 fHrVelocity; - FfxFloat32 fDepthClipFactor; - FfxFloat32 fDilatedReactiveFactor; - FfxFloat32 fAccumulationMask; - - FfxBoolean bIsResetFrame; - FfxBoolean bIsExistingSample; - FfxBoolean bIsNewSample; -}; - -struct LockState -{ - FfxBoolean NewLock; //Set for both unique new and re-locked new - FfxBoolean WasLockedPrevFrame; //Set to identify if the pixel was already locked (relock) -}; - -void InitializeNewLockSample(FFX_PARAMETER_OUT FfxFloat32x2 fLockStatus) -{ - fLockStatus = FfxFloat32x2(0, 0); -} - -#if FFX_HALF -void InitializeNewLockSample(FFX_PARAMETER_OUT FFX_MIN16_F2 fLockStatus) -{ - fLockStatus = FFX_MIN16_F2(0, 0); -} -#endif - - -void KillLock(FFX_PARAMETER_INOUT FfxFloat32x2 fLockStatus) -{ - fLockStatus[LOCK_LIFETIME_REMAINING] = 0; -} - -#if FFX_HALF -void KillLock(FFX_PARAMETER_INOUT FFX_MIN16_F2 fLockStatus) -{ - fLockStatus[LOCK_LIFETIME_REMAINING] = FFX_MIN16_F(0); -} -#endif - -struct RectificationBox -{ - FfxFloat32x3 boxCenter; - FfxFloat32x3 boxVec; - FfxFloat32x3 aabbMin; - FfxFloat32x3 aabbMax; - FfxFloat32 fBoxCenterWeight; -}; -#if FFX_HALF -struct RectificationBoxMin16 -{ - FFX_MIN16_F3 boxCenter; - FFX_MIN16_F3 boxVec; - FFX_MIN16_F3 aabbMin; - FFX_MIN16_F3 aabbMax; - FFX_MIN16_F fBoxCenterWeight; -}; -#endif - -void RectificationBoxReset(FFX_PARAMETER_INOUT RectificationBox rectificationBox) -{ - rectificationBox.fBoxCenterWeight = FfxFloat32(0); - - rectificationBox.boxCenter = FfxFloat32x3(0, 0, 0); - rectificationBox.boxVec = FfxFloat32x3(0, 0, 0); - rectificationBox.aabbMin = FfxFloat32x3(FSR3UPSCALER_FLT_MAX, FSR3UPSCALER_FLT_MAX, FSR3UPSCALER_FLT_MAX); - rectificationBox.aabbMax = -FfxFloat32x3(FSR3UPSCALER_FLT_MAX, FSR3UPSCALER_FLT_MAX, FSR3UPSCALER_FLT_MAX); -} -#if FFX_HALF -void RectificationBoxReset(FFX_PARAMETER_INOUT RectificationBoxMin16 rectificationBox) -{ - rectificationBox.fBoxCenterWeight = FFX_MIN16_F(0); - - rectificationBox.boxCenter = FFX_MIN16_F3(0, 0, 0); - rectificationBox.boxVec = FFX_MIN16_F3(0, 0, 0); - rectificationBox.aabbMin = FFX_MIN16_F3(FSR3UPSCALER_FP16_MAX, FSR3UPSCALER_FP16_MAX, FSR3UPSCALER_FP16_MAX); - rectificationBox.aabbMax = -FFX_MIN16_F3(FSR3UPSCALER_FP16_MAX, FSR3UPSCALER_FP16_MAX, FSR3UPSCALER_FP16_MAX); -} -#endif - -void RectificationBoxAddInitialSample(FFX_PARAMETER_INOUT RectificationBox rectificationBox, const FfxFloat32x3 colorSample, const FfxFloat32 fSampleWeight) -{ - rectificationBox.aabbMin = colorSample; - rectificationBox.aabbMax = colorSample; - - FfxFloat32x3 weightedSample = colorSample * fSampleWeight; - rectificationBox.boxCenter = weightedSample; - rectificationBox.boxVec = colorSample * weightedSample; - rectificationBox.fBoxCenterWeight = fSampleWeight; -} - -void RectificationBoxAddSample(FfxBoolean bInitialSample, FFX_PARAMETER_INOUT RectificationBox rectificationBox, const FfxFloat32x3 colorSample, const FfxFloat32 fSampleWeight) -{ - if (bInitialSample) { - RectificationBoxAddInitialSample(rectificationBox, colorSample, fSampleWeight); - } else { - rectificationBox.aabbMin = ffxMin(rectificationBox.aabbMin, colorSample); - rectificationBox.aabbMax = ffxMax(rectificationBox.aabbMax, colorSample); - - FfxFloat32x3 weightedSample = colorSample * fSampleWeight; - rectificationBox.boxCenter += weightedSample; - rectificationBox.boxVec += colorSample * weightedSample; - rectificationBox.fBoxCenterWeight += fSampleWeight; - } -} -#if FFX_HALF -void RectificationBoxAddInitialSample(FFX_PARAMETER_INOUT RectificationBoxMin16 rectificationBox, const FFX_MIN16_F3 colorSample, const FFX_MIN16_F fSampleWeight) -{ - rectificationBox.aabbMin = colorSample; - rectificationBox.aabbMax = colorSample; - - FFX_MIN16_F3 weightedSample = colorSample * fSampleWeight; - rectificationBox.boxCenter = weightedSample; - rectificationBox.boxVec = colorSample * weightedSample; - rectificationBox.fBoxCenterWeight = fSampleWeight; -} - -void RectificationBoxAddSample(FfxBoolean bInitialSample, FFX_PARAMETER_INOUT RectificationBoxMin16 rectificationBox, const FFX_MIN16_F3 colorSample, const FFX_MIN16_F fSampleWeight) -{ - if (bInitialSample) { - RectificationBoxAddInitialSample(rectificationBox, colorSample, fSampleWeight); - } else { - rectificationBox.aabbMin = ffxMin(rectificationBox.aabbMin, colorSample); - rectificationBox.aabbMax = ffxMax(rectificationBox.aabbMax, colorSample); - - FFX_MIN16_F3 weightedSample = colorSample * fSampleWeight; - rectificationBox.boxCenter += weightedSample; - rectificationBox.boxVec += colorSample * weightedSample; - rectificationBox.fBoxCenterWeight += fSampleWeight; - } -} -#endif - -void RectificationBoxComputeVarianceBoxData(FFX_PARAMETER_INOUT RectificationBox rectificationBox) -{ - rectificationBox.fBoxCenterWeight = (abs(rectificationBox.fBoxCenterWeight) > FfxFloat32(FSR3UPSCALER_EPSILON) ? rectificationBox.fBoxCenterWeight : FfxFloat32(1.f)); - rectificationBox.boxCenter /= rectificationBox.fBoxCenterWeight; - rectificationBox.boxVec /= rectificationBox.fBoxCenterWeight; - FfxFloat32x3 stdDev = sqrt(abs(rectificationBox.boxVec - rectificationBox.boxCenter * rectificationBox.boxCenter)); - rectificationBox.boxVec = stdDev; -} -#if FFX_HALF -void RectificationBoxComputeVarianceBoxData(FFX_PARAMETER_INOUT RectificationBoxMin16 rectificationBox) -{ - rectificationBox.fBoxCenterWeight = (abs(rectificationBox.fBoxCenterWeight) > FFX_MIN16_F(FSR3UPSCALER_EPSILON) ? rectificationBox.fBoxCenterWeight : FFX_MIN16_F(1.f)); - rectificationBox.boxCenter /= rectificationBox.fBoxCenterWeight; - rectificationBox.boxVec /= rectificationBox.fBoxCenterWeight; - FFX_MIN16_F3 stdDev = sqrt(abs(rectificationBox.boxVec - rectificationBox.boxCenter * rectificationBox.boxCenter)); - rectificationBox.boxVec = stdDev; -} -#endif - -FfxFloat32x3 SafeRcp3(FfxFloat32x3 v) -{ - return (all(FFX_NOT_EQUAL(v, FfxFloat32x3(0, 0, 0)))) ? (FfxFloat32x3(1, 1, 1) / v) : FfxFloat32x3(0, 0, 0); -} -#if FFX_HALF -FFX_MIN16_F3 SafeRcp3(FFX_MIN16_F3 v) -{ - return (all(FFX_NOT_EQUAL(v, FFX_MIN16_F3(0, 0, 0)))) ? (FFX_MIN16_F3(1, 1, 1) / v) : FFX_MIN16_F3(0, 0, 0); -} -#endif - -FfxFloat32 MinDividedByMax(const FfxFloat32 v0, const FfxFloat32 v1) -{ - const FfxFloat32 m = ffxMax(v0, v1); - return m != 0 ? ffxMin(v0, v1) / m : 0; -} - -#if FFX_HALF -FFX_MIN16_F MinDividedByMax(const FFX_MIN16_F v0, const FFX_MIN16_F v1) -{ - const FFX_MIN16_F m = ffxMax(v0, v1); - return m != FFX_MIN16_F(0) ? ffxMin(v0, v1) / m : FFX_MIN16_F(0); -} -#endif - -FfxFloat32x3 YCoCgToRGB(FfxFloat32x3 fYCoCg) -{ - FfxFloat32x3 fRgb; - - fRgb = FfxFloat32x3( - fYCoCg.x + fYCoCg.y - fYCoCg.z, - fYCoCg.x + fYCoCg.z, - fYCoCg.x - fYCoCg.y - fYCoCg.z); - - return fRgb; -} -#if FFX_HALF -FFX_MIN16_F3 YCoCgToRGB(FFX_MIN16_F3 fYCoCg) -{ - FFX_MIN16_F3 fRgb; - - fRgb = FFX_MIN16_F3( - fYCoCg.x + fYCoCg.y - fYCoCg.z, - fYCoCg.x + fYCoCg.z, - fYCoCg.x - fYCoCg.y - fYCoCg.z); - - return fRgb; -} -#endif - -FfxFloat32x3 RGBToYCoCg(FfxFloat32x3 fRgb) -{ - FfxFloat32x3 fYCoCg; - - fYCoCg = FfxFloat32x3( - 0.25f * fRgb.r + 0.5f * fRgb.g + 0.25f * fRgb.b, - 0.5f * fRgb.r - 0.5f * fRgb.b, - -0.25f * fRgb.r + 0.5f * fRgb.g - 0.25f * fRgb.b); - - return fYCoCg; -} -#if FFX_HALF -FFX_MIN16_F3 RGBToYCoCg(FFX_MIN16_F3 fRgb) -{ - FFX_MIN16_F3 fYCoCg; - - fYCoCg = FFX_MIN16_F3( - 0.25 * fRgb.r + 0.5 * fRgb.g + 0.25 * fRgb.b, - 0.5 * fRgb.r - 0.5 * fRgb.b, - -0.25 * fRgb.r + 0.5 * fRgb.g - 0.25 * fRgb.b); - - return fYCoCg; -} -#endif - -FfxFloat32 RGBToLuma(FfxFloat32x3 fLinearRgb) -{ - return dot(fLinearRgb, FfxFloat32x3(0.2126f, 0.7152f, 0.0722f)); -} -#if FFX_HALF -FFX_MIN16_F RGBToLuma(FFX_MIN16_F3 fLinearRgb) -{ - return dot(fLinearRgb, FFX_MIN16_F3(0.2126f, 0.7152f, 0.0722f)); -} -#endif - -FfxFloat32 RGBToPerceivedLuma(FfxFloat32x3 fLinearRgb) -{ - FfxFloat32 fLuminance = RGBToLuma(fLinearRgb); - - FfxFloat32 fPercievedLuminance = 0; - if (fLuminance <= 216.0f / 24389.0f) { - fPercievedLuminance = fLuminance * (24389.0f / 27.0f); - } - else { - fPercievedLuminance = ffxPow(fLuminance, 1.0f / 3.0f) * 116.0f - 16.0f; - } - - return fPercievedLuminance * 0.01f; -} -#if FFX_HALF -FFX_MIN16_F RGBToPerceivedLuma(FFX_MIN16_F3 fLinearRgb) -{ - FFX_MIN16_F fLuminance = RGBToLuma(fLinearRgb); - - FFX_MIN16_F fPercievedLuminance = FFX_MIN16_F(0); - if (fLuminance <= FFX_MIN16_F(216.0f / 24389.0f)) { - fPercievedLuminance = fLuminance * FFX_MIN16_F(24389.0f / 27.0f); - } - else { - fPercievedLuminance = ffxPow(fLuminance, FFX_MIN16_F(1.0f / 3.0f)) * FFX_MIN16_F(116.0f) - FFX_MIN16_F(16.0f); - } - - return fPercievedLuminance * FFX_MIN16_F(0.01f); -} -#endif - -FfxFloat32x3 Tonemap(FfxFloat32x3 fRgb) -{ - return fRgb / (ffxMax(ffxMax(0.f, fRgb.r), ffxMax(fRgb.g, fRgb.b)) + 1.f).xxx; -} - -FfxFloat32x3 InverseTonemap(FfxFloat32x3 fRgb) -{ - return fRgb / ffxMax(FSR3UPSCALER_TONEMAP_EPSILON, 1.f - ffxMax(fRgb.r, ffxMax(fRgb.g, fRgb.b))).xxx; -} - -#if FFX_HALF -FFX_MIN16_F3 Tonemap(FFX_MIN16_F3 fRgb) -{ - return fRgb / (ffxMax(ffxMax(FFX_MIN16_F(0.f), fRgb.r), ffxMax(fRgb.g, fRgb.b)) + FFX_MIN16_F(1.f)).xxx; -} - -FFX_MIN16_F3 InverseTonemap(FFX_MIN16_F3 fRgb) -{ - return fRgb / ffxMax(FFX_MIN16_F(FSR3UPSCALER_TONEMAP_EPSILON), FFX_MIN16_F(1.f) - ffxMax(fRgb.r, ffxMax(fRgb.g, fRgb.b))).xxx; -} -#endif - -FfxInt32x2 ClampLoad(FfxInt32x2 iPxSample, FfxInt32x2 iPxOffset, FfxInt32x2 iTextureSize) -{ - FfxInt32x2 result = iPxSample + iPxOffset; - result.x = (iPxOffset.x < 0) ? ffxMax(result.x, 0) : result.x; - result.x = (iPxOffset.x > 0) ? ffxMin(result.x, iTextureSize.x - 1) : result.x; - result.y = (iPxOffset.y < 0) ? ffxMax(result.y, 0) : result.y; - result.y = (iPxOffset.y > 0) ? ffxMin(result.y, iTextureSize.y - 1) : result.y; - return result; - - // return ffxMed3(iPxSample + iPxOffset, FfxInt32x2(0, 0), iTextureSize - FfxInt32x2(1, 1)); -} -#if FFX_HALF -FFX_MIN16_I2 ClampLoad(FFX_MIN16_I2 iPxSample, FFX_MIN16_I2 iPxOffset, FFX_MIN16_I2 iTextureSize) -{ - FFX_MIN16_I2 result = iPxSample + iPxOffset; - result.x = (iPxOffset.x < 0) ? ffxMax(result.x, FFX_MIN16_I(0)) : result.x; - result.x = (iPxOffset.x > 0) ? ffxMin(result.x, iTextureSize.x - FFX_MIN16_I(1)) : result.x; - result.y = (iPxOffset.y < 0) ? ffxMax(result.y, FFX_MIN16_I(0)) : result.y; - result.y = (iPxOffset.y > 0) ? ffxMin(result.y, iTextureSize.y - FFX_MIN16_I(1)) : result.y; - return result; - - // return ffxMed3Half(iPxSample + iPxOffset, FFX_MIN16_I2(0, 0), iTextureSize - FFX_MIN16_I2(1, 1)); -} -#endif - -FfxFloat32x2 ClampUv(FfxFloat32x2 fUv, FfxInt32x2 iTextureSize, FfxInt32x2 iResourceSize) -{ - const FfxFloat32x2 fSampleLocation = fUv * iTextureSize; - const FfxFloat32x2 fClampedLocation = ffxMax(FfxFloat32x2(0.5f, 0.5f), ffxMin(fSampleLocation, FfxFloat32x2(iTextureSize) - FfxFloat32x2(0.5f, 0.5f))); - const FfxFloat32x2 fClampedUv = fClampedLocation / FfxFloat32x2(iResourceSize); - - return fClampedUv; -} - -FfxBoolean IsOnScreen(FfxInt32x2 pos, FfxInt32x2 size) -{ - return all(FFX_LESS_THAN(FfxUInt32x2(pos), FfxUInt32x2(size))); -} -#if FFX_HALF -FfxBoolean IsOnScreen(FFX_MIN16_I2 pos, FFX_MIN16_I2 size) -{ - return all(FFX_LESS_THAN(FFX_MIN16_U2(pos), FFX_MIN16_U2(size))); -} -#endif - -FfxFloat32 ComputeAutoExposureFromLavg(FfxFloat32 Lavg) -{ - Lavg = exp(Lavg); - - const FfxFloat32 S = 100.0f; //ISO arithmetic speed - const FfxFloat32 K = 12.5f; - FfxFloat32 ExposureISO100 = log2((Lavg * S) / K); - - const FfxFloat32 q = 0.65f; - FfxFloat32 Lmax = (78.0f / (q * S)) * ffxPow(2.0f, ExposureISO100); - - return 1 / Lmax; -} -#if FFX_HALF -FFX_MIN16_F ComputeAutoExposureFromLavg(FFX_MIN16_F Lavg) -{ - Lavg = exp(Lavg); - - const FFX_MIN16_F S = FFX_MIN16_F(100.0f); //ISO arithmetic speed - const FFX_MIN16_F K = FFX_MIN16_F(12.5f); - const FFX_MIN16_F ExposureISO100 = log2((Lavg * S) / K); - - const FFX_MIN16_F q = FFX_MIN16_F(0.65f); - const FFX_MIN16_F Lmax = (FFX_MIN16_F(78.0f) / (q * S)) * ffxPow(FFX_MIN16_F(2.0f), ExposureISO100); - - return FFX_MIN16_F(1) / Lmax; -} -#endif - -FfxInt32x2 ComputeHrPosFromLrPos(FfxInt32x2 iPxLrPos) -{ - FfxFloat32x2 fSrcJitteredPos = FfxFloat32x2(iPxLrPos) + 0.5f - Jitter(); - FfxFloat32x2 fLrPosInHr = (fSrcJitteredPos / RenderSize()) * DisplaySize(); - FfxInt32x2 iPxHrPos = FfxInt32x2(floor(fLrPosInHr)); - return iPxHrPos; -} -#if FFX_HALF -FFX_MIN16_I2 ComputeHrPosFromLrPos(FFX_MIN16_I2 iPxLrPos) -{ - FFX_MIN16_F2 fSrcJitteredPos = FFX_MIN16_F2(iPxLrPos) + FFX_MIN16_F(0.5f) - FFX_MIN16_F2(Jitter()); - FFX_MIN16_F2 fLrPosInHr = (fSrcJitteredPos / FFX_MIN16_F2(RenderSize())) * FFX_MIN16_F2(DisplaySize()); - FFX_MIN16_I2 iPxHrPos = FFX_MIN16_I2(floor(fLrPosInHr)); - return iPxHrPos; -} -#endif - -FfxFloat32x2 ComputeNdc(FfxFloat32x2 fPxPos, FfxInt32x2 iSize) -{ - return fPxPos / FfxFloat32x2(iSize) * FfxFloat32x2(2.0f, -2.0f) + FfxFloat32x2(-1.0f, 1.0f); -} - -FfxFloat32 GetViewSpaceDepth(FfxFloat32 fDeviceDepth) -{ - const FfxFloat32x4 fDeviceToViewDepth = DeviceToViewSpaceTransformFactors(); - - // fDeviceToViewDepth details found in ffx_fsr3upscaler.cpp - return (fDeviceToViewDepth[1] / (fDeviceDepth - fDeviceToViewDepth[0])); -} - -FfxFloat32 GetViewSpaceDepthInMeters(FfxFloat32 fDeviceDepth) -{ - return GetViewSpaceDepth(fDeviceDepth) * ViewSpaceToMetersFactor(); -} - -FfxFloat32x3 GetViewSpacePosition(FfxInt32x2 iViewportPos, FfxInt32x2 iViewportSize, FfxFloat32 fDeviceDepth) -{ - const FfxFloat32x4 fDeviceToViewDepth = DeviceToViewSpaceTransformFactors(); - - const FfxFloat32 Z = GetViewSpaceDepth(fDeviceDepth); - - const FfxFloat32x2 fNdcPos = ComputeNdc(iViewportPos, iViewportSize); - const FfxFloat32 X = fDeviceToViewDepth[2] * fNdcPos.x * Z; - const FfxFloat32 Y = fDeviceToViewDepth[3] * fNdcPos.y * Z; - - return FfxFloat32x3(X, Y, Z); -} - -FfxFloat32x3 GetViewSpacePositionInMeters(FfxInt32x2 iViewportPos, FfxInt32x2 iViewportSize, FfxFloat32 fDeviceDepth) -{ - return GetViewSpacePosition(iViewportPos, iViewportSize, fDeviceDepth) * ViewSpaceToMetersFactor(); -} - -FfxFloat32 GetMaxDistanceInMeters() -{ -#if FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - return GetViewSpaceDepth(0.0f) * ViewSpaceToMetersFactor(); -#else - return GetViewSpaceDepth(1.0f) * ViewSpaceToMetersFactor(); -#endif -} - -FfxFloat32x3 PrepareRgb(FfxFloat32x3 fRgb, FfxFloat32 fExposure, FfxFloat32 fPreExposure) -{ - fRgb /= fPreExposure; - fRgb *= fExposure; - - fRgb = clamp(fRgb, 0.0f, FSR3UPSCALER_FP16_MAX); - - return fRgb; -} - -FfxFloat32x3 UnprepareRgb(FfxFloat32x3 fRgb, FfxFloat32 fExposure) -{ - fRgb /= fExposure; - fRgb *= PreExposure(); - - return fRgb; -} - - -struct BilinearSamplingData -{ - FfxInt32x2 iOffsets[4]; - FfxFloat32 fWeights[4]; - FfxInt32x2 iBasePos; -}; - -BilinearSamplingData GetBilinearSamplingData(FfxFloat32x2 fUv, FfxInt32x2 iSize) -{ - BilinearSamplingData data; - - FfxFloat32x2 fPxSample = (fUv * iSize) - FfxFloat32x2(0.5f, 0.5f); - data.iBasePos = FfxInt32x2(floor(fPxSample)); - FfxFloat32x2 fPxFrac = ffxFract(fPxSample); - - data.iOffsets[0] = FfxInt32x2(0, 0); - data.iOffsets[1] = FfxInt32x2(1, 0); - data.iOffsets[2] = FfxInt32x2(0, 1); - data.iOffsets[3] = FfxInt32x2(1, 1); - - data.fWeights[0] = (1 - fPxFrac.x) * (1 - fPxFrac.y); - data.fWeights[1] = (fPxFrac.x) * (1 - fPxFrac.y); - data.fWeights[2] = (1 - fPxFrac.x) * (fPxFrac.y); - data.fWeights[3] = (fPxFrac.x) * (fPxFrac.y); - - return data; -} - -struct PlaneData -{ - FfxFloat32x3 fNormal; - FfxFloat32 fDistanceFromOrigin; -}; - -PlaneData GetPlaneFromPoints(FfxFloat32x3 fP0, FfxFloat32x3 fP1, FfxFloat32x3 fP2) -{ - PlaneData plane; - - FfxFloat32x3 v0 = fP0 - fP1; - FfxFloat32x3 v1 = fP0 - fP2; - plane.fNormal = normalize(cross(v0, v1)); - plane.fDistanceFromOrigin = -dot(fP0, plane.fNormal); - - return plane; -} - -FfxFloat32 PointToPlaneDistance(PlaneData plane, FfxFloat32x3 fPoint) -{ - return abs(dot(plane.fNormal, fPoint) + plane.fDistanceFromOrigin); -} - -#endif // #if defined(FFX_GPU) - -#endif //!defined(FFX_FSR3UPSCALER_COMMON_H) diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_common.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_common.h.meta deleted file mode 100644 index 08b2046..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_common.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 6a638bec681caac4fa8e2ca198726694 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_compute_luminance_pyramid.h b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_compute_luminance_pyramid.h deleted file mode 100644 index d26cf23..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_compute_luminance_pyramid.h +++ /dev/null @@ -1,176 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -FFX_GROUPSHARED FfxUInt32 spdCounter; - -void SpdIncreaseAtomicCounter(FfxUInt32 slice) -{ - SPD_IncreaseAtomicCounter(spdCounter); -} - -FfxUInt32 SpdGetAtomicCounter() -{ - return spdCounter; -} - -void SpdResetAtomicCounter(FfxUInt32 slice) -{ - SPD_ResetAtomicCounter(); -} - -#ifndef SPD_PACKED_ONLY -FFX_GROUPSHARED FfxFloat32 spdIntermediateR[16][16]; -FFX_GROUPSHARED FfxFloat32 spdIntermediateG[16][16]; -FFX_GROUPSHARED FfxFloat32 spdIntermediateB[16][16]; -FFX_GROUPSHARED FfxFloat32 spdIntermediateA[16][16]; - -FfxFloat32x4 SpdLoadSourceImage(FfxFloat32x2 tex, FfxUInt32 slice) -{ - FfxFloat32x2 fUv = (tex + 0.5f + Jitter()) / RenderSize(); - fUv = ClampUv(fUv, RenderSize(), InputColorResourceDimensions()); - FfxFloat32x3 fRgb = SampleInputColor(fUv); - - fRgb /= PreExposure(); - - //compute log luma - const FfxFloat32 fLogLuma = log(ffxMax(FSR3UPSCALER_EPSILON, RGBToLuma(fRgb))); - - // Make sure out of screen pixels contribute no value to the end result - const FfxFloat32 result = all(FFX_LESS_THAN(tex, RenderSize())) ? fLogLuma : 0.0f; - - return FfxFloat32x4(result, 0, 0, 0); -} - -FfxFloat32x4 SpdLoad(FfxInt32x2 tex, FfxUInt32 slice) -{ - return SPD_LoadMipmap5(tex); -} - -void SpdStore(FfxInt32x2 pix, FfxFloat32x4 outValue, FfxUInt32 index, FfxUInt32 slice) -{ - if (index == LumaMipLevelToUse() || index == 5) - { - SPD_SetMipmap(pix, index, outValue.r); - } - - if (index == MipCount() - 1) { //accumulate on 1x1 level - - if (all(FFX_EQUAL(pix, FfxInt32x2(0, 0)))) - { - FfxFloat32 prev = SPD_LoadExposureBuffer().y; - FfxFloat32 result = outValue.r; - - if (prev < resetAutoExposureAverageSmoothing) // Compare Lavg, so small or negative values - { - FfxFloat32 rate = 1.0f; - result = prev + (result - prev) * (1 - exp(-DeltaTime() * rate)); - } - FfxFloat32x2 spdOutput = FfxFloat32x2(ComputeAutoExposureFromLavg(result), result); - SPD_SetExposureBuffer(spdOutput); - } - } -} - -FfxFloat32x4 SpdLoadIntermediate(FfxUInt32 x, FfxUInt32 y) -{ - return FfxFloat32x4( - spdIntermediateR[x][y], - spdIntermediateG[x][y], - spdIntermediateB[x][y], - spdIntermediateA[x][y]); -} -void SpdStoreIntermediate(FfxUInt32 x, FfxUInt32 y, FfxFloat32x4 value) -{ - spdIntermediateR[x][y] = value.x; - spdIntermediateG[x][y] = value.y; - spdIntermediateB[x][y] = value.z; - spdIntermediateA[x][y] = value.w; -} -FfxFloat32x4 SpdReduce4(FfxFloat32x4 v0, FfxFloat32x4 v1, FfxFloat32x4 v2, FfxFloat32x4 v3) -{ - return (v0 + v1 + v2 + v3) * 0.25f; -} -#endif - -// define fetch and store functions Packed -#if FFX_HALF - -FFX_GROUPSHARED FfxFloat16x2 spdIntermediateRG[16][16]; -FFX_GROUPSHARED FfxFloat16x2 spdIntermediateBA[16][16]; - -FfxFloat16x4 SpdLoadSourceImageH(FfxFloat32x2 tex, FfxUInt32 slice) -{ - return FfxFloat16x4(0, 0, 0, 0); -} - -FfxFloat16x4 SpdLoadH(FfxInt32x2 p, FfxUInt32 slice) -{ - return FfxFloat16x4(0, 0, 0, 0); -} - -void SpdStoreH(FfxInt32x2 p, FfxFloat16x4 value, FfxUInt32 mip, FfxUInt32 slice) -{ -} - -FfxFloat16x4 SpdLoadIntermediateH(FfxUInt32 x, FfxUInt32 y) -{ - return FfxFloat16x4( - spdIntermediateRG[x][y].x, - spdIntermediateRG[x][y].y, - spdIntermediateBA[x][y].x, - spdIntermediateBA[x][y].y); -} - -void SpdStoreIntermediateH(FfxUInt32 x, FfxUInt32 y, FfxFloat16x4 value) -{ - spdIntermediateRG[x][y] = value.xy; - spdIntermediateBA[x][y] = value.zw; -} - -FfxFloat16x4 SpdReduce4H(FfxFloat16x4 v0, FfxFloat16x4 v1, FfxFloat16x4 v2, FfxFloat16x4 v3) -{ - return (v0 + v1 + v2 + v3) * FfxFloat16(0.25); -} -#endif - -#include "spd/ffx_spd.h" - -void ComputeAutoExposure(FfxUInt32x3 WorkGroupId, FfxUInt32 LocalThreadIndex) -{ -#if FFX_HALF - SpdDownsampleH( - FfxUInt32x2(WorkGroupId.xy), - FfxUInt32(LocalThreadIndex), - FfxUInt32(MipCount()), - FfxUInt32(NumWorkGroups()), - FfxUInt32(WorkGroupId.z), - FfxUInt32x2(WorkGroupOffset())); -#else - SpdDownsample( - FfxUInt32x2(WorkGroupId.xy), - FfxUInt32(LocalThreadIndex), - FfxUInt32(MipCount()), - FfxUInt32(NumWorkGroups()), - FfxUInt32(WorkGroupId.z), - FfxUInt32x2(WorkGroupOffset())); -#endif -} diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_compute_luminance_pyramid.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_compute_luminance_pyramid.h.meta deleted file mode 100644 index 9fb7653..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_compute_luminance_pyramid.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 61bd10363d44ee2478461c9e9efbcb67 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_depth_clip.h b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_depth_clip.h deleted file mode 100644 index 53763c8..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_depth_clip.h +++ /dev/null @@ -1,259 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#ifndef FFX_FSR3UPSCALER_DEPTH_CLIP_H -#define FFX_FSR3UPSCALER_DEPTH_CLIP_H - -FFX_STATIC const FfxFloat32 DepthClipBaseScale = 4.0f; - -FfxFloat32 ComputeDepthClip(FfxFloat32x2 fUvSample, FfxFloat32 fCurrentDepthSample) -{ - FfxFloat32 fCurrentDepthViewSpace = GetViewSpaceDepth(fCurrentDepthSample); - BilinearSamplingData bilinearInfo = GetBilinearSamplingData(fUvSample, RenderSize()); - - FfxFloat32 fDilatedSum = 0.0f; - FfxFloat32 fDepth = 0.0f; - FfxFloat32 fWeightSum = 0.0f; - for (FfxInt32 iSampleIndex = 0; iSampleIndex < 4; iSampleIndex++) { - - const FfxInt32x2 iOffset = bilinearInfo.iOffsets[iSampleIndex]; - const FfxInt32x2 iSamplePos = bilinearInfo.iBasePos + iOffset; - - if (IsOnScreen(iSamplePos, RenderSize())) { - const FfxFloat32 fWeight = bilinearInfo.fWeights[iSampleIndex]; - if (fWeight > fReconstructedDepthBilinearWeightThreshold) { - - const FfxFloat32 fPrevDepthSample = LoadReconstructedPrevDepth(iSamplePos); - const FfxFloat32 fPrevNearestDepthViewSpace = GetViewSpaceDepth(fPrevDepthSample); - - const FfxFloat32 fDepthDiff = fCurrentDepthViewSpace - fPrevNearestDepthViewSpace; - - if (fDepthDiff > 0.0f) { - -#if FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - const FfxFloat32 fPlaneDepth = ffxMin(fPrevDepthSample, fCurrentDepthSample); -#else - const FfxFloat32 fPlaneDepth = ffxMax(fPrevDepthSample, fCurrentDepthSample); -#endif - - const FfxFloat32x3 fCenter = GetViewSpacePosition(FfxInt32x2(RenderSize() * 0.5f), RenderSize(), fPlaneDepth); - const FfxFloat32x3 fCorner = GetViewSpacePosition(FfxInt32x2(0, 0), RenderSize(), fPlaneDepth); - - const FfxFloat32 fHalfViewportWidth = length(FfxFloat32x2(RenderSize())); - const FfxFloat32 fDepthThreshold = ffxMax(fCurrentDepthViewSpace, fPrevNearestDepthViewSpace); - - const FfxFloat32 Ksep = 1.37e-05f; - const FfxFloat32 Kfov = length(fCorner) / length(fCenter); - const FfxFloat32 fRequiredDepthSeparation = Ksep * Kfov * fHalfViewportWidth * fDepthThreshold; - - const FfxFloat32 fResolutionFactor = ffxSaturate(length(FfxFloat32x2(RenderSize())) / length(FfxFloat32x2(1920.0f, 1080.0f))); - const FfxFloat32 fPower = ffxLerp(1.0f, 3.0f, fResolutionFactor); - fDepth += ffxPow(ffxSaturate(FfxFloat32(fRequiredDepthSeparation / fDepthDiff)), fPower) * fWeight; - fWeightSum += fWeight; - } - } - } - } - - return (fWeightSum > 0) ? ffxSaturate(1.0f - fDepth / fWeightSum) : 0.0f; -} - -FfxFloat32 ComputeMotionDivergence(FfxInt32x2 iPxPos, FfxInt32x2 iPxInputMotionVectorSize) -{ - FfxFloat32 minconvergence = 1.0f; - - FfxFloat32x2 fMotionVectorNucleus = LoadInputMotionVector(iPxPos); - FfxFloat32 fNucleusVelocityLr = length(fMotionVectorNucleus * RenderSize()); - FfxFloat32 fMaxVelocityUv = length(fMotionVectorNucleus); - - const FfxFloat32 MotionVectorVelocityEpsilon = 1e-02f; - - if (fNucleusVelocityLr > MotionVectorVelocityEpsilon) { - for (FfxInt32 y = -1; y <= 1; ++y) { - for (FfxInt32 x = -1; x <= 1; ++x) { - - FfxInt32x2 sp = ClampLoad(iPxPos, FfxInt32x2(x, y), iPxInputMotionVectorSize); - - FfxFloat32x2 fMotionVector = LoadInputMotionVector(sp); - FfxFloat32 fVelocityUv = length(fMotionVector); - - fMaxVelocityUv = ffxMax(fVelocityUv, fMaxVelocityUv); - fVelocityUv = ffxMax(fVelocityUv, fMaxVelocityUv); - minconvergence = ffxMin(minconvergence, dot(fMotionVector / fVelocityUv, fMotionVectorNucleus / fVelocityUv)); - } - } - } - - return ffxSaturate(1.0f - minconvergence) * ffxSaturate(fMaxVelocityUv / 0.01f); -} - -FfxFloat32 ComputeDepthDivergence(FfxInt32x2 iPxPos) -{ - const FfxFloat32 fMaxDistInMeters = GetMaxDistanceInMeters(); - FfxFloat32 fDepthMax = 0.0f; - FfxFloat32 fDepthMin = fMaxDistInMeters; - - FfxInt32 iMaxDistFound = 0; - - for (FfxInt32 y = -1; y < 2; y++) { - for (FfxInt32 x = -1; x < 2; x++) { - - const FfxInt32x2 iOffset = FfxInt32x2(x, y); - const FfxInt32x2 iSamplePos = iPxPos + iOffset; - - const FfxFloat32 fOnScreenFactor = IsOnScreen(iSamplePos, RenderSize()) ? 1.0f : 0.0f; - FfxFloat32 fDepth = GetViewSpaceDepthInMeters(LoadDilatedDepth(iSamplePos)) * fOnScreenFactor; - - iMaxDistFound |= FfxInt32(fMaxDistInMeters == fDepth); - - fDepthMin = ffxMin(fDepthMin, fDepth); - fDepthMax = ffxMax(fDepthMax, fDepth); - } - } - - return (1.0f - fDepthMin / fDepthMax) * (FfxBoolean(iMaxDistFound) ? 0.0f : 1.0f); -} - -FfxFloat32 ComputeTemporalMotionDivergence(FfxInt32x2 iPxPos) -{ - const FfxFloat32x2 fUv = FfxFloat32x2(iPxPos + 0.5f) / RenderSize(); - - FfxFloat32x2 fMotionVector = LoadDilatedMotionVector(iPxPos); - FfxFloat32x2 fReprojectedUv = fUv + fMotionVector; - fReprojectedUv = ClampUv(fReprojectedUv, RenderSize(), MaxRenderSize()); - FfxFloat32x2 fPrevMotionVector = SamplePreviousDilatedMotionVector(fReprojectedUv); - - float fPxDistance = length(fMotionVector * DisplaySize()); - return fPxDistance > 1.0f ? ffxLerp(0.0f, 1.0f - ffxSaturate(length(fPrevMotionVector) / length(fMotionVector)), ffxSaturate(ffxPow(fPxDistance / 20.0f, 3.0f))) : 0; -} - -void PreProcessReactiveMasks(FfxInt32x2 iPxLrPos, FfxFloat32 fMotionDivergence) -{ - // Compensate for bilinear sampling in accumulation pass - - FfxFloat32x3 fReferenceColor = LoadInputColor(iPxLrPos).xyz; - FfxFloat32x2 fReactiveFactor = FfxFloat32x2(0.0f, fMotionDivergence); - - float fMasksSum = 0.0f; - - FfxFloat32x3 fColorSamples[9]; - FfxFloat32 fReactiveSamples[9]; - FfxFloat32 fTransparencyAndCompositionSamples[9]; - - FFX_UNROLL - for (FfxInt32 y = -1; y < 2; y++) { - FFX_UNROLL - for (FfxInt32 x = -1; x < 2; x++) { - - const FfxInt32x2 sampleCoord = ClampLoad(iPxLrPos, FfxInt32x2(x, y), FfxInt32x2(RenderSize())); - - FfxInt32 sampleIdx = (y + 1) * 3 + x + 1; - - FfxFloat32x3 fColorSample = LoadInputColor(sampleCoord).xyz; - FfxFloat32 fReactiveSample = LoadReactiveMask(sampleCoord); - FfxFloat32 fTransparencyAndCompositionSample = LoadTransparencyAndCompositionMask(sampleCoord); - - fColorSamples[sampleIdx] = fColorSample; - fReactiveSamples[sampleIdx] = fReactiveSample; - fTransparencyAndCompositionSamples[sampleIdx] = fTransparencyAndCompositionSample; - - fMasksSum += (fReactiveSample + fTransparencyAndCompositionSample); - } - } - - if (fMasksSum > 0) - { - for (FfxInt32 sampleIdx = 0; sampleIdx < 9; sampleIdx++) - { - FfxFloat32x3 fColorSample = fColorSamples[sampleIdx]; - FfxFloat32 fReactiveSample = fReactiveSamples[sampleIdx]; - FfxFloat32 fTransparencyAndCompositionSample = fTransparencyAndCompositionSamples[sampleIdx]; - - const FfxFloat32 fMaxLenSq = ffxMax(dot(fReferenceColor, fReferenceColor), dot(fColorSample, fColorSample)); - const FfxFloat32 fSimilarity = dot(fReferenceColor, fColorSample) / fMaxLenSq; - - // Increase power for non-similar samples - const FfxFloat32 fPowerBiasMax = 6.0f; - const FfxFloat32 fSimilarityPower = 1.0f + (fPowerBiasMax - fSimilarity * fPowerBiasMax); - const FfxFloat32 fWeightedReactiveSample = ffxPow(fReactiveSample, fSimilarityPower); - const FfxFloat32 fWeightedTransparencyAndCompositionSample = ffxPow(fTransparencyAndCompositionSample, fSimilarityPower); - - fReactiveFactor = ffxMax(fReactiveFactor, FfxFloat32x2(fWeightedReactiveSample, fWeightedTransparencyAndCompositionSample)); - } - } - - StoreDilatedReactiveMasks(iPxLrPos, fReactiveFactor); -} - -FfxFloat32x3 ComputePreparedInputColor(FfxInt32x2 iPxLrPos) -{ - //We assume linear data. if non-linear input (sRGB, ...), - //then we should convert to linear first and back to sRGB on output. - FfxFloat32x3 fRgb = ffxMax(FfxFloat32x3(0, 0, 0), LoadInputColor(iPxLrPos)); - - fRgb = PrepareRgb(fRgb, Exposure(), PreExposure()); - - const FfxFloat32x3 fPreparedYCoCg = RGBToYCoCg(fRgb); - - return fPreparedYCoCg; -} - -FfxFloat32 EvaluateSurface(FfxInt32x2 iPxPos, FfxFloat32x2 fMotionVector) -{ - FfxFloat32 d0 = GetViewSpaceDepth(LoadReconstructedPrevDepth(iPxPos + FfxInt32x2(0, -1))); - FfxFloat32 d1 = GetViewSpaceDepth(LoadReconstructedPrevDepth(iPxPos + FfxInt32x2(0, 0))); - FfxFloat32 d2 = GetViewSpaceDepth(LoadReconstructedPrevDepth(iPxPos + FfxInt32x2(0, 1))); - - return 1.0f - FfxFloat32(((d0 - d1) > (d1 * 0.01f)) && ((d1 - d2) > (d2 * 0.01f))); -} - -void DepthClip(FfxInt32x2 iPxPos) -{ - FfxFloat32x2 fDepthUv = (iPxPos + 0.5f) / RenderSize(); - FfxFloat32x2 fMotionVector = LoadDilatedMotionVector(iPxPos); - - // Discard tiny mvs - fMotionVector *= FfxFloat32(length(fMotionVector * DisplaySize()) > 0.01f); - - const FfxFloat32x2 fDilatedUv = fDepthUv + fMotionVector; - const FfxFloat32 fDilatedDepth = LoadDilatedDepth(iPxPos); - const FfxFloat32 fCurrentDepthViewSpace = GetViewSpaceDepth(LoadInputDepth(iPxPos)); - - // Compute prepared input color and depth clip - FfxFloat32 fDepthClip = ComputeDepthClip(fDilatedUv, fDilatedDepth) * EvaluateSurface(iPxPos, fMotionVector); - FfxFloat32x3 fPreparedYCoCg = ComputePreparedInputColor(iPxPos); - StorePreparedInputColor(iPxPos, FfxFloat32x4(fPreparedYCoCg, fDepthClip)); - - // Compute dilated reactive mask -#if FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS - FfxInt32x2 iSamplePos = iPxPos; -#else - FfxInt32x2 iSamplePos = ComputeHrPosFromLrPos(iPxPos); -#endif - - FfxFloat32 fMotionDivergence = ComputeMotionDivergence(iSamplePos, RenderSize()); - FfxFloat32 fTemporalMotionDifference = ffxSaturate(ComputeTemporalMotionDivergence(iPxPos) - ComputeDepthDivergence(iPxPos)); - - PreProcessReactiveMasks(iPxPos, ffxMax(fTemporalMotionDifference, fMotionDivergence)); -} - -#endif //!defined( FFX_FSR3UPSCALER_DEPTH_CLIPH ) diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_depth_clip.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_depth_clip.h.meta deleted file mode 100644 index 891d3d1..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_depth_clip.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 7c662249d70c4434da4f2da00e432c38 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_lock.h b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_lock.h deleted file mode 100644 index e1a0d06..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_lock.h +++ /dev/null @@ -1,116 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#ifndef FFX_FSR3UPSCALER_LOCK_H -#define FFX_FSR3UPSCALER_LOCK_H - -void ClearResourcesForNextFrame(in FfxInt32x2 iPxHrPos) -{ - if (all(FFX_LESS_THAN(iPxHrPos, FfxInt32x2(RenderSize())))) - { -#if FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - const FfxUInt32 farZ = 0x0; -#else - const FfxUInt32 farZ = 0x3f800000; -#endif - SetReconstructedDepth(iPxHrPos, farZ); - } -} - -FfxBoolean ComputeThinFeatureConfidence(FfxInt32x2 pos) -{ - const FfxInt32 RADIUS = 1; - - FfxFloat32 fNucleus = LoadLockInputLuma(pos); - - FfxFloat32 similar_threshold = 1.05f; - FfxFloat32 dissimilarLumaMin = FSR3UPSCALER_FLT_MAX; - FfxFloat32 dissimilarLumaMax = 0; - - /* - 0 1 2 - 3 4 5 - 6 7 8 - */ - - #define SETBIT(x) (1U << x) - - FfxUInt32 mask = SETBIT(4); //flag fNucleus as similar - - const FfxUInt32 uNumRejectionMasks = 4; - const FfxUInt32 uRejectionMasks[uNumRejectionMasks] = { - SETBIT(0) | SETBIT(1) | SETBIT(3) | SETBIT(4), //Upper left - SETBIT(1) | SETBIT(2) | SETBIT(4) | SETBIT(5), //Upper right - SETBIT(3) | SETBIT(4) | SETBIT(6) | SETBIT(7), //Lower left - SETBIT(4) | SETBIT(5) | SETBIT(7) | SETBIT(8), //Lower right - }; - - FfxInt32 idx = 0; - FFX_UNROLL - for (FfxInt32 y = -RADIUS; y <= RADIUS; y++) { - FFX_UNROLL - for (FfxInt32 x = -RADIUS; x <= RADIUS; x++, idx++) { - if (x == 0 && y == 0) continue; - - FfxInt32x2 samplePos = ClampLoad(pos, FfxInt32x2(x, y), FfxInt32x2(RenderSize())); - - FfxFloat32 sampleLuma = LoadLockInputLuma(samplePos); - FfxFloat32 difference = ffxMax(sampleLuma, fNucleus) / ffxMin(sampleLuma, fNucleus); - - if (difference > 0 && (difference < similar_threshold)) { - mask |= SETBIT(idx); - } else { - dissimilarLumaMin = ffxMin(dissimilarLumaMin, sampleLuma); - dissimilarLumaMax = ffxMax(dissimilarLumaMax, sampleLuma); - } - } - } - - FfxBoolean isRidge = fNucleus > dissimilarLumaMax || fNucleus < dissimilarLumaMin; - - if (FFX_FALSE == isRidge) { - - return false; - } - - FFX_UNROLL - for (FfxInt32 i = 0; i < 4; i++) { - - if ((mask & uRejectionMasks[i]) == uRejectionMasks[i]) { - return false; - } - } - - return true; -} - -void ComputeLock(FfxInt32x2 iPxLrPos) -{ - if (ComputeThinFeatureConfidence(iPxLrPos)) - { - StoreNewLocks(ComputeHrPosFromLrPos(iPxLrPos), 1.f); - } - - // ClearResourcesForNextFrame(iPxLrPos); -} - -#endif // FFX_FSR3UPSCALER_LOCK_H diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_lock.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_lock.h.meta deleted file mode 100644 index 4013169..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_lock.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: c7e9f53dd040b2645af5ccd936a94b0e -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_postprocess_lock_status.h b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_postprocess_lock_status.h deleted file mode 100644 index 3709113..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_postprocess_lock_status.h +++ /dev/null @@ -1,107 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#ifndef FFX_FSR3UPSCALER_POSTPROCESS_LOCK_STATUS_H -#define FFX_FSR3UPSCALER_POSTPROCESS_LOCK_STATUS_H - -FfxFloat32x4 WrapShadingChangeLuma(FfxInt32x2 iPxSample) -{ - return FfxFloat32x4(LoadMipLuma(iPxSample, LumaMipLevelToUse()), 0, 0, 0); -} - -#if FFX_HALF -FFX_MIN16_F4 WrapShadingChangeLuma(FFX_MIN16_I2 iPxSample) -{ - return FFX_MIN16_F4(LoadMipLuma(iPxSample, LumaMipLevelToUse()), 0, 0, 0); -} -#endif - -#if FFX_FSR3UPSCALER_OPTION_POSTPROCESSLOCKSTATUS_SAMPLERS_USE_DATA_HALF && FFX_HALF -DeclareCustomFetchBilinearSamplesMin16(FetchShadingChangeLumaSamples, WrapShadingChangeLuma) -#else -DeclareCustomFetchBicubicSamples(FetchShadingChangeLumaSamples, WrapShadingChangeLuma) -#endif -DeclareCustomTextureSample(ShadingChangeLumaSample, Lanczos2, FetchShadingChangeLumaSamples) - -FfxFloat32 GetShadingChangeLuma(FfxInt32x2 iPxHrPos, FfxFloat32x2 fUvCoord) -{ - FfxFloat32 fShadingChangeLuma = 0; - -#if 0 - fShadingChangeLuma = Exposure() * exp(ShadingChangeLumaSample(fUvCoord, LumaMipDimensions()).x); -#else - - const FfxFloat32 fDiv = FfxFloat32(2u << LumaMipLevelToUse()); - FfxInt32x2 iMipRenderSize = FfxInt32x2(RenderSize() / fDiv); - - fUvCoord = ClampUv(fUvCoord, iMipRenderSize, LumaMipDimensions()); - fShadingChangeLuma = Exposure() * exp(FfxFloat32(SampleMipLuma(fUvCoord, LumaMipLevelToUse()))); -#endif - - fShadingChangeLuma = ffxPow(fShadingChangeLuma, 1.0f / 6.0f); - - return fShadingChangeLuma; -} - -void UpdateLockStatus(AccumulationPassCommonParams params, - FFX_PARAMETER_INOUT FfxFloat32 fReactiveFactor, LockState state, - FFX_PARAMETER_INOUT FfxFloat32x2 fLockStatus, - FFX_PARAMETER_OUT FfxFloat32 fLockContributionThisFrame, - FFX_PARAMETER_OUT FfxFloat32 fLuminanceDiff) { - - const FfxFloat32 fShadingChangeLuma = GetShadingChangeLuma(params.iPxHrPos, params.fHrUv); - - //init temporal shading change factor, init to -1 or so in reproject to know if "true new"? - fLockStatus[LOCK_TEMPORAL_LUMA] = (fLockStatus[LOCK_TEMPORAL_LUMA] == FfxFloat32(0.0f)) ? fShadingChangeLuma : fLockStatus[LOCK_TEMPORAL_LUMA]; - - FfxFloat32 fPreviousShadingChangeLuma = fLockStatus[LOCK_TEMPORAL_LUMA]; - - fLuminanceDiff = 1.0f - MinDividedByMax(fPreviousShadingChangeLuma, fShadingChangeLuma); - - if (state.NewLock) { - fLockStatus[LOCK_TEMPORAL_LUMA] = fShadingChangeLuma; - - fLockStatus[LOCK_LIFETIME_REMAINING] = (fLockStatus[LOCK_LIFETIME_REMAINING] != 0.0f) ? 2.0f : 1.0f; - } - else if(fLockStatus[LOCK_LIFETIME_REMAINING] <= 1.0f) { - fLockStatus[LOCK_TEMPORAL_LUMA] = ffxLerp(fLockStatus[LOCK_TEMPORAL_LUMA], FfxFloat32(fShadingChangeLuma), 0.5f); - } - else { - if (fLuminanceDiff > 0.1f) { - KillLock(fLockStatus); - } - } - - fReactiveFactor = ffxMax(fReactiveFactor, ffxSaturate((fLuminanceDiff - 0.1f) * 10.0f)); - fLockStatus[LOCK_LIFETIME_REMAINING] *= (1.0f - fReactiveFactor); - - fLockStatus[LOCK_LIFETIME_REMAINING] *= ffxSaturate(1.0f - params.fAccumulationMask); - fLockStatus[LOCK_LIFETIME_REMAINING] *= FfxFloat32(params.fDepthClipFactor < 0.1f); - - // Compute this frame lock contribution - const FfxFloat32 fLifetimeContribution = ffxSaturate(fLockStatus[LOCK_LIFETIME_REMAINING] - 1.0f); - const FfxFloat32 fShadingChangeContribution = ffxSaturate(MinDividedByMax(fLockStatus[LOCK_TEMPORAL_LUMA], fShadingChangeLuma)); - - fLockContributionThisFrame = ffxSaturate(ffxSaturate(fLifetimeContribution * 4.0f) * fShadingChangeContribution); -} - -#endif //!defined( FFX_FSR3UPSCALER_POSTPROCESS_LOCK_STATUS_H ) diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_postprocess_lock_status.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_postprocess_lock_status.h.meta deleted file mode 100644 index 8c8bf49..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_postprocess_lock_status.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 67a8b72ceb93d634f883b086fdccb348 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_rcas.h b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_rcas.h deleted file mode 100644 index 77619a5..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_rcas.h +++ /dev/null @@ -1,67 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#define GROUP_SIZE 8 -#define FSR_RCAS_DENOISE 1 - -#include "ffx_core.h" - -void WriteUpscaledOutput(FFX_MIN16_U2 iPxHrPos, FfxFloat32x3 fUpscaledColor) -{ - StoreUpscaledOutput(FFX_MIN16_I2(iPxHrPos), fUpscaledColor); -} - -#define FSR_RCAS_F 1 -FfxFloat32x4 FsrRcasLoadF(FfxInt32x2 p) -{ - FfxFloat32x4 fColor = LoadRCAS_Input(p); - - fColor.rgb = PrepareRgb(fColor.rgb, Exposure(), PreExposure()); - - return fColor; -} -void FsrRcasInputF(inout FfxFloat32 r, inout FfxFloat32 g, inout FfxFloat32 b) {} - -#include "fsr1/ffx_fsr1.h" - -void CurrFilter(FFX_MIN16_U2 pos) -{ - FfxFloat32x3 c; - FsrRcasF(c.r, c.g, c.b, pos, RCASConfig()); - - c = UnprepareRgb(c, Exposure()); - - WriteUpscaledOutput(pos, c); -} - -void RCAS(FfxUInt32x3 LocalThreadId, FfxUInt32x3 WorkGroupId, FfxUInt32x3 Dtid) -{ - // Do remapping of local xy in workgroup for a more PS-like swizzle pattern. - FfxUInt32x2 gxy = ffxRemapForQuad(LocalThreadId.x) + FfxUInt32x2(WorkGroupId.x << 4u, WorkGroupId.y << 4u); - CurrFilter(FFX_MIN16_U2(gxy)); - gxy.x += 8u; - CurrFilter(FFX_MIN16_U2(gxy)); - gxy.y += 8u; - CurrFilter(FFX_MIN16_U2(gxy)); - gxy.x -= 8u; - CurrFilter(FFX_MIN16_U2(gxy)); -} diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_rcas.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_rcas.h.meta deleted file mode 100644 index a315002..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_rcas.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 692efb7cec0df67408a583a7ff34146a -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reconstruct_dilated_velocity_and_previous_depth.h b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reconstruct_dilated_velocity_and_previous_depth.h deleted file mode 100644 index a822dfc..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reconstruct_dilated_velocity_and_previous_depth.h +++ /dev/null @@ -1,146 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#ifndef FFX_FSR3UPSCALER_RECONSTRUCT_DILATED_VELOCITY_AND_PREVIOUS_DEPTH_H -#define FFX_FSR3UPSCALER_RECONSTRUCT_DILATED_VELOCITY_AND_PREVIOUS_DEPTH_H - -void ReconstructPrevDepth(FfxInt32x2 iPxPos, FfxFloat32 fDepth, FfxFloat32x2 fMotionVector, FfxInt32x2 iPxDepthSize) -{ - fMotionVector *= FfxFloat32(length(fMotionVector * DisplaySize()) > 0.1f); - - FfxFloat32x2 fUv = (iPxPos + FfxFloat32(0.5)) / iPxDepthSize; - FfxFloat32x2 fReprojectedUv = fUv + fMotionVector; - - BilinearSamplingData bilinearInfo = GetBilinearSamplingData(fReprojectedUv, RenderSize()); - - // Project current depth into previous frame locations. - // Push to all pixels having some contribution if reprojection is using bilinear logic. - for (FfxInt32 iSampleIndex = 0; iSampleIndex < 4; iSampleIndex++) { - - const FfxInt32x2 iOffset = bilinearInfo.iOffsets[iSampleIndex]; - FfxFloat32 fWeight = bilinearInfo.fWeights[iSampleIndex]; - - if (fWeight > fReconstructedDepthBilinearWeightThreshold) { - - FfxInt32x2 iStorePos = bilinearInfo.iBasePos + iOffset; - if (IsOnScreen(iStorePos, iPxDepthSize)) { - StoreReconstructedDepth(iStorePos, fDepth); - } - } - } -} - -void FindNearestDepth(FFX_PARAMETER_IN FfxInt32x2 iPxPos, FFX_PARAMETER_IN FfxInt32x2 iPxSize, FFX_PARAMETER_OUT FfxFloat32 fNearestDepth, FFX_PARAMETER_OUT FfxInt32x2 fNearestDepthCoord) -{ - const FfxInt32 iSampleCount = 9; - const FfxInt32x2 iSampleOffsets[iSampleCount] = { - FfxInt32x2(+0, +0), - FfxInt32x2(+1, +0), - FfxInt32x2(+0, +1), - FfxInt32x2(+0, -1), - FfxInt32x2(-1, +0), - FfxInt32x2(-1, +1), - FfxInt32x2(+1, +1), - FfxInt32x2(-1, -1), - FfxInt32x2(+1, -1), - }; - - // pull out the depth loads to allow SC to batch them - FfxFloat32 depth[9]; - FfxInt32 iSampleIndex = 0; - FFX_UNROLL - for (iSampleIndex = 0; iSampleIndex < iSampleCount; ++iSampleIndex) { - - FfxInt32x2 iPos = iPxPos + iSampleOffsets[iSampleIndex]; - depth[iSampleIndex] = LoadInputDepth(iPos); - } - - // find closest depth - fNearestDepthCoord = iPxPos; - fNearestDepth = depth[0]; - FFX_UNROLL - for (iSampleIndex = 1; iSampleIndex < iSampleCount; ++iSampleIndex) { - - FfxInt32x2 iPos = iPxPos + iSampleOffsets[iSampleIndex]; - if (IsOnScreen(iPos, iPxSize)) { - - FfxFloat32 fNdDepth = depth[iSampleIndex]; -#if FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - if (fNdDepth > fNearestDepth) { -#else - if (fNdDepth < fNearestDepth) { -#endif - fNearestDepthCoord = iPos; - fNearestDepth = fNdDepth; - } - } - } -} - -FfxFloat32 ComputeLockInputLuma(FfxInt32x2 iPxLrPos) -{ - //We assume linear data. if non-linear input (sRGB, ...), - //then we should convert to linear first and back to sRGB on output. - FfxFloat32x3 fRgb = ffxMax(FfxFloat32x3(0, 0, 0), LoadInputColor(iPxLrPos)); - - // Use internal auto exposure for locking logic - fRgb /= PreExposure(); - fRgb *= Exposure(); - -#if FFX_FSR3UPSCALER_OPTION_HDR_COLOR_INPUT - fRgb = Tonemap(fRgb); -#endif - - //compute luma used to lock pixels, if used elsewhere the ffxPow must be moved! - const FfxFloat32 fLockInputLuma = ffxPow(RGBToPerceivedLuma(fRgb), FfxFloat32(1.0 / 6.0)); - - return fLockInputLuma; -} - -void ReconstructAndDilate(FfxInt32x2 iPxLrPos) -{ - FfxFloat32 fDilatedDepth; - FfxInt32x2 iNearestDepthCoord; - - FindNearestDepth(iPxLrPos, RenderSize(), fDilatedDepth, iNearestDepthCoord); - -#if FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS - FfxInt32x2 iSamplePos = iPxLrPos; - FfxInt32x2 iMotionVectorPos = iNearestDepthCoord; -#else - FfxInt32x2 iSamplePos = ComputeHrPosFromLrPos(iPxLrPos); - FfxInt32x2 iMotionVectorPos = ComputeHrPosFromLrPos(iNearestDepthCoord); -#endif - - FfxFloat32x2 fDilatedMotionVector = LoadInputMotionVector(iMotionVectorPos); - - StoreDilatedDepth(iPxLrPos, fDilatedDepth); - StoreDilatedMotionVector(iPxLrPos, fDilatedMotionVector); - - ReconstructPrevDepth(iPxLrPos, fDilatedDepth, fDilatedMotionVector, RenderSize()); - - FfxFloat32 fLockInputLuma = ComputeLockInputLuma(iPxLrPos); - StoreLockInputLuma(iPxLrPos, fLockInputLuma); -} - - -#endif //!defined( FFX_FSR3UPSCALER_RECONSTRUCT_DILATED_VELOCITY_AND_PREVIOUS_DEPTH_H ) diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reconstruct_dilated_velocity_and_previous_depth.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reconstruct_dilated_velocity_and_previous_depth.h.meta deleted file mode 100644 index a1fd018..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reconstruct_dilated_velocity_and_previous_depth.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: c8b3854bad30a8b40babc5a9805f294e -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reproject.h b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reproject.h deleted file mode 100644 index 29b7584..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reproject.h +++ /dev/null @@ -1,137 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#ifndef FFX_FSR3UPSCALER_REPROJECT_H -#define FFX_FSR3UPSCALER_REPROJECT_H - -#ifndef FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE -#define FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE 0 // Reference -#endif - -FfxFloat32x4 WrapHistory(FfxInt32x2 iPxSample) -{ - return LoadHistory(iPxSample); -} - -#if FFX_HALF -FFX_MIN16_F4 WrapHistory(FFX_MIN16_I2 iPxSample) -{ - return FFX_MIN16_F4(LoadHistory(iPxSample)); -} -#endif - - -#if FFX_FSR3UPSCALER_OPTION_REPROJECT_SAMPLERS_USE_DATA_HALF && FFX_HALF -DeclareCustomFetchBicubicSamplesMin16(FetchHistorySamples, WrapHistory) -DeclareCustomTextureSampleMin16(HistorySample, FFX_FSR3UPSCALER_GET_LANCZOS_SAMPLER1D(FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE), FetchHistorySamples) -#else -DeclareCustomFetchBicubicSamples(FetchHistorySamples, WrapHistory) -DeclareCustomTextureSample(HistorySample, FFX_FSR3UPSCALER_GET_LANCZOS_SAMPLER1D(FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE), FetchHistorySamples) -#endif - -FfxFloat32x4 WrapLockStatus(FfxInt32x2 iPxSample) -{ - FfxFloat32x4 fSample = FfxFloat32x4(LoadLockStatus(iPxSample), 0.0f, 0.0f); - return fSample; -} - -#if FFX_HALF -FFX_MIN16_F4 WrapLockStatus(FFX_MIN16_I2 iPxSample) -{ - FFX_MIN16_F4 fSample = FFX_MIN16_F4(LoadLockStatus(iPxSample), 0.0, 0.0); - - return fSample; -} -#endif - -#if 1 -#if FFX_FSR3UPSCALER_OPTION_REPROJECT_SAMPLERS_USE_DATA_HALF && FFX_HALF -DeclareCustomFetchBilinearSamplesMin16(FetchLockStatusSamples, WrapLockStatus) -DeclareCustomTextureSampleMin16(LockStatusSample, Bilinear, FetchLockStatusSamples) -#else -DeclareCustomFetchBilinearSamples(FetchLockStatusSamples, WrapLockStatus) -DeclareCustomTextureSample(LockStatusSample, Bilinear, FetchLockStatusSamples) -#endif -#else -#if FFX_FSR3UPSCALER_OPTION_REPROJECT_SAMPLERS_USE_DATA_HALF && FFX_HALF -DeclareCustomFetchBicubicSamplesMin16(FetchLockStatusSamples, WrapLockStatus) -DeclareCustomTextureSampleMin16(LockStatusSample, FFX_FSR3UPSCALER_GET_LANCZOS_SAMPLER1D(FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE), FetchLockStatusSamples) -#else -DeclareCustomFetchBicubicSamples(FetchLockStatusSamples, WrapLockStatus) -DeclareCustomTextureSample(LockStatusSample, FFX_FSR3UPSCALER_GET_LANCZOS_SAMPLER1D(FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE), FetchLockStatusSamples) -#endif -#endif - -FfxFloat32x2 GetMotionVector(FfxInt32x2 iPxHrPos, FfxFloat32x2 fHrUv) -{ -#if FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS - FfxFloat32x2 fDilatedMotionVector = LoadDilatedMotionVector(FFX_MIN16_I2(fHrUv * RenderSize())); -#else - FfxFloat32x2 fDilatedMotionVector = LoadInputMotionVector(iPxHrPos); -#endif - - return fDilatedMotionVector; -} - -FfxBoolean IsUvInside(FfxFloat32x2 fUv) -{ - return (fUv.x >= 0.0f && fUv.x <= 1.0f) && (fUv.y >= 0.0f && fUv.y <= 1.0f); -} - -void ComputeReprojectedUVs(const AccumulationPassCommonParams params, FFX_PARAMETER_OUT FfxFloat32x2 fReprojectedHrUv, FFX_PARAMETER_OUT FfxBoolean bIsExistingSample) -{ - fReprojectedHrUv = params.fHrUv + params.fMotionVector; - - bIsExistingSample = IsUvInside(fReprojectedHrUv); -} - -void ReprojectHistoryColor(const AccumulationPassCommonParams params, FFX_PARAMETER_OUT FfxFloat32x3 fHistoryColor, FFX_PARAMETER_OUT FfxFloat32 fTemporalReactiveFactor, FFX_PARAMETER_OUT FfxBoolean bInMotionLastFrame) -{ - FfxFloat32x4 fHistory = HistorySample(params.fReprojectedHrUv, DisplaySize()); - - fHistoryColor = PrepareRgb(fHistory.rgb, Exposure(), PreviousFramePreExposure()); - - fHistoryColor = RGBToYCoCg(fHistoryColor); - - //Compute temporal reactivity info - fTemporalReactiveFactor = ffxSaturate(abs(fHistory.w)); - bInMotionLastFrame = (fHistory.w < 0.0f); -} - -LockState ReprojectHistoryLockStatus(const AccumulationPassCommonParams params, FFX_PARAMETER_OUT FfxFloat32x2 fReprojectedLockStatus) -{ - LockState state = { FFX_FALSE, FFX_FALSE }; - const FfxFloat32 fNewLockIntensity = LoadRwNewLocks(params.iPxHrPos); - state.NewLock = fNewLockIntensity > (127.0f / 255.0f); - - FfxFloat32 fInPlaceLockLifetime = state.NewLock ? fNewLockIntensity : 0; - - fReprojectedLockStatus = SampleLockStatus(params.fReprojectedHrUv); - - if (fReprojectedLockStatus[LOCK_LIFETIME_REMAINING] != FfxFloat32(0.0f)) { - state.WasLockedPrevFrame = true; - } - - return state; -} - -#endif //!defined( FFX_FSR3UPSCALER_REPROJECT_H ) diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reproject.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reproject.h.meta deleted file mode 100644 index 3fa39db..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reproject.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 9d893016eebb2564f9a66b80afb0849f -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_resources.h b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_resources.h deleted file mode 100644 index d98cfcc..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_resources.h +++ /dev/null @@ -1,104 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#ifndef FFX_FSR3UPSCALER_RESOURCES_H -#define FFX_FSR3UPSCALER_RESOURCES_H - -#if defined(FFX_CPU) || defined(FFX_GPU) -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_NULL 0 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INPUT_OPAQUE_ONLY 1 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INPUT_COLOR 2 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INPUT_MOTION_VECTORS 3 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INPUT_DEPTH 4 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INPUT_EXPOSURE 5 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INPUT_REACTIVE_MASK 6 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INPUT_TRANSPARENCY_AND_COMPOSITION_MASK 7 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_RECONSTRUCTED_PREVIOUS_NEAREST_DEPTH 8 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_DILATED_MOTION_VECTORS 9 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_DILATED_DEPTH 10 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INTERNAL_UPSCALED_COLOR 11 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LOCK_STATUS 12 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_NEW_LOCKS 13 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_PREPARED_INPUT_COLOR 14 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LUMA_HISTORY 15 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_DEBUG_OUTPUT 16 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LANCZOS_LUT 17 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SPD_ATOMIC_COUNT 18 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_UPSCALED_OUTPUT 19 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_RCAS_INPUT 20 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LOCK_STATUS_1 21 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LOCK_STATUS_2 22 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INTERNAL_UPSCALED_COLOR_1 23 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INTERNAL_UPSCALED_COLOR_2 24 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INTERNAL_DEFAULT_REACTIVITY 25 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INTERNAL_DEFAULT_TRANSPARENCY_AND_COMPOSITION 26 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTITIER_UPSAMPLE_MAXIMUM_BIAS_LUT 27 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_DILATED_REACTIVE_MASKS 28 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE 29 // same as FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_0 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_0 29 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_1 30 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_2 31 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_3 32 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_4 33 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_5 34 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_6 35 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_7 36 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_8 37 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_9 38 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_10 39 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_11 40 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_12 41 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INTERNAL_DEFAULT_EXPOSURE 42 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_AUTO_EXPOSURE 43 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_AUTOREACTIVE 44 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_AUTOCOMPOSITION_DEPRECATED 45 - -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_PREV_PRE_ALPHA_COLOR 46 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_PREV_POST_ALPHA_COLOR 47 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_PREV_PRE_ALPHA_COLOR_1 48 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_PREV_POST_ALPHA_COLOR_1 49 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_PREV_PRE_ALPHA_COLOR_2 50 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_PREV_POST_ALPHA_COLOR_2 51 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_PREVIOUS_DILATED_MOTION_VECTORS 52 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LUMA_HISTORY_1 53 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LUMA_HISTORY_2 54 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LOCK_INPUT_LUMA 55 - -// Shading change detection mip level setting, value must be in the range [FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_0, FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_12] -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_SHADING_CHANGE FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_4 -#define FFX_FSR3UPSCALER_SHADING_CHANGE_MIP_LEVEL (FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_SHADING_CHANGE - FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE) - -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_COUNT 56 - -#define FFX_FSR3UPSCALER_CONSTANTBUFFER_IDENTIFIER_FSR3UPSCALER 0 -#define FFX_FSR3UPSCALER_CONSTANTBUFFER_IDENTIFIER_SPD 1 -#define FFX_FSR3UPSCALER_CONSTANTBUFFER_IDENTIFIER_RCAS 2 -#define FFX_FSR3UPSCALER_CONSTANTBUFFER_IDENTIFIER_GENREACTIVE 3 - -#define FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_APPLY_TONEMAP 1 -#define FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_APPLY_INVERSETONEMAP 2 -#define FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_APPLY_THRESHOLD 4 -#define FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_USE_COMPONENTS_MAX 8 - -#endif // #if defined(FFX_CPU) || defined(FFX_GPU) - -#endif //!defined( FFX_FSR3UPSCALER_RESOURCES_H ) diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_resources.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_resources.h.meta deleted file mode 100644 index 8e862df..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_resources.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: b5a95a38dcfaf3946a5095bbbc42939a -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_sample.h b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_sample.h deleted file mode 100644 index d33f70c..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_sample.h +++ /dev/null @@ -1,606 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#ifndef FFX_FSR3UPSCALER_SAMPLE_H -#define FFX_FSR3UPSCALER_SAMPLE_H - -// suppress warnings -#ifdef FFX_HLSL -#pragma warning(disable: 4008) // potentially divide by zero -#endif //FFX_HLSL - -struct FetchedBilinearSamples { - - FfxFloat32x4 fColor00; - FfxFloat32x4 fColor10; - - FfxFloat32x4 fColor01; - FfxFloat32x4 fColor11; -}; - -struct FetchedBicubicSamples { - - FfxFloat32x4 fColor00; - FfxFloat32x4 fColor10; - FfxFloat32x4 fColor20; - FfxFloat32x4 fColor30; - - FfxFloat32x4 fColor01; - FfxFloat32x4 fColor11; - FfxFloat32x4 fColor21; - FfxFloat32x4 fColor31; - - FfxFloat32x4 fColor02; - FfxFloat32x4 fColor12; - FfxFloat32x4 fColor22; - FfxFloat32x4 fColor32; - - FfxFloat32x4 fColor03; - FfxFloat32x4 fColor13; - FfxFloat32x4 fColor23; - FfxFloat32x4 fColor33; -}; - -#if FFX_HALF -struct FetchedBilinearSamplesMin16 { - - FFX_MIN16_F4 fColor00; - FFX_MIN16_F4 fColor10; - - FFX_MIN16_F4 fColor01; - FFX_MIN16_F4 fColor11; -}; - -struct FetchedBicubicSamplesMin16 { - - FFX_MIN16_F4 fColor00; - FFX_MIN16_F4 fColor10; - FFX_MIN16_F4 fColor20; - FFX_MIN16_F4 fColor30; - - FFX_MIN16_F4 fColor01; - FFX_MIN16_F4 fColor11; - FFX_MIN16_F4 fColor21; - FFX_MIN16_F4 fColor31; - - FFX_MIN16_F4 fColor02; - FFX_MIN16_F4 fColor12; - FFX_MIN16_F4 fColor22; - FFX_MIN16_F4 fColor32; - - FFX_MIN16_F4 fColor03; - FFX_MIN16_F4 fColor13; - FFX_MIN16_F4 fColor23; - FFX_MIN16_F4 fColor33; -}; -#else //FFX_HALF -#define FetchedBicubicSamplesMin16 FetchedBicubicSamples -#define FetchedBilinearSamplesMin16 FetchedBilinearSamples -#endif //FFX_HALF - -FfxFloat32x4 Linear(FfxFloat32x4 A, FfxFloat32x4 B, FfxFloat32 t) -{ - return A + (B - A) * t; -} - -FfxFloat32x4 Bilinear(FetchedBilinearSamples BilinearSamples, FfxFloat32x2 fPxFrac) -{ - FfxFloat32x4 fColorX0 = Linear(BilinearSamples.fColor00, BilinearSamples.fColor10, fPxFrac.x); - FfxFloat32x4 fColorX1 = Linear(BilinearSamples.fColor01, BilinearSamples.fColor11, fPxFrac.x); - FfxFloat32x4 fColorXY = Linear(fColorX0, fColorX1, fPxFrac.y); - return fColorXY; -} - -#if FFX_HALF -FFX_MIN16_F4 Linear(FFX_MIN16_F4 A, FFX_MIN16_F4 B, FFX_MIN16_F t) -{ - return A + (B - A) * t; -} - -FFX_MIN16_F4 Bilinear(FetchedBilinearSamplesMin16 BilinearSamples, FFX_MIN16_F2 fPxFrac) -{ - FFX_MIN16_F4 fColorX0 = Linear(BilinearSamples.fColor00, BilinearSamples.fColor10, fPxFrac.x); - FFX_MIN16_F4 fColorX1 = Linear(BilinearSamples.fColor01, BilinearSamples.fColor11, fPxFrac.x); - FFX_MIN16_F4 fColorXY = Linear(fColorX0, fColorX1, fPxFrac.y); - return fColorXY; -} -#endif - -FfxFloat32 Lanczos2NoClamp(FfxFloat32 x) -{ - const FfxFloat32 PI = 3.141592653589793f; // TODO: share SDK constants - return abs(x) < FSR3UPSCALER_EPSILON ? 1.f : (sin(PI * x) / (PI * x)) * (sin(0.5f * PI * x) / (0.5f * PI * x)); -} - -FfxFloat32 Lanczos2(FfxFloat32 x) -{ - x = ffxMin(abs(x), 2.0f); - return Lanczos2NoClamp(x); -} - -#if FFX_HALF - -#if 0 -FFX_MIN16_F Lanczos2NoClamp(FFX_MIN16_F x) -{ - const FFX_MIN16_F PI = FFX_MIN16_F(3.141592653589793f); // TODO: share SDK constants - return abs(x) < FFX_MIN16_F(FSR3UPSCALER_EPSILON) ? FFX_MIN16_F(1.f) : (sin(PI * x) / (PI * x)) * (sin(FFX_MIN16_F(0.5f) * PI * x) / (FFX_MIN16_F(0.5f) * PI * x)); -} -#endif - -FFX_MIN16_F Lanczos2(FFX_MIN16_F x) -{ - x = ffxMin(abs(x), FFX_MIN16_F(2.0f)); - return FFX_MIN16_F(Lanczos2NoClamp(x)); -} -#endif //FFX_HALF - -// FSR1 lanczos approximation. Input is x*x and must be <= 4. -FfxFloat32 Lanczos2ApproxSqNoClamp(FfxFloat32 x2) -{ - FfxFloat32 a = (2.0f / 5.0f) * x2 - 1; - FfxFloat32 b = (1.0f / 4.0f) * x2 - 1; - return ((25.0f / 16.0f) * a * a - (25.0f / 16.0f - 1)) * (b * b); -} - -#if FFX_HALF -FFX_MIN16_F Lanczos2ApproxSqNoClamp(FFX_MIN16_F x2) -{ - FFX_MIN16_F a = FFX_MIN16_F(2.0f / 5.0f) * x2 - FFX_MIN16_F(1); - FFX_MIN16_F b = FFX_MIN16_F(1.0f / 4.0f) * x2 - FFX_MIN16_F(1); - return (FFX_MIN16_F(25.0f / 16.0f) * a * a - FFX_MIN16_F(25.0f / 16.0f - 1)) * (b * b); -} -#endif //FFX_HALF - -FfxFloat32 Lanczos2ApproxSq(FfxFloat32 x2) -{ - x2 = ffxMin(x2, 4.0f); - return Lanczos2ApproxSqNoClamp(x2); -} - -#if FFX_HALF -FFX_MIN16_F Lanczos2ApproxSq(FFX_MIN16_F x2) -{ - x2 = ffxMin(x2, FFX_MIN16_F(4.0f)); - return Lanczos2ApproxSqNoClamp(x2); -} -#endif //FFX_HALF - -FfxFloat32 Lanczos2ApproxNoClamp(FfxFloat32 x) -{ - return Lanczos2ApproxSqNoClamp(x * x); -} - -#if FFX_HALF -FFX_MIN16_F Lanczos2ApproxNoClamp(FFX_MIN16_F x) -{ - return Lanczos2ApproxSqNoClamp(x * x); -} -#endif //FFX_HALF - -FfxFloat32 Lanczos2Approx(FfxFloat32 x) -{ - return Lanczos2ApproxSq(x * x); -} - -#if FFX_HALF -FFX_MIN16_F Lanczos2Approx(FFX_MIN16_F x) -{ - return Lanczos2ApproxSq(x * x); -} -#endif //FFX_HALF - -FfxFloat32 Lanczos2_UseLUT(FfxFloat32 x) -{ - return SampleLanczos2Weight(abs(x)); -} - -#if FFX_HALF -FFX_MIN16_F Lanczos2_UseLUT(FFX_MIN16_F x) -{ - return FFX_MIN16_F(SampleLanczos2Weight(abs(x))); -} -#endif //FFX_HALF - -FfxFloat32x4 Lanczos2_UseLUT(FfxFloat32x4 fColor0, FfxFloat32x4 fColor1, FfxFloat32x4 fColor2, FfxFloat32x4 fColor3, FfxFloat32 t) -{ - FfxFloat32 fWeight0 = Lanczos2_UseLUT(-1.f - t); - FfxFloat32 fWeight1 = Lanczos2_UseLUT(-0.f - t); - FfxFloat32 fWeight2 = Lanczos2_UseLUT(+1.f - t); - FfxFloat32 fWeight3 = Lanczos2_UseLUT(+2.f - t); - return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); -} -#if FFX_HALF -FFX_MIN16_F4 Lanczos2_UseLUT(FFX_MIN16_F4 fColor0, FFX_MIN16_F4 fColor1, FFX_MIN16_F4 fColor2, FFX_MIN16_F4 fColor3, FFX_MIN16_F t) -{ - FFX_MIN16_F fWeight0 = Lanczos2_UseLUT(FFX_MIN16_F(-1.f) - t); - FFX_MIN16_F fWeight1 = Lanczos2_UseLUT(FFX_MIN16_F(-0.f) - t); - FFX_MIN16_F fWeight2 = Lanczos2_UseLUT(FFX_MIN16_F(+1.f) - t); - FFX_MIN16_F fWeight3 = Lanczos2_UseLUT(FFX_MIN16_F(+2.f) - t); - return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); -} -#endif - -FfxFloat32x4 Lanczos2(FfxFloat32x4 fColor0, FfxFloat32x4 fColor1, FfxFloat32x4 fColor2, FfxFloat32x4 fColor3, FfxFloat32 t) -{ - FfxFloat32 fWeight0 = Lanczos2(-1.f - t); - FfxFloat32 fWeight1 = Lanczos2(-0.f - t); - FfxFloat32 fWeight2 = Lanczos2(+1.f - t); - FfxFloat32 fWeight3 = Lanczos2(+2.f - t); - return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); -} - -FfxFloat32x4 Lanczos2(FetchedBicubicSamples Samples, FfxFloat32x2 fPxFrac) -{ - FfxFloat32x4 fColorX0 = Lanczos2(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); - FfxFloat32x4 fColorX1 = Lanczos2(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); - FfxFloat32x4 fColorX2 = Lanczos2(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); - FfxFloat32x4 fColorX3 = Lanczos2(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); - FfxFloat32x4 fColorXY = Lanczos2(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); - - // Deringing - - // TODO: only use 4 by checking jitter - const FfxInt32 iDeringingSampleCount = 4; - const FfxFloat32x4 fDeringingSamples[4] = { - Samples.fColor11, - Samples.fColor21, - Samples.fColor12, - Samples.fColor22, - }; - - FfxFloat32x4 fDeringingMin = fDeringingSamples[0]; - FfxFloat32x4 fDeringingMax = fDeringingSamples[0]; - - FFX_UNROLL - for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) { - - fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); - fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); - } - - fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); - - return fColorXY; -} - -#if FFX_HALF -FFX_MIN16_F4 Lanczos2(FFX_MIN16_F4 fColor0, FFX_MIN16_F4 fColor1, FFX_MIN16_F4 fColor2, FFX_MIN16_F4 fColor3, FFX_MIN16_F t) -{ - FFX_MIN16_F fWeight0 = Lanczos2(FFX_MIN16_F(-1.f) - t); - FFX_MIN16_F fWeight1 = Lanczos2(FFX_MIN16_F(-0.f) - t); - FFX_MIN16_F fWeight2 = Lanczos2(FFX_MIN16_F(+1.f) - t); - FFX_MIN16_F fWeight3 = Lanczos2(FFX_MIN16_F(+2.f) - t); - return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); -} - -FFX_MIN16_F4 Lanczos2(FetchedBicubicSamplesMin16 Samples, FFX_MIN16_F2 fPxFrac) -{ - FFX_MIN16_F4 fColorX0 = Lanczos2(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); - FFX_MIN16_F4 fColorX1 = Lanczos2(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); - FFX_MIN16_F4 fColorX2 = Lanczos2(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); - FFX_MIN16_F4 fColorX3 = Lanczos2(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); - FFX_MIN16_F4 fColorXY = Lanczos2(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); - - // Deringing - - // TODO: only use 4 by checking jitter - const FfxInt32 iDeringingSampleCount = 4; - const FFX_MIN16_F4 fDeringingSamples[4] = { - Samples.fColor11, - Samples.fColor21, - Samples.fColor12, - Samples.fColor22, - }; - - FFX_MIN16_F4 fDeringingMin = fDeringingSamples[0]; - FFX_MIN16_F4 fDeringingMax = fDeringingSamples[0]; - - FFX_UNROLL - for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) - { - fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); - fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); - } - - fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); - - return fColorXY; -} -#endif //FFX_HALF - - -FfxFloat32x4 Lanczos2LUT(FetchedBicubicSamples Samples, FfxFloat32x2 fPxFrac) -{ - FfxFloat32x4 fColorX0 = Lanczos2_UseLUT(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); - FfxFloat32x4 fColorX1 = Lanczos2_UseLUT(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); - FfxFloat32x4 fColorX2 = Lanczos2_UseLUT(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); - FfxFloat32x4 fColorX3 = Lanczos2_UseLUT(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); - FfxFloat32x4 fColorXY = Lanczos2_UseLUT(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); - - // Deringing - - // TODO: only use 4 by checking jitter - const FfxInt32 iDeringingSampleCount = 4; - const FfxFloat32x4 fDeringingSamples[4] = { - Samples.fColor11, - Samples.fColor21, - Samples.fColor12, - Samples.fColor22, - }; - - FfxFloat32x4 fDeringingMin = fDeringingSamples[0]; - FfxFloat32x4 fDeringingMax = fDeringingSamples[0]; - - FFX_UNROLL - for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) { - - fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); - fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); - } - - fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); - - return fColorXY; -} - -#if FFX_HALF -FFX_MIN16_F4 Lanczos2LUT(FetchedBicubicSamplesMin16 Samples, FFX_MIN16_F2 fPxFrac) -{ - FFX_MIN16_F4 fColorX0 = Lanczos2_UseLUT(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); - FFX_MIN16_F4 fColorX1 = Lanczos2_UseLUT(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); - FFX_MIN16_F4 fColorX2 = Lanczos2_UseLUT(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); - FFX_MIN16_F4 fColorX3 = Lanczos2_UseLUT(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); - FFX_MIN16_F4 fColorXY = Lanczos2_UseLUT(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); - - // Deringing - - // TODO: only use 4 by checking jitter - const FfxInt32 iDeringingSampleCount = 4; - const FFX_MIN16_F4 fDeringingSamples[4] = { - Samples.fColor11, - Samples.fColor21, - Samples.fColor12, - Samples.fColor22, - }; - - FFX_MIN16_F4 fDeringingMin = fDeringingSamples[0]; - FFX_MIN16_F4 fDeringingMax = fDeringingSamples[0]; - - FFX_UNROLL - for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) - { - fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); - fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); - } - - fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); - - return fColorXY; -} -#endif //FFX_HALF - - - -FfxFloat32x4 Lanczos2Approx(FfxFloat32x4 fColor0, FfxFloat32x4 fColor1, FfxFloat32x4 fColor2, FfxFloat32x4 fColor3, FfxFloat32 t) -{ - FfxFloat32 fWeight0 = Lanczos2ApproxNoClamp(-1.f - t); - FfxFloat32 fWeight1 = Lanczos2ApproxNoClamp(-0.f - t); - FfxFloat32 fWeight2 = Lanczos2ApproxNoClamp(+1.f - t); - FfxFloat32 fWeight3 = Lanczos2ApproxNoClamp(+2.f - t); - return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); -} - -#if FFX_HALF -FFX_MIN16_F4 Lanczos2Approx(FFX_MIN16_F4 fColor0, FFX_MIN16_F4 fColor1, FFX_MIN16_F4 fColor2, FFX_MIN16_F4 fColor3, FFX_MIN16_F t) -{ - FFX_MIN16_F fWeight0 = Lanczos2ApproxNoClamp(FFX_MIN16_F(-1.f) - t); - FFX_MIN16_F fWeight1 = Lanczos2ApproxNoClamp(FFX_MIN16_F(-0.f) - t); - FFX_MIN16_F fWeight2 = Lanczos2ApproxNoClamp(FFX_MIN16_F(+1.f) - t); - FFX_MIN16_F fWeight3 = Lanczos2ApproxNoClamp(FFX_MIN16_F(+2.f) - t); - return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); -} -#endif //FFX_HALF - -FfxFloat32x4 Lanczos2Approx(FetchedBicubicSamples Samples, FfxFloat32x2 fPxFrac) -{ - FfxFloat32x4 fColorX0 = Lanczos2Approx(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); - FfxFloat32x4 fColorX1 = Lanczos2Approx(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); - FfxFloat32x4 fColorX2 = Lanczos2Approx(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); - FfxFloat32x4 fColorX3 = Lanczos2Approx(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); - FfxFloat32x4 fColorXY = Lanczos2Approx(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); - - // Deringing - - // TODO: only use 4 by checking jitter - const FfxInt32 iDeringingSampleCount = 4; - const FfxFloat32x4 fDeringingSamples[4] = { - Samples.fColor11, - Samples.fColor21, - Samples.fColor12, - Samples.fColor22, - }; - - FfxFloat32x4 fDeringingMin = fDeringingSamples[0]; - FfxFloat32x4 fDeringingMax = fDeringingSamples[0]; - - FFX_UNROLL - for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) - { - fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); - fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); - } - - fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); - - return fColorXY; -} - -#if FFX_HALF -FFX_MIN16_F4 Lanczos2Approx(FetchedBicubicSamplesMin16 Samples, FFX_MIN16_F2 fPxFrac) -{ - FFX_MIN16_F4 fColorX0 = Lanczos2Approx(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); - FFX_MIN16_F4 fColorX1 = Lanczos2Approx(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); - FFX_MIN16_F4 fColorX2 = Lanczos2Approx(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); - FFX_MIN16_F4 fColorX3 = Lanczos2Approx(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); - FFX_MIN16_F4 fColorXY = Lanczos2Approx(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); - - // Deringing - - // TODO: only use 4 by checking jitter - const FfxInt32 iDeringingSampleCount = 4; - const FFX_MIN16_F4 fDeringingSamples[4] = { - Samples.fColor11, - Samples.fColor21, - Samples.fColor12, - Samples.fColor22, - }; - - FFX_MIN16_F4 fDeringingMin = fDeringingSamples[0]; - FFX_MIN16_F4 fDeringingMax = fDeringingSamples[0]; - - FFX_UNROLL - for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) - { - fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); - fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); - } - - fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); - - return fColorXY; -} -#endif - -// Clamp by offset direction. Assuming iPxSample is already in range and iPxOffset is compile time constant. -FfxInt32x2 ClampCoord(FfxInt32x2 iPxSample, FfxInt32x2 iPxOffset, FfxInt32x2 iTextureSize) -{ - FfxInt32x2 result = iPxSample + iPxOffset; - result.x = (iPxOffset.x < 0) ? ffxMax(result.x, 0) : result.x; - result.x = (iPxOffset.x > 0) ? ffxMin(result.x, iTextureSize.x - 1) : result.x; - result.y = (iPxOffset.y < 0) ? ffxMax(result.y, 0) : result.y; - result.y = (iPxOffset.y > 0) ? ffxMin(result.y, iTextureSize.y - 1) : result.y; - return result; -} -#if FFX_HALF -FFX_MIN16_I2 ClampCoord(FFX_MIN16_I2 iPxSample, FFX_MIN16_I2 iPxOffset, FFX_MIN16_I2 iTextureSize) -{ - FFX_MIN16_I2 result = iPxSample + iPxOffset; - result.x = (iPxOffset.x < FFX_MIN16_I(0)) ? ffxMax(result.x, FFX_MIN16_I(0)) : result.x; - result.x = (iPxOffset.x > FFX_MIN16_I(0)) ? ffxMin(result.x, iTextureSize.x - FFX_MIN16_I(1)) : result.x; - result.y = (iPxOffset.y < FFX_MIN16_I(0)) ? ffxMax(result.y, FFX_MIN16_I(0)) : result.y; - result.y = (iPxOffset.y > FFX_MIN16_I(0)) ? ffxMin(result.y, iTextureSize.y - FFX_MIN16_I(1)) : result.y; - return result; -} -#endif //FFX_HALF - - -#define DeclareCustomFetchBicubicSamplesWithType(SampleType, TextureType, AddrType, Name, LoadTexture) \ - SampleType Name(AddrType iPxSample, AddrType iTextureSize) \ - { \ - SampleType Samples; \ - \ - Samples.fColor00 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(-1, -1), iTextureSize))); \ - Samples.fColor10 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, -1), iTextureSize))); \ - Samples.fColor20 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, -1), iTextureSize))); \ - Samples.fColor30 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+2, -1), iTextureSize))); \ - \ - Samples.fColor01 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(-1, +0), iTextureSize))); \ - Samples.fColor11 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, +0), iTextureSize))); \ - Samples.fColor21 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, +0), iTextureSize))); \ - Samples.fColor31 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+2, +0), iTextureSize))); \ - \ - Samples.fColor02 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(-1, +1), iTextureSize))); \ - Samples.fColor12 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, +1), iTextureSize))); \ - Samples.fColor22 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, +1), iTextureSize))); \ - Samples.fColor32 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+2, +1), iTextureSize))); \ - \ - Samples.fColor03 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(-1, +2), iTextureSize))); \ - Samples.fColor13 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, +2), iTextureSize))); \ - Samples.fColor23 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, +2), iTextureSize))); \ - Samples.fColor33 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+2, +2), iTextureSize))); \ - \ - return Samples; \ - } - -#define DeclareCustomFetchBicubicSamples(Name, LoadTexture) \ - DeclareCustomFetchBicubicSamplesWithType(FetchedBicubicSamples, FfxFloat32x4, FfxInt32x2, Name, LoadTexture) - -#define DeclareCustomFetchBicubicSamplesMin16(Name, LoadTexture) \ - DeclareCustomFetchBicubicSamplesWithType(FetchedBicubicSamplesMin16, FFX_MIN16_F4, FfxInt32x2, Name, LoadTexture) - -#define DeclareCustomFetchBilinearSamplesWithType(SampleType, TextureType,AddrType, Name, LoadTexture) \ - SampleType Name(AddrType iPxSample, AddrType iTextureSize) \ - { \ - SampleType Samples; \ - Samples.fColor00 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, +0), iTextureSize))); \ - Samples.fColor10 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, +0), iTextureSize))); \ - Samples.fColor01 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, +1), iTextureSize))); \ - Samples.fColor11 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, +1), iTextureSize))); \ - return Samples; \ - } - -#define DeclareCustomFetchBilinearSamples(Name, LoadTexture) \ - DeclareCustomFetchBilinearSamplesWithType(FetchedBilinearSamples, FfxFloat32x4, FfxInt32x2, Name, LoadTexture) - -#define DeclareCustomFetchBilinearSamplesMin16(Name, LoadTexture) \ - DeclareCustomFetchBilinearSamplesWithType(FetchedBilinearSamplesMin16, FFX_MIN16_F4, FfxInt32x2, Name, LoadTexture) - -// BE CAREFUL: there is some precision issues and (3253, 125) leading to (3252.9989778, 125.001102) -// is common, so iPxSample can "jitter" -#define DeclareCustomTextureSample(Name, InterpolateSamples, FetchSamples) \ - FfxFloat32x4 Name(FfxFloat32x2 fUvSample, FfxInt32x2 iTextureSize) \ - { \ - FfxFloat32x2 fPxSample = (fUvSample * FfxFloat32x2(iTextureSize)) - FfxFloat32x2(0.5f, 0.5f); \ - /* Clamp base coords */ \ - fPxSample.x = ffxMax(0.0f, ffxMin(FfxFloat32(iTextureSize.x), fPxSample.x)); \ - fPxSample.y = ffxMax(0.0f, ffxMin(FfxFloat32(iTextureSize.y), fPxSample.y)); \ - /* */ \ - FfxInt32x2 iPxSample = FfxInt32x2(floor(fPxSample)); \ - FfxFloat32x2 fPxFrac = ffxFract(fPxSample); \ - FfxFloat32x4 fColorXY = FfxFloat32x4(InterpolateSamples(FetchSamples(iPxSample, iTextureSize), fPxFrac)); \ - return fColorXY; \ - } - -#define DeclareCustomTextureSampleMin16(Name, InterpolateSamples, FetchSamples) \ - FFX_MIN16_F4 Name(FfxFloat32x2 fUvSample, FfxInt32x2 iTextureSize) \ - { \ - FfxFloat32x2 fPxSample = (fUvSample * FfxFloat32x2(iTextureSize)) - FfxFloat32x2(0.5f, 0.5f); \ - /* Clamp base coords */ \ - fPxSample.x = ffxMax(0.0f, ffxMin(FfxFloat32(iTextureSize.x), fPxSample.x)); \ - fPxSample.y = ffxMax(0.0f, ffxMin(FfxFloat32(iTextureSize.y), fPxSample.y)); \ - /* */ \ - FfxInt32x2 iPxSample = FfxInt32x2(floor(fPxSample)); \ - FFX_MIN16_F2 fPxFrac = FFX_MIN16_F2(ffxFract(fPxSample)); \ - FFX_MIN16_F4 fColorXY = FFX_MIN16_F4(InterpolateSamples(FetchSamples(iPxSample, iTextureSize), fPxFrac)); \ - return fColorXY; \ - } - -#define FFX_FSR3UPSCALER_CONCAT_ID(x, y) x ## y -#define FFX_FSR3UPSCALER_CONCAT(x, y) FFX_FSR3UPSCALER_CONCAT_ID(x, y) -#define FFX_FSR3UPSCALER_SAMPLER_1D_0 Lanczos2 -#define FFX_FSR3UPSCALER_SAMPLER_1D_1 Lanczos2LUT -#define FFX_FSR3UPSCALER_SAMPLER_1D_2 Lanczos2Approx - -#define FFX_FSR3UPSCALER_GET_LANCZOS_SAMPLER1D(x) FFX_FSR3UPSCALER_CONCAT(FFX_FSR3UPSCALER_SAMPLER_1D_, x) - -#endif //!defined( FFX_FSR3UPSCALER_SAMPLE_H ) diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_sample.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_sample.h.meta deleted file mode 100644 index 8e3cf72..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_sample.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: dcb900c9deecd06419a8a4c10c305890 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_tcr_autogen.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_tcr_autogen.h.meta deleted file mode 100644 index bdeb970..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_tcr_autogen.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: f01d5a8fbd1f34a4ea8d971755a21b6c -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_upsample.h b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_upsample.h deleted file mode 100644 index 47e7ccf..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_upsample.h +++ /dev/null @@ -1,195 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#ifndef FFX_FSR3UPSCALER_UPSAMPLE_H -#define FFX_FSR3UPSCALER_UPSAMPLE_H - -FFX_STATIC const FfxUInt32 iLanczos2SampleCount = 16; - -void Deringing(RectificationBox clippingBox, FFX_PARAMETER_INOUT FfxFloat32x3 fColor) -{ - fColor = clamp(fColor, clippingBox.aabbMin, clippingBox.aabbMax); -} -#if FFX_HALF -void Deringing(RectificationBoxMin16 clippingBox, FFX_PARAMETER_INOUT FFX_MIN16_F3 fColor) -{ - fColor = clamp(fColor, clippingBox.aabbMin, clippingBox.aabbMax); -} -#endif - -#ifndef FFX_FSR3UPSCALER_OPTION_UPSAMPLE_USE_LANCZOS_TYPE -#define FFX_FSR3UPSCALER_OPTION_UPSAMPLE_USE_LANCZOS_TYPE 2 // Approximate -#endif - -FfxFloat32 GetUpsampleLanczosWeight(FfxFloat32x2 fSrcSampleOffset, FfxFloat32 fKernelWeight) -{ - FfxFloat32x2 fSrcSampleOffsetBiased = fSrcSampleOffset * fKernelWeight.xx; -#if FFX_FSR3UPSCALER_OPTION_UPSAMPLE_USE_LANCZOS_TYPE == 0 // LANCZOS_TYPE_REFERENCE - FfxFloat32 fSampleWeight = Lanczos2(length(fSrcSampleOffsetBiased)); -#elif FFX_FSR3UPSCALER_OPTION_UPSAMPLE_USE_LANCZOS_TYPE == 1 // LANCZOS_TYPE_LUT - FfxFloat32 fSampleWeight = Lanczos2_UseLUT(length(fSrcSampleOffsetBiased)); -#elif FFX_FSR3UPSCALER_OPTION_UPSAMPLE_USE_LANCZOS_TYPE == 2 // LANCZOS_TYPE_APPROXIMATE - FfxFloat32 fSampleWeight = Lanczos2ApproxSq(dot(fSrcSampleOffsetBiased, fSrcSampleOffsetBiased)); -#else -#error "Invalid Lanczos type" -#endif - return fSampleWeight; -} - -#if FFX_HALF -FFX_MIN16_F GetUpsampleLanczosWeight(FFX_MIN16_F2 fSrcSampleOffset, FFX_MIN16_F fKernelWeight) -{ - FFX_MIN16_F2 fSrcSampleOffsetBiased = fSrcSampleOffset * fKernelWeight.xx; -#if FFX_FSR3UPSCALER_OPTION_UPSAMPLE_USE_LANCZOS_TYPE == 0 // LANCZOS_TYPE_REFERENCE - FFX_MIN16_F fSampleWeight = Lanczos2(length(fSrcSampleOffsetBiased)); -#elif FFX_FSR3UPSCALER_OPTION_UPSAMPLE_USE_LANCZOS_TYPE == 1 // LANCZOS_TYPE_LUT - FFX_MIN16_F fSampleWeight = Lanczos2_UseLUT(length(fSrcSampleOffsetBiased)); -#elif FFX_FSR3UPSCALER_OPTION_UPSAMPLE_USE_LANCZOS_TYPE == 2 // LANCZOS_TYPE_APPROXIMATE - FFX_MIN16_F fSampleWeight = Lanczos2ApproxSq(dot(fSrcSampleOffsetBiased, fSrcSampleOffsetBiased)); - - // To Test: Save reciproqual sqrt compute - // FfxFloat32 fSampleWeight = Lanczos2Sq_UseLUT(dot(fSrcSampleOffsetBiased, fSrcSampleOffsetBiased)); -#else -#error "Invalid Lanczos type" -#endif - return fSampleWeight; -} -#endif - -FfxFloat32 ComputeMaxKernelWeight() { - const FfxFloat32 fKernelSizeBias = 1.0f; - - FfxFloat32 fKernelWeight = FfxFloat32(1) + (FfxFloat32(1.0f) / FfxFloat32x2(DownscaleFactor()) - FfxFloat32(1)).x * FfxFloat32(fKernelSizeBias); - - return ffxMin(FfxFloat32(1.99f), fKernelWeight); -} - -FfxFloat32x4 ComputeUpsampledColorAndWeight(const AccumulationPassCommonParams params, - FFX_PARAMETER_INOUT RectificationBox clippingBox, FfxFloat32 fReactiveFactor) -{ - #if FFX_FSR3UPSCALER_OPTION_UPSAMPLE_SAMPLERS_USE_DATA_HALF && FFX_HALF - #include "ffx_fsr3upscaler_force16_begin.h" - #endif - // We compute a sliced lanczos filter with 2 lobes (other slices are accumulated temporaly) - FfxFloat32x2 fDstOutputPos = FfxFloat32x2(params.iPxHrPos) + FFX_BROADCAST_FLOAT32X2(0.5f); // Destination resolution output pixel center position - FfxFloat32x2 fSrcOutputPos = fDstOutputPos * DownscaleFactor(); // Source resolution output pixel center position - FfxInt32x2 iSrcInputPos = FfxInt32x2(floor(fSrcOutputPos)); // TODO: what about weird upscale factors... - - #if FFX_FSR3UPSCALER_OPTION_UPSAMPLE_SAMPLERS_USE_DATA_HALF && FFX_HALF - #include "ffx_fsr3upscaler_force16_end.h" - #endif - - FfxFloat32x3 fSamples[iLanczos2SampleCount]; - - FfxFloat32x2 fSrcUnjitteredPos = (FfxFloat32x2(iSrcInputPos) + FfxFloat32x2(0.5f, 0.5f)) - Jitter(); // This is the un-jittered position of the sample at offset 0,0 - - FfxInt32x2 offsetTL; - offsetTL.x = (fSrcUnjitteredPos.x > fSrcOutputPos.x) ? FfxInt32(-2) : FfxInt32(-1); - offsetTL.y = (fSrcUnjitteredPos.y > fSrcOutputPos.y) ? FfxInt32(-2) : FfxInt32(-1); - - //Load samples - // If fSrcUnjitteredPos.y > fSrcOutputPos.y, indicates offsetTL.y = -2, sample offset Y will be [-2, 1], clipbox will be rows [1, 3]. - // Flip row# for sampling offset in this case, so first 0~2 rows in the sampled array can always be used for computing the clipbox. - // This reduces branch or cmove on sampled colors, but moving this overhead to sample position / weight calculation time which apply to less values. - const FfxBoolean bFlipRow = fSrcUnjitteredPos.y > fSrcOutputPos.y; - const FfxBoolean bFlipCol = fSrcUnjitteredPos.x > fSrcOutputPos.x; - - FfxFloat32x2 fOffsetTL = FfxFloat32x2(offsetTL); - - FFX_UNROLL - for (FfxInt32 row = 0; row < 3; row++) { - - FFX_UNROLL - for (FfxInt32 col = 0; col < 3; col++) { - FfxInt32 iSampleIndex = col + (row << 2); - - FfxInt32x2 sampleColRow = FfxInt32x2(bFlipCol ? (3 - col) : col, bFlipRow ? (3 - row) : row); - FfxInt32x2 iSrcSamplePos = FfxInt32x2(iSrcInputPos) + offsetTL + sampleColRow; - - const FfxInt32x2 sampleCoord = ClampLoad(iSrcSamplePos, FfxInt32x2(0, 0), FfxInt32x2(RenderSize())); - - fSamples[iSampleIndex] = LoadPreparedInputColor(FfxInt32x2(sampleCoord)); - } - } - - FfxFloat32x4 fColorAndWeight = FfxFloat32x4(0.0f, 0.0f, 0.0f, 0.0f); - - FfxFloat32x2 fBaseSampleOffset = FfxFloat32x2(fSrcUnjitteredPos - fSrcOutputPos); - - // Identify how much of each upsampled color to be used for this frame - const FfxFloat32 fKernelReactiveFactor = ffxMax(fReactiveFactor, FfxFloat32(params.bIsNewSample)); - const FfxFloat32 fKernelBiasMax = ComputeMaxKernelWeight() * (1.0f - fKernelReactiveFactor); - - const FfxFloat32 fKernelBiasMin = ffxMax(1.0f, ((1.0f + fKernelBiasMax) * 0.3f)); - const FfxFloat32 fKernelBiasFactor = ffxMax(0.0f, ffxMax(0.25f * params.fDepthClipFactor, fKernelReactiveFactor)); - const FfxFloat32 fKernelBias = ffxLerp(fKernelBiasMax, fKernelBiasMin, fKernelBiasFactor); - - const FfxFloat32 fRectificationCurveBias = ffxLerp(-2.0f, -3.0f, ffxSaturate(params.fHrVelocity / 50.0f)); - - FFX_UNROLL - for (FfxInt32 row = 0; row < 3; row++) { - FFX_UNROLL - for (FfxInt32 col = 0; col < 3; col++) { - FfxInt32 iSampleIndex = col + (row << 2); - - const FfxInt32x2 sampleColRow = FfxInt32x2(bFlipCol ? (3 - col) : col, bFlipRow ? (3 - row) : row); - const FfxFloat32x2 fOffset = fOffsetTL + FfxFloat32x2(sampleColRow); - FfxFloat32x2 fSrcSampleOffset = fBaseSampleOffset + fOffset; - - FfxInt32x2 iSrcSamplePos = FfxInt32x2(iSrcInputPos) + FfxInt32x2(offsetTL) + sampleColRow; - - const FfxFloat32 fOnScreenFactor = FfxFloat32(IsOnScreen(FfxInt32x2(iSrcSamplePos), FfxInt32x2(RenderSize()))); - FfxFloat32 fSampleWeight = fOnScreenFactor * FfxFloat32(GetUpsampleLanczosWeight(fSrcSampleOffset, fKernelBias)); - - fColorAndWeight += FfxFloat32x4(fSamples[iSampleIndex] * fSampleWeight, fSampleWeight); - - // Update rectification box - { - const FfxFloat32 fSrcSampleOffsetSq = dot(fSrcSampleOffset, fSrcSampleOffset); - const FfxFloat32 fBoxSampleWeight = exp(fRectificationCurveBias * fSrcSampleOffsetSq); - - const FfxBoolean bInitialSample = (row == 0) && (col == 0); - RectificationBoxAddSample(bInitialSample, clippingBox, fSamples[iSampleIndex], fBoxSampleWeight); - } - } - } - - RectificationBoxComputeVarianceBoxData(clippingBox); - - fColorAndWeight.w *= FfxFloat32(fColorAndWeight.w > FSR3UPSCALER_EPSILON); - - if (fColorAndWeight.w > FSR3UPSCALER_EPSILON) { - // Normalize for deringing (we need to compare colors) - fColorAndWeight.xyz = fColorAndWeight.xyz / fColorAndWeight.w; - fColorAndWeight.w *= fUpsampleLanczosWeightScale; - - Deringing(clippingBox, fColorAndWeight.xyz); - } - - #if FFX_FSR3UPSCALER_OPTION_UPSAMPLE_SAMPLERS_USE_DATA_HALF && FFX_HALF - #include "ffx_fsr3upscaler_force16_end.h" - #endif - - return fColorAndWeight; -} - -#endif //!defined( FFX_FSR3UPSCALER_UPSAMPLE_H ) diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_upsample.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_upsample.h.meta deleted file mode 100644 index dc0a729..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_upsample.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 3e7832c4a9154414f9eaa125acfe6cd5 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/fsr1.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/fsr1.meta deleted file mode 100644 index 731c94f..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/fsr1.meta +++ /dev/null @@ -1,8 +0,0 @@ -fileFormatVersion: 2 -guid: 09438bc445e66204f970dc99ca8dae5a -folderAsset: yes -DefaultImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/fsr1/ffx_fsr1.h b/Assets/Shaders/FSR3/shaders/fsr3upscaler/fsr1/ffx_fsr1.h deleted file mode 100644 index e780995..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/fsr1/ffx_fsr1.h +++ /dev/null @@ -1,1252 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -/// @defgroup FfxGPUFsr1 FidelityFX FSR1 -/// FidelityFX Super Resolution 1 GPU documentation -/// -/// @ingroup FfxGPUEffects - -/// Setup required constant values for EASU (works on CPU or GPU). -/// -/// @param [out] con0 -/// @param [out] con1 -/// @param [out] con2 -/// @param [out] con3 -/// @param [in] inputViewportInPixelsX The rendered image resolution being upscaled in X dimension. -/// @param [in] inputViewportInPixelsY The rendered image resolution being upscaled in Y dimension. -/// @param [in] inputSizeInPixelsX The resolution of the resource containing the input image (useful for dynamic resolution) in X dimension. -/// @param [in] inputSizeInPixelsY The resolution of the resource containing the input image (useful for dynamic resolution) in Y dimension. -/// @param [in] outputSizeInPixelsX The display resolution which the input image gets upscaled to in X dimension. -/// @param [in] outputSizeInPixelsY The display resolution which the input image gets upscaled to in Y dimension. -/// -/// @ingroup FfxGPUFsr1 -FFX_STATIC void ffxFsrPopulateEasuConstants( - FFX_PARAMETER_INOUT FfxUInt32x4 con0, - FFX_PARAMETER_INOUT FfxUInt32x4 con1, - FFX_PARAMETER_INOUT FfxUInt32x4 con2, - FFX_PARAMETER_INOUT FfxUInt32x4 con3, - FFX_PARAMETER_IN FfxFloat32 inputViewportInPixelsX, - FFX_PARAMETER_IN FfxFloat32 inputViewportInPixelsY, - FFX_PARAMETER_IN FfxFloat32 inputSizeInPixelsX, - FFX_PARAMETER_IN FfxFloat32 inputSizeInPixelsY, - FFX_PARAMETER_IN FfxFloat32 outputSizeInPixelsX, - FFX_PARAMETER_IN FfxFloat32 outputSizeInPixelsY) -{ - // Output integer position to a pixel position in viewport. - con0[0] = ffxAsUInt32(inputViewportInPixelsX * ffxReciprocal(outputSizeInPixelsX)); - con0[1] = ffxAsUInt32(inputViewportInPixelsY * ffxReciprocal(outputSizeInPixelsY)); - con0[2] = ffxAsUInt32(FfxFloat32(0.5) * inputViewportInPixelsX * ffxReciprocal(outputSizeInPixelsX) - FfxFloat32(0.5)); - con0[3] = ffxAsUInt32(FfxFloat32(0.5) * inputViewportInPixelsY * ffxReciprocal(outputSizeInPixelsY) - FfxFloat32(0.5)); - - // Viewport pixel position to normalized image space. - // This is used to get upper-left of 'F' tap. - con1[0] = ffxAsUInt32(ffxReciprocal(inputSizeInPixelsX)); - con1[1] = ffxAsUInt32(ffxReciprocal(inputSizeInPixelsY)); - - // Centers of gather4, first offset from upper-left of 'F'. - // +---+---+ - // | | | - // +--(0)--+ - // | b | c | - // +---F---+---+---+ - // | e | f | g | h | - // +--(1)--+--(2)--+ - // | i | j | k | l | - // +---+---+---+---+ - // | n | o | - // +--(3)--+ - // | | | - // +---+---+ - con1[2] = ffxAsUInt32(FfxFloat32(1.0) * ffxReciprocal(inputSizeInPixelsX)); - con1[3] = ffxAsUInt32(FfxFloat32(-1.0) * ffxReciprocal(inputSizeInPixelsY)); - - // These are from (0) instead of 'F'. - con2[0] = ffxAsUInt32(FfxFloat32(-1.0) * ffxReciprocal(inputSizeInPixelsX)); - con2[1] = ffxAsUInt32(FfxFloat32(2.0) * ffxReciprocal(inputSizeInPixelsY)); - con2[2] = ffxAsUInt32(FfxFloat32(1.0) * ffxReciprocal(inputSizeInPixelsX)); - con2[3] = ffxAsUInt32(FfxFloat32(2.0) * ffxReciprocal(inputSizeInPixelsY)); - con3[0] = ffxAsUInt32(FfxFloat32(0.0) * ffxReciprocal(inputSizeInPixelsX)); - con3[1] = ffxAsUInt32(FfxFloat32(4.0) * ffxReciprocal(inputSizeInPixelsY)); - con3[2] = con3[3] = 0; -} - -/// Setup required constant values for EASU (works on CPU or GPU). -/// -/// @param [out] con0 -/// @param [out] con1 -/// @param [out] con2 -/// @param [out] con3 -/// @param [in] inputViewportInPixelsX The resolution of the input in the X dimension. -/// @param [in] inputViewportInPixelsY The resolution of the input in the Y dimension. -/// @param [in] inputSizeInPixelsX The input size in pixels in the X dimension. -/// @param [in] inputSizeInPixelsY The input size in pixels in the Y dimension. -/// @param [in] outputSizeInPixelsX The output size in pixels in the X dimension. -/// @param [in] outputSizeInPixelsY The output size in pixels in the Y dimension. -/// @param [in] inputOffsetInPixelsX The input image offset in the X dimension into the resource containing it (useful for dynamic resolution). -/// @param [in] inputOffsetInPixelsY The input image offset in the Y dimension into the resource containing it (useful for dynamic resolution). -/// -/// @ingroup FfxGPUFsr1 -FFX_STATIC void ffxFsrPopulateEasuConstantsOffset( - FFX_PARAMETER_INOUT FfxUInt32x4 con0, - FFX_PARAMETER_INOUT FfxUInt32x4 con1, - FFX_PARAMETER_INOUT FfxUInt32x4 con2, - FFX_PARAMETER_INOUT FfxUInt32x4 con3, - FFX_PARAMETER_IN FfxFloat32 inputViewportInPixelsX, - FFX_PARAMETER_IN FfxFloat32 inputViewportInPixelsY, - FFX_PARAMETER_IN FfxFloat32 inputSizeInPixelsX, - FFX_PARAMETER_IN FfxFloat32 inputSizeInPixelsY, - FFX_PARAMETER_IN FfxFloat32 outputSizeInPixelsX, - FFX_PARAMETER_IN FfxFloat32 outputSizeInPixelsY, - FFX_PARAMETER_IN FfxFloat32 inputOffsetInPixelsX, - FFX_PARAMETER_IN FfxFloat32 inputOffsetInPixelsY) -{ - ffxFsrPopulateEasuConstants( - con0, - con1, - con2, - con3, - inputViewportInPixelsX, - inputViewportInPixelsY, - inputSizeInPixelsX, - inputSizeInPixelsY, - outputSizeInPixelsX, - outputSizeInPixelsY); - - // override - con0[2] = ffxAsUInt32(FfxFloat32(0.5) * inputViewportInPixelsX * ffxReciprocal(outputSizeInPixelsX) - FfxFloat32(0.5) + inputOffsetInPixelsX); - con0[3] = ffxAsUInt32(FfxFloat32(0.5) * inputViewportInPixelsY * ffxReciprocal(outputSizeInPixelsY) - FfxFloat32(0.5) + inputOffsetInPixelsY); -} - -#if defined(FFX_GPU) && defined(FFX_FSR_EASU_FLOAT) -// Input callback prototypes, need to be implemented by calling shader -FfxFloat32x4 FsrEasuRF(FfxFloat32x2 p); -FfxFloat32x4 FsrEasuGF(FfxFloat32x2 p); -FfxFloat32x4 FsrEasuBF(FfxFloat32x2 p); - -// Filtering for a given tap for the scalar. -void fsrEasuTapFloat( - FFX_PARAMETER_INOUT FfxFloat32x3 accumulatedColor, // Accumulated color, with negative lobe. - FFX_PARAMETER_INOUT FfxFloat32 accumulatedWeight, // Accumulated weight. - FFX_PARAMETER_IN FfxFloat32x2 pixelOffset, // Pixel offset from resolve position to tap. - FFX_PARAMETER_IN FfxFloat32x2 gradientDirection, // Gradient direction. - FFX_PARAMETER_IN FfxFloat32x2 length, // Length. - FFX_PARAMETER_IN FfxFloat32 negativeLobeStrength, // Negative lobe strength. - FFX_PARAMETER_IN FfxFloat32 clippingPoint, // Clipping point. - FFX_PARAMETER_IN FfxFloat32x3 color) // Tap color. -{ - // Rotate offset by direction. - FfxFloat32x2 rotatedOffset; - rotatedOffset.x = (pixelOffset.x * (gradientDirection.x)) + (pixelOffset.y * gradientDirection.y); - rotatedOffset.y = (pixelOffset.x * (-gradientDirection.y)) + (pixelOffset.y * gradientDirection.x); - - // Anisotropy. - rotatedOffset *= length; - - // Compute distance^2. - FfxFloat32 distanceSquared = rotatedOffset.x * rotatedOffset.x + rotatedOffset.y * rotatedOffset.y; - - // Limit to the window as at corner, 2 taps can easily be outside. - distanceSquared = ffxMin(distanceSquared, clippingPoint); - - // Approximation of lancos2 without sin() or rcp(), or sqrt() to get x. - // (25/16 * (2/5 * x^2 - 1)^2 - (25/16 - 1)) * (1/4 * x^2 - 1)^2 - // |_______________________________________| |_______________| - // base window - // The general form of the 'base' is, - // (a*(b*x^2-1)^2-(a-1)) - // Where 'a=1/(2*b-b^2)' and 'b' moves around the negative lobe. - FfxFloat32 weightB = FfxFloat32(2.0 / 5.0) * distanceSquared + FfxFloat32(-1.0); - FfxFloat32 weightA = negativeLobeStrength * distanceSquared + FfxFloat32(-1.0); - weightB *= weightB; - weightA *= weightA; - weightB = FfxFloat32(25.0 / 16.0) * weightB + FfxFloat32(-(25.0 / 16.0 - 1.0)); - FfxFloat32 weight = weightB * weightA; - - // Do weighted average. - accumulatedColor += color * weight; - accumulatedWeight += weight; -} - -// Accumulate direction and length. -void fsrEasuSetFloat( - FFX_PARAMETER_INOUT FfxFloat32x2 direction, - FFX_PARAMETER_INOUT FfxFloat32 length, - FFX_PARAMETER_IN FfxFloat32x2 pp, - FFX_PARAMETER_IN FfxBoolean biS, - FFX_PARAMETER_IN FfxBoolean biT, - FFX_PARAMETER_IN FfxBoolean biU, - FFX_PARAMETER_IN FfxBoolean biV, - FFX_PARAMETER_IN FfxFloat32 lA, - FFX_PARAMETER_IN FfxFloat32 lB, - FFX_PARAMETER_IN FfxFloat32 lC, - FFX_PARAMETER_IN FfxFloat32 lD, - FFX_PARAMETER_IN FfxFloat32 lE) -{ - // Compute bilinear weight, branches factor out as predicates are compiler time immediates. - // s t - // u v - FfxFloat32 weight = FfxFloat32(0.0); - if (biS) - weight = (FfxFloat32(1.0) - pp.x) * (FfxFloat32(1.0) - pp.y); - if (biT) - weight = pp.x * (FfxFloat32(1.0) - pp.y); - if (biU) - weight = (FfxFloat32(1.0) - pp.x) * pp.y; - if (biV) - weight = pp.x * pp.y; - - // Direction is the '+' diff. - // a - // b c d - // e - // Then takes magnitude from abs average of both sides of 'c'. - // Length converts gradient reversal to 0, smoothly to non-reversal at 1, shaped, then adding horz and vert terms. - FfxFloat32 dc = lD - lC; - FfxFloat32 cb = lC - lB; - FfxFloat32 lengthX = max(abs(dc), abs(cb)); - lengthX = ffxApproximateReciprocal(lengthX); - FfxFloat32 directionX = lD - lB; - direction.x += directionX * weight; - lengthX = ffxSaturate(abs(directionX) * lengthX); - lengthX *= lengthX; - length += lengthX * weight; - - // Repeat for the y axis. - FfxFloat32 ec = lE - lC; - FfxFloat32 ca = lC - lA; - FfxFloat32 lengthY = max(abs(ec), abs(ca)); - lengthY = ffxApproximateReciprocal(lengthY); - FfxFloat32 directionY = lE - lA; - direction.y += directionY * weight; - lengthY = ffxSaturate(abs(directionY) * lengthY); - lengthY *= lengthY; - length += lengthY * weight; -} - -/// Apply edge-aware spatial upsampling using 32bit floating point precision calculations. -/// -/// @param [out] outPixel The computed color of a pixel. -/// @param [in] integerPosition Integer pixel position within the output. -/// @param [in] con0 The first constant value generated by ffxFsrPopulateEasuConstants. -/// @param [in] con1 The second constant value generated by ffxFsrPopulateEasuConstants. -/// @param [in] con2 The third constant value generated by ffxFsrPopulateEasuConstants. -/// @param [in] con3 The fourth constant value generated by ffxFsrPopulateEasuConstants. -/// -/// @ingroup FSR -void ffxFsrEasuFloat( - FFX_PARAMETER_OUT FfxFloat32x3 pix, - FFX_PARAMETER_IN FfxUInt32x2 ip, - FFX_PARAMETER_IN FfxUInt32x4 con0, - FFX_PARAMETER_IN FfxUInt32x4 con1, - FFX_PARAMETER_IN FfxUInt32x4 con2, - FFX_PARAMETER_IN FfxUInt32x4 con3) -{ - // Get position of 'f'. - FfxFloat32x2 pp = FfxFloat32x2(ip) * ffxAsFloat(con0.xy) + ffxAsFloat(con0.zw); - FfxFloat32x2 fp = floor(pp); - pp -= fp; - - // 12-tap kernel. - // b c - // e f g h - // i j k l - // n o - // Gather 4 ordering. - // a b - // r g - // For packed FP16, need either {rg} or {ab} so using the following setup for gather in all versions, - // a b <- unused (z) - // r g - // a b a b - // r g r g - // a b - // r g <- unused (z) - // Allowing dead-code removal to remove the 'z's. - FfxFloat32x2 p0 = fp * ffxAsFloat(con1.xy) + ffxAsFloat(con1.zw); - - // These are from p0 to avoid pulling two constants on pre-Navi hardware. - FfxFloat32x2 p1 = p0 + ffxAsFloat(con2.xy); - FfxFloat32x2 p2 = p0 + ffxAsFloat(con2.zw); - FfxFloat32x2 p3 = p0 + ffxAsFloat(con3.xy); - FfxFloat32x4 bczzR = FsrEasuRF(p0); - FfxFloat32x4 bczzG = FsrEasuGF(p0); - FfxFloat32x4 bczzB = FsrEasuBF(p0); - FfxFloat32x4 ijfeR = FsrEasuRF(p1); - FfxFloat32x4 ijfeG = FsrEasuGF(p1); - FfxFloat32x4 ijfeB = FsrEasuBF(p1); - FfxFloat32x4 klhgR = FsrEasuRF(p2); - FfxFloat32x4 klhgG = FsrEasuGF(p2); - FfxFloat32x4 klhgB = FsrEasuBF(p2); - FfxFloat32x4 zzonR = FsrEasuRF(p3); - FfxFloat32x4 zzonG = FsrEasuGF(p3); - FfxFloat32x4 zzonB = FsrEasuBF(p3); - - // Simplest multi-channel approximate luma possible (luma times 2, in 2 FMA/MAD). - FfxFloat32x4 bczzL = bczzB * ffxBroadcast4(0.5) + (bczzR * ffxBroadcast4(0.5) + bczzG); - FfxFloat32x4 ijfeL = ijfeB * ffxBroadcast4(0.5) + (ijfeR * ffxBroadcast4(0.5) + ijfeG); - FfxFloat32x4 klhgL = klhgB * ffxBroadcast4(0.5) + (klhgR * ffxBroadcast4(0.5) + klhgG); - FfxFloat32x4 zzonL = zzonB * ffxBroadcast4(0.5) + (zzonR * ffxBroadcast4(0.5) + zzonG); - - // Rename. - FfxFloat32 bL = bczzL.x; - FfxFloat32 cL = bczzL.y; - FfxFloat32 iL = ijfeL.x; - FfxFloat32 jL = ijfeL.y; - FfxFloat32 fL = ijfeL.z; - FfxFloat32 eL = ijfeL.w; - FfxFloat32 kL = klhgL.x; - FfxFloat32 lL = klhgL.y; - FfxFloat32 hL = klhgL.z; - FfxFloat32 gL = klhgL.w; - FfxFloat32 oL = zzonL.z; - FfxFloat32 nL = zzonL.w; - - // Accumulate for bilinear interpolation. - FfxFloat32x2 dir = ffxBroadcast2(0.0); - FfxFloat32 len = FfxFloat32(0.0); - fsrEasuSetFloat(dir, len, pp, FFX_TRUE, FFX_FALSE, FFX_FALSE, FFX_FALSE, bL, eL, fL, gL, jL); - fsrEasuSetFloat(dir, len, pp, FFX_FALSE, FFX_TRUE, FFX_FALSE, FFX_FALSE, cL, fL, gL, hL, kL); - fsrEasuSetFloat(dir, len, pp, FFX_FALSE, FFX_FALSE, FFX_TRUE, FFX_FALSE, fL, iL, jL, kL, nL); - fsrEasuSetFloat(dir, len, pp, FFX_FALSE, FFX_FALSE, FFX_FALSE, FFX_TRUE, gL, jL, kL, lL, oL); - - // Normalize with approximation, and cleanup close to zero. - FfxFloat32x2 dir2 = dir * dir; - FfxFloat32 dirR = dir2.x + dir2.y; - FfxBoolean zro = dirR < FfxFloat32(1.0 / 32768.0); - dirR = ffxApproximateReciprocalSquareRoot(dirR); - dirR = zro ? FfxFloat32(1.0) : dirR; - dir.x = zro ? FfxFloat32(1.0) : dir.x; - dir *= ffxBroadcast2(dirR); - - // Transform from {0 to 2} to {0 to 1} range, and shape with square. - len = len * FfxFloat32(0.5); - len *= len; - - // Stretch kernel {1.0 vert|horz, to sqrt(2.0) on diagonal}. - FfxFloat32 stretch = (dir.x * dir.x + dir.y * dir.y) * ffxApproximateReciprocal(max(abs(dir.x), abs(dir.y))); - - // Anisotropic length after rotation, - // x := 1.0 lerp to 'stretch' on edges - // y := 1.0 lerp to 2x on edges - FfxFloat32x2 len2 = FfxFloat32x2(FfxFloat32(1.0) + (stretch - FfxFloat32(1.0)) * len, FfxFloat32(1.0) + FfxFloat32(-0.5) * len); - - // Based on the amount of 'edge', - // the window shifts from +/-{sqrt(2.0) to slightly beyond 2.0}. - FfxFloat32 lob = FfxFloat32(0.5) + FfxFloat32((1.0 / 4.0 - 0.04) - 0.5) * len; - - // Set distance^2 clipping point to the end of the adjustable window. - FfxFloat32 clp = ffxApproximateReciprocal(lob); - - // Accumulation mixed with min/max of 4 nearest. - // b c - // e f g h - // i j k l - // n o - FfxFloat32x3 min4 = - ffxMin(ffxMin3(FfxFloat32x3(ijfeR.z, ijfeG.z, ijfeB.z), FfxFloat32x3(klhgR.w, klhgG.w, klhgB.w), FfxFloat32x3(ijfeR.y, ijfeG.y, ijfeB.y)), - FfxFloat32x3(klhgR.x, klhgG.x, klhgB.x)); - FfxFloat32x3 max4 = - max(ffxMax3(FfxFloat32x3(ijfeR.z, ijfeG.z, ijfeB.z), FfxFloat32x3(klhgR.w, klhgG.w, klhgB.w), FfxFloat32x3(ijfeR.y, ijfeG.y, ijfeB.y)), FfxFloat32x3(klhgR.x, klhgG.x, klhgB.x)); - - // Accumulation. - FfxFloat32x3 aC = ffxBroadcast3(0.0); - FfxFloat32 aW = FfxFloat32(0.0); - fsrEasuTapFloat(aC, aW, FfxFloat32x2(0.0, -1.0) - pp, dir, len2, lob, clp, FfxFloat32x3(bczzR.x, bczzG.x, bczzB.x)); // b - fsrEasuTapFloat(aC, aW, FfxFloat32x2(1.0, -1.0) - pp, dir, len2, lob, clp, FfxFloat32x3(bczzR.y, bczzG.y, bczzB.y)); // c - fsrEasuTapFloat(aC, aW, FfxFloat32x2(-1.0, 1.0) - pp, dir, len2, lob, clp, FfxFloat32x3(ijfeR.x, ijfeG.x, ijfeB.x)); // i - fsrEasuTapFloat(aC, aW, FfxFloat32x2(0.0, 1.0) - pp, dir, len2, lob, clp, FfxFloat32x3(ijfeR.y, ijfeG.y, ijfeB.y)); // j - fsrEasuTapFloat(aC, aW, FfxFloat32x2(0.0, 0.0) - pp, dir, len2, lob, clp, FfxFloat32x3(ijfeR.z, ijfeG.z, ijfeB.z)); // f - fsrEasuTapFloat(aC, aW, FfxFloat32x2(-1.0, 0.0) - pp, dir, len2, lob, clp, FfxFloat32x3(ijfeR.w, ijfeG.w, ijfeB.w)); // e - fsrEasuTapFloat(aC, aW, FfxFloat32x2(1.0, 1.0) - pp, dir, len2, lob, clp, FfxFloat32x3(klhgR.x, klhgG.x, klhgB.x)); // k - fsrEasuTapFloat(aC, aW, FfxFloat32x2(2.0, 1.0) - pp, dir, len2, lob, clp, FfxFloat32x3(klhgR.y, klhgG.y, klhgB.y)); // l - fsrEasuTapFloat(aC, aW, FfxFloat32x2(2.0, 0.0) - pp, dir, len2, lob, clp, FfxFloat32x3(klhgR.z, klhgG.z, klhgB.z)); // h - fsrEasuTapFloat(aC, aW, FfxFloat32x2(1.0, 0.0) - pp, dir, len2, lob, clp, FfxFloat32x3(klhgR.w, klhgG.w, klhgB.w)); // g - fsrEasuTapFloat(aC, aW, FfxFloat32x2(1.0, 2.0) - pp, dir, len2, lob, clp, FfxFloat32x3(zzonR.z, zzonG.z, zzonB.z)); // o - fsrEasuTapFloat(aC, aW, FfxFloat32x2(0.0, 2.0) - pp, dir, len2, lob, clp, FfxFloat32x3(zzonR.w, zzonG.w, zzonB.w)); // n - - // Normalize and dering. - pix = ffxMin(max4, max(min4, aC * ffxBroadcast3(rcp(aW)))); -} -#endif // #if defined(FFX_GPU) && defined(FFX_FSR_EASU_FLOAT) - -#if defined(FFX_GPU) && FFX_HALF == 1 && defined(FFX_FSR_EASU_HALF) -// Input callback prototypes, need to be implemented by calling shader -FfxFloat16x4 FsrEasuRH(FfxFloat32x2 p); -FfxFloat16x4 FsrEasuGH(FfxFloat32x2 p); -FfxFloat16x4 FsrEasuBH(FfxFloat32x2 p); - -// This runs 2 taps in parallel. -void FsrEasuTapH( - FFX_PARAMETER_INOUT FfxFloat16x2 aCR, - FFX_PARAMETER_INOUT FfxFloat16x2 aCG, - FFX_PARAMETER_INOUT FfxFloat16x2 aCB, - FFX_PARAMETER_INOUT FfxFloat16x2 aW, - FFX_PARAMETER_IN FfxFloat16x2 offX, - FFX_PARAMETER_IN FfxFloat16x2 offY, - FFX_PARAMETER_IN FfxFloat16x2 dir, - FFX_PARAMETER_IN FfxFloat16x2 len, - FFX_PARAMETER_IN FfxFloat16 lob, - FFX_PARAMETER_IN FfxFloat16 clp, - FFX_PARAMETER_IN FfxFloat16x2 cR, - FFX_PARAMETER_IN FfxFloat16x2 cG, - FFX_PARAMETER_IN FfxFloat16x2 cB) -{ - FfxFloat16x2 vX, vY; - vX = offX * dir.xx + offY * dir.yy; - vY = offX * (-dir.yy) + offY * dir.xx; - vX *= len.x; - vY *= len.y; - FfxFloat16x2 d2 = vX * vX + vY * vY; - d2 = min(d2, FFX_BROADCAST_FLOAT16X2(clp)); - FfxFloat16x2 wB = FFX_BROADCAST_FLOAT16X2(2.0 / 5.0) * d2 + FFX_BROADCAST_FLOAT16X2(-1.0); - FfxFloat16x2 wA = FFX_BROADCAST_FLOAT16X2(lob) * d2 + FFX_BROADCAST_FLOAT16X2(-1.0); - wB *= wB; - wA *= wA; - wB = FFX_BROADCAST_FLOAT16X2(25.0 / 16.0) * wB + FFX_BROADCAST_FLOAT16X2(-(25.0 / 16.0 - 1.0)); - FfxFloat16x2 w = wB * wA; - aCR += cR * w; - aCG += cG * w; - aCB += cB * w; - aW += w; -} - -// This runs 2 taps in parallel. -void FsrEasuSetH( - FFX_PARAMETER_INOUT FfxFloat16x2 dirPX, - FFX_PARAMETER_INOUT FfxFloat16x2 dirPY, - FFX_PARAMETER_INOUT FfxFloat16x2 lenP, - FFX_PARAMETER_IN FfxFloat16x2 pp, - FFX_PARAMETER_IN FfxBoolean biST, - FFX_PARAMETER_IN FfxBoolean biUV, - FFX_PARAMETER_IN FfxFloat16x2 lA, - FFX_PARAMETER_IN FfxFloat16x2 lB, - FFX_PARAMETER_IN FfxFloat16x2 lC, - FFX_PARAMETER_IN FfxFloat16x2 lD, - FFX_PARAMETER_IN FfxFloat16x2 lE) -{ - FfxFloat16x2 w = FFX_BROADCAST_FLOAT16X2(0.0); - - if (biST) - w = (FfxFloat16x2(1.0, 0.0) + FfxFloat16x2(-pp.x, pp.x)) * FFX_BROADCAST_FLOAT16X2(FFX_BROADCAST_FLOAT16(1.0) - pp.y); - - if (biUV) - w = (FfxFloat16x2(1.0, 0.0) + FfxFloat16x2(-pp.x, pp.x)) * FFX_BROADCAST_FLOAT16X2(pp.y); - - // ABS is not free in the packed FP16 path. - FfxFloat16x2 dc = lD - lC; - FfxFloat16x2 cb = lC - lB; - FfxFloat16x2 lenX = max(abs(dc), abs(cb)); - lenX = ffxReciprocalHalf(lenX); - - FfxFloat16x2 dirX = lD - lB; - dirPX += dirX * w; - lenX = FfxFloat16x2(ffxSaturate(abs(dirX) * lenX)); - lenX *= lenX; - lenP += lenX * w; - FfxFloat16x2 ec = lE - lC; - FfxFloat16x2 ca = lC - lA; - FfxFloat16x2 lenY = max(abs(ec), abs(ca)); - lenY = ffxReciprocalHalf(lenY); - FfxFloat16x2 dirY = lE - lA; - dirPY += dirY * w; - lenY = FfxFloat16x2(ffxSaturate(abs(dirY) * lenY)); - lenY *= lenY; - lenP += lenY * w; -} - -void FsrEasuH( - FFX_PARAMETER_OUT FfxFloat16x3 pix, - FFX_PARAMETER_IN FfxUInt32x2 ip, - FFX_PARAMETER_IN FfxUInt32x4 con0, - FFX_PARAMETER_IN FfxUInt32x4 con1, - FFX_PARAMETER_IN FfxUInt32x4 con2, - FFX_PARAMETER_IN FfxUInt32x4 con3) -{ - FfxFloat32x2 pp = FfxFloat32x2(ip) * ffxAsFloat(con0.xy) + ffxAsFloat(con0.zw); - FfxFloat32x2 fp = floor(pp); - pp -= fp; - FfxFloat16x2 ppp = FfxFloat16x2(pp); - - FfxFloat32x2 p0 = fp * ffxAsFloat(con1.xy) + ffxAsFloat(con1.zw); - FfxFloat32x2 p1 = p0 + ffxAsFloat(con2.xy); - FfxFloat32x2 p2 = p0 + ffxAsFloat(con2.zw); - FfxFloat32x2 p3 = p0 + ffxAsFloat(con3.xy); - FfxFloat16x4 bczzR = FsrEasuRH(p0); - FfxFloat16x4 bczzG = FsrEasuGH(p0); - FfxFloat16x4 bczzB = FsrEasuBH(p0); - FfxFloat16x4 ijfeR = FsrEasuRH(p1); - FfxFloat16x4 ijfeG = FsrEasuGH(p1); - FfxFloat16x4 ijfeB = FsrEasuBH(p1); - FfxFloat16x4 klhgR = FsrEasuRH(p2); - FfxFloat16x4 klhgG = FsrEasuGH(p2); - FfxFloat16x4 klhgB = FsrEasuBH(p2); - FfxFloat16x4 zzonR = FsrEasuRH(p3); - FfxFloat16x4 zzonG = FsrEasuGH(p3); - FfxFloat16x4 zzonB = FsrEasuBH(p3); - - FfxFloat16x4 bczzL = bczzB * FFX_BROADCAST_FLOAT16X4(0.5) + (bczzR * FFX_BROADCAST_FLOAT16X4(0.5) + bczzG); - FfxFloat16x4 ijfeL = ijfeB * FFX_BROADCAST_FLOAT16X4(0.5) + (ijfeR * FFX_BROADCAST_FLOAT16X4(0.5) + ijfeG); - FfxFloat16x4 klhgL = klhgB * FFX_BROADCAST_FLOAT16X4(0.5) + (klhgR * FFX_BROADCAST_FLOAT16X4(0.5) + klhgG); - FfxFloat16x4 zzonL = zzonB * FFX_BROADCAST_FLOAT16X4(0.5) + (zzonR * FFX_BROADCAST_FLOAT16X4(0.5) + zzonG); - FfxFloat16 bL = bczzL.x; - FfxFloat16 cL = bczzL.y; - FfxFloat16 iL = ijfeL.x; - FfxFloat16 jL = ijfeL.y; - FfxFloat16 fL = ijfeL.z; - FfxFloat16 eL = ijfeL.w; - FfxFloat16 kL = klhgL.x; - FfxFloat16 lL = klhgL.y; - FfxFloat16 hL = klhgL.z; - FfxFloat16 gL = klhgL.w; - FfxFloat16 oL = zzonL.z; - FfxFloat16 nL = zzonL.w; - - // This part is different, accumulating 2 taps in parallel. - FfxFloat16x2 dirPX = FFX_BROADCAST_FLOAT16X2(0.0); - FfxFloat16x2 dirPY = FFX_BROADCAST_FLOAT16X2(0.0); - FfxFloat16x2 lenP = FFX_BROADCAST_FLOAT16X2(0.0); - FsrEasuSetH(dirPX, - dirPY, - lenP, - ppp, - FfxBoolean(true), - FfxBoolean(false), - FfxFloat16x2(bL, cL), - FfxFloat16x2(eL, fL), - FfxFloat16x2(fL, gL), - FfxFloat16x2(gL, hL), - FfxFloat16x2(jL, kL)); - FsrEasuSetH(dirPX, - dirPY, - lenP, - ppp, - FfxBoolean(false), - FfxBoolean(true), - FfxFloat16x2(fL, gL), - FfxFloat16x2(iL, jL), - FfxFloat16x2(jL, kL), - FfxFloat16x2(kL, lL), - FfxFloat16x2(nL, oL)); - FfxFloat16x2 dir = FfxFloat16x2(dirPX.r + dirPX.g, dirPY.r + dirPY.g); - FfxFloat16 len = lenP.r + lenP.g; - - FfxFloat16x2 dir2 = dir * dir; - FfxFloat16 dirR = dir2.x + dir2.y; - FfxUInt32 zro = FfxUInt32(dirR < FFX_BROADCAST_FLOAT16(1.0 / 32768.0)); - dirR = ffxApproximateReciprocalSquareRootHalf(dirR); - dirR = (zro > 0) ? FFX_BROADCAST_FLOAT16(1.0) : dirR; - dir.x = (zro > 0) ? FFX_BROADCAST_FLOAT16(1.0) : dir.x; - dir *= FFX_BROADCAST_FLOAT16X2(dirR); - len = len * FFX_BROADCAST_FLOAT16(0.5); - len *= len; - FfxFloat16 stretch = (dir.x * dir.x + dir.y * dir.y) * ffxApproximateReciprocalHalf(max(abs(dir.x), abs(dir.y))); - FfxFloat16x2 len2 = - FfxFloat16x2(FFX_BROADCAST_FLOAT16(1.0) + (stretch - FFX_BROADCAST_FLOAT16(1.0)) * len, FFX_BROADCAST_FLOAT16(1.0) + FFX_BROADCAST_FLOAT16(-0.5) * len); - FfxFloat16 lob = FFX_BROADCAST_FLOAT16(0.5) + FFX_BROADCAST_FLOAT16((1.0 / 4.0 - 0.04) - 0.5) * len; - FfxFloat16 clp = ffxApproximateReciprocalHalf(lob); - - // FP16 is different, using packed trick to do min and max in same operation. - FfxFloat16x2 bothR = - max(max(FfxFloat16x2(-ijfeR.z, ijfeR.z), FfxFloat16x2(-klhgR.w, klhgR.w)), max(FfxFloat16x2(-ijfeR.y, ijfeR.y), FfxFloat16x2(-klhgR.x, klhgR.x))); - FfxFloat16x2 bothG = - max(max(FfxFloat16x2(-ijfeG.z, ijfeG.z), FfxFloat16x2(-klhgG.w, klhgG.w)), max(FfxFloat16x2(-ijfeG.y, ijfeG.y), FfxFloat16x2(-klhgG.x, klhgG.x))); - FfxFloat16x2 bothB = - max(max(FfxFloat16x2(-ijfeB.z, ijfeB.z), FfxFloat16x2(-klhgB.w, klhgB.w)), max(FfxFloat16x2(-ijfeB.y, ijfeB.y), FfxFloat16x2(-klhgB.x, klhgB.x))); - - // This part is different for FP16, working pairs of taps at a time. - FfxFloat16x2 pR = FFX_BROADCAST_FLOAT16X2(0.0); - FfxFloat16x2 pG = FFX_BROADCAST_FLOAT16X2(0.0); - FfxFloat16x2 pB = FFX_BROADCAST_FLOAT16X2(0.0); - FfxFloat16x2 pW = FFX_BROADCAST_FLOAT16X2(0.0); - FsrEasuTapH(pR, pG, pB, pW, FfxFloat16x2(0.0, 1.0) - ppp.xx, FfxFloat16x2(-1.0, -1.0) - ppp.yy, dir, len2, lob, clp, bczzR.xy, bczzG.xy, bczzB.xy); - FsrEasuTapH(pR, pG, pB, pW, FfxFloat16x2(-1.0, 0.0) - ppp.xx, FfxFloat16x2(1.0, 1.0) - ppp.yy, dir, len2, lob, clp, ijfeR.xy, ijfeG.xy, ijfeB.xy); - FsrEasuTapH(pR, pG, pB, pW, FfxFloat16x2(0.0, -1.0) - ppp.xx, FfxFloat16x2(0.0, 0.0) - ppp.yy, dir, len2, lob, clp, ijfeR.zw, ijfeG.zw, ijfeB.zw); - FsrEasuTapH(pR, pG, pB, pW, FfxFloat16x2(1.0, 2.0) - ppp.xx, FfxFloat16x2(1.0, 1.0) - ppp.yy, dir, len2, lob, clp, klhgR.xy, klhgG.xy, klhgB.xy); - FsrEasuTapH(pR, pG, pB, pW, FfxFloat16x2(2.0, 1.0) - ppp.xx, FfxFloat16x2(0.0, 0.0) - ppp.yy, dir, len2, lob, clp, klhgR.zw, klhgG.zw, klhgB.zw); - FsrEasuTapH(pR, pG, pB, pW, FfxFloat16x2(1.0, 0.0) - ppp.xx, FfxFloat16x2(2.0, 2.0) - ppp.yy, dir, len2, lob, clp, zzonR.zw, zzonG.zw, zzonB.zw); - FfxFloat16x3 aC = FfxFloat16x3(pR.x + pR.y, pG.x + pG.y, pB.x + pB.y); - FfxFloat16 aW = pW.x + pW.y; - - // Slightly different for FP16 version due to combined min and max. - pix = min(FfxFloat16x3(bothR.y, bothG.y, bothB.y), max(-FfxFloat16x3(bothR.x, bothG.x, bothB.x), aC * FFX_BROADCAST_FLOAT16X3(ffxReciprocalHalf(aW)))); -} -#endif // #if defined(FFX_GPU) && defined(FFX_HALF) && defined(FFX_FSR_EASU_HALF) - -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//_____________________________________________________________/\_______________________________________________________________ -//============================================================================================================================== -// -// FSR - [RCAS] ROBUST CONTRAST ADAPTIVE SHARPENING -// -//------------------------------------------------------------------------------------------------------------------------------ -// CAS uses a simplified mechanism to convert local contrast into a variable amount of sharpness. -// RCAS uses a more exact mechanism, solving for the maximum local sharpness possible before clipping. -// RCAS also has a built in process to limit sharpening of what it detects as possible noise. -// RCAS sharper does not support scaling, as it should be applied after EASU scaling. -// Pass EASU output straight into RCAS, no color conversions necessary. -//------------------------------------------------------------------------------------------------------------------------------ -// RCAS is based on the following logic. -// RCAS uses a 5 tap filter in a cross pattern (same as CAS), -// w n -// w 1 w for taps w m e -// w s -// Where 'w' is the negative lobe weight. -// output = (w*(n+e+w+s)+m)/(4*w+1) -// RCAS solves for 'w' by seeing where the signal might clip out of the {0 to 1} input range, -// 0 == (w*(n+e+w+s)+m)/(4*w+1) -> w = -m/(n+e+w+s) -// 1 == (w*(n+e+w+s)+m)/(4*w+1) -> w = (1-m)/(n+e+w+s-4*1) -// Then chooses the 'w' which results in no clipping, limits 'w', and multiplies by the 'sharp' amount. -// This solution above has issues with MSAA input as the steps along the gradient cause edge detection issues. -// So RCAS uses 4x the maximum and 4x the minimum (depending on equation)in place of the individual taps. -// As well as switching from 'm' to either the minimum or maximum (depending on side), to help in energy conservation. -// This stabilizes RCAS. -// RCAS does a simple highpass which is normalized against the local contrast then shaped, -// 0.25 -// 0.25 -1 0.25 -// 0.25 -// This is used as a noise detection filter, to reduce the effect of RCAS on grain, and focus on real edges. -// -// GLSL example for the required callbacks : -// -// FfxFloat16x4 FsrRcasLoadH(FfxInt16x2 p){return FfxFloat16x4(imageLoad(imgSrc,FfxInt32x2(p)));} -// void FsrRcasInputH(inout FfxFloat16 r,inout FfxFloat16 g,inout FfxFloat16 b) -// { -// //do any simple input color conversions here or leave empty if none needed -// } -// -// FsrRcasCon need to be called from the CPU or GPU to set up constants. -// Including a GPU example here, the 'con' value would be stored out to a constant buffer. -// -// FfxUInt32x4 con; -// FsrRcasCon(con, -// 0.0); // The scale is {0.0 := maximum sharpness, to N>0, where N is the number of stops (halving) of the reduction of sharpness}. -// --------------- -// RCAS sharpening supports a CAS-like pass-through alpha via, -// #define FSR_RCAS_PASSTHROUGH_ALPHA 1 -// RCAS also supports a define to enable a more expensive path to avoid some sharpening of noise. -// Would suggest it is better to apply film grain after RCAS sharpening (and after scaling) instead of using this define, -// #define FSR_RCAS_DENOISE 1 -//============================================================================================================================== -// This is set at the limit of providing unnatural results for sharpening. -#define FSR_RCAS_LIMIT (0.25-(1.0/16.0)) -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//_____________________________________________________________/\_______________________________________________________________ -//============================================================================================================================== -// CONSTANT SETUP -//============================================================================================================================== -// Call to setup required constant values (works on CPU or GPU). - FFX_STATIC void FsrRcasCon(FfxUInt32x4 con, - // The scale is {0.0 := maximum, to N>0, where N is the number of stops (halving) of the reduction of sharpness}. - FfxFloat32 sharpness) - { - // Transform from stops to linear value. - sharpness = exp2(-sharpness); - FfxFloat32x2 hSharp = {sharpness, sharpness}; - con[0] = ffxAsUInt32(sharpness); - con[1] = packHalf2x16(hSharp); - con[2] = 0; - con[3] = 0; - } - //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//_____________________________________________________________/\_______________________________________________________________ -//============================================================================================================================== -// NON-PACKED 32-BIT VERSION -//============================================================================================================================== -#if defined(FFX_GPU)&&defined(FSR_RCAS_F) - // Input callback prototypes that need to be implemented by calling shader - FfxFloat32x4 FsrRcasLoadF(FfxInt32x2 p); - void FsrRcasInputF(inout FfxFloat32 r,inout FfxFloat32 g,inout FfxFloat32 b); -//------------------------------------------------------------------------------------------------------------------------------ - void FsrRcasF(out FfxFloat32 pixR, // Output values, non-vector so port between RcasFilter() and RcasFilterH() is easy. - out FfxFloat32 pixG, - out FfxFloat32 pixB, -#ifdef FSR_RCAS_PASSTHROUGH_ALPHA - out FfxFloat32 pixA, -#endif - FfxUInt32x2 ip, // Integer pixel position in output. - FfxUInt32x4 con) - { // Constant generated by RcasSetup(). - // Algorithm uses minimal 3x3 pixel neighborhood. - // b - // d e f - // h - FfxInt32x2 sp = FfxInt32x2(ip); - FfxFloat32x3 b = FsrRcasLoadF(sp + FfxInt32x2(0, -1)).rgb; - FfxFloat32x3 d = FsrRcasLoadF(sp + FfxInt32x2(-1, 0)).rgb; -#ifdef FSR_RCAS_PASSTHROUGH_ALPHA - FfxFloat32x4 ee = FsrRcasLoadF(sp); - FfxFloat32x3 e = ee.rgb; - pixA = ee.a; -#else - FfxFloat32x3 e = FsrRcasLoadF(sp).rgb; -#endif - FfxFloat32x3 f = FsrRcasLoadF(sp + FfxInt32x2(1, 0)).rgb; - FfxFloat32x3 h = FsrRcasLoadF(sp + FfxInt32x2(0, 1)).rgb; - // Rename (32-bit) or regroup (16-bit). - FfxFloat32 bR = b.r; - FfxFloat32 bG = b.g; - FfxFloat32 bB = b.b; - FfxFloat32 dR = d.r; - FfxFloat32 dG = d.g; - FfxFloat32 dB = d.b; - FfxFloat32 eR = e.r; - FfxFloat32 eG = e.g; - FfxFloat32 eB = e.b; - FfxFloat32 fR = f.r; - FfxFloat32 fG = f.g; - FfxFloat32 fB = f.b; - FfxFloat32 hR = h.r; - FfxFloat32 hG = h.g; - FfxFloat32 hB = h.b; - // Run optional input transform. - FsrRcasInputF(bR, bG, bB); - FsrRcasInputF(dR, dG, dB); - FsrRcasInputF(eR, eG, eB); - FsrRcasInputF(fR, fG, fB); - FsrRcasInputF(hR, hG, hB); - // Luma times 2. - FfxFloat32 bL = bB * FfxFloat32(0.5) + (bR * FfxFloat32(0.5) + bG); - FfxFloat32 dL = dB * FfxFloat32(0.5) + (dR * FfxFloat32(0.5) + dG); - FfxFloat32 eL = eB * FfxFloat32(0.5) + (eR * FfxFloat32(0.5) + eG); - FfxFloat32 fL = fB * FfxFloat32(0.5) + (fR * FfxFloat32(0.5) + fG); - FfxFloat32 hL = hB * FfxFloat32(0.5) + (hR * FfxFloat32(0.5) + hG); - // Noise detection. - FfxFloat32 nz = FfxFloat32(0.25) * bL + FfxFloat32(0.25) * dL + FfxFloat32(0.25) * fL + FfxFloat32(0.25) * hL - eL; - nz = ffxSaturate(abs(nz) * ffxApproximateReciprocalMedium(ffxMax3(ffxMax3(bL, dL, eL), fL, hL) - ffxMin3(ffxMin3(bL, dL, eL), fL, hL))); - nz = FfxFloat32(-0.5) * nz + FfxFloat32(1.0); - // Min and max of ring. - FfxFloat32 mn4R = ffxMin(ffxMin3(bR, dR, fR), hR); - FfxFloat32 mn4G = ffxMin(ffxMin3(bG, dG, fG), hG); - FfxFloat32 mn4B = ffxMin(ffxMin3(bB, dB, fB), hB); - FfxFloat32 mx4R = max(ffxMax3(bR, dR, fR), hR); - FfxFloat32 mx4G = max(ffxMax3(bG, dG, fG), hG); - FfxFloat32 mx4B = max(ffxMax3(bB, dB, fB), hB); - // Immediate constants for peak range. - FfxFloat32x2 peakC = FfxFloat32x2(1.0, -1.0 * 4.0); - // Limiters, these need to be high precision RCPs. - FfxFloat32 hitMinR = mn4R * rcp(FfxFloat32(4.0) * mx4R); - FfxFloat32 hitMinG = mn4G * rcp(FfxFloat32(4.0) * mx4G); - FfxFloat32 hitMinB = mn4B * rcp(FfxFloat32(4.0) * mx4B); - FfxFloat32 hitMaxR = (peakC.x - mx4R) * rcp(FfxFloat32(4.0) * mn4R + peakC.y); - FfxFloat32 hitMaxG = (peakC.x - mx4G) * rcp(FfxFloat32(4.0) * mn4G + peakC.y); - FfxFloat32 hitMaxB = (peakC.x - mx4B) * rcp(FfxFloat32(4.0) * mn4B + peakC.y); - FfxFloat32 lobeR = max(-hitMinR, hitMaxR); - FfxFloat32 lobeG = max(-hitMinG, hitMaxG); - FfxFloat32 lobeB = max(-hitMinB, hitMaxB); - FfxFloat32 lobe = max(FfxFloat32(-FSR_RCAS_LIMIT), ffxMin(ffxMax3(lobeR, lobeG, lobeB), FfxFloat32(0.0))) * ffxAsFloat - (con.x); - // Apply noise removal. -#ifdef FSR_RCAS_DENOISE - lobe *= nz; -#endif - // Resolve, which needs the medium precision rcp approximation to avoid visible tonality changes. - FfxFloat32 rcpL = ffxApproximateReciprocalMedium(FfxFloat32(4.0) * lobe + FfxFloat32(1.0)); - pixR = (lobe * bR + lobe * dR + lobe * hR + lobe * fR + eR) * rcpL; - pixG = (lobe * bG + lobe * dG + lobe * hG + lobe * fG + eG) * rcpL; - pixB = (lobe * bB + lobe * dB + lobe * hB + lobe * fB + eB) * rcpL; - return; - } -#endif -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//_____________________________________________________________/\_______________________________________________________________ -//============================================================================================================================== -// NON-PACKED 16-BIT VERSION -//============================================================================================================================== -#if defined(FFX_GPU) && FFX_HALF == 1 && defined(FSR_RCAS_H) - // Input callback prototypes that need to be implemented by calling shader - FfxFloat16x4 FsrRcasLoadH(FfxInt16x2 p); - void FsrRcasInputH(inout FfxFloat16 r,inout FfxFloat16 g,inout FfxFloat16 b); -//------------------------------------------------------------------------------------------------------------------------------ - void FsrRcasH( - out FfxFloat16 pixR, // Output values, non-vector so port between RcasFilter() and RcasFilterH() is easy. - out FfxFloat16 pixG, - out FfxFloat16 pixB, - #ifdef FSR_RCAS_PASSTHROUGH_ALPHA - out FfxFloat16 pixA, - #endif - FfxUInt32x2 ip, // Integer pixel position in output. - FfxUInt32x4 con){ // Constant generated by RcasSetup(). - // Sharpening algorithm uses minimal 3x3 pixel neighborhood. - // b - // d e f - // h - FfxInt16x2 sp=FfxInt16x2(ip); - FfxFloat16x3 b=FsrRcasLoadH(sp+FfxInt16x2( 0,-1)).rgb; - FfxFloat16x3 d=FsrRcasLoadH(sp+FfxInt16x2(-1, 0)).rgb; - #ifdef FSR_RCAS_PASSTHROUGH_ALPHA - FfxFloat16x4 ee=FsrRcasLoadH(sp); - FfxFloat16x3 e=ee.rgb;pixA=ee.a; - #else - FfxFloat16x3 e=FsrRcasLoadH(sp).rgb; - #endif - FfxFloat16x3 f=FsrRcasLoadH(sp+FfxInt16x2( 1, 0)).rgb; - FfxFloat16x3 h=FsrRcasLoadH(sp+FfxInt16x2( 0, 1)).rgb; - // Rename (32-bit) or regroup (16-bit). - FfxFloat16 bR=b.r; - FfxFloat16 bG=b.g; - FfxFloat16 bB=b.b; - FfxFloat16 dR=d.r; - FfxFloat16 dG=d.g; - FfxFloat16 dB=d.b; - FfxFloat16 eR=e.r; - FfxFloat16 eG=e.g; - FfxFloat16 eB=e.b; - FfxFloat16 fR=f.r; - FfxFloat16 fG=f.g; - FfxFloat16 fB=f.b; - FfxFloat16 hR=h.r; - FfxFloat16 hG=h.g; - FfxFloat16 hB=h.b; - // Run optional input transform. - FsrRcasInputH(bR,bG,bB); - FsrRcasInputH(dR,dG,dB); - FsrRcasInputH(eR,eG,eB); - FsrRcasInputH(fR,fG,fB); - FsrRcasInputH(hR,hG,hB); - // Luma times 2. - FfxFloat16 bL=bB*FFX_BROADCAST_FLOAT16(0.5)+(bR*FFX_BROADCAST_FLOAT16(0.5)+bG); - FfxFloat16 dL=dB*FFX_BROADCAST_FLOAT16(0.5)+(dR*FFX_BROADCAST_FLOAT16(0.5)+dG); - FfxFloat16 eL=eB*FFX_BROADCAST_FLOAT16(0.5)+(eR*FFX_BROADCAST_FLOAT16(0.5)+eG); - FfxFloat16 fL=fB*FFX_BROADCAST_FLOAT16(0.5)+(fR*FFX_BROADCAST_FLOAT16(0.5)+fG); - FfxFloat16 hL=hB*FFX_BROADCAST_FLOAT16(0.5)+(hR*FFX_BROADCAST_FLOAT16(0.5)+hG); - // Noise detection. - FfxFloat16 nz=FFX_BROADCAST_FLOAT16(0.25)*bL+FFX_BROADCAST_FLOAT16(0.25)*dL+FFX_BROADCAST_FLOAT16(0.25)*fL+FFX_BROADCAST_FLOAT16(0.25)*hL-eL; - nz=FfxFloat16(ffxSaturate(abs(nz)*ffxApproximateReciprocalMediumHalf(ffxMax3Half(ffxMax3Half(bL,dL,eL),fL,hL)-ffxMin3Half(ffxMin3Half(bL,dL,eL),fL,hL)))); - nz=FFX_BROADCAST_FLOAT16(-0.5)*nz+FFX_BROADCAST_FLOAT16(1.0); - // Min and max of ring. - FfxFloat16 mn4R=min(ffxMin3Half(bR,dR,fR),hR); - FfxFloat16 mn4G=min(ffxMin3Half(bG,dG,fG),hG); - FfxFloat16 mn4B=min(ffxMin3Half(bB,dB,fB),hB); - FfxFloat16 mx4R=max(ffxMax3Half(bR,dR,fR),hR); - FfxFloat16 mx4G=max(ffxMax3Half(bG,dG,fG),hG); - FfxFloat16 mx4B=max(ffxMax3Half(bB,dB,fB),hB); - // Immediate constants for peak range. - FfxFloat16x2 peakC=FfxFloat16x2(1.0,-1.0*4.0); - // Limiters, these need to be high precision RCPs. - FfxFloat16 hitMinR=mn4R*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16(4.0)*mx4R); - FfxFloat16 hitMinG=mn4G*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16(4.0)*mx4G); - FfxFloat16 hitMinB=mn4B*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16(4.0)*mx4B); - FfxFloat16 hitMaxR=(peakC.x-mx4R)*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16(4.0)*mn4R+peakC.y); - FfxFloat16 hitMaxG=(peakC.x-mx4G)*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16(4.0)*mn4G+peakC.y); - FfxFloat16 hitMaxB=(peakC.x-mx4B)*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16(4.0)*mn4B+peakC.y); - FfxFloat16 lobeR=max(-hitMinR,hitMaxR); - FfxFloat16 lobeG=max(-hitMinG,hitMaxG); - FfxFloat16 lobeB=max(-hitMinB,hitMaxB); - FfxFloat16 lobe=max(FFX_BROADCAST_FLOAT16(-FSR_RCAS_LIMIT),min(ffxMax3Half(lobeR,lobeG,lobeB),FFX_BROADCAST_FLOAT16(0.0)))*FFX_UINT32_TO_FLOAT16X2(con.y).x; - // Apply noise removal. - #ifdef FSR_RCAS_DENOISE - lobe*=nz; - #endif - // Resolve, which needs the medium precision rcp approximation to avoid visible tonality changes. - FfxFloat16 rcpL=ffxApproximateReciprocalMediumHalf(FFX_BROADCAST_FLOAT16(4.0)*lobe+FFX_BROADCAST_FLOAT16(1.0)); - pixR=(lobe*bR+lobe*dR+lobe*hR+lobe*fR+eR)*rcpL; - pixG=(lobe*bG+lobe*dG+lobe*hG+lobe*fG+eG)*rcpL; - pixB=(lobe*bB+lobe*dB+lobe*hB+lobe*fB+eB)*rcpL; -} -#endif -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//_____________________________________________________________/\_______________________________________________________________ -//============================================================================================================================== -// PACKED 16-BIT VERSION -//============================================================================================================================== -#if defined(FFX_GPU)&& FFX_HALF == 1 && defined(FSR_RCAS_HX2) - // Input callback prototypes that need to be implemented by the calling shader - FfxFloat16x4 FsrRcasLoadHx2(FfxInt16x2 p); - void FsrRcasInputHx2(inout FfxFloat16x2 r,inout FfxFloat16x2 g,inout FfxFloat16x2 b); -//------------------------------------------------------------------------------------------------------------------------------ - // Can be used to convert from packed Structures of Arrays to Arrays of Structures for store. - void FsrRcasDepackHx2(out FfxFloat16x4 pix0,out FfxFloat16x4 pix1,FfxFloat16x2 pixR,FfxFloat16x2 pixG,FfxFloat16x2 pixB){ - #ifdef FFX_HLSL - // Invoke a slower path for DX only, since it won't allow uninitialized values. - pix0.a=pix1.a=0.0; - #endif - pix0.rgb=FfxFloat16x3(pixR.x,pixG.x,pixB.x); - pix1.rgb=FfxFloat16x3(pixR.y,pixG.y,pixB.y);} -//------------------------------------------------------------------------------------------------------------------------------ - void FsrRcasHx2( - // Output values are for 2 8x8 tiles in a 16x8 region. - // pix.x = left 8x8 tile - // pix.y = right 8x8 tile - // This enables later processing to easily be packed as well. - out FfxFloat16x2 pixR, - out FfxFloat16x2 pixG, - out FfxFloat16x2 pixB, - #ifdef FSR_RCAS_PASSTHROUGH_ALPHA - out FfxFloat16x2 pixA, - #endif - FfxUInt32x2 ip, // Integer pixel position in output. - FfxUInt32x4 con){ // Constant generated by RcasSetup(). - // No scaling algorithm uses minimal 3x3 pixel neighborhood. - FfxInt16x2 sp0=FfxInt16x2(ip); - FfxFloat16x3 b0=FsrRcasLoadHx2(sp0+FfxInt16x2( 0,-1)).rgb; - FfxFloat16x3 d0=FsrRcasLoadHx2(sp0+FfxInt16x2(-1, 0)).rgb; - #ifdef FSR_RCAS_PASSTHROUGH_ALPHA - FfxFloat16x4 ee0=FsrRcasLoadHx2(sp0); - FfxFloat16x3 e0=ee0.rgb;pixA.r=ee0.a; - #else - FfxFloat16x3 e0=FsrRcasLoadHx2(sp0).rgb; - #endif - FfxFloat16x3 f0=FsrRcasLoadHx2(sp0+FfxInt16x2( 1, 0)).rgb; - FfxFloat16x3 h0=FsrRcasLoadHx2(sp0+FfxInt16x2( 0, 1)).rgb; - FfxInt16x2 sp1=sp0+FfxInt16x2(8,0); - FfxFloat16x3 b1=FsrRcasLoadHx2(sp1+FfxInt16x2( 0,-1)).rgb; - FfxFloat16x3 d1=FsrRcasLoadHx2(sp1+FfxInt16x2(-1, 0)).rgb; - #ifdef FSR_RCAS_PASSTHROUGH_ALPHA - FfxFloat16x4 ee1=FsrRcasLoadHx2(sp1); - FfxFloat16x3 e1=ee1.rgb;pixA.g=ee1.a; - #else - FfxFloat16x3 e1=FsrRcasLoadHx2(sp1).rgb; - #endif - FfxFloat16x3 f1=FsrRcasLoadHx2(sp1+FfxInt16x2( 1, 0)).rgb; - FfxFloat16x3 h1=FsrRcasLoadHx2(sp1+FfxInt16x2( 0, 1)).rgb; - // Arrays of Structures to Structures of Arrays conversion. - FfxFloat16x2 bR=FfxFloat16x2(b0.r,b1.r); - FfxFloat16x2 bG=FfxFloat16x2(b0.g,b1.g); - FfxFloat16x2 bB=FfxFloat16x2(b0.b,b1.b); - FfxFloat16x2 dR=FfxFloat16x2(d0.r,d1.r); - FfxFloat16x2 dG=FfxFloat16x2(d0.g,d1.g); - FfxFloat16x2 dB=FfxFloat16x2(d0.b,d1.b); - FfxFloat16x2 eR=FfxFloat16x2(e0.r,e1.r); - FfxFloat16x2 eG=FfxFloat16x2(e0.g,e1.g); - FfxFloat16x2 eB=FfxFloat16x2(e0.b,e1.b); - FfxFloat16x2 fR=FfxFloat16x2(f0.r,f1.r); - FfxFloat16x2 fG=FfxFloat16x2(f0.g,f1.g); - FfxFloat16x2 fB=FfxFloat16x2(f0.b,f1.b); - FfxFloat16x2 hR=FfxFloat16x2(h0.r,h1.r); - FfxFloat16x2 hG=FfxFloat16x2(h0.g,h1.g); - FfxFloat16x2 hB=FfxFloat16x2(h0.b,h1.b); - // Run optional input transform. - FsrRcasInputHx2(bR,bG,bB); - FsrRcasInputHx2(dR,dG,dB); - FsrRcasInputHx2(eR,eG,eB); - FsrRcasInputHx2(fR,fG,fB); - FsrRcasInputHx2(hR,hG,hB); - // Luma times 2. - FfxFloat16x2 bL=bB*FFX_BROADCAST_FLOAT16X2(0.5)+(bR*FFX_BROADCAST_FLOAT16X2(0.5)+bG); - FfxFloat16x2 dL=dB*FFX_BROADCAST_FLOAT16X2(0.5)+(dR*FFX_BROADCAST_FLOAT16X2(0.5)+dG); - FfxFloat16x2 eL=eB*FFX_BROADCAST_FLOAT16X2(0.5)+(eR*FFX_BROADCAST_FLOAT16X2(0.5)+eG); - FfxFloat16x2 fL=fB*FFX_BROADCAST_FLOAT16X2(0.5)+(fR*FFX_BROADCAST_FLOAT16X2(0.5)+fG); - FfxFloat16x2 hL=hB*FFX_BROADCAST_FLOAT16X2(0.5)+(hR*FFX_BROADCAST_FLOAT16X2(0.5)+hG); - // Noise detection. - FfxFloat16x2 nz=FFX_BROADCAST_FLOAT16X2(0.25)*bL+FFX_BROADCAST_FLOAT16X2(0.25)*dL+FFX_BROADCAST_FLOAT16X2(0.25)*fL+FFX_BROADCAST_FLOAT16X2(0.25)*hL-eL; - nz=ffxSaturate(abs(nz)*ffxApproximateReciprocalMediumHalf(ffxMax3Half(ffxMax3Half(bL,dL,eL),fL,hL)-ffxMin3Half(ffxMin3Half(bL,dL,eL),fL,hL))); - nz=FFX_BROADCAST_FLOAT16X2(-0.5)*nz+FFX_BROADCAST_FLOAT16X2(1.0); - // Min and max of ring. - FfxFloat16x2 mn4R=min(ffxMin3Half(bR,dR,fR),hR); - FfxFloat16x2 mn4G=min(ffxMin3Half(bG,dG,fG),hG); - FfxFloat16x2 mn4B=min(ffxMin3Half(bB,dB,fB),hB); - FfxFloat16x2 mx4R=max(ffxMax3Half(bR,dR,fR),hR); - FfxFloat16x2 mx4G=max(ffxMax3Half(bG,dG,fG),hG); - FfxFloat16x2 mx4B=max(ffxMax3Half(bB,dB,fB),hB); - // Immediate constants for peak range. - FfxFloat16x2 peakC=FfxFloat16x2(1.0,-1.0*4.0); - // Limiters, these need to be high precision RCPs. - FfxFloat16x2 hitMinR=mn4R*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16X2(4.0)*mx4R); - FfxFloat16x2 hitMinG=mn4G*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16X2(4.0)*mx4G); - FfxFloat16x2 hitMinB=mn4B*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16X2(4.0)*mx4B); - FfxFloat16x2 hitMaxR=(peakC.x-mx4R)*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16X2(4.0)*mn4R+peakC.y); - FfxFloat16x2 hitMaxG=(peakC.x-mx4G)*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16X2(4.0)*mn4G+peakC.y); - FfxFloat16x2 hitMaxB=(peakC.x-mx4B)*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16X2(4.0)*mn4B+peakC.y); - FfxFloat16x2 lobeR=max(-hitMinR,hitMaxR); - FfxFloat16x2 lobeG=max(-hitMinG,hitMaxG); - FfxFloat16x2 lobeB=max(-hitMinB,hitMaxB); - FfxFloat16x2 lobe=max(FFX_BROADCAST_FLOAT16X2(-FSR_RCAS_LIMIT),min(ffxMax3Half(lobeR,lobeG,lobeB),FFX_BROADCAST_FLOAT16X2(0.0)))*FFX_BROADCAST_FLOAT16X2(FFX_UINT32_TO_FLOAT16X2(con.y).x); - // Apply noise removal. - #ifdef FSR_RCAS_DENOISE - lobe*=nz; - #endif - // Resolve, which needs the medium precision rcp approximation to avoid visible tonality changes. - FfxFloat16x2 rcpL=ffxApproximateReciprocalMediumHalf(FFX_BROADCAST_FLOAT16X2(4.0)*lobe+FFX_BROADCAST_FLOAT16X2(1.0)); - pixR=(lobe*bR+lobe*dR+lobe*hR+lobe*fR+eR)*rcpL; - pixG=(lobe*bG+lobe*dG+lobe*hG+lobe*fG+eG)*rcpL; - pixB=(lobe*bB+lobe*dB+lobe*hB+lobe*fB+eB)*rcpL;} -#endif -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//_____________________________________________________________/\_______________________________________________________________ -//============================================================================================================================== -// -// FSR - [LFGA] LINEAR FILM GRAIN APPLICATOR -// -//------------------------------------------------------------------------------------------------------------------------------ -// Adding output-resolution film grain after scaling is a good way to mask both rendering and scaling artifacts. -// Suggest using tiled blue noise as film grain input, with peak noise frequency set for a specific look and feel. -// The 'Lfga*()' functions provide a convenient way to introduce grain. -// These functions limit grain based on distance to signal limits. -// This is done so that the grain is temporally energy preserving, and thus won't modify image tonality. -// Grain application should be done in a linear colorspace. -// The grain should be temporally changing, but have a temporal sum per pixel that adds to zero (non-biased). -//------------------------------------------------------------------------------------------------------------------------------ -// Usage, -// FsrLfga*( -// color, // In/out linear colorspace color {0 to 1} ranged. -// grain, // Per pixel grain texture value {-0.5 to 0.5} ranged, input is 3-channel to support colored grain. -// amount); // Amount of grain (0 to 1} ranged. -//------------------------------------------------------------------------------------------------------------------------------ -// Example if grain texture is monochrome: 'FsrLfgaF(color,ffxBroadcast3(grain),amount)' -//============================================================================================================================== -#if defined(FFX_GPU) - // Maximum grain is the minimum distance to the signal limit. - void FsrLfgaF(inout FfxFloat32x3 c, FfxFloat32x3 t, FfxFloat32 a) - { - c += (t * ffxBroadcast3(a)) * ffxMin(ffxBroadcast3(1.0) - c, c); - } -#endif -//============================================================================================================================== -#if defined(FFX_GPU)&& FFX_HALF == 1 - // Half precision version (slower). - void FsrLfgaH(inout FfxFloat16x3 c, FfxFloat16x3 t, FfxFloat16 a) - { - c += (t * FFX_BROADCAST_FLOAT16X3(a)) * min(FFX_BROADCAST_FLOAT16X3(1.0) - c, c); - } - //------------------------------------------------------------------------------------------------------------------------------ - // Packed half precision version (faster). - void FsrLfgaHx2(inout FfxFloat16x2 cR,inout FfxFloat16x2 cG,inout FfxFloat16x2 cB,FfxFloat16x2 tR,FfxFloat16x2 tG,FfxFloat16x2 tB,FfxFloat16 a){ - cR+=(tR*FFX_BROADCAST_FLOAT16X2(a))*min(FFX_BROADCAST_FLOAT16X2(1.0)-cR,cR);cG+=(tG*FFX_BROADCAST_FLOAT16X2(a))*min(FFX_BROADCAST_FLOAT16X2(1.0)-cG,cG);cB+=(tB*FFX_BROADCAST_FLOAT16X2(a))*min(FFX_BROADCAST_FLOAT16X2(1.0)-cB,cB);} -#endif -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//_____________________________________________________________/\_______________________________________________________________ -//============================================================================================================================== -// -// FSR - [SRTM] SIMPLE REVERSIBLE TONE-MAPPER -// -//------------------------------------------------------------------------------------------------------------------------------ -// This provides a way to take linear HDR color {0 to FP16_MAX} and convert it into a temporary {0 to 1} ranged post-tonemapped linear. -// The tonemapper preserves RGB ratio, which helps maintain HDR color bleed during filtering. -//------------------------------------------------------------------------------------------------------------------------------ -// Reversible tonemapper usage, -// FsrSrtm*(color); // {0 to FP16_MAX} converted to {0 to 1}. -// FsrSrtmInv*(color); // {0 to 1} converted into {0 to 32768, output peak safe for FP16}. -//============================================================================================================================== -#if defined(FFX_GPU) - void FsrSrtmF(inout FfxFloat32x3 c) - { - c *= ffxBroadcast3(rcp(ffxMax3(c.r, c.g, c.b) + FfxFloat32(1.0))); - } - // The extra max solves the c=1.0 case (which is a /0). - void FsrSrtmInvF(inout FfxFloat32x3 c){c*=ffxBroadcast3(rcp(max(FfxFloat32(1.0/32768.0),FfxFloat32(1.0)-ffxMax3(c.r,c.g,c.b))));} -#endif -//============================================================================================================================== -#if defined(FFX_GPU )&& FFX_HALF == 1 - void FsrSrtmH(inout FfxFloat16x3 c) - { - c *= FFX_BROADCAST_FLOAT16X3(ffxReciprocalHalf(ffxMax3Half(c.r, c.g, c.b) + FFX_BROADCAST_FLOAT16(1.0))); - } - void FsrSrtmInvH(inout FfxFloat16x3 c) - { - c *= FFX_BROADCAST_FLOAT16X3(ffxReciprocalHalf(max(FFX_BROADCAST_FLOAT16(1.0 / 32768.0), FFX_BROADCAST_FLOAT16(1.0) - ffxMax3Half(c.r, c.g, c.b)))); - } - //------------------------------------------------------------------------------------------------------------------------------ - void FsrSrtmHx2(inout FfxFloat16x2 cR, inout FfxFloat16x2 cG, inout FfxFloat16x2 cB) - { - FfxFloat16x2 rcp = ffxReciprocalHalf(ffxMax3Half(cR, cG, cB) + FFX_BROADCAST_FLOAT16X2(1.0)); - cR *= rcp; - cG *= rcp; - cB *= rcp; - } - void FsrSrtmInvHx2(inout FfxFloat16x2 cR,inout FfxFloat16x2 cG,inout FfxFloat16x2 cB) - { - FfxFloat16x2 rcp=ffxReciprocalHalf(max(FFX_BROADCAST_FLOAT16X2(1.0/32768.0),FFX_BROADCAST_FLOAT16X2(1.0)-ffxMax3Half(cR,cG,cB))); - cR*=rcp; - cG*=rcp; - cB*=rcp; - } -#endif -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//_____________________________________________________________/\_______________________________________________________________ -//============================================================================================================================== -// -// FSR - [TEPD] TEMPORAL ENERGY PRESERVING DITHER -// -//------------------------------------------------------------------------------------------------------------------------------ -// Temporally energy preserving dithered {0 to 1} linear to gamma 2.0 conversion. -// Gamma 2.0 is used so that the conversion back to linear is just to square the color. -// The conversion comes in 8-bit and 10-bit modes, designed for output to 8-bit UNORM or 10:10:10:2 respectively. -// Given good non-biased temporal blue noise as dither input, -// the output dither will temporally conserve energy. -// This is done by choosing the linear nearest step point instead of perceptual nearest. -// See code below for details. -//------------------------------------------------------------------------------------------------------------------------------ -// DX SPEC RULES FOR FLOAT->UNORM 8-BIT CONVERSION -// =============================================== -// - Output is 'FfxUInt32(floor(saturate(n)*255.0+0.5))'. -// - Thus rounding is to nearest. -// - NaN gets converted to zero. -// - INF is clamped to {0.0 to 1.0}. -//============================================================================================================================== -#if defined(FFX_GPU) - // Hand tuned integer position to dither value, with more values than simple checkerboard. - // Only 32-bit has enough precision for this compddation. - // Output is {0 to <1}. - FfxFloat32 FsrTepdDitF(FfxUInt32x2 p, FfxUInt32 f) - { - FfxFloat32 x = FfxFloat32(p.x + f); - FfxFloat32 y = FfxFloat32(p.y); - // The 1.61803 golden ratio. - FfxFloat32 a = FfxFloat32((1.0 + ffxSqrt(5.0f)) / 2.0); - // Number designed to provide a good visual pattern. - FfxFloat32 b = FfxFloat32(1.0 / 3.69); - x = x * a + (y * b); - return ffxFract(x); - } - //------------------------------------------------------------------------------------------------------------------------------ - // This version is 8-bit gamma 2.0. - // The 'c' input is {0 to 1}. - // Output is {0 to 1} ready for image store. - void FsrTepdC8F(inout FfxFloat32x3 c, FfxFloat32 dit) - { - FfxFloat32x3 n = ffxSqrt(c); - n = floor(n * ffxBroadcast3(255.0)) * ffxBroadcast3(1.0 / 255.0); - FfxFloat32x3 a = n * n; - FfxFloat32x3 b = n + ffxBroadcast3(1.0 / 255.0); - b = b * b; - // Ratio of 'a' to 'b' required to produce 'c'. - // ffxApproximateReciprocal() won't work here (at least for very high dynamic ranges). - // ffxApproximateReciprocalMedium() is an IADD,FMA,MUL. - FfxFloat32x3 r = (c - b) * ffxApproximateReciprocalMedium(a - b); - // Use the ratio as a cutoff to choose 'a' or 'b'. - // ffxIsGreaterThanZero() is a MUL. - c = ffxSaturate(n + ffxIsGreaterThanZero(ffxBroadcast3(dit) - r) * ffxBroadcast3(1.0 / 255.0)); - } - //------------------------------------------------------------------------------------------------------------------------------ - // This version is 10-bit gamma 2.0. - // The 'c' input is {0 to 1}. - // Output is {0 to 1} ready for image store. - void FsrTepdC10F(inout FfxFloat32x3 c, FfxFloat32 dit) - { - FfxFloat32x3 n = ffxSqrt(c); - n = floor(n * ffxBroadcast3(1023.0)) * ffxBroadcast3(1.0 / 1023.0); - FfxFloat32x3 a = n * n; - FfxFloat32x3 b = n + ffxBroadcast3(1.0 / 1023.0); - b = b * b; - FfxFloat32x3 r = (c - b) * ffxApproximateReciprocalMedium(a - b); - c = ffxSaturate(n + ffxIsGreaterThanZero(ffxBroadcast3(dit) - r) * ffxBroadcast3(1.0 / 1023.0)); - } -#endif -//============================================================================================================================== -#if defined(FFX_GPU)&& FFX_HALF == 1 - FfxFloat16 FsrTepdDitH(FfxUInt32x2 p, FfxUInt32 f) - { - FfxFloat32 x = FfxFloat32(p.x + f); - FfxFloat32 y = FfxFloat32(p.y); - FfxFloat32 a = FfxFloat32((1.0 + ffxSqrt(5.0f)) / 2.0); - FfxFloat32 b = FfxFloat32(1.0 / 3.69); - x = x * a + (y * b); - return FfxFloat16(ffxFract(x)); - } - //------------------------------------------------------------------------------------------------------------------------------ - void FsrTepdC8H(inout FfxFloat16x3 c, FfxFloat16 dit) - { - FfxFloat16x3 n = sqrt(c); - n = floor(n * FFX_BROADCAST_FLOAT16X3(255.0)) * FFX_BROADCAST_FLOAT16X3(1.0 / 255.0); - FfxFloat16x3 a = n * n; - FfxFloat16x3 b = n + FFX_BROADCAST_FLOAT16X3(1.0 / 255.0); - b = b * b; - FfxFloat16x3 r = (c - b) * ffxApproximateReciprocalMediumHalf(a - b); - c = FfxFloat16x3(ffxSaturate(n + ffxIsGreaterThanZeroHalf(FFX_BROADCAST_FLOAT16X3(dit) - r) * FFX_BROADCAST_FLOAT16X3(1.0 / 255.0))); - } - //------------------------------------------------------------------------------------------------------------------------------ - void FsrTepdC10H(inout FfxFloat16x3 c, FfxFloat16 dit) - { - FfxFloat16x3 n = sqrt(c); - n = floor(n * FFX_BROADCAST_FLOAT16X3(1023.0)) * FFX_BROADCAST_FLOAT16X3(1.0 / 1023.0); - FfxFloat16x3 a = n * n; - FfxFloat16x3 b = n + FFX_BROADCAST_FLOAT16X3(1.0 / 1023.0); - b = b * b; - FfxFloat16x3 r = (c - b) * ffxApproximateReciprocalMediumHalf(a - b); - c = FfxFloat16x3(ffxSaturate(n + ffxIsGreaterThanZeroHalf(FFX_BROADCAST_FLOAT16X3(dit) - r) * FFX_BROADCAST_FLOAT16X3(1.0 / 1023.0))); - } - //============================================================================================================================== - // This computes dither for positions 'p' and 'p+{8,0}'. - FfxFloat16x2 FsrTepdDitHx2(FfxUInt32x2 p, FfxUInt32 f) - { - FfxFloat32x2 x; - x.x = FfxFloat32(p.x + f); - x.y = x.x + FfxFloat32(8.0); - FfxFloat32 y = FfxFloat32(p.y); - FfxFloat32 a = FfxFloat32((1.0 + ffxSqrt(5.0f)) / 2.0); - FfxFloat32 b = FfxFloat32(1.0 / 3.69); - x = x * ffxBroadcast2(a) + ffxBroadcast2(y * b); - return FfxFloat16x2(ffxFract(x)); - } - //------------------------------------------------------------------------------------------------------------------------------ - void FsrTepdC8Hx2(inout FfxFloat16x2 cR, inout FfxFloat16x2 cG, inout FfxFloat16x2 cB, FfxFloat16x2 dit) - { - FfxFloat16x2 nR = sqrt(cR); - FfxFloat16x2 nG = sqrt(cG); - FfxFloat16x2 nB = sqrt(cB); - nR = floor(nR * FFX_BROADCAST_FLOAT16X2(255.0)) * FFX_BROADCAST_FLOAT16X2(1.0 / 255.0); - nG = floor(nG * FFX_BROADCAST_FLOAT16X2(255.0)) * FFX_BROADCAST_FLOAT16X2(1.0 / 255.0); - nB = floor(nB * FFX_BROADCAST_FLOAT16X2(255.0)) * FFX_BROADCAST_FLOAT16X2(1.0 / 255.0); - FfxFloat16x2 aR = nR * nR; - FfxFloat16x2 aG = nG * nG; - FfxFloat16x2 aB = nB * nB; - FfxFloat16x2 bR = nR + FFX_BROADCAST_FLOAT16X2(1.0 / 255.0); - bR = bR * bR; - FfxFloat16x2 bG = nG + FFX_BROADCAST_FLOAT16X2(1.0 / 255.0); - bG = bG * bG; - FfxFloat16x2 bB = nB + FFX_BROADCAST_FLOAT16X2(1.0 / 255.0); - bB = bB * bB; - FfxFloat16x2 rR = (cR - bR) * ffxApproximateReciprocalMediumHalf(aR - bR); - FfxFloat16x2 rG = (cG - bG) * ffxApproximateReciprocalMediumHalf(aG - bG); - FfxFloat16x2 rB = (cB - bB) * ffxApproximateReciprocalMediumHalf(aB - bB); - cR = FfxFloat16x2(ffxSaturate(nR + ffxIsGreaterThanZeroHalf(dit - rR) * FFX_BROADCAST_FLOAT16X2(1.0 / 255.0))); - cG = FfxFloat16x2(ffxSaturate(nG + ffxIsGreaterThanZeroHalf(dit - rG) * FFX_BROADCAST_FLOAT16X2(1.0 / 255.0))); - cB = FfxFloat16x2(ffxSaturate(nB + ffxIsGreaterThanZeroHalf(dit - rB) * FFX_BROADCAST_FLOAT16X2(1.0 / 255.0))); - } - //------------------------------------------------------------------------------------------------------------------------------ - void FsrTepdC10Hx2(inout FfxFloat16x2 cR,inout FfxFloat16x2 cG,inout FfxFloat16x2 cB,FfxFloat16x2 dit){ - FfxFloat16x2 nR=sqrt(cR); - FfxFloat16x2 nG=sqrt(cG); - FfxFloat16x2 nB=sqrt(cB); - nR=floor(nR*FFX_BROADCAST_FLOAT16X2(1023.0))*FFX_BROADCAST_FLOAT16X2(1.0/1023.0); - nG=floor(nG*FFX_BROADCAST_FLOAT16X2(1023.0))*FFX_BROADCAST_FLOAT16X2(1.0/1023.0); - nB=floor(nB*FFX_BROADCAST_FLOAT16X2(1023.0))*FFX_BROADCAST_FLOAT16X2(1.0/1023.0); - FfxFloat16x2 aR=nR*nR; - FfxFloat16x2 aG=nG*nG; - FfxFloat16x2 aB=nB*nB; - FfxFloat16x2 bR=nR+FFX_BROADCAST_FLOAT16X2(1.0/1023.0);bR=bR*bR; - FfxFloat16x2 bG=nG+FFX_BROADCAST_FLOAT16X2(1.0/1023.0);bG=bG*bG; - FfxFloat16x2 bB=nB+FFX_BROADCAST_FLOAT16X2(1.0/1023.0);bB=bB*bB; - FfxFloat16x2 rR=(cR-bR)*ffxApproximateReciprocalMediumHalf(aR-bR); - FfxFloat16x2 rG=(cG-bG)*ffxApproximateReciprocalMediumHalf(aG-bG); - FfxFloat16x2 rB=(cB-bB)*ffxApproximateReciprocalMediumHalf(aB-bB); - cR=FfxFloat16x2(ffxSaturate(nR+ffxIsGreaterThanZeroHalf(dit-rR)*FFX_BROADCAST_FLOAT16X2(1.0/1023.0))); - cG=FfxFloat16x2(ffxSaturate(nG+ffxIsGreaterThanZeroHalf(dit-rG)*FFX_BROADCAST_FLOAT16X2(1.0/1023.0))); - cB=FfxFloat16x2(ffxSaturate(nB + ffxIsGreaterThanZeroHalf(dit - rB) * FFX_BROADCAST_FLOAT16X2(1.0 / 1023.0))); -} -#endif diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/fsr1/ffx_fsr1.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/fsr1/ffx_fsr1.h.meta deleted file mode 100644 index 3c97f69..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/fsr1/ffx_fsr1.h.meta +++ /dev/null @@ -1,67 +0,0 @@ -fileFormatVersion: 2 -guid: 628e23510f46ef44bbf0035ce9a63be0 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - CPU: AnyCPU - DefaultValueInitialized: true - OS: AnyOS - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/spd.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/spd.meta deleted file mode 100644 index 0b775af..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/spd.meta +++ /dev/null @@ -1,8 +0,0 @@ -fileFormatVersion: 2 -guid: 0f03de1579ac3294595ae4f40106b7a2 -folderAsset: yes -DefaultImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/spd/ffx_spd.h b/Assets/Shaders/FSR3/shaders/fsr3upscaler/spd/ffx_spd.h deleted file mode 100644 index 6441419..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/spd/ffx_spd.h +++ /dev/null @@ -1,1009 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -/// @defgroup FfxGPUSpd FidelityFX SPD -/// FidelityFX Single Pass Downsampler 2.0 GPU documentation -/// -/// @ingroup FfxGPUEffects - -/// Setup required constant values for SPD (CPU). -/// -/// @param [out] dispatchThreadGroupCountXY CPU side: dispatch thread group count xy. z is number of slices of the input texture -/// @param [out] workGroupOffset GPU side: pass in as constant -/// @param [out] numWorkGroupsAndMips GPU side: pass in as constant -/// @param [in] rectInfo left, top, width, height -/// @param [in] mips optional: if -1, calculate based on rect width and height -/// -/// @ingroup FfxGPUSpd -#if defined(FFX_CPU) -FFX_STATIC void ffxSpdSetup(FfxUInt32x2 dispatchThreadGroupCountXY, - FfxUInt32x2 workGroupOffset, - FfxUInt32x2 numWorkGroupsAndMips, - FfxUInt32x4 rectInfo, - FfxInt32 mips) -{ - // determines the offset of the first tile to downsample based on - // left (rectInfo[0]) and top (rectInfo[1]) of the subregion. - workGroupOffset[0] = rectInfo[0] / 64; - workGroupOffset[1] = rectInfo[1] / 64; - - FfxUInt32 endIndexX = (rectInfo[0] + rectInfo[2] - 1) / 64; // rectInfo[0] = left, rectInfo[2] = width - FfxUInt32 endIndexY = (rectInfo[1] + rectInfo[3] - 1) / 64; // rectInfo[1] = top, rectInfo[3] = height - - // we only need to dispatch as many thread groups as tiles we need to downsample - // number of tiles per slice depends on the subregion to downsample - dispatchThreadGroupCountXY[0] = endIndexX + 1 - workGroupOffset[0]; - dispatchThreadGroupCountXY[1] = endIndexY + 1 - workGroupOffset[1]; - - // number of thread groups per slice - numWorkGroupsAndMips[0] = (dispatchThreadGroupCountXY[0]) * (dispatchThreadGroupCountXY[1]); - - if (mips >= 0) - { - numWorkGroupsAndMips[1] = FfxUInt32(mips); - } - else - { - // calculate based on rect width and height - FfxUInt32 resolution = ffxMax(rectInfo[2], rectInfo[3]); - numWorkGroupsAndMips[1] = FfxUInt32((ffxMin(floor(log2(FfxFloat32(resolution))), FfxFloat32(12)))); - } -} - -/// Setup required constant values for SPD (CPU). -/// -/// @param [out] dispatchThreadGroupCountXY CPU side: dispatch thread group count xy. z is number of slices of the input texture -/// @param [out] workGroupOffset GPU side: pass in as constant -/// @param [out] numWorkGroupsAndMips GPU side: pass in as constant -/// @param [in] rectInfo left, top, width, height -/// -/// @ingroup FfxGPUSpd -FFX_STATIC void ffxSpdSetup(FfxUInt32x2 dispatchThreadGroupCountXY, - FfxUInt32x2 workGroupOffset, - FfxUInt32x2 numWorkGroupsAndMips, - FfxUInt32x4 rectInfo) -{ - ffxSpdSetup(dispatchThreadGroupCountXY, workGroupOffset, numWorkGroupsAndMips, rectInfo, -1); -} -#endif // #if defined(FFX_CPU) - - -//============================================================================================================================== -// NON-PACKED VERSION -//============================================================================================================================== -#if defined(FFX_GPU) -#if defined(FFX_SPD_PACKED_ONLY) -// Avoid compiler errors by including default implementations of these callbacks. -FfxFloat32x4 SpdLoadSourceImage(FfxInt32x2 p, FfxUInt32 slice) -{ - return FfxFloat32x4(0.0, 0.0, 0.0, 0.0); -} - -FfxFloat32x4 SpdLoad(FfxInt32x2 p, FfxUInt32 slice) -{ - return FfxFloat32x4(0.0, 0.0, 0.0, 0.0); -} -void SpdStore(FfxInt32x2 p, FfxFloat32x4 value, FfxUInt32 mip, FfxUInt32 slice) -{ -} -FfxFloat32x4 SpdLoadIntermediate(FfxUInt32 x, FfxUInt32 y) -{ - return FfxFloat32x4(0.0, 0.0, 0.0, 0.0); -} -void SpdStoreIntermediate(FfxUInt32 x, FfxUInt32 y, FfxFloat32x4 value) -{ -} -FfxFloat32x4 SpdReduce4(FfxFloat32x4 v0, FfxFloat32x4 v1, FfxFloat32x4 v2, FfxFloat32x4 v3) -{ - return FfxFloat32x4(0.0, 0.0, 0.0, 0.0); -} -#endif // #if FFX_SPD_PACKED_ONLY - -//_____________________________________________________________/\_______________________________________________________________ - -void ffxSpdWorkgroupShuffleBarrier() -{ - FFX_GROUP_MEMORY_BARRIER(); -} - -// Only last active workgroup should proceed -bool SpdExitWorkgroup(FfxUInt32 numWorkGroups, FfxUInt32 localInvocationIndex, FfxUInt32 slice) -{ - // global atomic counter - if (localInvocationIndex == 0) - { - SpdIncreaseAtomicCounter(slice); - } - - ffxSpdWorkgroupShuffleBarrier(); - return (SpdGetAtomicCounter() != (numWorkGroups - 1)); -} - -// User defined: FfxFloat32x4 SpdReduce4(FfxFloat32x4 v0, FfxFloat32x4 v1, FfxFloat32x4 v2, FfxFloat32x4 v3); -FfxFloat32x4 SpdReduceQuad(FfxFloat32x4 v) -{ -#if defined(FFX_GLSL) && !defined(FFX_SPD_NO_WAVE_OPERATIONS) - - FfxFloat32x4 v0 = v; - FfxFloat32x4 v1 = subgroupQuadSwapHorizontal(v); - FfxFloat32x4 v2 = subgroupQuadSwapVertical(v); - FfxFloat32x4 v3 = subgroupQuadSwapDiagonal(v); - return SpdReduce4(v0, v1, v2, v3); - -#elif defined(FFX_HLSL) && !defined(FFX_SPD_NO_WAVE_OPERATIONS) - - // requires SM6.0 - FfxUInt32 quad = WaveGetLaneIndex() & (~0x3); - FfxFloat32x4 v0 = v; - FfxFloat32x4 v1 = WaveReadLaneAt(v, quad | 1); - FfxFloat32x4 v2 = WaveReadLaneAt(v, quad | 2); - FfxFloat32x4 v3 = WaveReadLaneAt(v, quad | 3); - return SpdReduce4(v0, v1, v2, v3); -/* - // if SM6.0 is not available, you can use the AMD shader intrinsics - // the AMD shader intrinsics are available in AMD GPU Services (AGS) library: - // https://gpuopen.com/amd-gpu-services-ags-library/ - // works for DX11 - FfxFloat32x4 v0 = v; - FfxFloat32x4 v1; - v1.x = AmdExtD3DShaderIntrinsics_SwizzleF(v.x, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); - v1.y = AmdExtD3DShaderIntrinsics_SwizzleF(v.y, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); - v1.z = AmdExtD3DShaderIntrinsics_SwizzleF(v.z, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); - v1.w = AmdExtD3DShaderIntrinsics_SwizzleF(v.w, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); - FfxFloat32x4 v2; - v2.x = AmdExtD3DShaderIntrinsics_SwizzleF(v.x, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); - v2.y = AmdExtD3DShaderIntrinsics_SwizzleF(v.y, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); - v2.z = AmdExtD3DShaderIntrinsics_SwizzleF(v.z, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); - v2.w = AmdExtD3DShaderIntrinsics_SwizzleF(v.w, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); - FfxFloat32x4 v3; - v3.x = AmdExtD3DShaderIntrinsics_SwizzleF(v.x, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); - v3.y = AmdExtD3DShaderIntrinsics_SwizzleF(v.y, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); - v3.z = AmdExtD3DShaderIntrinsics_SwizzleF(v.z, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); - v3.w = AmdExtD3DShaderIntrinsics_SwizzleF(v.w, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); - return SpdReduce4(v0, v1, v2, v3); - */ -#endif - return v; -} - -FfxFloat32x4 SpdReduceIntermediate(FfxUInt32x2 i0, FfxUInt32x2 i1, FfxUInt32x2 i2, FfxUInt32x2 i3) -{ - FfxFloat32x4 v0 = SpdLoadIntermediate(i0.x, i0.y); - FfxFloat32x4 v1 = SpdLoadIntermediate(i1.x, i1.y); - FfxFloat32x4 v2 = SpdLoadIntermediate(i2.x, i2.y); - FfxFloat32x4 v3 = SpdLoadIntermediate(i3.x, i3.y); - return SpdReduce4(v0, v1, v2, v3); -} - -FfxFloat32x4 SpdReduceLoad4(FfxUInt32x2 i0, FfxUInt32x2 i1, FfxUInt32x2 i2, FfxUInt32x2 i3, FfxUInt32 slice) -{ - FfxFloat32x4 v0 = SpdLoad(FfxInt32x2(i0), slice); - FfxFloat32x4 v1 = SpdLoad(FfxInt32x2(i1), slice); - FfxFloat32x4 v2 = SpdLoad(FfxInt32x2(i2), slice); - FfxFloat32x4 v3 = SpdLoad(FfxInt32x2(i3), slice); - return SpdReduce4(v0, v1, v2, v3); -} - -FfxFloat32x4 SpdReduceLoad4(FfxUInt32x2 base, FfxUInt32 slice) -{ - return SpdReduceLoad4(FfxUInt32x2(base + FfxUInt32x2(0, 0)), FfxUInt32x2(base + FfxUInt32x2(0, 1)), FfxUInt32x2(base + FfxUInt32x2(1, 0)), FfxUInt32x2(base + FfxUInt32x2(1, 1)), slice); -} - -FfxFloat32x4 SpdReduceLoadSourceImage4(FfxUInt32x2 i0, FfxUInt32x2 i1, FfxUInt32x2 i2, FfxUInt32x2 i3, FfxUInt32 slice) -{ - FfxFloat32x4 v0 = SpdLoadSourceImage(FfxInt32x2(i0), slice); - FfxFloat32x4 v1 = SpdLoadSourceImage(FfxInt32x2(i1), slice); - FfxFloat32x4 v2 = SpdLoadSourceImage(FfxInt32x2(i2), slice); - FfxFloat32x4 v3 = SpdLoadSourceImage(FfxInt32x2(i3), slice); - return SpdReduce4(v0, v1, v2, v3); -} - -FfxFloat32x4 SpdReduceLoadSourceImage(FfxUInt32x2 base, FfxUInt32 slice) -{ -#if defined(SPD_LINEAR_SAMPLER) - return SpdLoadSourceImage(FfxInt32x2(base), slice); -#else - return SpdReduceLoadSourceImage4(FfxUInt32x2(base + FfxUInt32x2(0, 0)), FfxUInt32x2(base + FfxUInt32x2(0, 1)), FfxUInt32x2(base + FfxUInt32x2(1, 0)), FfxUInt32x2(base + FfxUInt32x2(1, 1)), slice); -#endif -} - -void SpdDownsampleMips_0_1_Intrinsics(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) -{ - FfxFloat32x4 v[4]; - - FfxInt32x2 tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2); - FfxInt32x2 pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y); - v[0] = SpdReduceLoadSourceImage(tex, slice); - SpdStore(pix, v[0], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y); - v[1] = SpdReduceLoadSourceImage(tex, slice); - SpdStore(pix, v[1], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2 + 32); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y + 16); - v[2] = SpdReduceLoadSourceImage(tex, slice); - SpdStore(pix, v[2], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2 + 32); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y + 16); - v[3] = SpdReduceLoadSourceImage(tex, slice); - SpdStore(pix, v[3], 0, slice); - - if (mip <= 1) - return; - - v[0] = SpdReduceQuad(v[0]); - v[1] = SpdReduceQuad(v[1]); - v[2] = SpdReduceQuad(v[2]); - v[3] = SpdReduceQuad(v[3]); - - if ((localInvocationIndex % 4) == 0) - { - SpdStore(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2, y / 2), v[0], 1, slice); - SpdStoreIntermediate(x / 2, y / 2, v[0]); - - SpdStore(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2 + 8, y / 2), v[1], 1, slice); - SpdStoreIntermediate(x / 2 + 8, y / 2, v[1]); - - SpdStore(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2, y / 2 + 8), v[2], 1, slice); - SpdStoreIntermediate(x / 2, y / 2 + 8, v[2]); - - SpdStore(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2 + 8, y / 2 + 8), v[3], 1, slice); - SpdStoreIntermediate(x / 2 + 8, y / 2 + 8, v[3]); - } -} - -void SpdDownsampleMips_0_1_LDS(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) -{ - FfxFloat32x4 v[4]; - - FfxInt32x2 tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2); - FfxInt32x2 pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y); - v[0] = SpdReduceLoadSourceImage(tex, slice); - SpdStore(pix, v[0], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y); - v[1] = SpdReduceLoadSourceImage(tex, slice); - SpdStore(pix, v[1], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2 + 32); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y + 16); - v[2] = SpdReduceLoadSourceImage(tex, slice); - SpdStore(pix, v[2], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2 + 32); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y + 16); - v[3] = SpdReduceLoadSourceImage(tex, slice); - SpdStore(pix, v[3], 0, slice); - - if (mip <= 1) - return; - - for (FfxUInt32 i = 0; i < 4; i++) - { - SpdStoreIntermediate(x, y, v[i]); - ffxSpdWorkgroupShuffleBarrier(); - if (localInvocationIndex < 64) - { - v[i] = SpdReduceIntermediate(FfxUInt32x2(x * 2 + 0, y * 2 + 0), FfxUInt32x2(x * 2 + 1, y * 2 + 0), FfxUInt32x2(x * 2 + 0, y * 2 + 1), FfxUInt32x2(x * 2 + 1, y * 2 + 1)); - SpdStore(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x + (i % 2) * 8, y + (i / 2) * 8), v[i], 1, slice); - } - ffxSpdWorkgroupShuffleBarrier(); - } - - if (localInvocationIndex < 64) - { - SpdStoreIntermediate(x + 0, y + 0, v[0]); - SpdStoreIntermediate(x + 8, y + 0, v[1]); - SpdStoreIntermediate(x + 0, y + 8, v[2]); - SpdStoreIntermediate(x + 8, y + 8, v[3]); - } -} - -void SpdDownsampleMips_0_1(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) -{ -#if defined(FFX_SPD_NO_WAVE_OPERATIONS) - SpdDownsampleMips_0_1_LDS(x, y, workGroupID, localInvocationIndex, mip, slice); -#else - SpdDownsampleMips_0_1_Intrinsics(x, y, workGroupID, localInvocationIndex, mip, slice); -#endif -} - - -void SpdDownsampleMip_2(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) -{ -#if defined(FFX_SPD_NO_WAVE_OPERATIONS) - if (localInvocationIndex < 64) - { - FfxFloat32x4 v = SpdReduceIntermediate(FfxUInt32x2(x * 2 + 0, y * 2 + 0), FfxUInt32x2(x * 2 + 1, y * 2 + 0), FfxUInt32x2(x * 2 + 0, y * 2 + 1), FfxUInt32x2(x * 2 + 1, y * 2 + 1)); - SpdStore(FfxInt32x2(workGroupID.xy * 8) + FfxInt32x2(x, y), v, mip, slice); - // store to LDS, try to reduce bank conflicts - // x 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 - // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - // 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 x - // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - // x 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 - // ... - // x 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 - SpdStoreIntermediate(x * 2 + y % 2, y * 2, v); - } -#else - FfxFloat32x4 v = SpdLoadIntermediate(x, y); - v = SpdReduceQuad(v); - // quad index 0 stores result - if (localInvocationIndex % 4 == 0) - { - SpdStore(FfxInt32x2(workGroupID.xy * 8) + FfxInt32x2(x / 2, y / 2), v, mip, slice); - SpdStoreIntermediate(x + (y / 2) % 2, y, v); - } -#endif -} - -void SpdDownsampleMip_3(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) -{ -#if defined(FFX_SPD_NO_WAVE_OPERATIONS) - if (localInvocationIndex < 16) - { - // x 0 x 0 - // 0 0 0 0 - // 0 x 0 x - // 0 0 0 0 - FfxFloat32x4 v = - SpdReduceIntermediate(FfxUInt32x2(x * 4 + 0 + 0, y * 4 + 0), FfxUInt32x2(x * 4 + 2 + 0, y * 4 + 0), FfxUInt32x2(x * 4 + 0 + 1, y * 4 + 2), FfxUInt32x2(x * 4 + 2 + 1, y * 4 + 2)); - SpdStore(FfxInt32x2(workGroupID.xy * 4) + FfxInt32x2(x, y), v, mip, slice); - // store to LDS - // x 0 0 0 x 0 0 0 x 0 0 0 x 0 0 0 - // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - // 0 x 0 0 0 x 0 0 0 x 0 0 0 x 0 0 - // ... - // 0 0 x 0 0 0 x 0 0 0 x 0 0 0 x 0 - // ... - // 0 0 0 x 0 0 0 x 0 0 0 x 0 0 0 x - // ... - SpdStoreIntermediate(x * 4 + y, y * 4, v); - } -#else - if (localInvocationIndex < 64) - { - FfxFloat32x4 v = SpdLoadIntermediate(x * 2 + y % 2, y * 2); - v = SpdReduceQuad(v); - // quad index 0 stores result - if (localInvocationIndex % 4 == 0) - { - SpdStore(FfxInt32x2(workGroupID.xy * 4) + FfxInt32x2(x / 2, y / 2), v, mip, slice); - SpdStoreIntermediate(x * 2 + y / 2, y * 2, v); - } - } -#endif -} - -void SpdDownsampleMip_4(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) -{ -#if defined(FFX_SPD_NO_WAVE_OPERATIONS) - if (localInvocationIndex < 4) - { - // x 0 0 0 x 0 0 0 - // ... - // 0 x 0 0 0 x 0 0 - FfxFloat32x4 v = SpdReduceIntermediate(FfxUInt32x2(x * 8 + 0 + 0 + y * 2, y * 8 + 0), - FfxUInt32x2(x * 8 + 4 + 0 + y * 2, y * 8 + 0), - FfxUInt32x2(x * 8 + 0 + 1 + y * 2, y * 8 + 4), - FfxUInt32x2(x * 8 + 4 + 1 + y * 2, y * 8 + 4)); - SpdStore(FfxInt32x2(workGroupID.xy * 2) + FfxInt32x2(x, y), v, mip, slice); - // store to LDS - // x x x x 0 ... - // 0 ... - SpdStoreIntermediate(x + y * 2, 0, v); - } -#else - if (localInvocationIndex < 16) - { - FfxFloat32x4 v = SpdLoadIntermediate(x * 4 + y, y * 4); - v = SpdReduceQuad(v); - // quad index 0 stores result - if (localInvocationIndex % 4 == 0) - { - SpdStore(FfxInt32x2(workGroupID.xy * 2) + FfxInt32x2(x / 2, y / 2), v, mip, slice); - SpdStoreIntermediate(x / 2 + y, 0, v); - } - } -#endif -} - -void SpdDownsampleMip_5(FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) -{ -#if defined(FFX_SPD_NO_WAVE_OPERATIONS) - if (localInvocationIndex < 1) - { - // x x x x 0 ... - // 0 ... - FfxFloat32x4 v = SpdReduceIntermediate(FfxUInt32x2(0, 0), FfxUInt32x2(1, 0), FfxUInt32x2(2, 0), FfxUInt32x2(3, 0)); - SpdStore(FfxInt32x2(workGroupID.xy), v, mip, slice); - } -#else - if (localInvocationIndex < 4) - { - FfxFloat32x4 v = SpdLoadIntermediate(localInvocationIndex, 0); - v = SpdReduceQuad(v); - // quad index 0 stores result - if (localInvocationIndex % 4 == 0) - { - SpdStore(FfxInt32x2(workGroupID.xy), v, mip, slice); - } - } -#endif -} - -void SpdDownsampleMips_6_7(FfxUInt32 x, FfxUInt32 y, FfxUInt32 mips, FfxUInt32 slice) -{ - FfxInt32x2 tex = FfxInt32x2(x * 4 + 0, y * 4 + 0); - FfxInt32x2 pix = FfxInt32x2(x * 2 + 0, y * 2 + 0); - FfxFloat32x4 v0 = SpdReduceLoad4(tex, slice); - SpdStore(pix, v0, 6, slice); - - tex = FfxInt32x2(x * 4 + 2, y * 4 + 0); - pix = FfxInt32x2(x * 2 + 1, y * 2 + 0); - FfxFloat32x4 v1 = SpdReduceLoad4(tex, slice); - SpdStore(pix, v1, 6, slice); - - tex = FfxInt32x2(x * 4 + 0, y * 4 + 2); - pix = FfxInt32x2(x * 2 + 0, y * 2 + 1); - FfxFloat32x4 v2 = SpdReduceLoad4(tex, slice); - SpdStore(pix, v2, 6, slice); - - tex = FfxInt32x2(x * 4 + 2, y * 4 + 2); - pix = FfxInt32x2(x * 2 + 1, y * 2 + 1); - FfxFloat32x4 v3 = SpdReduceLoad4(tex, slice); - SpdStore(pix, v3, 6, slice); - - if (mips <= 7) - return; - // no barrier needed, working on values only from the same thread - - FfxFloat32x4 v = SpdReduce4(v0, v1, v2, v3); - SpdStore(FfxInt32x2(x, y), v, 7, slice); - SpdStoreIntermediate(x, y, v); -} - -void SpdDownsampleNextFour(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 baseMip, FfxUInt32 mips, FfxUInt32 slice) -{ - if (mips <= baseMip) - return; - ffxSpdWorkgroupShuffleBarrier(); - SpdDownsampleMip_2(x, y, workGroupID, localInvocationIndex, baseMip, slice); - - if (mips <= baseMip + 1) - return; - ffxSpdWorkgroupShuffleBarrier(); - SpdDownsampleMip_3(x, y, workGroupID, localInvocationIndex, baseMip + 1, slice); - - if (mips <= baseMip + 2) - return; - ffxSpdWorkgroupShuffleBarrier(); - SpdDownsampleMip_4(x, y, workGroupID, localInvocationIndex, baseMip + 2, slice); - - if (mips <= baseMip + 3) - return; - ffxSpdWorkgroupShuffleBarrier(); - SpdDownsampleMip_5(workGroupID, localInvocationIndex, baseMip + 3, slice); -} - -/// Downsamples a 64x64 tile based on the work group id. -/// If after downsampling it's the last active thread group, computes the remaining MIP levels. -/// -/// @param [in] workGroupID index of the work group / thread group -/// @param [in] localInvocationIndex index of the thread within the thread group in 1D -/// @param [in] mips the number of total MIP levels to compute for the input texture -/// @param [in] numWorkGroups the total number of dispatched work groups / thread groups for this slice -/// @param [in] slice the slice of the input texture -/// -/// @ingroup FfxGPUSpd -void SpdDownsample(FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mips, FfxUInt32 numWorkGroups, FfxUInt32 slice) -{ - // compute MIP level 0 and 1 - FfxUInt32x2 sub_xy = ffxRemapForWaveReduction(localInvocationIndex % 64); - FfxUInt32 x = sub_xy.x + 8 * ((localInvocationIndex >> 6) % 2); - FfxUInt32 y = sub_xy.y + 8 * ((localInvocationIndex >> 7)); - SpdDownsampleMips_0_1(x, y, workGroupID, localInvocationIndex, mips, slice); - - // compute MIP level 2, 3, 4, 5 - SpdDownsampleNextFour(x, y, workGroupID, localInvocationIndex, 2, mips, slice); - - if (mips <= 6) - return; - - // increase the global atomic counter for the given slice and check if it's the last remaining thread group: - // terminate if not, continue if yes. - if (SpdExitWorkgroup(numWorkGroups, localInvocationIndex, slice)) - return; - - // reset the global atomic counter back to 0 for the next spd dispatch - SpdResetAtomicCounter(slice); - - // After mip 5 there is only a single workgroup left that downsamples the remaining up to 64x64 texels. - // compute MIP level 6 and 7 - SpdDownsampleMips_6_7(x, y, mips, slice); - - // compute MIP level 8, 9, 10, 11 - SpdDownsampleNextFour(x, y, FfxUInt32x2(0, 0), localInvocationIndex, 8, mips, slice); -} -/// Downsamples a 64x64 tile based on the work group id and work group offset. -/// If after downsampling it's the last active thread group, computes the remaining MIP levels. -/// -/// @param [in] workGroupID index of the work group / thread group -/// @param [in] localInvocationIndex index of the thread within the thread group in 1D -/// @param [in] mips the number of total MIP levels to compute for the input texture -/// @param [in] numWorkGroups the total number of dispatched work groups / thread groups for this slice -/// @param [in] slice the slice of the input texture -/// @param [in] workGroupOffset the work group offset. it's (0,0) in case the entire input texture is downsampled. -/// -/// @ingroup FfxGPUSpd -void SpdDownsample(FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mips, FfxUInt32 numWorkGroups, FfxUInt32 slice, FfxUInt32x2 workGroupOffset) -{ - SpdDownsample(workGroupID + workGroupOffset, localInvocationIndex, mips, numWorkGroups, slice); -} - -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// - -//============================================================================================================================== -// PACKED VERSION -//============================================================================================================================== - -#if FFX_HALF - -FfxFloat16x4 SpdReduceQuadH(FfxFloat16x4 v) -{ -#if defined(FFX_GLSL) && !defined(FFX_SPD_NO_WAVE_OPERATIONS) - FfxFloat16x4 v0 = v; - FfxFloat16x4 v1 = subgroupQuadSwapHorizontal(v); - FfxFloat16x4 v2 = subgroupQuadSwapVertical(v); - FfxFloat16x4 v3 = subgroupQuadSwapDiagonal(v); - return SpdReduce4H(v0, v1, v2, v3); -#elif defined(FFX_HLSL) && !defined(FFX_SPD_NO_WAVE_OPERATIONS) - // requires SM6.0 - FfxUInt32 quad = WaveGetLaneIndex() & (~0x3); - FfxFloat16x4 v0 = v; - FfxFloat16x4 v1 = WaveReadLaneAt(v, quad | 1); - FfxFloat16x4 v2 = WaveReadLaneAt(v, quad | 2); - FfxFloat16x4 v3 = WaveReadLaneAt(v, quad | 3); - return SpdReduce4H(v0, v1, v2, v3); -/* - // if SM6.0 is not available, you can use the AMD shader intrinsics - // the AMD shader intrinsics are available in AMD GPU Services (AGS) library: - // https://gpuopen.com/amd-gpu-services-ags-library/ - // works for DX11 - FfxFloat16x4 v0 = v; - FfxFloat16x4 v1; - v1.x = AmdExtD3DShaderIntrinsics_SwizzleF(v.x, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); - v1.y = AmdExtD3DShaderIntrinsics_SwizzleF(v.y, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); - v1.z = AmdExtD3DShaderIntrinsics_SwizzleF(v.z, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); - v1.w = AmdExtD3DShaderIntrinsics_SwizzleF(v.w, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); - FfxFloat16x4 v2; - v2.x = AmdExtD3DShaderIntrinsics_SwizzleF(v.x, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); - v2.y = AmdExtD3DShaderIntrinsics_SwizzleF(v.y, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); - v2.z = AmdExtD3DShaderIntrinsics_SwizzleF(v.z, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); - v2.w = AmdExtD3DShaderIntrinsics_SwizzleF(v.w, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); - FfxFloat16x4 v3; - v3.x = AmdExtD3DShaderIntrinsics_SwizzleF(v.x, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); - v3.y = AmdExtD3DShaderIntrinsics_SwizzleF(v.y, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); - v3.z = AmdExtD3DShaderIntrinsics_SwizzleF(v.z, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); - v3.w = AmdExtD3DShaderIntrinsics_SwizzleF(v.w, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); - return SpdReduce4H(v0, v1, v2, v3); - */ -#endif - return FfxFloat16x4(0.0, 0.0, 0.0, 0.0); -} - -FfxFloat16x4 SpdReduceIntermediateH(FfxUInt32x2 i0, FfxUInt32x2 i1, FfxUInt32x2 i2, FfxUInt32x2 i3) -{ - FfxFloat16x4 v0 = SpdLoadIntermediateH(i0.x, i0.y); - FfxFloat16x4 v1 = SpdLoadIntermediateH(i1.x, i1.y); - FfxFloat16x4 v2 = SpdLoadIntermediateH(i2.x, i2.y); - FfxFloat16x4 v3 = SpdLoadIntermediateH(i3.x, i3.y); - return SpdReduce4H(v0, v1, v2, v3); -} - -FfxFloat16x4 SpdReduceLoad4H(FfxUInt32x2 i0, FfxUInt32x2 i1, FfxUInt32x2 i2, FfxUInt32x2 i3, FfxUInt32 slice) -{ - FfxFloat16x4 v0 = SpdLoadH(FfxInt32x2(i0), slice); - FfxFloat16x4 v1 = SpdLoadH(FfxInt32x2(i1), slice); - FfxFloat16x4 v2 = SpdLoadH(FfxInt32x2(i2), slice); - FfxFloat16x4 v3 = SpdLoadH(FfxInt32x2(i3), slice); - return SpdReduce4H(v0, v1, v2, v3); -} - -FfxFloat16x4 SpdReduceLoad4H(FfxUInt32x2 base, FfxUInt32 slice) -{ - return SpdReduceLoad4H(FfxUInt32x2(base + FfxUInt32x2(0, 0)), FfxUInt32x2(base + FfxUInt32x2(0, 1)), FfxUInt32x2(base + FfxUInt32x2(1, 0)), FfxUInt32x2(base + FfxUInt32x2(1, 1)), slice); -} - -FfxFloat16x4 SpdReduceLoadSourceImage4H(FfxUInt32x2 i0, FfxUInt32x2 i1, FfxUInt32x2 i2, FfxUInt32x2 i3, FfxUInt32 slice) -{ - FfxFloat16x4 v0 = SpdLoadSourceImageH(FfxInt32x2(i0), slice); - FfxFloat16x4 v1 = SpdLoadSourceImageH(FfxInt32x2(i1), slice); - FfxFloat16x4 v2 = SpdLoadSourceImageH(FfxInt32x2(i2), slice); - FfxFloat16x4 v3 = SpdLoadSourceImageH(FfxInt32x2(i3), slice); - return SpdReduce4H(v0, v1, v2, v3); -} - -FfxFloat16x4 SpdReduceLoadSourceImageH(FfxUInt32x2 base, FfxUInt32 slice) -{ -#if defined(SPD_LINEAR_SAMPLER) - return SpdLoadSourceImageH(FfxInt32x2(base), slice); -#else - return SpdReduceLoadSourceImage4H(FfxUInt32x2(base + FfxUInt32x2(0, 0)), FfxUInt32x2(base + FfxUInt32x2(0, 1)), FfxUInt32x2(base + FfxUInt32x2(1, 0)), FfxUInt32x2(base + FfxUInt32x2(1, 1)), slice); -#endif -} - -void SpdDownsampleMips_0_1_IntrinsicsH(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mips, FfxUInt32 slice) -{ - FfxFloat16x4 v[4]; - - FfxInt32x2 tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2); - FfxInt32x2 pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y); - v[0] = SpdReduceLoadSourceImageH(tex, slice); - SpdStoreH(pix, v[0], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y); - v[1] = SpdReduceLoadSourceImageH(tex, slice); - SpdStoreH(pix, v[1], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2 + 32); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y + 16); - v[2] = SpdReduceLoadSourceImageH(tex, slice); - SpdStoreH(pix, v[2], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2 + 32); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y + 16); - v[3] = SpdReduceLoadSourceImageH(tex, slice); - SpdStoreH(pix, v[3], 0, slice); - - if (mips <= 1) - return; - - v[0] = SpdReduceQuadH(v[0]); - v[1] = SpdReduceQuadH(v[1]); - v[2] = SpdReduceQuadH(v[2]); - v[3] = SpdReduceQuadH(v[3]); - - if ((localInvocationIndex % 4) == 0) - { - SpdStoreH(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2, y / 2), v[0], 1, slice); - SpdStoreIntermediateH(x / 2, y / 2, v[0]); - - SpdStoreH(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2 + 8, y / 2), v[1], 1, slice); - SpdStoreIntermediateH(x / 2 + 8, y / 2, v[1]); - - SpdStoreH(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2, y / 2 + 8), v[2], 1, slice); - SpdStoreIntermediateH(x / 2, y / 2 + 8, v[2]); - - SpdStoreH(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2 + 8, y / 2 + 8), v[3], 1, slice); - SpdStoreIntermediateH(x / 2 + 8, y / 2 + 8, v[3]); - } -} - -void SpdDownsampleMips_0_1_LDSH(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mips, FfxUInt32 slice) -{ - FfxFloat16x4 v[4]; - - FfxInt32x2 tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2); - FfxInt32x2 pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y); - v[0] = SpdReduceLoadSourceImageH(tex, slice); - SpdStoreH(pix, v[0], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y); - v[1] = SpdReduceLoadSourceImageH(tex, slice); - SpdStoreH(pix, v[1], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2 + 32); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y + 16); - v[2] = SpdReduceLoadSourceImageH(tex, slice); - SpdStoreH(pix, v[2], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2 + 32); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y + 16); - v[3] = SpdReduceLoadSourceImageH(tex, slice); - SpdStoreH(pix, v[3], 0, slice); - - if (mips <= 1) - return; - - for (FfxUInt32 i = 0; i < 4; i++) - { - SpdStoreIntermediateH(x, y, v[i]); - ffxSpdWorkgroupShuffleBarrier(); - if (localInvocationIndex < 64) - { - v[i] = SpdReduceIntermediateH(FfxUInt32x2(x * 2 + 0, y * 2 + 0), FfxUInt32x2(x * 2 + 1, y * 2 + 0), FfxUInt32x2(x * 2 + 0, y * 2 + 1), FfxUInt32x2(x * 2 + 1, y * 2 + 1)); - SpdStoreH(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x + (i % 2) * 8, y + (i / 2) * 8), v[i], 1, slice); - } - ffxSpdWorkgroupShuffleBarrier(); - } - - if (localInvocationIndex < 64) - { - SpdStoreIntermediateH(x + 0, y + 0, v[0]); - SpdStoreIntermediateH(x + 8, y + 0, v[1]); - SpdStoreIntermediateH(x + 0, y + 8, v[2]); - SpdStoreIntermediateH(x + 8, y + 8, v[3]); - } -} - -void SpdDownsampleMips_0_1H(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mips, FfxUInt32 slice) -{ -#if defined(FFX_SPD_NO_WAVE_OPERATIONS) - SpdDownsampleMips_0_1_LDSH(x, y, workGroupID, localInvocationIndex, mips, slice); -#else - SpdDownsampleMips_0_1_IntrinsicsH(x, y, workGroupID, localInvocationIndex, mips, slice); -#endif -} - - -void SpdDownsampleMip_2H(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) -{ -#if defined(FFX_SPD_NO_WAVE_OPERATIONS) - if (localInvocationIndex < 64) - { - FfxFloat16x4 v = SpdReduceIntermediateH(FfxUInt32x2(x * 2 + 0, y * 2 + 0), FfxUInt32x2(x * 2 + 1, y * 2 + 0), FfxUInt32x2(x * 2 + 0, y * 2 + 1), FfxUInt32x2(x * 2 + 1, y * 2 + 1)); - SpdStoreH(FfxInt32x2(workGroupID.xy * 8) + FfxInt32x2(x, y), v, mip, slice); - // store to LDS, try to reduce bank conflicts - // x 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 - // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - // 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 x - // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - // x 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 - // ... - // x 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 - SpdStoreIntermediateH(x * 2 + y % 2, y * 2, v); - } -#else - FfxFloat16x4 v = SpdLoadIntermediateH(x, y); - v = SpdReduceQuadH(v); - // quad index 0 stores result - if (localInvocationIndex % 4 == 0) - { - SpdStoreH(FfxInt32x2(workGroupID.xy * 8) + FfxInt32x2(x / 2, y / 2), v, mip, slice); - SpdStoreIntermediateH(x + (y / 2) % 2, y, v); - } -#endif -} - -void SpdDownsampleMip_3H(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) -{ -#if defined(FFX_SPD_NO_WAVE_OPERATIONS) - if (localInvocationIndex < 16) - { - // x 0 x 0 - // 0 0 0 0 - // 0 x 0 x - // 0 0 0 0 - FfxFloat16x4 v = - SpdReduceIntermediateH(FfxUInt32x2(x * 4 + 0 + 0, y * 4 + 0), FfxUInt32x2(x * 4 + 2 + 0, y * 4 + 0), FfxUInt32x2(x * 4 + 0 + 1, y * 4 + 2), FfxUInt32x2(x * 4 + 2 + 1, y * 4 + 2)); - SpdStoreH(FfxInt32x2(workGroupID.xy * 4) + FfxInt32x2(x, y), v, mip, slice); - // store to LDS - // x 0 0 0 x 0 0 0 x 0 0 0 x 0 0 0 - // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - // 0 x 0 0 0 x 0 0 0 x 0 0 0 x 0 0 - // ... - // 0 0 x 0 0 0 x 0 0 0 x 0 0 0 x 0 - // ... - // 0 0 0 x 0 0 0 x 0 0 0 x 0 0 0 x - // ... - SpdStoreIntermediateH(x * 4 + y, y * 4, v); - } -#else - if (localInvocationIndex < 64) - { - FfxFloat16x4 v = SpdLoadIntermediateH(x * 2 + y % 2, y * 2); - v = SpdReduceQuadH(v); - // quad index 0 stores result - if (localInvocationIndex % 4 == 0) - { - SpdStoreH(FfxInt32x2(workGroupID.xy * 4) + FfxInt32x2(x / 2, y / 2), v, mip, slice); - SpdStoreIntermediateH(x * 2 + y / 2, y * 2, v); - } - } -#endif -} - -void SpdDownsampleMip_4H(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) -{ -#if defined(FFX_SPD_NO_WAVE_OPERATIONS) - if (localInvocationIndex < 4) - { - // x 0 0 0 x 0 0 0 - // ... - // 0 x 0 0 0 x 0 0 - FfxFloat16x4 v = SpdReduceIntermediateH(FfxUInt32x2(x * 8 + 0 + 0 + y * 2, y * 8 + 0), - FfxUInt32x2(x * 8 + 4 + 0 + y * 2, y * 8 + 0), - FfxUInt32x2(x * 8 + 0 + 1 + y * 2, y * 8 + 4), - FfxUInt32x2(x * 8 + 4 + 1 + y * 2, y * 8 + 4)); - SpdStoreH(FfxInt32x2(workGroupID.xy * 2) + FfxInt32x2(x, y), v, mip, slice); - // store to LDS - // x x x x 0 ... - // 0 ... - SpdStoreIntermediateH(x + y * 2, 0, v); - } -#else - if (localInvocationIndex < 16) - { - FfxFloat16x4 v = SpdLoadIntermediateH(x * 4 + y, y * 4); - v = SpdReduceQuadH(v); - // quad index 0 stores result - if (localInvocationIndex % 4 == 0) - { - SpdStoreH(FfxInt32x2(workGroupID.xy * 2) + FfxInt32x2(x / 2, y / 2), v, mip, slice); - SpdStoreIntermediateH(x / 2 + y, 0, v); - } - } -#endif -} - -void SpdDownsampleMip_5H(FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) -{ -#if defined(FFX_SPD_NO_WAVE_OPERATIONS) - if (localInvocationIndex < 1) - { - // x x x x 0 ... - // 0 ... - FfxFloat16x4 v = SpdReduceIntermediateH(FfxUInt32x2(0, 0), FfxUInt32x2(1, 0), FfxUInt32x2(2, 0), FfxUInt32x2(3, 0)); - SpdStoreH(FfxInt32x2(workGroupID.xy), v, mip, slice); - } -#else - if (localInvocationIndex < 4) - { - FfxFloat16x4 v = SpdLoadIntermediateH(localInvocationIndex, 0); - v = SpdReduceQuadH(v); - // quad index 0 stores result - if (localInvocationIndex % 4 == 0) - { - SpdStoreH(FfxInt32x2(workGroupID.xy), v, mip, slice); - } - } -#endif -} - -void SpdDownsampleMips_6_7H(FfxUInt32 x, FfxUInt32 y, FfxUInt32 mips, FfxUInt32 slice) -{ - FfxInt32x2 tex = FfxInt32x2(x * 4 + 0, y * 4 + 0); - FfxInt32x2 pix = FfxInt32x2(x * 2 + 0, y * 2 + 0); - FfxFloat16x4 v0 = SpdReduceLoad4H(tex, slice); - SpdStoreH(pix, v0, 6, slice); - - tex = FfxInt32x2(x * 4 + 2, y * 4 + 0); - pix = FfxInt32x2(x * 2 + 1, y * 2 + 0); - FfxFloat16x4 v1 = SpdReduceLoad4H(tex, slice); - SpdStoreH(pix, v1, 6, slice); - - tex = FfxInt32x2(x * 4 + 0, y * 4 + 2); - pix = FfxInt32x2(x * 2 + 0, y * 2 + 1); - FfxFloat16x4 v2 = SpdReduceLoad4H(tex, slice); - SpdStoreH(pix, v2, 6, slice); - - tex = FfxInt32x2(x * 4 + 2, y * 4 + 2); - pix = FfxInt32x2(x * 2 + 1, y * 2 + 1); - FfxFloat16x4 v3 = SpdReduceLoad4H(tex, slice); - SpdStoreH(pix, v3, 6, slice); - - if (mips < 8) - return; - // no barrier needed, working on values only from the same thread - - FfxFloat16x4 v = SpdReduce4H(v0, v1, v2, v3); - SpdStoreH(FfxInt32x2(x, y), v, 7, slice); - SpdStoreIntermediateH(x, y, v); -} - -void SpdDownsampleNextFourH(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 baseMip, FfxUInt32 mips, FfxUInt32 slice) -{ - if (mips <= baseMip) - return; - ffxSpdWorkgroupShuffleBarrier(); - SpdDownsampleMip_2H(x, y, workGroupID, localInvocationIndex, baseMip, slice); - - if (mips <= baseMip + 1) - return; - ffxSpdWorkgroupShuffleBarrier(); - SpdDownsampleMip_3H(x, y, workGroupID, localInvocationIndex, baseMip + 1, slice); - - if (mips <= baseMip + 2) - return; - ffxSpdWorkgroupShuffleBarrier(); - SpdDownsampleMip_4H(x, y, workGroupID, localInvocationIndex, baseMip + 2, slice); - - if (mips <= baseMip + 3) - return; - ffxSpdWorkgroupShuffleBarrier(); - SpdDownsampleMip_5H(workGroupID, localInvocationIndex, baseMip + 3, slice); -} - -/// Downsamples a 64x64 tile based on the work group id and work group offset. -/// If after downsampling it's the last active thread group, computes the remaining MIP levels. -/// Uses half types. -/// -/// @param [in] workGroupID index of the work group / thread group -/// @param [in] localInvocationIndex index of the thread within the thread group in 1D -/// @param [in] mips the number of total MIP levels to compute for the input texture -/// @param [in] numWorkGroups the total number of dispatched work groups / thread groups for this slice -/// @param [in] slice the slice of the input texture -/// -/// @ingroup FfxGPUSpd -void SpdDownsampleH(FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mips, FfxUInt32 numWorkGroups, FfxUInt32 slice) -{ - FfxUInt32x2 sub_xy = ffxRemapForWaveReduction(localInvocationIndex % 64); - FfxUInt32 x = sub_xy.x + 8 * ((localInvocationIndex >> 6) % 2); - FfxUInt32 y = sub_xy.y + 8 * ((localInvocationIndex >> 7)); - - // compute MIP level 0 and 1 - SpdDownsampleMips_0_1H(x, y, workGroupID, localInvocationIndex, mips, slice); - - // compute MIP level 2, 3, 4, 5 - SpdDownsampleNextFourH(x, y, workGroupID, localInvocationIndex, 2, mips, slice); - - if (mips < 7) - return; - - // increase the global atomic counter for the given slice and check if it's the last remaining thread group: - // terminate if not, continue if yes. - if (SpdExitWorkgroup(numWorkGroups, localInvocationIndex, slice)) - return; - - // reset the global atomic counter back to 0 for the next spd dispatch - SpdResetAtomicCounter(slice); - - // After mip 5 there is only a single workgroup left that downsamples the remaining up to 64x64 texels. - // compute MIP level 6 and 7 - SpdDownsampleMips_6_7H(x, y, mips, slice); - - // compute MIP level 8, 9, 10, 11 - SpdDownsampleNextFourH(x, y, FfxUInt32x2(0, 0), localInvocationIndex, 8, mips, slice); -} - -/// Downsamples a 64x64 tile based on the work group id and work group offset. -/// If after downsampling it's the last active thread group, computes the remaining MIP levels. -/// Uses half types. -/// -/// @param [in] workGroupID index of the work group / thread group -/// @param [in] localInvocationIndex index of the thread within the thread group in 1D -/// @param [in] mips the number of total MIP levels to compute for the input texture -/// @param [in] numWorkGroups the total number of dispatched work groups / thread groups for this slice -/// @param [in] slice the slice of the input texture -/// @param [in] workGroupOffset the work group offset. it's (0,0) in case the entire input texture is downsampled. -/// -/// @ingroup FfxGPUSpd -void SpdDownsampleH(FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mips, FfxUInt32 numWorkGroups, FfxUInt32 slice, FfxUInt32x2 workGroupOffset) -{ - SpdDownsampleH(workGroupID + workGroupOffset, localInvocationIndex, mips, numWorkGroups, slice); -} - -#endif // #if FFX_HALF -#endif // #if defined(FFX_GPU) diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/spd/ffx_spd.h.meta b/Assets/Shaders/FSR3/shaders/fsr3upscaler/spd/ffx_spd.h.meta deleted file mode 100644 index 2741ab1..0000000 --- a/Assets/Shaders/FSR3/shaders/fsr3upscaler/spd/ffx_spd.h.meta +++ /dev/null @@ -1,67 +0,0 @@ -fileFormatVersion: 2 -guid: face65176ee3b82498bd0b8fed0ddacd -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - CPU: AnyCPU - DefaultValueInitialized: true - OS: AnyOS - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/LICENSE.txt b/LICENSE.txt index cbe5a70..01b3e51 100644 --- a/LICENSE.txt +++ b/LICENSE.txt @@ -1,19 +1,19 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. +Copyright (c) 2024 Nico de Poel + +Permission is hereby granted, free of charge, to any person obtaining a copy +of this software and associated documentation files (the "Software"), to deal +in the Software without restriction, including without limitation the rights +to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +copies of the Software, and to permit persons to whom the Software is +furnished to do so, subject to the following conditions: + +The above copyright notice and this permission notice shall be included in all +copies or substantial portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +THE SOFTWARE. diff --git a/com.unity.postprocessing/.signature b/Packages/com.unity.postprocessing/.signature similarity index 100% rename from com.unity.postprocessing/.signature rename to Packages/com.unity.postprocessing/.signature diff --git a/com.unity.postprocessing/CHANGELOG.md b/Packages/com.unity.postprocessing/CHANGELOG.md similarity index 100% rename from com.unity.postprocessing/CHANGELOG.md rename to Packages/com.unity.postprocessing/CHANGELOG.md diff --git a/com.unity.postprocessing/CHANGELOG.md.meta b/Packages/com.unity.postprocessing/CHANGELOG.md.meta similarity index 100% rename from com.unity.postprocessing/CHANGELOG.md.meta rename to Packages/com.unity.postprocessing/CHANGELOG.md.meta diff --git a/com.unity.postprocessing/Documentation~/Ambient-Occlusion.md b/Packages/com.unity.postprocessing/Documentation~/Ambient-Occlusion.md similarity index 100% rename from com.unity.postprocessing/Documentation~/Ambient-Occlusion.md rename to Packages/com.unity.postprocessing/Documentation~/Ambient-Occlusion.md diff --git a/com.unity.postprocessing/Documentation~/Anti-aliasing.md b/Packages/com.unity.postprocessing/Documentation~/Anti-aliasing.md similarity index 100% rename from com.unity.postprocessing/Documentation~/Anti-aliasing.md rename to Packages/com.unity.postprocessing/Documentation~/Anti-aliasing.md diff --git a/com.unity.postprocessing/Documentation~/Auto-Exposure.md b/Packages/com.unity.postprocessing/Documentation~/Auto-Exposure.md similarity index 100% rename from com.unity.postprocessing/Documentation~/Auto-Exposure.md rename to Packages/com.unity.postprocessing/Documentation~/Auto-Exposure.md diff --git a/com.unity.postprocessing/Documentation~/Bloom.md b/Packages/com.unity.postprocessing/Documentation~/Bloom.md similarity index 100% rename from com.unity.postprocessing/Documentation~/Bloom.md rename to Packages/com.unity.postprocessing/Documentation~/Bloom.md diff --git a/com.unity.postprocessing/Documentation~/Chromatic-Aberration.md b/Packages/com.unity.postprocessing/Documentation~/Chromatic-Aberration.md similarity index 100% rename from com.unity.postprocessing/Documentation~/Chromatic-Aberration.md rename to Packages/com.unity.postprocessing/Documentation~/Chromatic-Aberration.md diff --git a/com.unity.postprocessing/Documentation~/Color-Grading.md b/Packages/com.unity.postprocessing/Documentation~/Color-Grading.md similarity index 100% rename from com.unity.postprocessing/Documentation~/Color-Grading.md rename to Packages/com.unity.postprocessing/Documentation~/Color-Grading.md diff --git a/com.unity.postprocessing/Documentation~/Debugging-Post-processing-effects.md b/Packages/com.unity.postprocessing/Documentation~/Debugging-Post-processing-effects.md similarity index 100% rename from com.unity.postprocessing/Documentation~/Debugging-Post-processing-effects.md rename to Packages/com.unity.postprocessing/Documentation~/Debugging-Post-processing-effects.md diff --git a/com.unity.postprocessing/Documentation~/Deferred-Fog.md b/Packages/com.unity.postprocessing/Documentation~/Deferred-Fog.md similarity index 100% rename from com.unity.postprocessing/Documentation~/Deferred-Fog.md rename to Packages/com.unity.postprocessing/Documentation~/Deferred-Fog.md diff --git a/com.unity.postprocessing/Documentation~/Depth-of-Field.md b/Packages/com.unity.postprocessing/Documentation~/Depth-of-Field.md similarity index 100% rename from com.unity.postprocessing/Documentation~/Depth-of-Field.md rename to Packages/com.unity.postprocessing/Documentation~/Depth-of-Field.md diff --git a/com.unity.postprocessing/Documentation~/Grain.md b/Packages/com.unity.postprocessing/Documentation~/Grain.md similarity index 100% rename from com.unity.postprocessing/Documentation~/Grain.md rename to Packages/com.unity.postprocessing/Documentation~/Grain.md diff --git a/com.unity.postprocessing/Documentation~/Installation.md b/Packages/com.unity.postprocessing/Documentation~/Installation.md similarity index 100% rename from com.unity.postprocessing/Documentation~/Installation.md rename to Packages/com.unity.postprocessing/Documentation~/Installation.md diff --git a/com.unity.postprocessing/Documentation~/Lens-Distortion.md b/Packages/com.unity.postprocessing/Documentation~/Lens-Distortion.md similarity index 100% rename from com.unity.postprocessing/Documentation~/Lens-Distortion.md rename to Packages/com.unity.postprocessing/Documentation~/Lens-Distortion.md diff --git a/com.unity.postprocessing/Documentation~/Manipulating-the-Stack.md b/Packages/com.unity.postprocessing/Documentation~/Manipulating-the-Stack.md similarity index 100% rename from com.unity.postprocessing/Documentation~/Manipulating-the-Stack.md rename to Packages/com.unity.postprocessing/Documentation~/Manipulating-the-Stack.md diff --git a/com.unity.postprocessing/Documentation~/Motion-Blur.md b/Packages/com.unity.postprocessing/Documentation~/Motion-Blur.md similarity index 100% rename from com.unity.postprocessing/Documentation~/Motion-Blur.md rename to Packages/com.unity.postprocessing/Documentation~/Motion-Blur.md diff --git a/com.unity.postprocessing/Documentation~/Quick-start.md b/Packages/com.unity.postprocessing/Documentation~/Quick-start.md similarity index 100% rename from com.unity.postprocessing/Documentation~/Quick-start.md rename to Packages/com.unity.postprocessing/Documentation~/Quick-start.md diff --git a/com.unity.postprocessing/Documentation~/Screen-Space-Reflections.md b/Packages/com.unity.postprocessing/Documentation~/Screen-Space-Reflections.md similarity index 100% rename from com.unity.postprocessing/Documentation~/Screen-Space-Reflections.md rename to Packages/com.unity.postprocessing/Documentation~/Screen-Space-Reflections.md diff --git a/com.unity.postprocessing/Documentation~/TableOfContents.md b/Packages/com.unity.postprocessing/Documentation~/TableOfContents.md similarity index 100% rename from com.unity.postprocessing/Documentation~/TableOfContents.md rename to Packages/com.unity.postprocessing/Documentation~/TableOfContents.md diff --git a/com.unity.postprocessing/Documentation~/Vignette.md b/Packages/com.unity.postprocessing/Documentation~/Vignette.md similarity index 100% rename from com.unity.postprocessing/Documentation~/Vignette.md rename to Packages/com.unity.postprocessing/Documentation~/Vignette.md diff --git a/com.unity.postprocessing/Documentation~/Writing-Custom-Effects.md b/Packages/com.unity.postprocessing/Documentation~/Writing-Custom-Effects.md similarity index 100% rename from com.unity.postprocessing/Documentation~/Writing-Custom-Effects.md rename to Packages/com.unity.postprocessing/Documentation~/Writing-Custom-Effects.md diff --git a/com.unity.postprocessing/Documentation~/images/Grain_image_0.png b/Packages/com.unity.postprocessing/Documentation~/images/Grain_image_0.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/Grain_image_0.png rename to Packages/com.unity.postprocessing/Documentation~/images/Grain_image_0.png diff --git a/com.unity.postprocessing/Documentation~/images/Grain_image_1.png b/Packages/com.unity.postprocessing/Documentation~/images/Grain_image_1.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/Grain_image_1.png rename to Packages/com.unity.postprocessing/Documentation~/images/Grain_image_1.png diff --git a/com.unity.postprocessing/Documentation~/images/PostProcessing-Bloom-0.png b/Packages/com.unity.postprocessing/Documentation~/images/PostProcessing-Bloom-0.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/PostProcessing-Bloom-0.png rename to Packages/com.unity.postprocessing/Documentation~/images/PostProcessing-Bloom-0.png diff --git a/com.unity.postprocessing/Documentation~/images/PostProcessing-ChromaticAberration-0.png b/Packages/com.unity.postprocessing/Documentation~/images/PostProcessing-ChromaticAberration-0.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/PostProcessing-ChromaticAberration-0.png rename to Packages/com.unity.postprocessing/Documentation~/images/PostProcessing-ChromaticAberration-0.png diff --git a/com.unity.postprocessing/Documentation~/images/PostProcessing-ChromaticAberration-1.png b/Packages/com.unity.postprocessing/Documentation~/images/PostProcessing-ChromaticAberration-1.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/PostProcessing-ChromaticAberration-1.png rename to Packages/com.unity.postprocessing/Documentation~/images/PostProcessing-ChromaticAberration-1.png diff --git a/com.unity.postprocessing/Documentation~/images/PostProcessing-Vignette-1.png b/Packages/com.unity.postprocessing/Documentation~/images/PostProcessing-Vignette-1.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/PostProcessing-Vignette-1.png rename to Packages/com.unity.postprocessing/Documentation~/images/PostProcessing-Vignette-1.png diff --git a/com.unity.postprocessing/Documentation~/images/PostProcessing-Vignette-2.png b/Packages/com.unity.postprocessing/Documentation~/images/PostProcessing-Vignette-2.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/PostProcessing-Vignette-2.png rename to Packages/com.unity.postprocessing/Documentation~/images/PostProcessing-Vignette-2.png diff --git a/com.unity.postprocessing/Documentation~/images/Ppv2 _ Debugging_Light meter_Graph.png b/Packages/com.unity.postprocessing/Documentation~/images/Ppv2 _ Debugging_Light meter_Graph.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/Ppv2 _ Debugging_Light meter_Graph.png rename to Packages/com.unity.postprocessing/Documentation~/images/Ppv2 _ Debugging_Light meter_Graph.png diff --git a/com.unity.postprocessing/Documentation~/images/Ppv2 _ Debugging_Vectorscope_Graph.png b/Packages/com.unity.postprocessing/Documentation~/images/Ppv2 _ Debugging_Vectorscope_Graph.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/Ppv2 _ Debugging_Vectorscope_Graph.png rename to Packages/com.unity.postprocessing/Documentation~/images/Ppv2 _ Debugging_Vectorscope_Graph.png diff --git a/com.unity.postprocessing/Documentation~/images/Ppv2-Debugging-Histogram-Graph.png b/Packages/com.unity.postprocessing/Documentation~/images/Ppv2-Debugging-Histogram-Graph.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/Ppv2-Debugging-Histogram-Graph.png rename to Packages/com.unity.postprocessing/Documentation~/images/Ppv2-Debugging-Histogram-Graph.png diff --git a/com.unity.postprocessing/Documentation~/images/Ppv2-Debugging-Light-Meter-Graph.png b/Packages/com.unity.postprocessing/Documentation~/images/Ppv2-Debugging-Light-Meter-Graph.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/Ppv2-Debugging-Light-Meter-Graph.png rename to Packages/com.unity.postprocessing/Documentation~/images/Ppv2-Debugging-Light-Meter-Graph.png diff --git a/com.unity.postprocessing/Documentation~/images/Ppv2-Debugging-Vectorscope-Graph.png b/Packages/com.unity.postprocessing/Documentation~/images/Ppv2-Debugging-Vectorscope-Graph.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/Ppv2-Debugging-Vectorscope-Graph.png rename to Packages/com.unity.postprocessing/Documentation~/images/Ppv2-Debugging-Vectorscope-Graph.png diff --git a/com.unity.postprocessing/Documentation~/images/Ppv2-Debugging-Waveform-Graph.png b/Packages/com.unity.postprocessing/Documentation~/images/Ppv2-Debugging-Waveform-Graph.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/Ppv2-Debugging-Waveform-Graph.png rename to Packages/com.unity.postprocessing/Documentation~/images/Ppv2-Debugging-Waveform-Graph.png diff --git a/com.unity.postprocessing/Documentation~/images/Ppv2_Debugging_Histogram.png b/Packages/com.unity.postprocessing/Documentation~/images/Ppv2_Debugging_Histogram.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/Ppv2_Debugging_Histogram.png rename to Packages/com.unity.postprocessing/Documentation~/images/Ppv2_Debugging_Histogram.png diff --git a/com.unity.postprocessing/Documentation~/images/Ppv2_Debugging_Waveform_Graph.png b/Packages/com.unity.postprocessing/Documentation~/images/Ppv2_Debugging_Waveform_Graph.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/Ppv2_Debugging_Waveform_Graph.png rename to Packages/com.unity.postprocessing/Documentation~/images/Ppv2_Debugging_Waveform_Graph.png diff --git a/com.unity.postprocessing/Documentation~/images/Ppv2_Light Meter.png b/Packages/com.unity.postprocessing/Documentation~/images/Ppv2_Light Meter.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/Ppv2_Light Meter.png rename to Packages/com.unity.postprocessing/Documentation~/images/Ppv2_Light Meter.png diff --git a/com.unity.postprocessing/Documentation~/images/Ppv2_Post-processing-Debug.png b/Packages/com.unity.postprocessing/Documentation~/images/Ppv2_Post-processing-Debug.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/Ppv2_Post-processing-Debug.png rename to Packages/com.unity.postprocessing/Documentation~/images/Ppv2_Post-processing-Debug.png diff --git a/com.unity.postprocessing/Documentation~/images/aa-1.png b/Packages/com.unity.postprocessing/Documentation~/images/aa-1.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/aa-1.png rename to Packages/com.unity.postprocessing/Documentation~/images/aa-1.png diff --git a/com.unity.postprocessing/Documentation~/images/aa-2.png b/Packages/com.unity.postprocessing/Documentation~/images/aa-2.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/aa-2.png rename to Packages/com.unity.postprocessing/Documentation~/images/aa-2.png diff --git a/com.unity.postprocessing/Documentation~/images/aa-3.png b/Packages/com.unity.postprocessing/Documentation~/images/aa-3.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/aa-3.png rename to Packages/com.unity.postprocessing/Documentation~/images/aa-3.png diff --git a/com.unity.postprocessing/Documentation~/images/ao-off.png b/Packages/com.unity.postprocessing/Documentation~/images/ao-off.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/ao-off.png rename to Packages/com.unity.postprocessing/Documentation~/images/ao-off.png diff --git a/com.unity.postprocessing/Documentation~/images/ao-on.png b/Packages/com.unity.postprocessing/Documentation~/images/ao-on.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/ao-on.png rename to Packages/com.unity.postprocessing/Documentation~/images/ao-on.png diff --git a/com.unity.postprocessing/Documentation~/images/auto-exposure-off.png b/Packages/com.unity.postprocessing/Documentation~/images/auto-exposure-off.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/auto-exposure-off.png rename to Packages/com.unity.postprocessing/Documentation~/images/auto-exposure-off.png diff --git a/com.unity.postprocessing/Documentation~/images/auto-exposure-on.png b/Packages/com.unity.postprocessing/Documentation~/images/auto-exposure-on.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/auto-exposure-on.png rename to Packages/com.unity.postprocessing/Documentation~/images/auto-exposure-on.png diff --git a/com.unity.postprocessing/Documentation~/images/autoexposure.png b/Packages/com.unity.postprocessing/Documentation~/images/autoexposure.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/autoexposure.png rename to Packages/com.unity.postprocessing/Documentation~/images/autoexposure.png diff --git a/com.unity.postprocessing/Documentation~/images/bloom.png b/Packages/com.unity.postprocessing/Documentation~/images/bloom.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/bloom.png rename to Packages/com.unity.postprocessing/Documentation~/images/bloom.png diff --git a/com.unity.postprocessing/Documentation~/images/chroma.png b/Packages/com.unity.postprocessing/Documentation~/images/chroma.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/chroma.png rename to Packages/com.unity.postprocessing/Documentation~/images/chroma.png diff --git a/com.unity.postprocessing/Documentation~/images/custom-effect-sorting.png b/Packages/com.unity.postprocessing/Documentation~/images/custom-effect-sorting.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/custom-effect-sorting.png rename to Packages/com.unity.postprocessing/Documentation~/images/custom-effect-sorting.png diff --git a/com.unity.postprocessing/Documentation~/images/deferredfog.png b/Packages/com.unity.postprocessing/Documentation~/images/deferredfog.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/deferredfog.png rename to Packages/com.unity.postprocessing/Documentation~/images/deferredfog.png diff --git a/com.unity.postprocessing/Documentation~/images/dof.png b/Packages/com.unity.postprocessing/Documentation~/images/dof.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/dof.png rename to Packages/com.unity.postprocessing/Documentation~/images/dof.png diff --git a/com.unity.postprocessing/Documentation~/images/grading-1.png b/Packages/com.unity.postprocessing/Documentation~/images/grading-1.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/grading-1.png rename to Packages/com.unity.postprocessing/Documentation~/images/grading-1.png diff --git a/com.unity.postprocessing/Documentation~/images/grading-10.png b/Packages/com.unity.postprocessing/Documentation~/images/grading-10.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/grading-10.png rename to Packages/com.unity.postprocessing/Documentation~/images/grading-10.png diff --git a/com.unity.postprocessing/Documentation~/images/grading-11.png b/Packages/com.unity.postprocessing/Documentation~/images/grading-11.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/grading-11.png rename to Packages/com.unity.postprocessing/Documentation~/images/grading-11.png diff --git a/com.unity.postprocessing/Documentation~/images/grading-2.png b/Packages/com.unity.postprocessing/Documentation~/images/grading-2.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/grading-2.png rename to Packages/com.unity.postprocessing/Documentation~/images/grading-2.png diff --git a/com.unity.postprocessing/Documentation~/images/grading-3.png b/Packages/com.unity.postprocessing/Documentation~/images/grading-3.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/grading-3.png rename to Packages/com.unity.postprocessing/Documentation~/images/grading-3.png diff --git a/com.unity.postprocessing/Documentation~/images/grading-4.png b/Packages/com.unity.postprocessing/Documentation~/images/grading-4.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/grading-4.png rename to Packages/com.unity.postprocessing/Documentation~/images/grading-4.png diff --git a/com.unity.postprocessing/Documentation~/images/grading-5.png b/Packages/com.unity.postprocessing/Documentation~/images/grading-5.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/grading-5.png rename to Packages/com.unity.postprocessing/Documentation~/images/grading-5.png diff --git a/com.unity.postprocessing/Documentation~/images/grading-6.png b/Packages/com.unity.postprocessing/Documentation~/images/grading-6.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/grading-6.png rename to Packages/com.unity.postprocessing/Documentation~/images/grading-6.png diff --git a/com.unity.postprocessing/Documentation~/images/grading-7.png b/Packages/com.unity.postprocessing/Documentation~/images/grading-7.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/grading-7.png rename to Packages/com.unity.postprocessing/Documentation~/images/grading-7.png diff --git a/com.unity.postprocessing/Documentation~/images/grading-8.png b/Packages/com.unity.postprocessing/Documentation~/images/grading-8.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/grading-8.png rename to Packages/com.unity.postprocessing/Documentation~/images/grading-8.png diff --git a/com.unity.postprocessing/Documentation~/images/grading-9.png b/Packages/com.unity.postprocessing/Documentation~/images/grading-9.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/grading-9.png rename to Packages/com.unity.postprocessing/Documentation~/images/grading-9.png diff --git a/com.unity.postprocessing/Documentation~/images/grain.png b/Packages/com.unity.postprocessing/Documentation~/images/grain.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/grain.png rename to Packages/com.unity.postprocessing/Documentation~/images/grain.png diff --git a/com.unity.postprocessing/Documentation~/images/home-after.png b/Packages/com.unity.postprocessing/Documentation~/images/home-after.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/home-after.png rename to Packages/com.unity.postprocessing/Documentation~/images/home-after.png diff --git a/com.unity.postprocessing/Documentation~/images/home-before.png b/Packages/com.unity.postprocessing/Documentation~/images/home-before.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/home-before.png rename to Packages/com.unity.postprocessing/Documentation~/images/home-before.png diff --git a/com.unity.postprocessing/Documentation~/images/hue-vs-hue.png b/Packages/com.unity.postprocessing/Documentation~/images/hue-vs-hue.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/hue-vs-hue.png rename to Packages/com.unity.postprocessing/Documentation~/images/hue-vs-hue.png diff --git a/com.unity.postprocessing/Documentation~/images/hue-vs-sat.png b/Packages/com.unity.postprocessing/Documentation~/images/hue-vs-sat.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/hue-vs-sat.png rename to Packages/com.unity.postprocessing/Documentation~/images/hue-vs-sat.png diff --git a/com.unity.postprocessing/Documentation~/images/lens-distortion.png b/Packages/com.unity.postprocessing/Documentation~/images/lens-distortion.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/lens-distortion.png rename to Packages/com.unity.postprocessing/Documentation~/images/lens-distortion.png diff --git a/com.unity.postprocessing/Documentation~/images/lensdistortion.png b/Packages/com.unity.postprocessing/Documentation~/images/lensdistortion.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/lensdistortion.png rename to Packages/com.unity.postprocessing/Documentation~/images/lensdistortion.png diff --git a/com.unity.postprocessing/Documentation~/images/lum-vs-sat.png b/Packages/com.unity.postprocessing/Documentation~/images/lum-vs-sat.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/lum-vs-sat.png rename to Packages/com.unity.postprocessing/Documentation~/images/lum-vs-sat.png diff --git a/com.unity.postprocessing/Documentation~/images/motionblur.png b/Packages/com.unity.postprocessing/Documentation~/images/motionblur.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/motionblur.png rename to Packages/com.unity.postprocessing/Documentation~/images/motionblur.png diff --git a/com.unity.postprocessing/Documentation~/images/no-lens-distortion.png b/Packages/com.unity.postprocessing/Documentation~/images/no-lens-distortion.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/no-lens-distortion.png rename to Packages/com.unity.postprocessing/Documentation~/images/no-lens-distortion.png diff --git a/com.unity.postprocessing/Documentation~/images/quickstart-1.png b/Packages/com.unity.postprocessing/Documentation~/images/quickstart-1.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/quickstart-1.png rename to Packages/com.unity.postprocessing/Documentation~/images/quickstart-1.png diff --git a/com.unity.postprocessing/Documentation~/images/quickstart-2.png b/Packages/com.unity.postprocessing/Documentation~/images/quickstart-2.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/quickstart-2.png rename to Packages/com.unity.postprocessing/Documentation~/images/quickstart-2.png diff --git a/com.unity.postprocessing/Documentation~/images/quickstart-3.png b/Packages/com.unity.postprocessing/Documentation~/images/quickstart-3.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/quickstart-3.png rename to Packages/com.unity.postprocessing/Documentation~/images/quickstart-3.png diff --git a/com.unity.postprocessing/Documentation~/images/sat-vs-sat.png b/Packages/com.unity.postprocessing/Documentation~/images/sat-vs-sat.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/sat-vs-sat.png rename to Packages/com.unity.postprocessing/Documentation~/images/sat-vs-sat.png diff --git a/com.unity.postprocessing/Documentation~/images/ssao-1.png b/Packages/com.unity.postprocessing/Documentation~/images/ssao-1.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/ssao-1.png rename to Packages/com.unity.postprocessing/Documentation~/images/ssao-1.png diff --git a/com.unity.postprocessing/Documentation~/images/ssao-2.png b/Packages/com.unity.postprocessing/Documentation~/images/ssao-2.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/ssao-2.png rename to Packages/com.unity.postprocessing/Documentation~/images/ssao-2.png diff --git a/com.unity.postprocessing/Documentation~/images/ssr.png b/Packages/com.unity.postprocessing/Documentation~/images/ssr.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/ssr.png rename to Packages/com.unity.postprocessing/Documentation~/images/ssr.png diff --git a/com.unity.postprocessing/Documentation~/images/tonemapping.png b/Packages/com.unity.postprocessing/Documentation~/images/tonemapping.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/tonemapping.png rename to Packages/com.unity.postprocessing/Documentation~/images/tonemapping.png diff --git a/com.unity.postprocessing/Documentation~/images/trackballs.png b/Packages/com.unity.postprocessing/Documentation~/images/trackballs.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/trackballs.png rename to Packages/com.unity.postprocessing/Documentation~/images/trackballs.png diff --git a/com.unity.postprocessing/Documentation~/images/vignette-1.png b/Packages/com.unity.postprocessing/Documentation~/images/vignette-1.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/vignette-1.png rename to Packages/com.unity.postprocessing/Documentation~/images/vignette-1.png diff --git a/com.unity.postprocessing/Documentation~/images/vignette-2.png b/Packages/com.unity.postprocessing/Documentation~/images/vignette-2.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/vignette-2.png rename to Packages/com.unity.postprocessing/Documentation~/images/vignette-2.png diff --git a/com.unity.postprocessing/Documentation~/images/yrgb-curves.png b/Packages/com.unity.postprocessing/Documentation~/images/yrgb-curves.png similarity index 100% rename from com.unity.postprocessing/Documentation~/images/yrgb-curves.png rename to Packages/com.unity.postprocessing/Documentation~/images/yrgb-curves.png diff --git a/com.unity.postprocessing/Documentation~/index.md b/Packages/com.unity.postprocessing/Documentation~/index.md similarity index 100% rename from com.unity.postprocessing/Documentation~/index.md rename to Packages/com.unity.postprocessing/Documentation~/index.md diff --git a/com.unity.postprocessing/Documentation~/known-issues.md b/Packages/com.unity.postprocessing/Documentation~/known-issues.md similarity index 100% rename from com.unity.postprocessing/Documentation~/known-issues.md rename to Packages/com.unity.postprocessing/Documentation~/known-issues.md diff --git a/com.unity.postprocessing/Documentation~/requirements.md b/Packages/com.unity.postprocessing/Documentation~/requirements.md similarity index 100% rename from com.unity.postprocessing/Documentation~/requirements.md rename to Packages/com.unity.postprocessing/Documentation~/requirements.md diff --git a/com.unity.postprocessing/LICENSE.md b/Packages/com.unity.postprocessing/LICENSE.md similarity index 100% rename from com.unity.postprocessing/LICENSE.md rename to Packages/com.unity.postprocessing/LICENSE.md diff --git a/com.unity.postprocessing/LICENSE.md.meta b/Packages/com.unity.postprocessing/LICENSE.md.meta similarity index 100% rename from com.unity.postprocessing/LICENSE.md.meta rename to Packages/com.unity.postprocessing/LICENSE.md.meta diff --git a/com.unity.postprocessing/PostProcessing.meta b/Packages/com.unity.postprocessing/PostProcessing.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing.meta rename to Packages/com.unity.postprocessing/PostProcessing.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Attributes.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Attributes.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Attributes.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Attributes.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Attributes/DecoratorAttribute.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Attributes/DecoratorAttribute.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Attributes/DecoratorAttribute.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Attributes/DecoratorAttribute.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Attributes/DecoratorAttribute.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Attributes/DecoratorAttribute.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Attributes/DecoratorAttribute.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Attributes/DecoratorAttribute.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Attributes/PostProcessEditorAttribute.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Attributes/PostProcessEditorAttribute.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Attributes/PostProcessEditorAttribute.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Attributes/PostProcessEditorAttribute.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Attributes/PostProcessEditorAttribute.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Attributes/PostProcessEditorAttribute.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Attributes/PostProcessEditorAttribute.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Attributes/PostProcessEditorAttribute.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/BaseEditor.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/BaseEditor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/BaseEditor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/BaseEditor.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/BaseEditor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/BaseEditor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/BaseEditor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/BaseEditor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Decorators.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Decorators.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Decorators.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Decorators.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Decorators/AttributeDecorator.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Decorators/AttributeDecorator.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Decorators/AttributeDecorator.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Decorators/AttributeDecorator.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Decorators/AttributeDecorator.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Decorators/AttributeDecorator.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Decorators/AttributeDecorator.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Decorators/AttributeDecorator.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Decorators/Decorators.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Decorators/Decorators.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Decorators/Decorators.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Decorators/Decorators.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Decorators/Decorators.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Decorators/Decorators.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Decorators/Decorators.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Decorators/Decorators.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Decorators/TrackballDecorator.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Decorators/TrackballDecorator.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Decorators/TrackballDecorator.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Decorators/TrackballDecorator.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Decorators/TrackballDecorator.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Decorators/TrackballDecorator.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Decorators/TrackballDecorator.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Decorators/TrackballDecorator.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/EffectListEditor.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/EffectListEditor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/EffectListEditor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/EffectListEditor.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/EffectListEditor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/EffectListEditor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/EffectListEditor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/EffectListEditor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects/AmbientOcclusionEditor.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/AmbientOcclusionEditor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects/AmbientOcclusionEditor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/AmbientOcclusionEditor.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects/AmbientOcclusionEditor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/AmbientOcclusionEditor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects/AmbientOcclusionEditor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/AmbientOcclusionEditor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects/AutoExposureEditor.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/AutoExposureEditor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects/AutoExposureEditor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/AutoExposureEditor.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects/AutoExposureEditor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/AutoExposureEditor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects/AutoExposureEditor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/AutoExposureEditor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects/BloomEditor.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/BloomEditor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects/BloomEditor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/BloomEditor.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects/BloomEditor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/BloomEditor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects/BloomEditor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/BloomEditor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects/ChromaticAberrationEditor.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/ChromaticAberrationEditor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects/ChromaticAberrationEditor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/ChromaticAberrationEditor.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects/ChromaticAberrationEditor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/ChromaticAberrationEditor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects/ChromaticAberrationEditor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/ChromaticAberrationEditor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects/ColorGradingEditor.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/ColorGradingEditor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects/ColorGradingEditor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/ColorGradingEditor.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects/ColorGradingEditor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/ColorGradingEditor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects/ColorGradingEditor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/ColorGradingEditor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects/DefaultPostProcessEffectEditor.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/DefaultPostProcessEffectEditor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects/DefaultPostProcessEffectEditor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/DefaultPostProcessEffectEditor.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects/DefaultPostProcessEffectEditor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/DefaultPostProcessEffectEditor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects/DefaultPostProcessEffectEditor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/DefaultPostProcessEffectEditor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects/DepthOfFieldEditor.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/DepthOfFieldEditor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects/DepthOfFieldEditor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/DepthOfFieldEditor.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects/DepthOfFieldEditor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/DepthOfFieldEditor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects/DepthOfFieldEditor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/DepthOfFieldEditor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects/LensDistortionEditor.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/LensDistortionEditor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects/LensDistortionEditor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/LensDistortionEditor.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects/LensDistortionEditor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/LensDistortionEditor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects/LensDistortionEditor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/LensDistortionEditor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects/MotionBlurEditor.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/MotionBlurEditor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects/MotionBlurEditor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/MotionBlurEditor.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects/MotionBlurEditor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/MotionBlurEditor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects/MotionBlurEditor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/MotionBlurEditor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects/ScreenSpaceReflectionsEditor.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/ScreenSpaceReflectionsEditor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects/ScreenSpaceReflectionsEditor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/ScreenSpaceReflectionsEditor.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects/ScreenSpaceReflectionsEditor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/ScreenSpaceReflectionsEditor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects/ScreenSpaceReflectionsEditor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/ScreenSpaceReflectionsEditor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects/VignetteEditor.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/VignetteEditor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects/VignetteEditor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/VignetteEditor.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Effects/VignetteEditor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/VignetteEditor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Effects/VignetteEditor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Effects/VignetteEditor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/PostProcessDebugEditor.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessDebugEditor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/PostProcessDebugEditor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessDebugEditor.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/PostProcessDebugEditor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessDebugEditor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/PostProcessDebugEditor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessDebugEditor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/PostProcessEffectBaseEditor.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessEffectBaseEditor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/PostProcessEffectBaseEditor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessEffectBaseEditor.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/PostProcessEffectBaseEditor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessEffectBaseEditor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/PostProcessEffectBaseEditor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessEffectBaseEditor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/PostProcessEffectEditor.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessEffectEditor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/PostProcessEffectEditor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessEffectEditor.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/PostProcessEffectEditor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessEffectEditor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/PostProcessEffectEditor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessEffectEditor.cs.meta diff --git a/Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessLayerEditor.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessLayerEditor.cs new file mode 100644 index 0000000..6fdf634 --- /dev/null +++ b/Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessLayerEditor.cs @@ -0,0 +1,475 @@ +using System; +using System.Collections.Generic; +using System.Linq; +using UnityEngine; +using UnityEngine.Rendering.PostProcessing; +using UnityEditorInternal; +using System.IO; + +namespace UnityEditor.Rendering.PostProcessing +{ + using SerializedBundleRef = PostProcessLayer.SerializedBundleRef; + using EXRFlags = Texture2D.EXRFlags; + + [CanEditMultipleObjects, CustomEditor(typeof(PostProcessLayer))] + sealed class PostProcessLayerEditor : BaseEditor + { + SerializedProperty m_StopNaNPropagation; +#pragma warning disable 414 + SerializedProperty m_DirectToCameraTarget; +#pragma warning restore 414 + SerializedProperty m_VolumeTrigger; + SerializedProperty m_VolumeLayer; + + SerializedProperty m_AntialiasingMode; + SerializedProperty m_TaaJitterSpread; + SerializedProperty m_TaaSharpness; + SerializedProperty m_TaaStationaryBlending; + SerializedProperty m_TaaMotionBlending; + SerializedProperty m_SmaaQuality; + SerializedProperty m_FxaaFastMode; + SerializedProperty m_FxaaKeepAlpha; + + SerializedProperty m_FsrQualityMode; + SerializedProperty m_FsrPerformSharpen; + SerializedProperty m_FsrSharpness; + SerializedProperty m_FsrEnableFP16; + SerializedProperty m_FsrExposureSource; + SerializedProperty m_FsrExposureTexture; + SerializedProperty m_FsrPreExposure; + SerializedProperty m_FsrDebugView; + SerializedProperty m_FsrAutoReactive; + SerializedProperty m_FsrAutoReactiveParams; + SerializedProperty m_FsrReactiveMaskTexture; + SerializedProperty m_FsrAutoTcr; + SerializedProperty m_FsrAutoTcrParams; + SerializedProperty m_FsrTcrMaskTexture; + + SerializedProperty m_FogEnabled; + SerializedProperty m_FogExcludeSkybox; + + SerializedProperty m_ShowToolkit; + SerializedProperty m_ShowCustomSorter; + + Dictionary m_CustomLists; + +#if UNITY_2017_3_OR_NEWER + Camera m_TargetCameraComponent; +#endif + + static GUIContent[] s_AntialiasingMethodNames = + { + new GUIContent("No Anti-aliasing"), + new GUIContent("Fast Approximate Anti-aliasing (FXAA)"), + new GUIContent("Subpixel Morphological Anti-aliasing (SMAA)"), + new GUIContent("Temporal Anti-aliasing (TAA)"), + new GUIContent("FidelityFX Super Resolution 3 (FSR3) Upscaler") + }; + + enum ExportMode + { + FullFrame, + DisablePost, + BreakBeforeColorGradingLinear, + BreakBeforeColorGradingLog + } + + void OnEnable() + { + m_StopNaNPropagation = FindProperty(x => x.stopNaNPropagation); + m_DirectToCameraTarget = FindProperty(x => x.finalBlitToCameraTarget); + m_VolumeTrigger = FindProperty(x => x.volumeTrigger); + m_VolumeLayer = FindProperty(x => x.volumeLayer); + + m_AntialiasingMode = FindProperty(x => x.antialiasingMode); + m_TaaJitterSpread = FindProperty(x => x.temporalAntialiasing.jitterSpread); + m_TaaSharpness = FindProperty(x => x.temporalAntialiasing.sharpness); + m_TaaStationaryBlending = FindProperty(x => x.temporalAntialiasing.stationaryBlending); + m_TaaMotionBlending = FindProperty(x => x.temporalAntialiasing.motionBlending); + m_SmaaQuality = FindProperty(x => x.subpixelMorphologicalAntialiasing.quality); + m_FxaaFastMode = FindProperty(x => x.fastApproximateAntialiasing.fastMode); + m_FxaaKeepAlpha = FindProperty(x => x.fastApproximateAntialiasing.keepAlpha); + + m_FsrQualityMode = FindProperty(x => x.superResolution.qualityMode); + m_FsrPerformSharpen = FindProperty(x => x.superResolution.performSharpenPass); + m_FsrSharpness = FindProperty(x => x.superResolution.sharpness); + m_FsrEnableFP16 = FindProperty(x => x.superResolution.enableFP16); + m_FsrExposureSource = FindProperty(x => x.superResolution.exposureSource); + m_FsrExposureTexture = FindProperty(x => x.superResolution.exposure); + m_FsrPreExposure = FindProperty(x => x.superResolution.preExposure); + m_FsrDebugView = FindProperty(x => x.superResolution.enableDebugView); + m_FsrAutoReactive = FindProperty(x => x.superResolution.autoGenerateReactiveMask); + m_FsrAutoReactiveParams = FindProperty(x => x.superResolution.generateReactiveParameters); + m_FsrReactiveMaskTexture = FindProperty(x => x.superResolution.reactiveMask); + m_FsrAutoTcr = FindProperty(x => x.superResolution.autoGenerateTransparencyAndComposition); + m_FsrAutoTcrParams = FindProperty(x => x.superResolution.generateTransparencyAndCompositionParameters); + m_FsrTcrMaskTexture = FindProperty(x => x.superResolution.transparencyAndCompositionMask); + + m_FogEnabled = FindProperty(x => x.fog.enabled); + m_FogExcludeSkybox = FindProperty(x => x.fog.excludeSkybox); + + m_ShowToolkit = serializedObject.FindProperty("m_ShowToolkit"); + m_ShowCustomSorter = serializedObject.FindProperty("m_ShowCustomSorter"); + +#if UNITY_2017_3_OR_NEWER + m_TargetCameraComponent = m_Target.GetComponent(); +#endif + } + + void OnDisable() + { + m_CustomLists = null; + } + + public override void OnInspectorGUI() + { + serializedObject.Update(); + + var camera = m_Target.GetComponent(); + + DoVolumeBlending(); + DoAntialiasing(); + DoFog(camera); + + EditorGUILayout.PropertyField(m_StopNaNPropagation, EditorUtilities.GetContent("Stop NaN Propagation|Automatically replaces NaN/Inf in shaders by a black pixel to avoid breaking some effects. This will slightly affect performances and should only be used if you experience NaN issues that you can't fix. Has no effect on GLES2 platforms.")); + +#if UNITY_2019_1_OR_NEWER + if (!RuntimeUtilities.scriptableRenderPipelineActive) + EditorGUILayout.PropertyField(m_DirectToCameraTarget, EditorUtilities.GetContent("Directly to Camera Target|Use the final blit to the camera render target for postprocessing. This has less overhead but breaks compatibility with legacy image effect that use OnRenderImage.")); +#endif + + EditorGUILayout.Space(); + + DoToolkit(); + DoCustomEffectSorter(); + + EditorUtilities.DrawSplitter(); + EditorGUILayout.Space(); + + serializedObject.ApplyModifiedProperties(); + } + + void DoVolumeBlending() + { + EditorGUILayout.LabelField(EditorUtilities.GetContent("Volume blending"), EditorStyles.boldLabel); + EditorGUI.indentLevel++; + { + // The layout system sort of break alignement when mixing inspector fields with + // custom layouted fields, do the layout manually instead + var indentOffset = EditorGUI.indentLevel * 15f; + var lineRect = GUILayoutUtility.GetRect(1, EditorGUIUtility.singleLineHeight); + var labelRect = new Rect(lineRect.x, lineRect.y, EditorGUIUtility.labelWidth - indentOffset, lineRect.height); + var fieldRect = new Rect(labelRect.xMax, lineRect.y, lineRect.width - labelRect.width - 60f, lineRect.height); + var buttonRect = new Rect(fieldRect.xMax, lineRect.y, 60f, lineRect.height); + + EditorGUI.PrefixLabel(labelRect, EditorUtilities.GetContent("Trigger|A transform that will act as a trigger for volume blending.")); + m_VolumeTrigger.objectReferenceValue = (Transform)EditorGUI.ObjectField(fieldRect, m_VolumeTrigger.objectReferenceValue, typeof(Transform), true); + if (GUI.Button(buttonRect, EditorUtilities.GetContent("This|Assigns the current GameObject as a trigger."), EditorStyles.miniButton)) + m_VolumeTrigger.objectReferenceValue = m_Target.transform; + + if (m_VolumeTrigger.objectReferenceValue == null) + EditorGUILayout.HelpBox("No trigger has been set, the camera will only be affected by global volumes.", MessageType.Info); + + EditorGUILayout.PropertyField(m_VolumeLayer, EditorUtilities.GetContent("Layer|This camera will only be affected by volumes in the selected scene-layers.")); + + int mask = m_VolumeLayer.intValue; + if (mask == 0) + EditorGUILayout.HelpBox("No layer has been set, the trigger will never be affected by volumes.", MessageType.Warning); + else if (mask == -1 || ((mask & 1) != 0)) + EditorGUILayout.HelpBox("Do not use \"Everything\" or \"Default\" as a layer mask as it will slow down the volume blending process! Put post-processing volumes in their own dedicated layer for best performances.", MessageType.Warning); + } + EditorGUI.indentLevel--; + + EditorGUILayout.Space(); + } + + void DoAntialiasing() + { + EditorGUILayout.LabelField(EditorUtilities.GetContent("Anti-aliasing"), EditorStyles.boldLabel); + EditorGUI.indentLevel++; + { + m_AntialiasingMode.intValue = EditorGUILayout.Popup(EditorUtilities.GetContent("Mode|The anti-aliasing method to use. FXAA is fast but low quality. SMAA works well for non-HDR scenes. TAA is a bit slower but higher quality and works well with HDR."), m_AntialiasingMode.intValue, s_AntialiasingMethodNames); + + if (m_AntialiasingMode.intValue == (int)PostProcessLayer.Antialiasing.TemporalAntialiasing) + { +#if !UNITY_2017_3_OR_NEWER + if (RuntimeUtilities.isSinglePassStereoSelected) + EditorGUILayout.HelpBox("TAA requires Unity 2017.3+ for Single-pass stereo rendering support.", MessageType.Warning); +#endif +#if UNITY_2017_3_OR_NEWER + if (m_TargetCameraComponent != null && RuntimeUtilities.IsDynamicResolutionEnabled(m_TargetCameraComponent)) + EditorGUILayout.HelpBox("TAA is not supported with Dynamic Resolution.", MessageType.Warning); +#endif + + EditorGUILayout.PropertyField(m_TaaJitterSpread); + EditorGUILayout.PropertyField(m_TaaStationaryBlending); + EditorGUILayout.PropertyField(m_TaaMotionBlending); + EditorGUILayout.PropertyField(m_TaaSharpness); + } + else if (m_AntialiasingMode.intValue == (int)PostProcessLayer.Antialiasing.SubpixelMorphologicalAntialiasing) + { + if (RuntimeUtilities.isSinglePassStereoSelected) + EditorGUILayout.HelpBox("SMAA doesn't work with Single-pass stereo rendering.", MessageType.Warning); + + EditorGUILayout.PropertyField(m_SmaaQuality); + + if (m_SmaaQuality.intValue != (int)SubpixelMorphologicalAntialiasing.Quality.Low && EditorUtilities.isTargetingConsolesOrMobiles) + EditorGUILayout.HelpBox("For performance reasons it is recommended to use Low Quality on mobile and console platforms.", MessageType.Warning); + } + else if (m_AntialiasingMode.intValue == (int)PostProcessLayer.Antialiasing.FastApproximateAntialiasing) + { + EditorGUILayout.PropertyField(m_FxaaFastMode); + EditorGUILayout.PropertyField(m_FxaaKeepAlpha); + + if (!m_FxaaFastMode.boolValue && EditorUtilities.isTargetingConsolesOrMobiles) + EditorGUILayout.HelpBox("For performance reasons it is recommended to use Fast Mode on mobile and console platforms.", MessageType.Warning); + } + else if (m_AntialiasingMode.intValue == (int)PostProcessLayer.Antialiasing.SuperResolution) + { + EditorGUILayout.PropertyField(m_FsrQualityMode); + EditorGUILayout.PropertyField(m_FsrPerformSharpen); + EditorGUILayout.PropertyField(m_FsrSharpness); + EditorGUILayout.PropertyField(m_FsrEnableFP16); + EditorGUILayout.PropertyField(m_FsrExposureSource); + if (m_FsrExposureSource.intValue == (int)SuperResolution.ExposureSource.Manual) EditorGUILayout.PropertyField(m_FsrExposureTexture); + EditorGUILayout.PropertyField(m_FsrPreExposure); + EditorGUILayout.PropertyField(m_FsrDebugView); + EditorGUILayout.PropertyField(m_FsrAutoReactive); + EditorGUILayout.PropertyField(m_FsrAutoReactive.boolValue ? m_FsrAutoReactiveParams : m_FsrReactiveMaskTexture); + EditorGUILayout.PropertyField(m_FsrAutoTcr); + EditorGUILayout.PropertyField(m_FsrAutoTcr.boolValue ? m_FsrAutoTcrParams : m_FsrTcrMaskTexture); + } + } + EditorGUI.indentLevel--; + + EditorGUILayout.Space(); + } + + void DoFog(Camera camera) + { + if (camera == null || camera.actualRenderingPath != RenderingPath.DeferredShading) + return; + + EditorGUILayout.LabelField(EditorUtilities.GetContent("Deferred Fog"), EditorStyles.boldLabel); + EditorGUI.indentLevel++; + { + EditorGUILayout.PropertyField(m_FogEnabled); + + if (m_FogEnabled.boolValue) + { + EditorGUILayout.PropertyField(m_FogExcludeSkybox); + EditorGUILayout.HelpBox("This adds fog compatibility to the deferred rendering path; actual fog settings should be set in the Lighting panel.", MessageType.Info); + } + } + EditorGUI.indentLevel--; + + EditorGUILayout.Space(); + } + + void DoToolkit() + { + EditorUtilities.DrawSplitter(); + m_ShowToolkit.boolValue = EditorUtilities.DrawHeader("Toolkit", m_ShowToolkit.boolValue); + + if (m_ShowToolkit.boolValue) + { + GUILayout.Space(2); + + if (GUILayout.Button(EditorUtilities.GetContent("Export frame to EXR..."), EditorStyles.miniButton)) + { + var menu = new GenericMenu(); + menu.AddItem(EditorUtilities.GetContent("Full Frame (as displayed)"), false, () => ExportFrameToExr(ExportMode.FullFrame)); + menu.AddItem(EditorUtilities.GetContent("Disable post-processing"), false, () => ExportFrameToExr(ExportMode.DisablePost)); + menu.AddItem(EditorUtilities.GetContent("Break before Color Grading (Linear)"), false, () => ExportFrameToExr(ExportMode.BreakBeforeColorGradingLinear)); + menu.AddItem(EditorUtilities.GetContent("Break before Color Grading (Log)"), false, () => ExportFrameToExr(ExportMode.BreakBeforeColorGradingLog)); + menu.ShowAsContext(); + } + + if (GUILayout.Button(EditorUtilities.GetContent("Select all layer volumes|Selects all the volumes that will influence this layer."), EditorStyles.miniButton)) + { + var volumes = RuntimeUtilities.GetAllSceneObjects() + .Where(x => (m_VolumeLayer.intValue & (1 << x.gameObject.layer)) != 0) + .Select(x => x.gameObject) + .Cast() + .ToArray(); + + if (volumes.Length > 0) + Selection.objects = volumes; + } + + if (GUILayout.Button(EditorUtilities.GetContent("Select all active volumes|Selects all volumes currently affecting the layer."), EditorStyles.miniButton)) + { + var volumes = new List(); + PostProcessManager.instance.GetActiveVolumes(m_Target, volumes); + + if (volumes.Count > 0) + { + Selection.objects = volumes + .Select(x => x.gameObject) + .Cast() + .ToArray(); + } + } + + GUILayout.Space(3); + } + } + + void DoCustomEffectSorter() + { + EditorUtilities.DrawSplitter(); + m_ShowCustomSorter.boolValue = EditorUtilities.DrawHeader("Custom Effect Sorting", m_ShowCustomSorter.boolValue); + + if (m_ShowCustomSorter.boolValue) + { + bool isInPrefab = false; + + // Init lists if needed + if (m_CustomLists == null) + { + // In some cases the editor will refresh before components which means + // components might not have been fully initialized yet. In this case we also + // need to make sure that we're not in a prefab as sorteBundles isn't a + // serializable object and won't exist until put on a scene. + if (m_Target.sortedBundles == null) + { + isInPrefab = string.IsNullOrEmpty(m_Target.gameObject.scene.name); + + if (!isInPrefab) + { + // sortedBundles will be initialized and ready to use on the next frame + Repaint(); + } + } + else + { + // Create a reorderable list for each injection event + m_CustomLists = new Dictionary(); + foreach (var evt in Enum.GetValues(typeof(PostProcessEvent)).Cast()) + { + var bundles = m_Target.sortedBundles[evt]; + var listName = ObjectNames.NicifyVariableName(evt.ToString()); + + var list = new ReorderableList(bundles, typeof(SerializedBundleRef), true, true, false, false); + + list.drawHeaderCallback = (rect) => + { + EditorGUI.LabelField(rect, listName); + }; + + list.drawElementCallback = (rect, index, isActive, isFocused) => + { + var sbr = (SerializedBundleRef)list.list[index]; + EditorGUI.LabelField(rect, sbr.bundle.attribute.menuItem); + }; + + list.onReorderCallback = (l) => + { + EditorUtility.SetDirty(m_Target); + }; + + m_CustomLists.Add(evt, list); + } + } + } + + GUILayout.Space(5); + + if (isInPrefab) + { + EditorGUILayout.HelpBox("Not supported in prefabs.", MessageType.Info); + GUILayout.Space(3); + return; + } + + bool anyList = false; + if (m_CustomLists != null) + { + foreach (var kvp in m_CustomLists) + { + var list = kvp.Value; + + // Skip empty lists to avoid polluting the inspector + if (list.count == 0) + continue; + + list.DoLayoutList(); + anyList = true; + } + } + + if (!anyList) + { + EditorGUILayout.HelpBox("No custom effect loaded.", MessageType.Info); + GUILayout.Space(3); + } + } + } + + void ExportFrameToExr(ExportMode mode) + { + string path = EditorUtility.SaveFilePanel("Export EXR...", "", "Frame", "exr"); + + if (string.IsNullOrEmpty(path)) + return; + + EditorUtility.DisplayProgressBar("Export EXR", "Rendering...", 0f); + + var camera = m_Target.GetComponent(); + var w = camera.pixelWidth; + var h = camera.pixelHeight; + + var texOut = new Texture2D(w, h, TextureFormat.RGBAFloat, false, true); + var target = RenderTexture.GetTemporary(w, h, 24, RenderTextureFormat.ARGBFloat, RenderTextureReadWrite.Linear); + + var lastActive = RenderTexture.active; + var lastTargetSet = camera.targetTexture; + var lastPostFXState = m_Target.enabled; + var lastBreakColorGradingState = m_Target.breakBeforeColorGrading; + + if (mode == ExportMode.DisablePost) + m_Target.enabled = false; + else if (mode == ExportMode.BreakBeforeColorGradingLinear || mode == ExportMode.BreakBeforeColorGradingLog) + m_Target.breakBeforeColorGrading = true; + + camera.targetTexture = target; + camera.Render(); + camera.targetTexture = lastTargetSet; + + EditorUtility.DisplayProgressBar("Export EXR", "Reading...", 0.25f); + + m_Target.enabled = lastPostFXState; + m_Target.breakBeforeColorGrading = lastBreakColorGradingState; + + if (mode == ExportMode.BreakBeforeColorGradingLog) + { + // Convert to log + var material = new Material(Shader.Find("Hidden/PostProcessing/Editor/ConvertToLog")); + var newTarget = RenderTexture.GetTemporary(w, h, 0, RenderTextureFormat.ARGBFloat, RenderTextureReadWrite.Linear); + Graphics.Blit(target, newTarget, material, 0); + RenderTexture.ReleaseTemporary(target); + DestroyImmediate(material); + target = newTarget; + } + + RenderTexture.active = target; + texOut.ReadPixels(new Rect(0, 0, w, h), 0, 0); + texOut.Apply(); + RenderTexture.active = lastActive; + + EditorUtility.DisplayProgressBar("Export EXR", "Encoding...", 0.5f); + + var bytes = texOut.EncodeToEXR(EXRFlags.OutputAsFloat | EXRFlags.CompressZIP); + + EditorUtility.DisplayProgressBar("Export EXR", "Saving...", 0.75f); + + File.WriteAllBytes(path, bytes); + + EditorUtility.ClearProgressBar(); + AssetDatabase.Refresh(); + + RenderTexture.ReleaseTemporary(target); + DestroyImmediate(texOut); + } + } +} diff --git a/com.unity.postprocessing/PostProcessing/Editor/PostProcessLayerEditor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessLayerEditor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/PostProcessLayerEditor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessLayerEditor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/PostProcessProfileEditor.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessProfileEditor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/PostProcessProfileEditor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessProfileEditor.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/PostProcessProfileEditor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessProfileEditor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/PostProcessProfileEditor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessProfileEditor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/PostProcessVolumeEditor.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessVolumeEditor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/PostProcessVolumeEditor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessVolumeEditor.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/PostProcessVolumeEditor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessVolumeEditor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/PostProcessVolumeEditor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/PostProcessVolumeEditor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Tools.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Tools.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Tools.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Tools.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Tools/CubeLutAssetFactory.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/CubeLutAssetFactory.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Tools/CubeLutAssetFactory.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/CubeLutAssetFactory.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Tools/CubeLutAssetFactory.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/CubeLutAssetFactory.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Tools/CubeLutAssetFactory.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/CubeLutAssetFactory.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Tools/CubeLutAssetImporter.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/CubeLutAssetImporter.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Tools/CubeLutAssetImporter.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/CubeLutAssetImporter.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Tools/CubeLutAssetImporter.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/CubeLutAssetImporter.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Tools/CubeLutAssetImporter.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/CubeLutAssetImporter.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Tools/DefineSetter.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/DefineSetter.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Tools/DefineSetter.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/DefineSetter.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Tools/DefineSetter.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/DefineSetter.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Tools/DefineSetter.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/DefineSetter.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Tools/ProfileFactory.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/ProfileFactory.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Tools/ProfileFactory.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/ProfileFactory.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Tools/ProfileFactory.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/ProfileFactory.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Tools/ProfileFactory.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/ProfileFactory.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Tools/ResourceAssetFactory.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/ResourceAssetFactory.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Tools/ResourceAssetFactory.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/ResourceAssetFactory.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Tools/ResourceAssetFactory.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/ResourceAssetFactory.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Tools/ResourceAssetFactory.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/ResourceAssetFactory.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Tools/VolumeFactory.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/VolumeFactory.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Tools/VolumeFactory.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/VolumeFactory.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Tools/VolumeFactory.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/VolumeFactory.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Tools/VolumeFactory.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Tools/VolumeFactory.cs.meta diff --git a/Packages/com.unity.postprocessing/PostProcessing/Editor/Unity.Postprocessing.Editor.asmdef b/Packages/com.unity.postprocessing/PostProcessing/Editor/Unity.Postprocessing.Editor.asmdef new file mode 100644 index 0000000..f90cfb2 --- /dev/null +++ b/Packages/com.unity.postprocessing/PostProcessing/Editor/Unity.Postprocessing.Editor.asmdef @@ -0,0 +1,37 @@ +{ + "name": "Unity.Postprocessing.Editor", + "rootNamespace": "", + "references": [ + "Unity.Postprocessing.Runtime", + "Unity.XR.Management", + "Unity.XR.Management.Editor", + "FidelityFX.FSR" + ], + "includePlatforms": [ + "Editor" + ], + "excludePlatforms": [], + "allowUnsafeCode": false, + "overrideReferences": false, + "precompiledReferences": [], + "autoReferenced": true, + "defineConstraints": [], + "versionDefines": [ + { + "name": "com.unity.xr.management", + "expression": "4.0.1", + "define": "XR_MANAGEMENT_4_0_1_OR_NEWER" + }, + { + "name": "com.unity.modules.vr", + "expression": "1.0.0", + "define": "ENABLE_VR_MODULE" + }, + { + "name": "com.unity.modules.xr", + "expression": "1.0.0", + "define": "ENABLE_XR_MODULE" + } + ], + "noEngineReferences": false +} diff --git a/com.unity.postprocessing/PostProcessing/Editor/Unity.Postprocessing.Editor.asmdef.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Unity.Postprocessing.Editor.asmdef.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Unity.Postprocessing.Editor.asmdef.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Unity.Postprocessing.Editor.asmdef.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Utils.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Utils.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Utils.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Utils.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Utils/CurveEditor.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Utils/CurveEditor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Utils/CurveEditor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Utils/CurveEditor.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Utils/CurveEditor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Utils/CurveEditor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Utils/CurveEditor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Utils/CurveEditor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Utils/EditorUtilities.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Utils/EditorUtilities.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Utils/EditorUtilities.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Utils/EditorUtilities.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Utils/EditorUtilities.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Utils/EditorUtilities.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Utils/EditorUtilities.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Utils/EditorUtilities.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Utils/GlobalSettings.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Utils/GlobalSettings.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Utils/GlobalSettings.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Utils/GlobalSettings.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Utils/GlobalSettings.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Utils/GlobalSettings.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Utils/GlobalSettings.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Utils/GlobalSettings.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Utils/SerializedParameterOverride.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Utils/SerializedParameterOverride.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Utils/SerializedParameterOverride.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Utils/SerializedParameterOverride.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Utils/SerializedParameterOverride.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Utils/SerializedParameterOverride.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Utils/SerializedParameterOverride.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Utils/SerializedParameterOverride.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Editor/Utils/Styling.cs b/Packages/com.unity.postprocessing/PostProcessing/Editor/Utils/Styling.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Utils/Styling.cs rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Utils/Styling.cs diff --git a/com.unity.postprocessing/PostProcessing/Editor/Utils/Styling.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Editor/Utils/Styling.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Editor/Utils/Styling.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Editor/Utils/Styling.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Gizmos.meta b/Packages/com.unity.postprocessing/PostProcessing/Gizmos.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Gizmos.meta rename to Packages/com.unity.postprocessing/PostProcessing/Gizmos.meta diff --git a/com.unity.postprocessing/PostProcessing/Gizmos/PostProcessLayer.png b/Packages/com.unity.postprocessing/PostProcessing/Gizmos/PostProcessLayer.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Gizmos/PostProcessLayer.png rename to Packages/com.unity.postprocessing/PostProcessing/Gizmos/PostProcessLayer.png diff --git a/com.unity.postprocessing/PostProcessing/Gizmos/PostProcessLayer.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Gizmos/PostProcessLayer.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Gizmos/PostProcessLayer.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Gizmos/PostProcessLayer.png.meta diff --git a/Packages/com.unity.postprocessing/PostProcessing/PostProcessResources.asset b/Packages/com.unity.postprocessing/PostProcessing/PostProcessResources.asset new file mode 100644 index 0000000..c9f5acb --- /dev/null +++ b/Packages/com.unity.postprocessing/PostProcessing/PostProcessResources.asset @@ -0,0 +1,141 @@ +%YAML 1.1 +%TAG !u! tag:unity3d.com,2011: +--- !u!114 &11400000 +MonoBehaviour: + m_ObjectHideFlags: 0 + m_CorrespondingSourceObject: {fileID: 0} + m_PrefabInstance: {fileID: 0} + m_PrefabAsset: {fileID: 0} + m_GameObject: {fileID: 0} + m_Enabled: 1 + m_EditorHideFlags: 0 + m_Script: {fileID: 11500000, guid: 30f4b897495c7ad40b2d47143e02aaba, type: 3} + m_Name: PostProcessResources + m_EditorClassIdentifier: + blueNoise64: + - {fileID: 2800000, guid: 50b54341495978843a6f85583ed4417d, type: 3} + - {fileID: 2800000, guid: 3c2f1fb7e4b66e74191b7c328ada52d9, type: 3} + - {fileID: 2800000, guid: a469f920b21fc7c4fb5b950917ce2fb2, type: 3} + - {fileID: 2800000, guid: 373f9bf6b0841af4ebf26d25e4a3f4e2, type: 3} + - {fileID: 2800000, guid: 6fa5cf178eaaa5f42b820f636bb6e0bd, type: 3} + - {fileID: 2800000, guid: a1ae041906217ae44a774d4ca139af50, type: 3} + - {fileID: 2800000, guid: 79b86f3419b87f3429164a956da8cfab, type: 3} + - {fileID: 2800000, guid: 3ac02e7e783571c468f9c086d2384ba7, type: 3} + - {fileID: 2800000, guid: c55042318a938344ab23cd7f09dd0076, type: 3} + - {fileID: 2800000, guid: 71583cfd8899717428d5b1a95fa39cda, type: 3} + - {fileID: 2800000, guid: afe1e502240079342a0a980484b6da8b, type: 3} + - {fileID: 2800000, guid: 771903fe7b4674445829e52e91cff019, type: 3} + - {fileID: 2800000, guid: 980acadb960f8424c94307ec0e585b4e, type: 3} + - {fileID: 2800000, guid: 68613e6a221be1a4b9f31d7fa1c2d1bf, type: 3} + - {fileID: 2800000, guid: f6439b54b28f3884eb67579dec0b6f21, type: 3} + - {fileID: 2800000, guid: 2ee161d8945169243b5698fec114e1b7, type: 3} + - {fileID: 2800000, guid: 153f7d6dfbe713d4884df0f1e243ba92, type: 3} + - {fileID: 2800000, guid: bf95b6fdc179b0e4f890c841406193fc, type: 3} + - {fileID: 2800000, guid: 74aca53eb7273624baffc2bf5e5cc173, type: 3} + - {fileID: 2800000, guid: 729a3ae164bcb3b4380459386adcf331, type: 3} + - {fileID: 2800000, guid: 6dda07f1420a968449cf4c6620c44d9f, type: 3} + - {fileID: 2800000, guid: b7f000750830ddb4bbc80065b9314ce9, type: 3} + - {fileID: 2800000, guid: df01d03f056c6f445b4b8a0ae054207c, type: 3} + - {fileID: 2800000, guid: bfe953600e8fb1849a804ee08ace7b4c, type: 3} + - {fileID: 2800000, guid: 32c6a5f7143b86c44bd5cdee2ff3f8ad, type: 3} + - {fileID: 2800000, guid: f4b8ab78b57749d4e96d36f6d8a395d0, type: 3} + - {fileID: 2800000, guid: 09f6c01f98a3ded4daf1afc52a3c260f, type: 3} + - {fileID: 2800000, guid: bdd06fb88ef36ed4a85dd506352c2d80, type: 3} + - {fileID: 2800000, guid: 02c0a84bd64c6f044954d8bde9b46ec8, type: 3} + - {fileID: 2800000, guid: aa80dc44aa4fe4c43bb9d51d90cf2958, type: 3} + - {fileID: 2800000, guid: 0fa10b21877c61b4db40ba5708815f81, type: 3} + - {fileID: 2800000, guid: 6b0a189df0bd4d5448eaefb4e673ace8, type: 3} + - {fileID: 2800000, guid: 87a5e40cc271ea648b583616f6ebe7fe, type: 3} + - {fileID: 2800000, guid: b71bb466b71fd13449dd736f63caeb67, type: 3} + - {fileID: 2800000, guid: 319b8e66db3faa4438cf6982e9c89b2f, type: 3} + - {fileID: 2800000, guid: 0a79c155edf9b2d429d4736abee5acdb, type: 3} + - {fileID: 2800000, guid: 351e95d0e20a54849bd4ce5f9b498934, type: 3} + - {fileID: 2800000, guid: 1d6958e30e40a254dbe5a54c573eeb3c, type: 3} + - {fileID: 2800000, guid: 9660a4ca1ca8425408ac25c641932977, type: 3} + - {fileID: 2800000, guid: 547dbd5f858c74047ba3f213e4408307, type: 3} + - {fileID: 2800000, guid: 1a9ce5640cde5934aae0022f020464a6, type: 3} + - {fileID: 2800000, guid: cd9006dc442cc244e89b3f492384d46a, type: 3} + - {fileID: 2800000, guid: b266511438fae724f9d3ce6bd26583e8, type: 3} + - {fileID: 2800000, guid: 71bc1b6b66e8b784b972199b7e90204e, type: 3} + - {fileID: 2800000, guid: 15e54aa23a938444389469d53765d741, type: 3} + - {fileID: 2800000, guid: b9960364038cbfa4aa49d7b2032d3110, type: 3} + - {fileID: 2800000, guid: 8ecbbcae4cc747a4abbc4adce795d25e, type: 3} + - {fileID: 2800000, guid: 1378a33cdd085d64c9da863d2484ff21, type: 3} + - {fileID: 2800000, guid: aff59c63d25d43f4c938f248837c30fb, type: 3} + - {fileID: 2800000, guid: 3f7c3687170b90e4a8d2ee6b142670f4, type: 3} + - {fileID: 2800000, guid: d8c290e38ff0425409d0ae6a98c95e41, type: 3} + - {fileID: 2800000, guid: d5a51525b27e3ee4aadbeb39cbcf0750, type: 3} + - {fileID: 2800000, guid: d2e8e90fac2e6a341a38e1c3963c218d, type: 3} + - {fileID: 2800000, guid: c94b57b5a32a22d43ade66e09f6a4bd2, type: 3} + - {fileID: 2800000, guid: 936dea238abb0864ab3985a995e16a29, type: 3} + - {fileID: 2800000, guid: 5e542d0126a2c7848b66bffc428905fd, type: 3} + - {fileID: 2800000, guid: 70f23eaf7d8ae9147aa542d20e93733b, type: 3} + - {fileID: 2800000, guid: e138166e7a7c70f49943be7edda35d35, type: 3} + - {fileID: 2800000, guid: 85a45a6d8b2ffb84987d2b028ecfb220, type: 3} + - {fileID: 2800000, guid: d96974690c77f50489eb60ec84bd8dac, type: 3} + - {fileID: 2800000, guid: 404fa8def46b1c447817e1ebdaa7144e, type: 3} + - {fileID: 2800000, guid: 119591e0bb084e848835d237546b3882, type: 3} + - {fileID: 2800000, guid: a03c400b0e3959f428ee99dfc6cfc263, type: 3} + - {fileID: 2800000, guid: 4a11d65ce13d5f542a0ff136cc2f3fba, type: 3} + blueNoise256: + - {fileID: 2800000, guid: 6017f374382d64245a0a4aab668e6f38, type: 3} + - {fileID: 2800000, guid: 0f8fa14b3731cda4e947062e734d5e1e, type: 3} + - {fileID: 2800000, guid: 1abfe0e165ca1e9428b455ffc9a2d9ef, type: 3} + - {fileID: 2800000, guid: c072b653e98a06e40857d76ca8c7eecd, type: 3} + - {fileID: 2800000, guid: b52d5033b68309943a2386c270a90f44, type: 3} + - {fileID: 2800000, guid: acde5141d5f4f7a4188394bd52c4dc38, type: 3} + - {fileID: 2800000, guid: 999434725cbc2be4eb54043b36efd4a8, type: 3} + - {fileID: 2800000, guid: 70d0a1182b29d6347ac70374c3593bba, type: 3} + smaaLuts: + area: {fileID: 2800000, guid: 73ec4ae984a0a0f44a2be737e41a6f2f, type: 3} + search: {fileID: 2800000, guid: d99701099481a2f489610e977df6dcbc, type: 3} + shaders: + bloom: {fileID: 4800000, guid: c1e1d3119c6fd4646aea0b4b74cacc1a, type: 3} + copy: {fileID: 4800000, guid: cdbdb71de5f9c454b980f6d0e87f0afb, type: 3} + copyStd: {fileID: 4800000, guid: 4bf4cff0d0bac3d43894e2e8839feb40, type: 3} + copyStdFromTexArray: {fileID: 4800000, guid: 02d2da9bc88d25c4d878c1ed4e0b3854, type: 3} + copyStdFromDoubleWide: {fileID: 4800000, guid: e8ce9961912f3214586fe8709b9012c1, type: 3} + discardAlpha: {fileID: 4800000, guid: 5ab0816423f0dfe45841cab3b05ec9ef, type: 3} + depthOfField: {fileID: 4800000, guid: 0ef78d24e85a44f4da9d5b5eaa00e50b, type: 3} + finalPass: {fileID: 4800000, guid: f75014305794b3948a3c6d5ccd550e05, type: 3} + grainBaker: {fileID: 4800000, guid: 0d8afcb51cc9f0349a6d190da929b838, type: 3} + motionBlur: {fileID: 4800000, guid: 2c459b89a7c8b1a4fbefe0d81341651c, type: 3} + temporalAntialiasing: {fileID: 4800000, guid: 51bcf79c50dc92e47ba87821b61100c3, type: 3} + subpixelMorphologicalAntialiasing: {fileID: 4800000, guid: 81af42a93ade3dd46a9b583d4eec76d6, type: 3} + texture2dLerp: {fileID: 4800000, guid: 34a819c9e33402547a81619693adc8d5, type: 3} + uber: {fileID: 4800000, guid: 382151503e2a43a4ebb7366d1632731d, type: 3} + lut2DBaker: {fileID: 4800000, guid: 7ad194cbe7d006f4bace915156972026, type: 3} + lightMeter: {fileID: 4800000, guid: b34a29e523cb9d545881e193a079f2df, type: 3} + gammaHistogram: {fileID: 4800000, guid: f7ea35cfb33fcad4ab8f2429ec103bef, type: 3} + waveform: {fileID: 4800000, guid: 3020ac7ece79a7f4eb789a236f8bd6c5, type: 3} + vectorscope: {fileID: 4800000, guid: a71093f2a4fe26a40805c22739e10e4a, type: 3} + debugOverlays: {fileID: 4800000, guid: b958ad1c92bd3d64c9e61318b8681dab, type: 3} + deferredFog: {fileID: 4800000, guid: 4117fce9491711c4094d33a048e36e73, type: 3} + scalableAO: {fileID: 4800000, guid: d7640629310e79646af0f46eb55ae466, type: 3} + multiScaleAO: {fileID: 4800000, guid: 67f9497810829eb4791ec19e95781e51, type: 3} + screenSpaceReflections: {fileID: 4800000, guid: f997a3dc9254c44459323cced085150c, type: 3} + computeShaders: + autoExposure: {fileID: 7200000, guid: 34845e0ca016b7448842e965db5890a5, type: 3} + exposureHistogram: {fileID: 7200000, guid: 8c2fcbdf9bc58664f89917f7b9d79501, type: 3} + lut3DBaker: {fileID: 7200000, guid: 42496b74c071f5749950ca1abe33e945, type: 3} + texture3dLerp: {fileID: 7200000, guid: 31e9175024adfd44aba2530ff9b77494, type: 3} + gammaHistogram: {fileID: 7200000, guid: 18183ebfeeab97749b43e38b928604a7, type: 3} + waveform: {fileID: 7200000, guid: 92c63830cd50c0b4fbb8233613839958, type: 3} + vectorscope: {fileID: 7200000, guid: e1efca7c36fd01840aae0dd10378de5c, type: 3} + multiScaleAODownsample1: {fileID: 7200000, guid: 4c63bc487e6c29a4a99f85a6c47b292b, type: 3} + multiScaleAODownsample2: {fileID: 7200000, guid: e4d3e4779e48a374f91d48d4c0aedb7b, type: 3} + multiScaleAORender: {fileID: 7200000, guid: 34a460e8a2e66c243a9c12024e5a798d, type: 3} + multiScaleAOUpsample: {fileID: 7200000, guid: 600d6212b59bb40409d19d750b5fd1e9, type: 3} + gaussianDownsample: {fileID: 7200000, guid: 6dba4103d23a7904fbc49099355aff3e, type: 3} + superResolution: + prepareInputsPass: {fileID: 7200000, guid: 4f59e5b9179d74844ae06a30ae1e0629, type: 3} + lumaPyramidPass: {fileID: 7200000, guid: d253be05abcdc80428503d3e4cce3a36, type: 3} + shadingChangePyramidPass: {fileID: 7200000, guid: 251e663738905fa4d8817001682d802f, type: 3} + shadingChangePass: {fileID: 7200000, guid: 9a2bff2f97619ed4989d9b0577ba0641, type: 3} + prepareReactivityPass: {fileID: 7200000, guid: 20e44016ed34b0d4b8de499d1b566c69, type: 3} + lumaInstabilityPass: {fileID: 7200000, guid: a135306e6d1857e43a86ef20db2a47fe, type: 3} + accumulatePass: {fileID: 7200000, guid: c9b45f0ae7673694ba57a4aadfe212e9, type: 3} + sharpenPass: {fileID: 7200000, guid: 7aaf5cfff022de2499e9b0412f947f6c, type: 3} + autoGenReactivePass: {fileID: 7200000, guid: 5716b91fdaa4e9e439df6b96a796fe6e, type: 3} + tcrAutoGenPass: {fileID: 7200000, guid: 75cdc6ef23f08ed498d4da511923fcea, type: 3} + debugViewPass: {fileID: 7200000, guid: cb24a71d54164c54eb5e86839acd48c5, type: 3} diff --git a/com.unity.postprocessing/PostProcessing/PostProcessResources.asset.meta b/Packages/com.unity.postprocessing/PostProcessing/PostProcessResources.asset.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/PostProcessResources.asset.meta rename to Packages/com.unity.postprocessing/PostProcessing/PostProcessResources.asset.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Attributes.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Attributes.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Attributes/DisplayNameAttribute.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/DisplayNameAttribute.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Attributes/DisplayNameAttribute.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/DisplayNameAttribute.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Attributes/DisplayNameAttribute.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/DisplayNameAttribute.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Attributes/DisplayNameAttribute.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/DisplayNameAttribute.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Attributes/MaxAttribute.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/MaxAttribute.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Attributes/MaxAttribute.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/MaxAttribute.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Attributes/MaxAttribute.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/MaxAttribute.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Attributes/MaxAttribute.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/MaxAttribute.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Attributes/MinAttribute.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/MinAttribute.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Attributes/MinAttribute.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/MinAttribute.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Attributes/MinAttribute.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/MinAttribute.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Attributes/MinAttribute.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/MinAttribute.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Attributes/MinMaxAttribute.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/MinMaxAttribute.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Attributes/MinMaxAttribute.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/MinMaxAttribute.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Attributes/MinMaxAttribute.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/MinMaxAttribute.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Attributes/MinMaxAttribute.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/MinMaxAttribute.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Attributes/PostProcessAttribute.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/PostProcessAttribute.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Attributes/PostProcessAttribute.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/PostProcessAttribute.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Attributes/PostProcessAttribute.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/PostProcessAttribute.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Attributes/PostProcessAttribute.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/PostProcessAttribute.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Attributes/TrackballAttribute.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/TrackballAttribute.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Attributes/TrackballAttribute.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/TrackballAttribute.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Attributes/TrackballAttribute.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/TrackballAttribute.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Attributes/TrackballAttribute.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Attributes/TrackballAttribute.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/AmbientOcclusion.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/AmbientOcclusion.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/AmbientOcclusion.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/AmbientOcclusion.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/AmbientOcclusion.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/AmbientOcclusion.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/AmbientOcclusion.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/AmbientOcclusion.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/AutoExposure.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/AutoExposure.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/AutoExposure.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/AutoExposure.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/AutoExposure.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/AutoExposure.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/AutoExposure.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/AutoExposure.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/Bloom.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/Bloom.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/Bloom.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/Bloom.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/Bloom.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/Bloom.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/Bloom.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/Bloom.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/ChromaticAberration.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/ChromaticAberration.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/ChromaticAberration.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/ChromaticAberration.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/ChromaticAberration.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/ChromaticAberration.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/ChromaticAberration.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/ChromaticAberration.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/ColorGrading.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/ColorGrading.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/ColorGrading.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/ColorGrading.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/ColorGrading.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/ColorGrading.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/ColorGrading.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/ColorGrading.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/DepthOfField.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/DepthOfField.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/DepthOfField.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/DepthOfField.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/DepthOfField.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/DepthOfField.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/DepthOfField.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/DepthOfField.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/Dithering.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/Dithering.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/Dithering.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/Dithering.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/Dithering.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/Dithering.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/Dithering.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/Dithering.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/FastApproximateAntialiasing.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/FastApproximateAntialiasing.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/FastApproximateAntialiasing.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/FastApproximateAntialiasing.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/FastApproximateAntialiasing.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/FastApproximateAntialiasing.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/FastApproximateAntialiasing.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/FastApproximateAntialiasing.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/Fog.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/Fog.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/Fog.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/Fog.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/Fog.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/Fog.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/Fog.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/Fog.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/Grain.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/Grain.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/Grain.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/Grain.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/Grain.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/Grain.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/Grain.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/Grain.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/LensDistortion.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/LensDistortion.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/LensDistortion.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/LensDistortion.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/LensDistortion.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/LensDistortion.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/LensDistortion.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/LensDistortion.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/MotionBlur.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/MotionBlur.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/MotionBlur.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/MotionBlur.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/MotionBlur.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/MotionBlur.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/MotionBlur.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/MotionBlur.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/MultiScaleVO.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/MultiScaleVO.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/MultiScaleVO.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/MultiScaleVO.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/MultiScaleVO.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/MultiScaleVO.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/MultiScaleVO.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/MultiScaleVO.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/ScalableAO.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/ScalableAO.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/ScalableAO.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/ScalableAO.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/ScalableAO.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/ScalableAO.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/ScalableAO.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/ScalableAO.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/ScreenSpaceReflections.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/ScreenSpaceReflections.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/ScreenSpaceReflections.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/ScreenSpaceReflections.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/ScreenSpaceReflections.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/ScreenSpaceReflections.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/ScreenSpaceReflections.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/ScreenSpaceReflections.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/SubpixelMorphologicalAntialiasing.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/SubpixelMorphologicalAntialiasing.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/SubpixelMorphologicalAntialiasing.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/SubpixelMorphologicalAntialiasing.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/SubpixelMorphologicalAntialiasing.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/SubpixelMorphologicalAntialiasing.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/SubpixelMorphologicalAntialiasing.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/SubpixelMorphologicalAntialiasing.cs.meta diff --git a/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/SuperResolution.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/SuperResolution.cs new file mode 100644 index 0000000..e997b3f --- /dev/null +++ b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/SuperResolution.cs @@ -0,0 +1,339 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +using System; +using System.Collections; +using System.Collections.Generic; +using UnityEngine; +using UnityEngine.Experimental.Rendering; +using FidelityFX; +using FidelityFX.FSR3; + +namespace UnityEngine.Rendering.PostProcessing +{ + [UnityEngine.Scripting.Preserve] + [Serializable] + public class SuperResolution + { + public Func callbacksFactory { get; set; } = (context) => new Fsr3UpscalerCallbacksBase(); + + [Tooltip("Standard scaling ratio presets.")] + public Fsr3Upscaler.QualityMode qualityMode = Fsr3Upscaler.QualityMode.Quality; + + [Tooltip("Apply RCAS sharpening to the image after upscaling.")] + public bool performSharpenPass = true; + [Tooltip("Strength of the sharpening effect.")] + [Range(0, 1)] public float sharpness = 0.8f; + + [Tooltip("Allow the use of half precision compute operations, potentially improving performance if the platform supports it.")] + public bool enableFP16 = false; + + [Tooltip("Choose where to get the exposure value from. Use auto-exposure from either FSR3 or Unity, provide a manual exposure texture, or use a default value.")] + public ExposureSource exposureSource = ExposureSource.Auto; + [Tooltip("Value by which the input signal will be divided, to get back to the original signal produced by the game.")] + public float preExposure = 1.0f; + [Tooltip("Optional 1x1 texture containing the exposure value for the current frame.")] + public Texture exposure = null; + + public enum ExposureSource + { + Default, + Auto, + Unity, + Manual, + } + + [Tooltip("Enable a debug view to analyze the upscaling process.")] + public bool enableDebugView = false; + + [Tooltip("Optional texture to control the influence of the current frame on the reconstructed output. If unset, either an auto-generated or a default cleared reactive mask will be used.")] + public Texture reactiveMask = null; + [Tooltip("Optional texture for marking areas of specialist rendering which should be accounted for during the upscaling process. If unset, a default cleared mask will be used.")] + public Texture transparencyAndCompositionMask = null; + [Tooltip("Automatically generate a reactive mask based on the difference between opaque-only render output and the final render output including alpha transparencies.")] + public bool autoGenerateReactiveMask = true; + [Tooltip("Parameters to control the process of auto-generating a reactive mask.")] + public GenerateReactiveParameters generateReactiveParameters = new GenerateReactiveParameters(); + + [Serializable] + public class GenerateReactiveParameters + { + [Tooltip("A value to scale the output")] + [Range(0, 2)] public float scale = 0.5f; + [Tooltip("A threshold value to generate a binary reactive mask")] + [Range(0, 1)] public float cutoffThreshold = 0.2f; + [Tooltip("A value to set for the binary reactive mask")] + [Range(0, 1)] public float binaryValue = 0.9f; + [Tooltip("Flags to determine how to generate the reactive mask")] + public Fsr3Upscaler.GenerateReactiveFlags flags = Fsr3Upscaler.GenerateReactiveFlags.ApplyTonemap | Fsr3Upscaler.GenerateReactiveFlags.ApplyThreshold | Fsr3Upscaler.GenerateReactiveFlags.UseComponentsMax; + } + + [Tooltip("(Experimental) Automatically generate and use Reactive mask and Transparency & composition mask internally.")] + public bool autoGenerateTransparencyAndComposition = false; + [Tooltip("Parameters to control the process of auto-generating transparency and composition masks.")] + public GenerateTcrParameters generateTransparencyAndCompositionParameters = new GenerateTcrParameters(); + + [Serializable] + public class GenerateTcrParameters + { + [Tooltip("Setting this value too small will cause visual instability. Larger values can cause ghosting.")] + [Range(0, 1)] public float autoTcThreshold = 0.05f; + [Tooltip("Smaller values will increase stability at hard edges of translucent objects.")] + [Range(0, 2)] public float autoTcScale = 1.0f; + [Tooltip("Larger values result in more reactive pixels.")] + [Range(0, 10)] public float autoReactiveScale = 5.0f; + [Tooltip("Maximum value reactivity can reach.")] + [Range(0, 1)] public float autoReactiveMax = 0.9f; + } + + public Vector2 jitter { get; private set; } + public Vector2Int renderSize => _maxRenderSize; + public Vector2Int displaySize => _displaySize; + public RenderTargetIdentifier colorOpaqueOnly { get; set; } + + private Fsr3UpscalerContext _fsrContext; + private Vector2Int _maxRenderSize; + private Vector2Int _displaySize; + private bool _resetHistory; + + private IFsr3UpscalerCallbacks _callbacks; + + private readonly Fsr3Upscaler.DispatchDescription _dispatchDescription = new Fsr3Upscaler.DispatchDescription(); + private readonly Fsr3Upscaler.GenerateReactiveDescription _genReactiveDescription = new Fsr3Upscaler.GenerateReactiveDescription(); + + private Fsr3Upscaler.QualityMode _prevQualityMode; + private ExposureSource _prevExposureSource; + private Vector2Int _prevDisplaySize; + + private Rect _originalRect; + + public bool IsSupported() + { + return SystemInfo.supportsComputeShaders && SystemInfo.supportsMotionVectors; + } + + public DepthTextureMode GetCameraFlags() + { + return DepthTextureMode.Depth | DepthTextureMode.MotionVectors; + } + + public void Release() + { + DestroyFsrContext(); + } + + public void ResetHistory() + { + _resetHistory = true; + } + + public void ConfigureJitteredProjectionMatrix(PostProcessRenderContext context) + { + ApplyJitter(context.camera); + } + + public void ConfigureCameraViewport(PostProcessRenderContext context) + { + var camera = context.camera; + _originalRect = camera.rect; + + // Determine the desired rendering and display resolutions + _displaySize = new Vector2Int(camera.pixelWidth, camera.pixelHeight); + Fsr3Upscaler.GetRenderResolutionFromQualityMode(out int maxRenderWidth, out int maxRenderHeight, _displaySize.x, _displaySize.y, qualityMode); + _maxRenderSize = new Vector2Int(maxRenderWidth, maxRenderHeight); + + // Render to a smaller portion of the screen by manipulating the camera's viewport rect + camera.aspect = (float)_displaySize.x / _displaySize.y; + camera.rect = new Rect(0, 0, _originalRect.width * _maxRenderSize.x / _displaySize.x, _originalRect.height * _maxRenderSize.y / _displaySize.y); + } + + public void ResetCameraViewport(PostProcessRenderContext context) + { + context.camera.rect = _originalRect; + } + + public void Render(PostProcessRenderContext context) + { + var cmd = context.command; + cmd.BeginSample("FSR3 Upscaler"); + + // Monitor for any resolution changes and recreate the FSR3 Upscaler context if necessary + // We can't create an FSR3 Upscaler context without info from the post-processing context, so delay the initial setup until here + if (_fsrContext == null || _displaySize.x != _prevDisplaySize.x || _displaySize.y != _prevDisplaySize.y || qualityMode != _prevQualityMode || exposureSource != _prevExposureSource) + { + DestroyFsrContext(); + CreateFsrContext(context); + } + + SetupDispatchDescription(context); + + if (autoGenerateReactiveMask) + { + SetupAutoReactiveDescription(context); + + var scaledRenderSize = _genReactiveDescription.RenderSize; + cmd.GetTemporaryRT(Fsr3ShaderIDs.UavAutoReactive, scaledRenderSize.x, scaledRenderSize.y, 0, default, GraphicsFormat.R8_UNorm, 1, true); + _fsrContext.GenerateReactiveMask(_genReactiveDescription, cmd); + _dispatchDescription.Reactive = new ResourceView(Fsr3ShaderIDs.UavAutoReactive); + } + + _fsrContext.Dispatch(_dispatchDescription, cmd); + + cmd.EndSample("FSR3 Upscaler"); + + _resetHistory = false; + } + + private void CreateFsrContext(PostProcessRenderContext context) + { + _prevQualityMode = qualityMode; + _prevExposureSource = exposureSource; + _prevDisplaySize = _displaySize; + + // Initialize FSR3 Upscaler context + Fsr3Upscaler.InitializationFlags flags = 0; + if (context.camera.allowHDR) flags |= Fsr3Upscaler.InitializationFlags.EnableHighDynamicRange; + if (enableFP16) flags |= Fsr3Upscaler.InitializationFlags.EnableFP16Usage; + if (exposureSource == ExposureSource.Auto) flags |= Fsr3Upscaler.InitializationFlags.EnableAutoExposure; + if (RuntimeUtilities.IsDynamicResolutionEnabled(context.camera)) flags |= Fsr3Upscaler.InitializationFlags.EnableDynamicResolution; + + _callbacks = callbacksFactory(context); + _fsrContext = Fsr3Upscaler.CreateContext(_displaySize, _maxRenderSize, context.resources.computeShaders.superResolution, flags); + + // Apply a mipmap bias so that textures retain their sharpness + float biasOffset = Fsr3Upscaler.GetMipmapBiasOffset(_maxRenderSize.x, _displaySize.x); + if (!float.IsNaN(biasOffset) && !float.IsInfinity(biasOffset)) + { + _callbacks.ApplyMipmapBias(biasOffset); + } + } + + private void DestroyFsrContext() + { + if (_fsrContext != null) + { + _fsrContext.Destroy(); + _fsrContext = null; + } + + if (_callbacks != null) + { + // Undo the current mipmap bias offset + _callbacks.UndoMipmapBias(); + _callbacks = null; + } + } + + private void ApplyJitter(Camera camera) + { + var scaledRenderSize = GetScaledRenderSize(camera); + + // Perform custom jittering of the camera's projection matrix according to FSR3's recipe + int jitterPhaseCount = Fsr3Upscaler.GetJitterPhaseCount(scaledRenderSize.x, _displaySize.x); + Fsr3Upscaler.GetJitterOffset(out float jitterX, out float jitterY, Time.frameCount, jitterPhaseCount); + + _dispatchDescription.JitterOffset = new Vector2(jitterX, jitterY); + + jitterX = 2.0f * jitterX / scaledRenderSize.x; + jitterY = 2.0f * jitterY / scaledRenderSize.y; + + var jitterTranslationMatrix = Matrix4x4.Translate(new Vector3(jitterX, jitterY, 0)); + camera.nonJitteredProjectionMatrix = camera.projectionMatrix; + camera.projectionMatrix = jitterTranslationMatrix * camera.nonJitteredProjectionMatrix; + camera.useJitteredProjectionMatrixForTransparentRendering = true; + + jitter = new Vector2(jitterX, jitterY); + } + + private void SetupDispatchDescription(PostProcessRenderContext context) + { + var camera = context.camera; + + // Set up the main FSR3 Upscaler dispatch parameters + _dispatchDescription.Color = new ResourceView(context.source); + _dispatchDescription.Depth = new ResourceView(BuiltinRenderTextureType.CameraTarget, RenderTextureSubElement.Depth); + _dispatchDescription.MotionVectors = new ResourceView(BuiltinRenderTextureType.MotionVectors); + _dispatchDescription.Exposure = ResourceView.Unassigned; + _dispatchDescription.Reactive = ResourceView.Unassigned; + _dispatchDescription.TransparencyAndComposition = ResourceView.Unassigned; + + if (exposureSource == ExposureSource.Manual && exposure != null) _dispatchDescription.Exposure = new ResourceView(exposure); + if (exposureSource == ExposureSource.Unity) _dispatchDescription.Exposure = new ResourceView(context.autoExposureTexture); + if (reactiveMask != null) _dispatchDescription.Reactive = new ResourceView(reactiveMask); + if (transparencyAndCompositionMask != null) _dispatchDescription.TransparencyAndComposition = new ResourceView(transparencyAndCompositionMask); + + var scaledRenderSize = GetScaledRenderSize(context.camera); + + _dispatchDescription.Output = new ResourceView(context.destination); + _dispatchDescription.PreExposure = preExposure; + _dispatchDescription.EnableSharpening = performSharpenPass; + _dispatchDescription.Sharpness = sharpness; + _dispatchDescription.MotionVectorScale.x = -scaledRenderSize.x; + _dispatchDescription.MotionVectorScale.y = -scaledRenderSize.y; + _dispatchDescription.RenderSize = scaledRenderSize; + _dispatchDescription.UpscaleSize = _displaySize; + _dispatchDescription.FrameTimeDelta = Time.unscaledDeltaTime; + _dispatchDescription.CameraNear = camera.nearClipPlane; + _dispatchDescription.CameraFar = camera.farClipPlane; + _dispatchDescription.CameraFovAngleVertical = camera.fieldOfView * Mathf.Deg2Rad; + _dispatchDescription.ViewSpaceToMetersFactor = 1.0f; // 1 unit is 1 meter in Unity + _dispatchDescription.Reset = _resetHistory; + _dispatchDescription.Flags = enableDebugView ? Fsr3Upscaler.DispatchFlags.DrawDebugView : 0; + + // Set up the parameters for the optional experimental auto-TCR feature + _dispatchDescription.EnableAutoReactive = autoGenerateTransparencyAndComposition; + if (autoGenerateTransparencyAndComposition) + { + _dispatchDescription.ColorOpaqueOnly = new ResourceView(colorOpaqueOnly); + _dispatchDescription.AutoTcThreshold = generateTransparencyAndCompositionParameters.autoTcThreshold; + _dispatchDescription.AutoTcScale = generateTransparencyAndCompositionParameters.autoTcScale; + _dispatchDescription.AutoReactiveScale = generateTransparencyAndCompositionParameters.autoReactiveScale; + _dispatchDescription.AutoReactiveMax = generateTransparencyAndCompositionParameters.autoReactiveMax; + } + + if (SystemInfo.usesReversedZBuffer) + { + // Swap the near and far clip plane distances as FSR3 expects this when using inverted depth + (_dispatchDescription.CameraNear, _dispatchDescription.CameraFar) = (_dispatchDescription.CameraFar, _dispatchDescription.CameraNear); + } + } + + private void SetupAutoReactiveDescription(PostProcessRenderContext context) + { + // Set up the parameters to auto-generate a reactive mask + _genReactiveDescription.ColorOpaqueOnly = new ResourceView(colorOpaqueOnly); + _genReactiveDescription.ColorPreUpscale = new ResourceView(context.source); + _genReactiveDescription.OutReactive = new ResourceView(Fsr3ShaderIDs.UavAutoReactive); + _genReactiveDescription.RenderSize = GetScaledRenderSize(context.camera); + _genReactiveDescription.Scale = generateReactiveParameters.scale; + _genReactiveDescription.CutoffThreshold = generateReactiveParameters.cutoffThreshold; + _genReactiveDescription.BinaryValue = generateReactiveParameters.binaryValue; + _genReactiveDescription.Flags = generateReactiveParameters.flags; + } + + internal Vector2Int GetScaledRenderSize(Camera camera) + { + if (!RuntimeUtilities.IsDynamicResolutionEnabled(camera)) + return _maxRenderSize; + + return new Vector2Int(Mathf.CeilToInt(_maxRenderSize.x * ScalableBufferManager.widthScaleFactor), Mathf.CeilToInt(_maxRenderSize.y * ScalableBufferManager.heightScaleFactor)); + } + } +} diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/SuperResolution.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/SuperResolution.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/SuperResolution.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/SuperResolution.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/TemporalAntialiasing.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/TemporalAntialiasing.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/TemporalAntialiasing.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/TemporalAntialiasing.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/TemporalAntialiasing.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/TemporalAntialiasing.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/TemporalAntialiasing.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/TemporalAntialiasing.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/Vignette.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/Vignette.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/Vignette.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/Vignette.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/Vignette.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/Vignette.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Effects/Vignette.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Effects/Vignette.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Monitors.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Monitors.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Monitors.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Monitors.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Monitors/HistogramMonitor.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Monitors/HistogramMonitor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Monitors/HistogramMonitor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Monitors/HistogramMonitor.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Monitors/HistogramMonitor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Monitors/HistogramMonitor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Monitors/HistogramMonitor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Monitors/HistogramMonitor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Monitors/LightMeterMonitor.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Monitors/LightMeterMonitor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Monitors/LightMeterMonitor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Monitors/LightMeterMonitor.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Monitors/LightMeterMonitor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Monitors/LightMeterMonitor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Monitors/LightMeterMonitor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Monitors/LightMeterMonitor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Monitors/Monitor.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Monitors/Monitor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Monitors/Monitor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Monitors/Monitor.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Monitors/Monitor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Monitors/Monitor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Monitors/Monitor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Monitors/Monitor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Monitors/VectorscopeMonitor.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Monitors/VectorscopeMonitor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Monitors/VectorscopeMonitor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Monitors/VectorscopeMonitor.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Monitors/VectorscopeMonitor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Monitors/VectorscopeMonitor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Monitors/VectorscopeMonitor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Monitors/VectorscopeMonitor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Monitors/WaveformMonitor.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Monitors/WaveformMonitor.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Monitors/WaveformMonitor.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Monitors/WaveformMonitor.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Monitors/WaveformMonitor.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Monitors/WaveformMonitor.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Monitors/WaveformMonitor.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Monitors/WaveformMonitor.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/ParameterOverride.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/ParameterOverride.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/ParameterOverride.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/ParameterOverride.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/ParameterOverride.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/ParameterOverride.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/ParameterOverride.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/ParameterOverride.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessBundle.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessBundle.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessBundle.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessBundle.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessBundle.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessBundle.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessBundle.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessBundle.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessDebug.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessDebug.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessDebug.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessDebug.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessDebug.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessDebug.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessDebug.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessDebug.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessDebugLayer.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessDebugLayer.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessDebugLayer.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessDebugLayer.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessDebugLayer.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessDebugLayer.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessDebugLayer.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessDebugLayer.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessEffectRenderer.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessEffectRenderer.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessEffectRenderer.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessEffectRenderer.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessEffectRenderer.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessEffectRenderer.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessEffectRenderer.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessEffectRenderer.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessEffectSettings.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessEffectSettings.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessEffectSettings.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessEffectSettings.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessEffectSettings.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessEffectSettings.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessEffectSettings.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessEffectSettings.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessEvent.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessEvent.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessEvent.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessEvent.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessEvent.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessEvent.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessEvent.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessEvent.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessLayer.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessLayer.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessLayer.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessLayer.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessLayer.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessLayer.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessLayer.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessLayer.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessManager.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessManager.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessManager.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessManager.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessManager.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessManager.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessManager.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessManager.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessProfile.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessProfile.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessProfile.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessProfile.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessProfile.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessProfile.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessProfile.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessProfile.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessRenderContext.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessRenderContext.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessRenderContext.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessRenderContext.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessRenderContext.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessRenderContext.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessRenderContext.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessRenderContext.cs.meta diff --git a/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessResources.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessResources.cs new file mode 100644 index 0000000..e745fa6 --- /dev/null +++ b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessResources.cs @@ -0,0 +1,293 @@ +using System; +using FidelityFX.FSR3; + +namespace UnityEngine.Rendering.PostProcessing +{ + /// + /// This asset is used to store references to shaders and other resources we might need at + /// runtime without having to use a `Resources` folder. This allows for better memory management, + /// better dependency tracking and better interoperability with asset bundles. + /// + public sealed class PostProcessResources : ScriptableObject + { + /// + /// All the shaders used by post-processing. + /// + [Serializable] + public sealed class Shaders + { + /// + /// The shader used for the bloom effect. + /// + public Shader bloom; + + /// + /// The shader used for internal blit copies. + /// + public Shader copy; + + /// + /// The shader used for built-in blit copies in the built-in pipeline. + /// + public Shader copyStd; + + /// + /// The shader used for built-in blit copies in the built-in pipeline when using a + /// texture array (for stereo rendering). + /// + public Shader copyStdFromTexArray; + + /// + /// The shader used for built-in blit copies in the built-in pipeline when using a + /// double-wide texture (for stereo rendering). + /// + public Shader copyStdFromDoubleWide; + + /// + /// The shader used to kill the alpha. + /// + public Shader discardAlpha; + + /// + /// The shader used for the depth of field effect. + /// + public Shader depthOfField; + + /// + /// The shader used for the final pass. + /// + public Shader finalPass; + + /// + /// The shader used to generate the grain texture. + /// + public Shader grainBaker; + + /// + /// The shader used for the motion blur effect. + /// + public Shader motionBlur; + + /// + /// The shader used for the temporal anti-aliasing effect. + /// + public Shader temporalAntialiasing; + + /// + /// The shader used for the sub-pixel morphological anti-aliasing effect. + /// + public Shader subpixelMorphologicalAntialiasing; + + /// + /// The shader use by the volume manager to interpolate between two 2D textures. + /// + public Shader texture2dLerp; + + /// + /// The uber shader that combine several effects into one. + /// + public Shader uber; + + /// + /// The shader used to bake the 2D lookup table for color grading. + /// + public Shader lut2DBaker; + + /// + /// The shader used to draw the light meter monitor. + /// + public Shader lightMeter; + + /// + /// The shader used to draw the histogram monitor. + /// + public Shader gammaHistogram; + + /// + /// The shader used to draw the waveform monitor. + /// + public Shader waveform; + + /// + /// The shader used to draw the vectorscope monitor. + /// + public Shader vectorscope; + + /// + /// The shader used to draw debug overlays. + /// + public Shader debugOverlays; + + /// + /// The shader used for the deferred fog effect. + /// + public Shader deferredFog; + + /// + /// The shader used for the scalable ambient occlusion effect. + /// + public Shader scalableAO; + + /// + /// The shader used for the multi-scale ambient occlusion effect. + /// + public Shader multiScaleAO; + + /// + /// The shader used for the screen-space reflection effect. + /// + public Shader screenSpaceReflections; + + /// + /// Returns a copy of this class and its content. + /// + /// A copy of this class and its content. + public Shaders Clone() + { + return (Shaders)MemberwiseClone(); + } + } + + /// + /// All the compute shaders used by post-processing. + /// + [Serializable] + public sealed class ComputeShaders + { + /// + /// The compute shader used for the auto-exposure effect. + /// + public ComputeShader autoExposure; + + /// + /// The compute shader used to compute an histogram of the current frame. + /// + public ComputeShader exposureHistogram; + + /// + /// The compute shader used to bake the 3D lookup table for color grading. + /// + public ComputeShader lut3DBaker; + + /// + /// The compute shader used by the volume manager to interpolate between two 3D textures. + /// + public ComputeShader texture3dLerp; + + /// + /// The compute shader used to compute the histogram monitor. + /// + public ComputeShader gammaHistogram; + + /// + /// The compute shader used to compute the waveform monitor. + /// + public ComputeShader waveform; + + /// + /// The compute shader used to compute the vectorscope monitor. + /// + public ComputeShader vectorscope; + + /// + /// The compute shader used for the first downsampling pass of MSVO. + /// + public ComputeShader multiScaleAODownsample1; + + /// + /// The compute shader used for the second downsampling pass of MSVO. + /// + public ComputeShader multiScaleAODownsample2; + + /// + /// The compute shader used for the render pass of MSVO. + /// + public ComputeShader multiScaleAORender; + + /// + /// The compute shader used for the upsampling pass of MSVO. + /// + public ComputeShader multiScaleAOUpsample; + + /// + /// The compute shader used to a fast gaussian downsample. + /// + public ComputeShader gaussianDownsample; + + /// + /// Compute shaders used by the FidelityFX Super Resolution 3 (FSR3) Upscaler. + /// + public Fsr3UpscalerShaders superResolution; + + /// + /// Returns a copy of this class and its content. + /// + /// A copy of this class and its content. + public ComputeShaders Clone() + { + return (ComputeShaders)MemberwiseClone(); + } + } + + /// + /// A set of textures needed by the sub-pixel morphological anti-aliasing effect. + /// + [Serializable] + public sealed class SMAALuts + { + /// + /// The area lookup table. + /// + public Texture2D area; + + /// + /// The search lookup table. + /// + public Texture2D search; + } + + /// + /// A set of 64x64, single-channel blue noise textures. + /// + public Texture2D[] blueNoise64; + + /// + /// A set of 256x256, single-channel blue noise textures. + /// + public Texture2D[] blueNoise256; + + /// + /// Lookup tables used by the sub-pixel morphological anti-aliasing effect. + /// + public SMAALuts smaaLuts; + + /// + /// All the shaders used by post-processing. + /// + public Shaders shaders; + + /// + /// All the compute shaders used by post-processing. + /// + public ComputeShaders computeShaders; + +#if UNITY_EDITOR + /// + /// A delegate used to track resource changes. + /// + public delegate void ChangeHandler(); + + /// + /// Set this callback to be notified of resource changes. + /// + public ChangeHandler changeHandler; + + void OnValidate() + { + if (changeHandler != null) + changeHandler(); + } + +#endif + } +} diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessResources.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessResources.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessResources.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessResources.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessVolume.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessVolume.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessVolume.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessVolume.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessVolume.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessVolume.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/PostProcessVolume.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/PostProcessVolume.cs.meta diff --git a/Packages/com.unity.postprocessing/PostProcessing/Runtime/Unity.Postprocessing.Runtime.asmdef b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Unity.Postprocessing.Runtime.asmdef new file mode 100644 index 0000000..8220b4a --- /dev/null +++ b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Unity.Postprocessing.Runtime.asmdef @@ -0,0 +1,37 @@ +{ + "name": "Unity.Postprocessing.Runtime", + "rootNamespace": "", + "references": [ + "FidelityFX.FSR" + ], + "includePlatforms": [], + "excludePlatforms": [], + "allowUnsafeCode": false, + "overrideReferences": false, + "precompiledReferences": [], + "autoReferenced": true, + "defineConstraints": [], + "versionDefines": [ + { + "name": "com.unity.render-pipelines.lightweight", + "expression": "1.0.0", + "define": "LWRP_1_0_0_OR_NEWER" + }, + { + "name": "com.unity.render-pipelines.universal", + "expression": "1.0.0", + "define": "UNIVERSAL_1_0_0_OR_NEWER" + }, + { + "name": "com.unity.modules.vr", + "expression": "1.0.0", + "define": "ENABLE_VR_MODULE" + }, + { + "name": "com.unity.modules.xr", + "expression": "1.0.0", + "define": "ENABLE_XR_MODULE" + } + ], + "noEngineReferences": false +} \ No newline at end of file diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Unity.Postprocessing.Runtime.asmdef.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Unity.Postprocessing.Runtime.asmdef.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Unity.Postprocessing.Runtime.asmdef.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Unity.Postprocessing.Runtime.asmdef.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/ColorUtilities.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/ColorUtilities.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/ColorUtilities.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/ColorUtilities.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/ColorUtilities.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/ColorUtilities.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/ColorUtilities.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/ColorUtilities.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/HableCurve.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/HableCurve.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/HableCurve.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/HableCurve.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/HableCurve.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/HableCurve.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/HableCurve.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/HableCurve.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/HaltonSeq.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/HaltonSeq.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/HaltonSeq.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/HaltonSeq.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/HaltonSeq.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/HaltonSeq.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/HaltonSeq.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/HaltonSeq.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/LogHistogram.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/LogHistogram.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/LogHistogram.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/LogHistogram.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/LogHistogram.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/LogHistogram.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/LogHistogram.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/LogHistogram.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/MeshUtilities.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/MeshUtilities.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/MeshUtilities.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/MeshUtilities.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/MeshUtilities.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/MeshUtilities.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/MeshUtilities.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/MeshUtilities.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/PostProcessEffectRendererExtensions.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/PostProcessEffectRendererExtensions.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/PostProcessEffectRendererExtensions.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/PostProcessEffectRendererExtensions.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/PostProcessEffectRendererExtensions.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/PostProcessEffectRendererExtensions.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/PostProcessEffectRendererExtensions.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/PostProcessEffectRendererExtensions.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/PropertySheet.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/PropertySheet.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/PropertySheet.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/PropertySheet.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/PropertySheet.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/PropertySheet.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/PropertySheet.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/PropertySheet.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/PropertySheetFactory.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/PropertySheetFactory.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/PropertySheetFactory.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/PropertySheetFactory.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/PropertySheetFactory.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/PropertySheetFactory.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/PropertySheetFactory.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/PropertySheetFactory.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/RuntimeUtilities.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/RuntimeUtilities.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/RuntimeUtilities.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/RuntimeUtilities.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/RuntimeUtilities.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/RuntimeUtilities.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/RuntimeUtilities.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/RuntimeUtilities.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/ShaderIDs.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/ShaderIDs.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/ShaderIDs.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/ShaderIDs.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/ShaderIDs.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/ShaderIDs.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/ShaderIDs.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/ShaderIDs.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/Spline.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/Spline.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/Spline.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/Spline.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/Spline.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/Spline.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/Spline.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/Spline.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/TargetPool.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/TargetPool.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/TargetPool.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/TargetPool.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/TargetPool.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/TargetPool.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/TargetPool.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/TargetPool.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/TextureFormatUtilities.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/TextureFormatUtilities.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/TextureFormatUtilities.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/TextureFormatUtilities.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/TextureFormatUtilities.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/TextureFormatUtilities.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/TextureFormatUtilities.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/TextureFormatUtilities.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/TextureLerper.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/TextureLerper.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/TextureLerper.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/TextureLerper.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/TextureLerper.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/TextureLerper.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/TextureLerper.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/TextureLerper.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/XRSettings.cs b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/XRSettings.cs similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/XRSettings.cs rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/XRSettings.cs diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Utils/XRSettings.cs.meta b/Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/XRSettings.cs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Runtime/Utils/XRSettings.cs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Runtime/Utils/XRSettings.cs.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/ACES.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/ACES.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/ACES.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/ACES.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/ACES.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/ACES.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/ACES.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/ACES.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API/D3D11.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API/D3D11.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API/D3D11.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API/D3D11.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API/D3D11.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API/D3D11.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API/D3D11.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API/D3D11.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API/D3D12.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API/D3D12.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API/D3D12.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API/D3D12.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API/D3D12.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API/D3D12.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API/D3D12.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API/D3D12.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API/D3D9.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API/D3D9.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API/D3D9.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API/D3D9.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API/D3D9.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API/D3D9.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API/D3D9.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API/D3D9.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API/Metal.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API/Metal.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API/Metal.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API/Metal.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API/Metal.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API/Metal.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API/Metal.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API/Metal.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API/OpenGL.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API/OpenGL.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API/OpenGL.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API/OpenGL.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API/OpenGL.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API/OpenGL.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API/OpenGL.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API/OpenGL.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API/PSP2.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API/PSP2.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API/PSP2.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API/PSP2.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API/PSP2.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API/PSP2.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API/PSP2.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API/PSP2.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API/PSSL.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API/PSSL.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API/PSSL.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API/PSSL.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API/PSSL.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API/PSSL.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API/PSSL.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API/PSSL.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API/Switch.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API/Switch.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API/Switch.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API/Switch.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API/Switch.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API/Switch.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API/Switch.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API/Switch.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API/Vulkan.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API/Vulkan.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API/Vulkan.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API/Vulkan.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API/Vulkan.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API/Vulkan.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API/Vulkan.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API/Vulkan.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API/WebGPU.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API/WebGPU.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API/WebGPU.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API/WebGPU.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API/WebGPU.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API/WebGPU.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API/WebGPU.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API/WebGPU.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API/XboxOne.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API/XboxOne.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API/XboxOne.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API/XboxOne.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/API/XboxOne.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/API/XboxOne.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/API/XboxOne.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/API/XboxOne.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/AutoExposure.compute b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/AutoExposure.compute similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/AutoExposure.compute rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/AutoExposure.compute diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/AutoExposure.compute.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/AutoExposure.compute.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/AutoExposure.compute.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/AutoExposure.compute.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Bloom.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Bloom.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/Bloom.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Bloom.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Bloom.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Bloom.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/Bloom.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Bloom.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Copy.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Copy.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/Copy.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Copy.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Copy.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Copy.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/Copy.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Copy.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStd.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStd.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStd.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStd.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStd.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStd.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStd.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStd.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStdFromDoubleWide.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStdFromDoubleWide.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStdFromDoubleWide.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStdFromDoubleWide.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStdFromDoubleWide.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStdFromDoubleWide.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStdFromDoubleWide.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStdFromDoubleWide.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStdFromTexArray.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStdFromTexArray.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStdFromTexArray.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStdFromTexArray.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStdFromTexArray.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStdFromTexArray.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStdFromTexArray.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/CopyStdFromTexArray.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DeferredFog.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DeferredFog.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/DeferredFog.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DeferredFog.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DeferredFog.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DeferredFog.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/DeferredFog.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DeferredFog.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DepthOfField.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DepthOfField.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/DepthOfField.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DepthOfField.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DepthOfField.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DepthOfField.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/DepthOfField.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DepthOfField.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DepthOfField.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DepthOfField.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/DepthOfField.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DepthOfField.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DepthOfField.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DepthOfField.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/DepthOfField.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DepthOfField.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DiscardAlpha.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DiscardAlpha.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/DiscardAlpha.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DiscardAlpha.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DiscardAlpha.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DiscardAlpha.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/DiscardAlpha.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DiscardAlpha.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DiskKernels.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DiskKernels.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/DiskKernels.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DiskKernels.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DiskKernels.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DiskKernels.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/DiskKernels.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/DiskKernels.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Distortion.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Distortion.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/Distortion.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Distortion.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Distortion.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Distortion.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/Distortion.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Distortion.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Dithering.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Dithering.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/Dithering.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Dithering.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Dithering.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Dithering.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/Dithering.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Dithering.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ExposureHistogram.compute b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ExposureHistogram.compute similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/ExposureHistogram.compute rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ExposureHistogram.compute diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ExposureHistogram.compute.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ExposureHistogram.compute.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/ExposureHistogram.compute.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ExposureHistogram.compute.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ExposureHistogram.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ExposureHistogram.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/ExposureHistogram.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ExposureHistogram.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ExposureHistogram.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ExposureHistogram.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/ExposureHistogram.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ExposureHistogram.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/FastApproximateAntialiasing.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/FastApproximateAntialiasing.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/FastApproximateAntialiasing.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/FastApproximateAntialiasing.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/FastApproximateAntialiasing.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/FastApproximateAntialiasing.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/FastApproximateAntialiasing.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/FastApproximateAntialiasing.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/FinalPass.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/FinalPass.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/FinalPass.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/FinalPass.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/FinalPass.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/FinalPass.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/FinalPass.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/FinalPass.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Fog.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Fog.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/Fog.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Fog.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Fog.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Fog.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/Fog.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Fog.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/GaussianDownsample.compute b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/GaussianDownsample.compute similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/GaussianDownsample.compute rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/GaussianDownsample.compute diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/GaussianDownsample.compute.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/GaussianDownsample.compute.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/GaussianDownsample.compute.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/GaussianDownsample.compute.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/GrainBaker.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/GrainBaker.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/GrainBaker.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/GrainBaker.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/GrainBaker.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/GrainBaker.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/GrainBaker.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/GrainBaker.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Lut2DBaker.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Lut2DBaker.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/Lut2DBaker.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Lut2DBaker.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Lut2DBaker.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Lut2DBaker.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/Lut2DBaker.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Lut2DBaker.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Lut3DBaker.compute b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Lut3DBaker.compute similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/Lut3DBaker.compute rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Lut3DBaker.compute diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Lut3DBaker.compute.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Lut3DBaker.compute.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/Lut3DBaker.compute.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Lut3DBaker.compute.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MotionBlur.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MotionBlur.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/MotionBlur.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MotionBlur.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MotionBlur.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MotionBlur.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/MotionBlur.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MotionBlur.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVO.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVO.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVO.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVO.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVO.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVO.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVO.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVO.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVODownsample1.compute b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVODownsample1.compute similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVODownsample1.compute rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVODownsample1.compute diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVODownsample1.compute.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVODownsample1.compute.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVODownsample1.compute.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVODownsample1.compute.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVODownsample2.compute b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVODownsample2.compute similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVODownsample2.compute rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVODownsample2.compute diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVODownsample2.compute.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVODownsample2.compute.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVODownsample2.compute.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVODownsample2.compute.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVORender.compute b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVORender.compute similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVORender.compute rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVORender.compute diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVORender.compute.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVORender.compute.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVORender.compute.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVORender.compute.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVOUpsample.compute b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVOUpsample.compute similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVOUpsample.compute rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVOUpsample.compute diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVOUpsample.compute.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVOUpsample.compute.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVOUpsample.compute.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/MultiScaleVOUpsample.compute.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScalableAO.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScalableAO.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScalableAO.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScalableAO.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScalableAO.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScalableAO.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScalableAO.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScalableAO.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScalableAO.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScalableAO.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScalableAO.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScalableAO.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScalableAO.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScalableAO.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScalableAO.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScalableAO.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScreenSpaceReflections.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScreenSpaceReflections.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScreenSpaceReflections.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScreenSpaceReflections.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScreenSpaceReflections.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScreenSpaceReflections.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScreenSpaceReflections.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScreenSpaceReflections.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScreenSpaceReflections.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScreenSpaceReflections.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScreenSpaceReflections.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScreenSpaceReflections.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScreenSpaceReflections.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScreenSpaceReflections.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScreenSpaceReflections.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/ScreenSpaceReflections.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasing.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasing.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasing.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasing.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasing.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasing.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasing.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasing.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasing.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasing.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasing.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasing.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasing.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasing.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasing.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasing.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasingBridge.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasingBridge.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasingBridge.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasingBridge.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasingBridge.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasingBridge.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasingBridge.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/SubpixelMorphologicalAntialiasingBridge.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/TemporalAntialiasing.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/TemporalAntialiasing.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/TemporalAntialiasing.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/TemporalAntialiasing.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/TemporalAntialiasing.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/TemporalAntialiasing.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/TemporalAntialiasing.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/TemporalAntialiasing.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Texture2DLerp.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Texture2DLerp.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/Texture2DLerp.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Texture2DLerp.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Texture2DLerp.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Texture2DLerp.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/Texture2DLerp.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Texture2DLerp.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Texture3DLerp.compute b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Texture3DLerp.compute similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/Texture3DLerp.compute rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Texture3DLerp.compute diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Texture3DLerp.compute.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Texture3DLerp.compute.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/Texture3DLerp.compute.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Texture3DLerp.compute.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Uber.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Uber.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/Uber.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Uber.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Uber.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Uber.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Builtins/Uber.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Builtins/Uber.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Colors.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Colors.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Colors.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Colors.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Colors.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Colors.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Colors.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Colors.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Debug.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Debug.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Debug/Histogram.compute b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Histogram.compute similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Debug/Histogram.compute rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Histogram.compute diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Debug/Histogram.compute.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Histogram.compute.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Debug/Histogram.compute.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Histogram.compute.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Debug/Histogram.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Histogram.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Debug/Histogram.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Histogram.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Debug/Histogram.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Histogram.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Debug/Histogram.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Histogram.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Debug/LightMeter.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/LightMeter.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Debug/LightMeter.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/LightMeter.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Debug/LightMeter.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/LightMeter.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Debug/LightMeter.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/LightMeter.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Debug/Overlays.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Overlays.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Debug/Overlays.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Overlays.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Debug/Overlays.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Overlays.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Debug/Overlays.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Overlays.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Debug/Vectorscope.compute b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Vectorscope.compute similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Debug/Vectorscope.compute rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Vectorscope.compute diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Debug/Vectorscope.compute.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Vectorscope.compute.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Debug/Vectorscope.compute.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Vectorscope.compute.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Debug/Vectorscope.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Vectorscope.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Debug/Vectorscope.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Vectorscope.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Debug/Vectorscope.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Vectorscope.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Debug/Vectorscope.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Vectorscope.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Debug/Waveform.compute b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Waveform.compute similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Debug/Waveform.compute rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Waveform.compute diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Debug/Waveform.compute.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Waveform.compute.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Debug/Waveform.compute.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Waveform.compute.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Debug/Waveform.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Waveform.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Debug/Waveform.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Waveform.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Debug/Waveform.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Waveform.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Debug/Waveform.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Debug/Waveform.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Editor.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Editor.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Editor.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Editor.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Editor/ConvertToLog.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Editor/ConvertToLog.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Editor/ConvertToLog.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Editor/ConvertToLog.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Editor/ConvertToLog.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Editor/ConvertToLog.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Editor/ConvertToLog.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Editor/ConvertToLog.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Editor/CurveGrid.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Editor/CurveGrid.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Editor/CurveGrid.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Editor/CurveGrid.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Editor/CurveGrid.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Editor/CurveGrid.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Editor/CurveGrid.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Editor/CurveGrid.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Editor/Trackball.shader b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Editor/Trackball.shader similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Editor/Trackball.shader rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Editor/Trackball.shader diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Editor/Trackball.shader.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Editor/Trackball.shader.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Editor/Trackball.shader.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Editor/Trackball.shader.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Sampling.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Sampling.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Sampling.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Sampling.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/Sampling.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/Sampling.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/Sampling.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/Sampling.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/StdLib.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/StdLib.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/StdLib.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/StdLib.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/StdLib.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/StdLib.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/StdLib.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/StdLib.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Shaders/xRLib.hlsl b/Packages/com.unity.postprocessing/PostProcessing/Shaders/xRLib.hlsl similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/xRLib.hlsl rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/xRLib.hlsl diff --git a/com.unity.postprocessing/PostProcessing/Shaders/xRLib.hlsl.meta b/Packages/com.unity.postprocessing/PostProcessing/Shaders/xRLib.hlsl.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Shaders/xRLib.hlsl.meta rename to Packages/com.unity.postprocessing/PostProcessing/Shaders/xRLib.hlsl.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_0.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_0.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_0.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_0.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_0.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_0.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_0.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_0.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_1.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_1.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_1.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_1.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_1.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_1.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_1.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_1.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_2.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_2.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_2.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_2.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_2.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_2.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_2.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_2.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_3.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_3.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_3.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_3.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_3.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_3.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_3.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_3.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_4.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_4.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_4.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_4.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_4.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_4.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_4.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_4.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_5.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_5.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_5.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_5.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_5.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_5.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_5.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_5.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_6.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_6.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_6.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_6.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_6.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_6.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_6.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_6.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_7.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_7.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_7.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_7.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_7.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_7.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_7.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise256px/LDR_LLL1_7.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_0.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_0.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_0.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_0.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_0.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_0.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_0.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_0.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_1.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_1.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_1.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_1.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_1.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_1.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_1.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_1.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_10.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_10.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_10.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_10.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_10.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_10.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_10.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_10.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_11.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_11.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_11.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_11.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_11.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_11.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_11.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_11.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_12.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_12.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_12.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_12.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_12.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_12.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_12.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_12.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_13.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_13.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_13.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_13.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_13.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_13.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_13.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_13.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_14.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_14.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_14.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_14.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_14.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_14.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_14.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_14.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_15.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_15.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_15.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_15.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_15.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_15.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_15.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_15.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_16.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_16.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_16.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_16.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_16.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_16.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_16.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_16.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_17.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_17.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_17.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_17.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_17.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_17.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_17.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_17.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_18.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_18.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_18.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_18.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_18.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_18.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_18.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_18.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_19.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_19.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_19.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_19.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_19.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_19.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_19.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_19.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_2.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_2.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_2.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_2.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_2.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_2.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_2.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_2.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_20.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_20.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_20.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_20.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_20.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_20.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_20.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_20.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_21.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_21.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_21.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_21.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_21.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_21.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_21.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_21.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_22.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_22.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_22.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_22.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_22.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_22.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_22.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_22.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_23.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_23.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_23.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_23.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_23.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_23.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_23.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_23.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_24.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_24.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_24.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_24.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_24.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_24.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_24.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_24.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_25.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_25.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_25.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_25.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_25.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_25.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_25.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_25.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_26.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_26.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_26.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_26.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_26.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_26.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_26.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_26.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_27.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_27.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_27.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_27.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_27.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_27.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_27.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_27.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_28.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_28.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_28.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_28.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_28.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_28.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_28.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_28.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_29.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_29.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_29.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_29.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_29.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_29.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_29.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_29.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_3.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_3.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_3.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_3.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_3.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_3.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_3.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_3.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_30.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_30.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_30.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_30.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_30.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_30.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_30.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_30.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_31.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_31.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_31.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_31.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_31.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_31.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_31.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_31.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_32.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_32.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_32.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_32.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_32.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_32.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_32.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_32.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_33.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_33.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_33.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_33.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_33.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_33.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_33.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_33.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_34.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_34.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_34.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_34.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_34.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_34.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_34.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_34.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_35.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_35.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_35.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_35.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_35.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_35.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_35.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_35.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_36.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_36.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_36.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_36.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_36.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_36.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_36.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_36.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_37.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_37.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_37.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_37.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_37.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_37.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_37.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_37.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_38.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_38.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_38.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_38.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_38.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_38.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_38.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_38.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_39.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_39.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_39.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_39.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_39.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_39.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_39.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_39.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_4.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_4.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_4.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_4.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_4.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_4.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_4.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_4.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_40.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_40.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_40.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_40.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_40.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_40.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_40.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_40.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_41.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_41.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_41.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_41.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_41.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_41.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_41.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_41.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_42.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_42.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_42.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_42.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_42.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_42.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_42.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_42.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_43.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_43.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_43.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_43.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_43.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_43.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_43.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_43.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_44.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_44.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_44.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_44.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_44.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_44.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_44.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_44.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_45.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_45.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_45.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_45.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_45.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_45.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_45.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_45.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_46.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_46.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_46.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_46.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_46.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_46.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_46.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_46.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_47.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_47.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_47.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_47.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_47.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_47.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_47.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_47.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_48.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_48.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_48.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_48.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_48.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_48.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_48.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_48.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_49.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_49.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_49.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_49.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_49.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_49.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_49.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_49.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_5.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_5.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_5.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_5.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_5.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_5.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_5.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_5.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_50.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_50.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_50.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_50.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_50.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_50.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_50.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_50.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_51.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_51.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_51.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_51.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_51.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_51.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_51.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_51.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_52.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_52.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_52.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_52.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_52.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_52.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_52.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_52.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_53.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_53.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_53.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_53.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_53.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_53.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_53.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_53.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_54.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_54.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_54.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_54.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_54.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_54.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_54.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_54.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_55.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_55.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_55.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_55.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_55.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_55.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_55.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_55.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_56.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_56.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_56.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_56.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_56.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_56.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_56.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_56.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_57.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_57.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_57.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_57.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_57.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_57.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_57.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_57.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_58.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_58.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_58.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_58.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_58.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_58.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_58.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_58.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_59.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_59.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_59.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_59.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_59.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_59.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_59.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_59.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_6.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_6.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_6.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_6.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_6.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_6.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_6.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_6.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_60.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_60.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_60.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_60.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_60.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_60.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_60.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_60.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_61.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_61.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_61.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_61.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_61.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_61.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_61.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_61.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_62.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_62.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_62.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_62.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_62.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_62.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_62.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_62.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_63.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_63.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_63.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_63.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_63.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_63.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_63.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_63.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_7.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_7.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_7.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_7.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_7.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_7.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_7.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_7.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_8.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_8.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_8.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_8.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_8.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_8.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_8.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_8.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_9.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_9.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_9.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_9.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_9.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_9.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_9.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/BlueNoise64px/LDR_LLL1_9.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/Cubes.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/Cubes.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/Cubes/Linear_to_Unity_Log_r1.cube b/Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/Linear_to_Unity_Log_r1.cube similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/Cubes/Linear_to_Unity_Log_r1.cube rename to Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/Linear_to_Unity_Log_r1.cube diff --git a/com.unity.postprocessing/PostProcessing/Textures/Cubes/Linear_to_Unity_Log_r1.cube.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/Linear_to_Unity_Log_r1.cube.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/Cubes/Linear_to_Unity_Log_r1.cube.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/Linear_to_Unity_Log_r1.cube.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/Cubes/Linear_to_sRGB_r1.cube b/Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/Linear_to_sRGB_r1.cube similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/Cubes/Linear_to_sRGB_r1.cube rename to Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/Linear_to_sRGB_r1.cube diff --git a/com.unity.postprocessing/PostProcessing/Textures/Cubes/Linear_to_sRGB_r1.cube.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/Linear_to_sRGB_r1.cube.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/Cubes/Linear_to_sRGB_r1.cube.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/Linear_to_sRGB_r1.cube.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/Cubes/Unity_Log_to_Linear_r1.cube b/Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/Unity_Log_to_Linear_r1.cube similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/Cubes/Unity_Log_to_Linear_r1.cube rename to Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/Unity_Log_to_Linear_r1.cube diff --git a/com.unity.postprocessing/PostProcessing/Textures/Cubes/Unity_Log_to_Linear_r1.cube.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/Unity_Log_to_Linear_r1.cube.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/Cubes/Unity_Log_to_Linear_r1.cube.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/Unity_Log_to_Linear_r1.cube.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/Cubes/Unity_Log_to_sRGB_r1.cube b/Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/Unity_Log_to_sRGB_r1.cube similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/Cubes/Unity_Log_to_sRGB_r1.cube rename to Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/Unity_Log_to_sRGB_r1.cube diff --git a/com.unity.postprocessing/PostProcessing/Textures/Cubes/Unity_Log_to_sRGB_r1.cube.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/Unity_Log_to_sRGB_r1.cube.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/Cubes/Unity_Log_to_sRGB_r1.cube.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/Unity_Log_to_sRGB_r1.cube.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/Cubes/sRGB_to_Linear_r1.cube b/Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/sRGB_to_Linear_r1.cube similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/Cubes/sRGB_to_Linear_r1.cube rename to Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/sRGB_to_Linear_r1.cube diff --git a/com.unity.postprocessing/PostProcessing/Textures/Cubes/sRGB_to_Linear_r1.cube.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/sRGB_to_Linear_r1.cube.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/Cubes/sRGB_to_Linear_r1.cube.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/sRGB_to_Linear_r1.cube.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/Cubes/sRGB_to_Unity_Log_r1.cube b/Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/sRGB_to_Unity_Log_r1.cube similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/Cubes/sRGB_to_Unity_Log_r1.cube rename to Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/sRGB_to_Unity_Log_r1.cube diff --git a/com.unity.postprocessing/PostProcessing/Textures/Cubes/sRGB_to_Unity_Log_r1.cube.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/sRGB_to_Unity_Log_r1.cube.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/Cubes/sRGB_to_Unity_Log_r1.cube.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/Cubes/sRGB_to_Unity_Log_r1.cube.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/LUTs.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/LUTs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/LUTs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/LUTs.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/LUTs/NeutralLdrLut.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/LUTs/NeutralLdrLut.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/LUTs/NeutralLdrLut.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/LUTs/NeutralLdrLut.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/LUTs/NeutralLdrLut.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/LUTs/NeutralLdrLut.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/LUTs/NeutralLdrLut.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/LUTs/NeutralLdrLut.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/LensDirt.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/LensDirt.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/LensDirt.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/LensDirt.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt00.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt00.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt00.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt00.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt00.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt00.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt00.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt00.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt01.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt01.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt01.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt01.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt01.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt01.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt01.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt01.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt02.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt02.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt02.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt02.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt02.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt02.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt02.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt02.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt03.png b/Packages/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt03.png similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt03.png rename to Packages/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt03.png diff --git a/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt03.png.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt03.png.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt03.png.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/LensDirt/LensDirt03.png.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/SMAA.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/SMAA.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/SMAA.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/SMAA.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/SMAA/AreaTex.tga b/Packages/com.unity.postprocessing/PostProcessing/Textures/SMAA/AreaTex.tga similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/SMAA/AreaTex.tga rename to Packages/com.unity.postprocessing/PostProcessing/Textures/SMAA/AreaTex.tga diff --git a/com.unity.postprocessing/PostProcessing/Textures/SMAA/AreaTex.tga.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/SMAA/AreaTex.tga.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/SMAA/AreaTex.tga.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/SMAA/AreaTex.tga.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/SMAA/SearchTex.tga b/Packages/com.unity.postprocessing/PostProcessing/Textures/SMAA/SearchTex.tga similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/SMAA/SearchTex.tga rename to Packages/com.unity.postprocessing/PostProcessing/Textures/SMAA/SearchTex.tga diff --git a/com.unity.postprocessing/PostProcessing/Textures/SMAA/SearchTex.tga.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/SMAA/SearchTex.tga.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/SMAA/SearchTex.tga.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/SMAA/SearchTex.tga.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_BlueRed.tga b/Packages/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_BlueRed.tga similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_BlueRed.tga rename to Packages/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_BlueRed.tga diff --git a/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_BlueRed.tga.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_BlueRed.tga.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_BlueRed.tga.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_BlueRed.tga.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_GreenPurple.tga b/Packages/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_GreenPurple.tga similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_GreenPurple.tga rename to Packages/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_GreenPurple.tga diff --git a/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_GreenPurple.tga.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_GreenPurple.tga.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_GreenPurple.tga.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_GreenPurple.tga.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_PurpleGreen.tga b/Packages/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_PurpleGreen.tga similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_PurpleGreen.tga rename to Packages/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_PurpleGreen.tga diff --git a/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_PurpleGreen.tga.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_PurpleGreen.tga.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_PurpleGreen.tga.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_PurpleGreen.tga.meta diff --git a/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_RedBlue.tga b/Packages/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_RedBlue.tga similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_RedBlue.tga rename to Packages/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_RedBlue.tga diff --git a/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_RedBlue.tga.meta b/Packages/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_RedBlue.tga.meta similarity index 100% rename from com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_RedBlue.tga.meta rename to Packages/com.unity.postprocessing/PostProcessing/Textures/SpectralLUTs/SpectralLut_RedBlue.tga.meta diff --git a/com.unity.postprocessing/Tests.meta b/Packages/com.unity.postprocessing/Tests.meta similarity index 100% rename from com.unity.postprocessing/Tests.meta rename to Packages/com.unity.postprocessing/Tests.meta diff --git a/com.unity.postprocessing/Tests/.tests.json b/Packages/com.unity.postprocessing/Tests/.tests.json similarity index 100% rename from com.unity.postprocessing/Tests/.tests.json rename to Packages/com.unity.postprocessing/Tests/.tests.json diff --git a/com.unity.postprocessing/Tests/Editor.meta b/Packages/com.unity.postprocessing/Tests/Editor.meta similarity index 100% rename from com.unity.postprocessing/Tests/Editor.meta rename to Packages/com.unity.postprocessing/Tests/Editor.meta diff --git a/com.unity.postprocessing/Tests/Editor/PostProcessEditorTests.cs b/Packages/com.unity.postprocessing/Tests/Editor/PostProcessEditorTests.cs similarity index 100% rename from com.unity.postprocessing/Tests/Editor/PostProcessEditorTests.cs rename to Packages/com.unity.postprocessing/Tests/Editor/PostProcessEditorTests.cs diff --git a/com.unity.postprocessing/Tests/Editor/PostProcessEditorTests.cs.meta b/Packages/com.unity.postprocessing/Tests/Editor/PostProcessEditorTests.cs.meta similarity index 100% rename from com.unity.postprocessing/Tests/Editor/PostProcessEditorTests.cs.meta rename to Packages/com.unity.postprocessing/Tests/Editor/PostProcessEditorTests.cs.meta diff --git a/com.unity.postprocessing/Tests/Editor/Unity.Postprocessing.Editor.Tests.asmdef b/Packages/com.unity.postprocessing/Tests/Editor/Unity.Postprocessing.Editor.Tests.asmdef similarity index 100% rename from com.unity.postprocessing/Tests/Editor/Unity.Postprocessing.Editor.Tests.asmdef rename to Packages/com.unity.postprocessing/Tests/Editor/Unity.Postprocessing.Editor.Tests.asmdef diff --git a/com.unity.postprocessing/Tests/Editor/Unity.Postprocessing.Editor.Tests.asmdef.meta b/Packages/com.unity.postprocessing/Tests/Editor/Unity.Postprocessing.Editor.Tests.asmdef.meta similarity index 100% rename from com.unity.postprocessing/Tests/Editor/Unity.Postprocessing.Editor.Tests.asmdef.meta rename to Packages/com.unity.postprocessing/Tests/Editor/Unity.Postprocessing.Editor.Tests.asmdef.meta diff --git a/com.unity.postprocessing/Tests/Runtime.meta b/Packages/com.unity.postprocessing/Tests/Runtime.meta similarity index 100% rename from com.unity.postprocessing/Tests/Runtime.meta rename to Packages/com.unity.postprocessing/Tests/Runtime.meta diff --git a/com.unity.postprocessing/Tests/Runtime/PostProcessingRuntimeTests.cs b/Packages/com.unity.postprocessing/Tests/Runtime/PostProcessingRuntimeTests.cs similarity index 100% rename from com.unity.postprocessing/Tests/Runtime/PostProcessingRuntimeTests.cs rename to Packages/com.unity.postprocessing/Tests/Runtime/PostProcessingRuntimeTests.cs diff --git a/com.unity.postprocessing/Tests/Runtime/PostProcessingRuntimeTests.cs.meta b/Packages/com.unity.postprocessing/Tests/Runtime/PostProcessingRuntimeTests.cs.meta similarity index 100% rename from com.unity.postprocessing/Tests/Runtime/PostProcessingRuntimeTests.cs.meta rename to Packages/com.unity.postprocessing/Tests/Runtime/PostProcessingRuntimeTests.cs.meta diff --git a/com.unity.postprocessing/Tests/Runtime/Unity.Postprocessing.Runtime.Tests.asmdef b/Packages/com.unity.postprocessing/Tests/Runtime/Unity.Postprocessing.Runtime.Tests.asmdef similarity index 100% rename from com.unity.postprocessing/Tests/Runtime/Unity.Postprocessing.Runtime.Tests.asmdef rename to Packages/com.unity.postprocessing/Tests/Runtime/Unity.Postprocessing.Runtime.Tests.asmdef diff --git a/com.unity.postprocessing/Tests/Runtime/Unity.Postprocessing.Runtime.Tests.asmdef.meta b/Packages/com.unity.postprocessing/Tests/Runtime/Unity.Postprocessing.Runtime.Tests.asmdef.meta similarity index 100% rename from com.unity.postprocessing/Tests/Runtime/Unity.Postprocessing.Runtime.Tests.asmdef.meta rename to Packages/com.unity.postprocessing/Tests/Runtime/Unity.Postprocessing.Runtime.Tests.asmdef.meta diff --git a/Packages/com.unity.postprocessing/package.json b/Packages/com.unity.postprocessing/package.json new file mode 100644 index 0000000..2cb7183 --- /dev/null +++ b/Packages/com.unity.postprocessing/package.json @@ -0,0 +1,11 @@ +{ + "name": "com.unity.postprocessing", + "version": "3.4.0", + "displayName": "Post Processing", + "unity": "2020.1", + "description": "The post-processing stack (v2) comes with a collection of effects and image filters you can apply to your cameras to improve the visuals of your games.", + "dependencies": { + "com.unity.modules.physics": "1.0.0", + "fidelityfx.fsr": "1.0.0" + } +} \ No newline at end of file diff --git a/com.unity.postprocessing/package.json.meta b/Packages/com.unity.postprocessing/package.json.meta similarity index 100% rename from com.unity.postprocessing/package.json.meta rename to Packages/com.unity.postprocessing/package.json.meta diff --git a/Packages/com.unity.render-pipelines.high-definition/Runtime/RenderPipeline/RenderPass/Fsr3Upscaler.meta b/Packages/com.unity.render-pipelines.high-definition/Runtime/RenderPipeline/RenderPass/Fsr3Upscaler.meta new file mode 100644 index 0000000..d1de5ce --- /dev/null +++ b/Packages/com.unity.render-pipelines.high-definition/Runtime/RenderPipeline/RenderPass/Fsr3Upscaler.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: 73e35266d635ae64f9b765169e998d8c +folderAsset: yes +DefaultImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/com.unity.render-pipelines.high-definition/Runtime/RenderPipeline/RenderPass/Fsr3Upscaler/Fsr3UpscalerPlugin.cs b/Packages/com.unity.render-pipelines.high-definition/Runtime/RenderPipeline/RenderPass/Fsr3Upscaler/Fsr3UpscalerPlugin.cs new file mode 100644 index 0000000..012cf05 --- /dev/null +++ b/Packages/com.unity.render-pipelines.high-definition/Runtime/RenderPipeline/RenderPass/Fsr3Upscaler/Fsr3UpscalerPlugin.cs @@ -0,0 +1,294 @@ +using System; +using System.Collections.Generic; +using FidelityFX; +using FidelityFX.FSR3; + +// We use an old C# trick here to override the UnityEngine.AMD namespace and force FSR2Pass to use code from this namespace instead. +// By exactly mimicking the interface of the AMDUnityPlugin module, we can replace Unity's implementation of FSR2 with one of our own. +// Here we redirect all calls to the open source FSR3 Upscaler port with a bit of glue logic, making the entire thing cross-platform compatible. +namespace UnityEngine.Rendering.HighDefinition.AMD +{ + public static class AMDUnityPlugin + { + internal static Fsr3UpscalerAssets Assets; + + static AMDUnityPlugin() + { + _ = Load(); + } + + public static bool Load() + { + Unload(); + + Assets = Resources.Load("FSR3 Upscaler Assets"); + return Assets != null; + } + + public static bool IsLoaded() => Assets != null; + + internal static void Unload() + { + if (Assets != null) + { + Resources.UnloadAsset(Assets); + Assets = null; + } + } + } + + public class GraphicsDevice + { + private static GraphicsDevice sGraphicsDevice; + + public static GraphicsDevice device => sGraphicsDevice; + + public static int version => 0x00; + + private readonly List _contexts = new(); + + public static GraphicsDevice CreateGraphicsDevice() + { + if (sGraphicsDevice != null) + { + sGraphicsDevice.Destroy(); + sGraphicsDevice = null; + } + + var graphicsDevice = new GraphicsDevice(); + if (graphicsDevice.Initialize()) + { + sGraphicsDevice = graphicsDevice; + return graphicsDevice; + } + + Debug.LogWarning("Failed to initialize FSR3 Upscaler Graphics Device"); + return null; + } + +#if UNITY_EDITOR + private GraphicsDevice() + { + // Ensure resources are properly cleaned up during an in-editor domain reload + UnityEditor.AssemblyReloadEvents.beforeAssemblyReload += Destroy; + } +#endif + + private bool Initialize() + { + return AMDUnityPlugin.Load(); + } + + private void Destroy() + { +#if UNITY_EDITOR + UnityEditor.AssemblyReloadEvents.beforeAssemblyReload -= Destroy; +#endif + + foreach (var context in _contexts) + { + context.Destroy(); + } + + _contexts.Clear(); + + AMDUnityPlugin.Unload(); + } + + public FSR2Context CreateFeature(CommandBuffer cmd, in FSR2CommandInitializationData initSettings) + { + var context = new FSR2Context(); + if (!context.Initialize(initSettings)) + return null; + + _contexts.Add(context); + return context; + } + + public void DestroyFeature(CommandBuffer cmd, FSR2Context fsrContext) + { + if (fsrContext == null) + return; + + fsrContext.Destroy(); + _contexts.Remove(fsrContext); + } + + public void ExecuteFSR2(CommandBuffer cmd, FSR2Context fsrContext, in FSR2TextureTable textures) + { + if (fsrContext == null) + return; + + fsrContext.Execute(cmd, textures); + } + + public bool GetRenderResolutionFromQualityMode(FSR2Quality qualityMode, uint displayWidth, uint displayHeight, out uint renderWidth, out uint renderHeight) + { + Fsr3Upscaler.GetRenderResolutionFromQualityMode(out int rw, out int rh, (int)displayWidth, (int)displayHeight, ConvertQualityMode(qualityMode)); + renderWidth = (uint)rw; + renderHeight = (uint)rh; + return true; + } + + public float GetUpscaleRatioFromQualityMode(FSR2Quality qualityMode) + { + return Fsr3Upscaler.GetUpscaleRatioFromQualityMode(ConvertQualityMode(qualityMode)); + } + + private static Fsr3Upscaler.QualityMode ConvertQualityMode(FSR2Quality qualityMode) + { + // FSR3 offers two more quality modes (Native AA and Ultra Quality) than standard FSR2, so a conversion is needed + const int diff = (int)Fsr3Upscaler.QualityMode.Quality - (int)FSR2Quality.Quality; + return (Fsr3Upscaler.QualityMode)((int)qualityMode + diff); + } + } + + public class FSR2Context + { + public static bool DrawDebugView { get; set; } = false; + + private FSR2CommandInitializationData _initData; + public ref readonly FSR2CommandInitializationData initData => ref _initData; + + private FSR2CommandExecutionData _executeData; + public ref FSR2CommandExecutionData executeData => ref _executeData; + + private readonly Fsr3UpscalerContext _context = new(); + private readonly Fsr3Upscaler.DispatchDescription _dispatchDescription = new(); + + internal bool Initialize(in FSR2CommandInitializationData initSettings) + { + _initData = initSettings; + _executeData = new FSR2CommandExecutionData(); + + if (AMDUnityPlugin.Assets == null || AMDUnityPlugin.Assets.shaders == null) + return false; + + Fsr3Upscaler.InitializationFlags flags = 0; + if (initSettings.GetFlag(FfxFsr2InitializationFlags.EnableHighDynamicRange)) flags |= Fsr3Upscaler.InitializationFlags.EnableHighDynamicRange; + if (initSettings.GetFlag(FfxFsr2InitializationFlags.EnableDisplayResolutionMotionVectors)) flags |= Fsr3Upscaler.InitializationFlags.EnableDisplayResolutionMotionVectors; + if (initSettings.GetFlag(FfxFsr2InitializationFlags.EnableMotionVectorsJitterCancellation)) flags |= Fsr3Upscaler.InitializationFlags.EnableMotionVectorsJitterCancellation; + if (initSettings.GetFlag(FfxFsr2InitializationFlags.DepthInverted)) flags |= Fsr3Upscaler.InitializationFlags.EnableDepthInverted; + if (initSettings.GetFlag(FfxFsr2InitializationFlags.EnableDepthInfinite)) flags |= Fsr3Upscaler.InitializationFlags.EnableDepthInfinite; + if (initSettings.GetFlag(FfxFsr2InitializationFlags.EnableAutoExposure)) flags |= Fsr3Upscaler.InitializationFlags.EnableAutoExposure; + if (initSettings.GetFlag(FfxFsr2InitializationFlags.EnableDynamicResolution)) flags |= Fsr3Upscaler.InitializationFlags.EnableDynamicResolution; + + var contextDescription = new Fsr3Upscaler.ContextDescription + { + Flags = flags, + MaxRenderSize = new Vector2Int((int)initSettings.maxRenderSizeWidth, (int)initSettings.maxRenderSizeHeight), + MaxUpscaleSize = new Vector2Int((int)initSettings.displaySizeWidth, (int)initSettings.displaySizeHeight), + Shaders = AMDUnityPlugin.Assets.shaders, + }; + + _context.Create(contextDescription); + return true; + } + + internal void Destroy() + { + _context.Destroy(); + } + + internal void Execute(CommandBuffer cmd, in FSR2TextureTable textures) + { + _dispatchDescription.Color = new ResourceView(textures.colorInput); + _dispatchDescription.Depth = new ResourceView(textures.depth); + _dispatchDescription.MotionVectors = new ResourceView(textures.motionVectors); + _dispatchDescription.Exposure = new ResourceView(textures.exposureTexture); + _dispatchDescription.Reactive = new ResourceView(textures.biasColorMask); + _dispatchDescription.TransparencyAndComposition = new ResourceView(textures.transparencyMask); + _dispatchDescription.Output = new ResourceView(textures.colorOutput); + _dispatchDescription.JitterOffset = new Vector2(_executeData.jitterOffsetX, _executeData.jitterOffsetY); + _dispatchDescription.MotionVectorScale = new Vector2(_executeData.MVScaleX, _executeData.MVScaleY); + _dispatchDescription.RenderSize = new Vector2Int((int)_executeData.renderSizeWidth, (int)_executeData.renderSizeHeight); + _dispatchDescription.UpscaleSize = new Vector2Int((int)_initData.displaySizeWidth, (int)_initData.displaySizeHeight); + _dispatchDescription.EnableSharpening = _executeData.enableSharpening != 0; + _dispatchDescription.Sharpness = _executeData.sharpness; + _dispatchDescription.FrameTimeDelta = _executeData.frameTimeDelta / 1000f; + _dispatchDescription.PreExposure = _executeData.preExposure; + _dispatchDescription.Reset = _executeData.reset != 0; + _dispatchDescription.CameraNear = _executeData.cameraNear; + _dispatchDescription.CameraFar = _executeData.cameraFar; + _dispatchDescription.CameraFovAngleVertical = _executeData.cameraFovAngleVertical; + _dispatchDescription.ViewSpaceToMetersFactor = 1.0f; // 1 unit == 1 meter in Unity + _dispatchDescription.Flags = DrawDebugView ? Fsr3Upscaler.DispatchFlags.DrawDebugView : 0; + _dispatchDescription.UseTextureArrays = TextureXR.useTexArray && textures.colorInput.dimension == TextureDimension.Tex2DArray; + + _context.Dispatch(_dispatchDescription, cmd); + } + } + + public struct FSR2CommandInitializationData + { + public uint displaySizeHeight; + public uint displaySizeWidth; + public FfxFsr2InitializationFlags ffxFsrFlags; + public uint maxRenderSizeHeight; + public uint maxRenderSizeWidth; + + public readonly bool GetFlag(FfxFsr2InitializationFlags flag) + { + return (ffxFsrFlags & flag) == flag; + } + + public void SetFlag(FfxFsr2InitializationFlags flag, bool value) + { + if (value) + ffxFsrFlags |= flag; + else + ffxFsrFlags &= ~flag; + } + } + + public struct FSR2CommandExecutionData + { + public float cameraFar; + public float cameraFovAngleVertical; + public float cameraNear; + public int enableSharpening; + public float frameTimeDelta; + public float jitterOffsetX; + public float jitterOffsetY; + public float MVScaleX; + public float MVScaleY; + public float preExposure; + public uint renderSizeHeight; + public uint renderSizeWidth; + public int reset; + public float sharpness; + } + + public struct FSR2TextureTable + { + public Texture biasColorMask; + public Texture colorInput; + public Texture colorOutput; + public Texture depth; + public Texture exposureTexture; + public Texture motionVectors; + public Texture reactiveMask; // Note: reactiveMask does not seem to be used at all by HDRP, instead we get a biasColorMask + public Texture transparencyMask; + } + + [Flags] + public enum FfxFsr2InitializationFlags + { + EnableHighDynamicRange = 1 << 0, + EnableDisplayResolutionMotionVectors = 1 << 1, + EnableMotionVectorsJitterCancellation = 1 << 2, + DepthInverted = 1 << 3, + EnableDepthInfinite = 1 << 4, + EnableAutoExposure = 1 << 5, + EnableDynamicResolution = 1 << 6, + EnableTexture1DUsage = 1 << 7, + } + + public enum FSR2Quality + { + Quality, + Balanced, + Performance, + UltraPerformance, + } +} diff --git a/Packages/com.unity.render-pipelines.high-definition/Runtime/RenderPipeline/RenderPass/Fsr3Upscaler/Fsr3UpscalerPlugin.cs.meta b/Packages/com.unity.render-pipelines.high-definition/Runtime/RenderPipeline/RenderPass/Fsr3Upscaler/Fsr3UpscalerPlugin.cs.meta new file mode 100644 index 0000000..d5d7074 --- /dev/null +++ b/Packages/com.unity.render-pipelines.high-definition/Runtime/RenderPipeline/RenderPass/Fsr3Upscaler/Fsr3UpscalerPlugin.cs.meta @@ -0,0 +1,2 @@ +fileFormatVersion: 2 +guid: 774d82037ae83c74baac61a4c4e8bce5 \ No newline at end of file diff --git a/Packages/com.unity.render-pipelines.high-definition/Runtime/RenderPipeline/RenderPass/Fsr3Upscaler/Resources.meta b/Packages/com.unity.render-pipelines.high-definition/Runtime/RenderPipeline/RenderPass/Fsr3Upscaler/Resources.meta new file mode 100644 index 0000000..d0892b2 --- /dev/null +++ b/Packages/com.unity.render-pipelines.high-definition/Runtime/RenderPipeline/RenderPass/Fsr3Upscaler/Resources.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: 0599fe7b8fac94a4b81d1cac815f887f +folderAsset: yes +DefaultImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/com.unity.render-pipelines.high-definition/Runtime/RenderPipeline/RenderPass/Fsr3Upscaler/Resources/FSR3 Upscaler Assets.asset b/Packages/com.unity.render-pipelines.high-definition/Runtime/RenderPipeline/RenderPass/Fsr3Upscaler/Resources/FSR3 Upscaler Assets.asset new file mode 100644 index 0000000..0b741dc --- /dev/null +++ b/Packages/com.unity.render-pipelines.high-definition/Runtime/RenderPipeline/RenderPass/Fsr3Upscaler/Resources/FSR3 Upscaler Assets.asset @@ -0,0 +1,30 @@ +%YAML 1.1 +%TAG !u! tag:unity3d.com,2011: +--- !u!114 &11400000 +MonoBehaviour: + m_ObjectHideFlags: 0 + m_CorrespondingSourceObject: {fileID: 0} + m_PrefabInstance: {fileID: 0} + m_PrefabAsset: {fileID: 0} + m_GameObject: {fileID: 0} + m_Enabled: 1 + m_EditorHideFlags: 0 + m_Script: {fileID: 11500000, guid: 2dbcc608a4754d049a14a0bcce2eb40b, type: 3} + m_Name: FSR3 Upscaler Assets + m_EditorClassIdentifier: + shaders: + prepareInputsPass: {fileID: 7200000, guid: 4f59e5b9179d74844ae06a30ae1e0629, type: 3} + lumaPyramidPass: {fileID: 7200000, guid: d253be05abcdc80428503d3e4cce3a36, type: 3} + shadingChangePyramidPass: {fileID: 7200000, guid: 251e663738905fa4d8817001682d802f, + type: 3} + shadingChangePass: {fileID: 7200000, guid: 9a2bff2f97619ed4989d9b0577ba0641, type: 3} + prepareReactivityPass: {fileID: 7200000, guid: 20e44016ed34b0d4b8de499d1b566c69, + type: 3} + lumaInstabilityPass: {fileID: 7200000, guid: a135306e6d1857e43a86ef20db2a47fe, + type: 3} + accumulatePass: {fileID: 7200000, guid: c9b45f0ae7673694ba57a4aadfe212e9, type: 3} + sharpenPass: {fileID: 7200000, guid: 7aaf5cfff022de2499e9b0412f947f6c, type: 3} + autoGenReactivePass: {fileID: 7200000, guid: 5716b91fdaa4e9e439df6b96a796fe6e, + type: 3} + tcrAutoGenPass: {fileID: 7200000, guid: 75cdc6ef23f08ed498d4da511923fcea, type: 3} + debugViewPass: {fileID: 7200000, guid: cb24a71d54164c54eb5e86839acd48c5, type: 3} diff --git a/Packages/com.unity.render-pipelines.high-definition/Runtime/RenderPipeline/RenderPass/Fsr3Upscaler/Resources/FSR3 Upscaler Assets.asset.meta b/Packages/com.unity.render-pipelines.high-definition/Runtime/RenderPipeline/RenderPass/Fsr3Upscaler/Resources/FSR3 Upscaler Assets.asset.meta new file mode 100644 index 0000000..4c6e899 --- /dev/null +++ b/Packages/com.unity.render-pipelines.high-definition/Runtime/RenderPipeline/RenderPass/Fsr3Upscaler/Resources/FSR3 Upscaler Assets.asset.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: 73e17b6fa270e2f4c90963a8d2fe5394 +NativeFormatImporter: + externalObjects: {} + mainObjectFileID: 11400000 + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/LICENSE.md b/Packages/fidelityfx.fsr/LICENSE.md new file mode 100644 index 0000000..01b3e51 --- /dev/null +++ b/Packages/fidelityfx.fsr/LICENSE.md @@ -0,0 +1,19 @@ +Copyright (c) 2024 Nico de Poel + +Permission is hereby granted, free of charge, to any person obtaining a copy +of this software and associated documentation files (the "Software"), to deal +in the Software without restriction, including without limitation the rights +to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +copies of the Software, and to permit persons to whom the Software is +furnished to do so, subject to the following conditions: + +The above copyright notice and this permission notice shall be included in all +copies or substantial portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +THE SOFTWARE. diff --git a/Packages/fidelityfx.fsr/LICENSE.md.meta b/Packages/fidelityfx.fsr/LICENSE.md.meta new file mode 100644 index 0000000..880be8e --- /dev/null +++ b/Packages/fidelityfx.fsr/LICENSE.md.meta @@ -0,0 +1,7 @@ +fileFormatVersion: 2 +guid: 09b65756f4c90b942b60816c53887dfa +TextScriptImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Runtime.meta b/Packages/fidelityfx.fsr/Runtime.meta new file mode 100644 index 0000000..a57d834 --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: b64c5aa2a705139438e2bcc551a02b97 +folderAsset: yes +DefaultImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Runtime/Common.meta b/Packages/fidelityfx.fsr/Runtime/Common.meta new file mode 100644 index 0000000..7ac2081 --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/Common.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: e87a0efd0d1d4644abdef16b5a28bd0f +folderAsset: yes +DefaultImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Runtime/Common/ResourceView.cs b/Packages/fidelityfx.fsr/Runtime/Common/ResourceView.cs new file mode 100644 index 0000000..e5946a3 --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/Common/ResourceView.cs @@ -0,0 +1,55 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +using UnityEngine.Rendering; + +namespace FidelityFX +{ + /// + /// An immutable structure wrapping all of the necessary information to bind a specific buffer or attachment of a render target to a compute shader. + /// + public readonly struct ResourceView + { + /// + /// This value is the equivalent of not setting any value at all; all struct fields will have their default values. + /// It does not refer to a valid texture, therefore any variable set to this value should be checked for IsValid and reassigned before being bound to a shader. + /// + public static readonly ResourceView Unassigned = new ResourceView(default); + + /// + /// This value contains a valid texture reference that can be bound to a shader, however it is just an empty placeholder texture. + /// Binding this to a shader can be seen as setting the texture variable inside the shader to null. + /// + public static readonly ResourceView None = new ResourceView(BuiltinRenderTextureType.None); + + public ResourceView(in RenderTargetIdentifier renderTarget, RenderTextureSubElement subElement = RenderTextureSubElement.Default, int mipLevel = 0) + { + RenderTarget = renderTarget; + SubElement = subElement; + MipLevel = mipLevel; + } + + public bool IsValid => !RenderTarget.Equals(default); + + public readonly RenderTargetIdentifier RenderTarget; + public readonly RenderTextureSubElement SubElement; + public readonly int MipLevel; + } +} diff --git a/Packages/fidelityfx.fsr/Runtime/Common/ResourceView.cs.meta b/Packages/fidelityfx.fsr/Runtime/Common/ResourceView.cs.meta new file mode 100644 index 0000000..901157b --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/Common/ResourceView.cs.meta @@ -0,0 +1,11 @@ +fileFormatVersion: 2 +guid: eb9fdfac33a070740b66520d88f43ab7 +MonoImporter: + externalObjects: {} + serializedVersion: 2 + defaultReferences: [] + executionOrder: 0 + icon: {instanceID: 0} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Runtime/FSR2.meta b/Packages/fidelityfx.fsr/Runtime/FSR2.meta new file mode 100644 index 0000000..696ee85 --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR2.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: cce3e495c6ece2145b041d0a6e43bb26 +folderAsset: yes +DefaultImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2.cs b/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2.cs new file mode 100644 index 0000000..4632b96 --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2.cs @@ -0,0 +1,302 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +using System; +using System.Runtime.InteropServices; +using UnityEngine; +using UnityEngine.Rendering; + +namespace FidelityFX.FSR2 +{ + /// + /// A collection of helper functions and data structures required by the FSR2 process. + /// + public static class Fsr2 + { + /// + /// Creates a new FSR2 context with standard parameters that are appropriate for the current platform. + /// + public static Fsr2Context CreateContext(Vector2Int displaySize, Vector2Int maxRenderSize, Fsr2Shaders shaders, InitializationFlags flags = 0) + { + if (SystemInfo.usesReversedZBuffer) + flags |= InitializationFlags.EnableDepthInverted; + else + flags &= ~InitializationFlags.EnableDepthInverted; + +#if UNITY_EDITOR || DEVELOPMENT_BUILD + flags |= InitializationFlags.EnableDebugChecking; +#endif + + Debug.Log($"Setting up FSR2 with render size: {maxRenderSize.x}x{maxRenderSize.y}, display size: {displaySize.x}x{displaySize.y}, flags: {flags}"); + + var contextDescription = new ContextDescription + { + Flags = flags, + DisplaySize = displaySize, + MaxRenderSize = maxRenderSize, + Shaders = shaders, + }; + + var context = new Fsr2Context(); + context.Create(contextDescription); + return context; + } + + public static float GetUpscaleRatioFromQualityMode(QualityMode qualityMode) + { + switch (qualityMode) + { + case QualityMode.NativeAA: + return 1.0f; + case QualityMode.UltraQuality: + return 1.2f; + case QualityMode.Quality: + return 1.5f; + case QualityMode.Balanced: + return 1.7f; + case QualityMode.Performance: + return 2.0f; + case QualityMode.UltraPerformance: + return 3.0f; + default: + return 1.0f; + } + } + + public static void GetRenderResolutionFromQualityMode( + out int renderWidth, out int renderHeight, + int displayWidth, int displayHeight, QualityMode qualityMode) + { + float ratio = GetUpscaleRatioFromQualityMode(qualityMode); + renderWidth = Mathf.RoundToInt(displayWidth / ratio); + renderHeight = Mathf.RoundToInt(displayHeight / ratio); + } + + public static float GetMipmapBiasOffset(int renderWidth, int displayWidth) + { + return Mathf.Log((float)renderWidth / displayWidth, 2.0f) - 1.0f; + } + + public static int GetJitterPhaseCount(int renderWidth, int displayWidth) + { + const float basePhaseCount = 8.0f; + int jitterPhaseCount = (int)(basePhaseCount * Mathf.Pow((float)displayWidth / renderWidth, 2.0f)); + return jitterPhaseCount; + } + + public static void GetJitterOffset(out float outX, out float outY, int index, int phaseCount) + { + outX = Halton((index % phaseCount) + 1, 2) - 0.5f; + outY = Halton((index % phaseCount) + 1, 3) - 0.5f; + } + + // Calculate halton number for index and base. + private static float Halton(int index, int @base) + { + float f = 1.0f, result = 0.0f; + + for (int currentIndex = index; currentIndex > 0;) { + + f /= @base; + result += f * (currentIndex % @base); + currentIndex = (int)Mathf.Floor((float)currentIndex / @base); + } + + return result; + } + + public static float Lanczos2(float value) + { + return Mathf.Abs(value) < Mathf.Epsilon ? 1.0f : Mathf.Sin(Mathf.PI * value) / (Mathf.PI * value) * (Mathf.Sin(0.5f * Mathf.PI * value) / (0.5f * Mathf.PI * value)); + } + +#if !UNITY_2021_1_OR_NEWER + internal static void SetBufferData(this CommandBuffer commandBuffer, ComputeBuffer computeBuffer, Array data) + { + commandBuffer.SetComputeBufferData(computeBuffer, data); + } +#endif + + public enum QualityMode + { + NativeAA = 0, + UltraQuality = 1, + Quality = 2, + Balanced = 3, + Performance = 4, + UltraPerformance = 5, + } + + [Flags] + public enum InitializationFlags + { + EnableHighDynamicRange = 1 << 0, + EnableDisplayResolutionMotionVectors = 1 << 1, + EnableMotionVectorsJitterCancellation = 1 << 2, + EnableDepthInverted = 1 << 3, + EnableDepthInfinite = 1 << 4, + EnableAutoExposure = 1 << 5, + EnableDynamicResolution = 1 << 6, + EnableFP16Usage = 1 << 7, + EnableDebugChecking = 1 << 8, + } + + /// + /// A structure encapsulating the parameters required to initialize FidelityFX Super Resolution 2 upscaling. + /// + public struct ContextDescription + { + public InitializationFlags Flags; + public Vector2Int MaxRenderSize; + public Vector2Int DisplaySize; + public Fsr2Shaders Shaders; + } + + /// + /// A structure encapsulating the parameters for dispatching the various passes of FidelityFX Super Resolution 2. + /// + public class DispatchDescription + { + public ResourceView Color; + public ResourceView Depth; + public ResourceView MotionVectors; + public ResourceView Exposure; // optional + public ResourceView Reactive; // optional + public ResourceView TransparencyAndComposition; // optional + public ResourceView Output; + public Vector2 JitterOffset; + public Vector2 MotionVectorScale; + public Vector2Int RenderSize; + public Vector2Int InputResourceSize; + public bool EnableSharpening; + public float Sharpness; + public float FrameTimeDelta; // in seconds + public float PreExposure; + public bool Reset; + public float CameraNear; + public float CameraFar; + public float CameraFovAngleVertical; + public float ViewSpaceToMetersFactor; + public bool UseTextureArrays; // Enable texture array bindings, primarily used for HDRP and XR + + // EXPERIMENTAL reactive mask generation parameters + public bool EnableAutoReactive; + public ResourceView ColorOpaqueOnly; + public float AutoTcThreshold = 0.05f; + public float AutoTcScale = 1.0f; + public float AutoReactiveScale = 5.0f; + public float AutoReactiveMax = 0.9f; + } + + /// + /// A structure encapsulating the parameters for automatic generation of a reactive mask. + /// The default values for Scale, CutoffThreshold, BinaryValue and Flags were taken from the FSR2 demo project. + /// + public class GenerateReactiveDescription + { + public ResourceView ColorOpaqueOnly; + public ResourceView ColorPreUpscale; + public ResourceView OutReactive; + public Vector2Int RenderSize; + public float Scale = 0.5f; + public float CutoffThreshold = 0.2f; + public float BinaryValue = 0.9f; + public GenerateReactiveFlags Flags = GenerateReactiveFlags.ApplyTonemap | GenerateReactiveFlags.ApplyThreshold | GenerateReactiveFlags.UseComponentsMax; + } + + [Flags] + public enum GenerateReactiveFlags + { + ApplyTonemap = 1 << 0, + ApplyInverseTonemap = 1 << 1, + ApplyThreshold = 1 << 2, + UseComponentsMax = 1 << 3, + } + + [Serializable, StructLayout(LayoutKind.Sequential)] + internal struct UpscalerConstants + { + public Vector2Int renderSize; + public Vector2Int maxRenderSize; + public Vector2Int displaySize; + public Vector2Int inputColorResourceDimensions; + public Vector2Int lumaMipDimensions; + public int lumaMipLevelToUse; + public int frameIndex; + + public Vector4 deviceToViewDepth; + public Vector2 jitterOffset; + public Vector2 motionVectorScale; + public Vector2 downscaleFactor; + public Vector2 motionVectorJitterCancellation; + public float preExposure; + public float previousFramePreExposure; + public float tanHalfFOV; + public float jitterPhaseCount; + public float deltaTime; + public float dynamicResChangeFactor; + public float viewSpaceToMetersFactor; + public float padding; + } + + [Serializable, StructLayout(LayoutKind.Sequential)] + internal struct SpdConstants + { + public uint mips; + public uint numWorkGroups; + public uint workGroupOffsetX, workGroupOffsetY; + public uint renderSizeX, renderSizeY; + } + + [Serializable, StructLayout(LayoutKind.Sequential)] + internal struct GenerateReactiveConstants + { + public float scale; + public float threshold; + public float binaryValue; + public uint flags; + } + + [Serializable, StructLayout(LayoutKind.Sequential)] + internal struct GenerateReactiveConstants2 + { + public float autoTcThreshold; + public float autoTcScale; + public float autoReactiveScale; + public float autoReactiveMax; + } + + [Serializable, StructLayout(LayoutKind.Sequential)] + internal struct RcasConstants + { + public RcasConstants(uint sharpness, uint halfSharp) + { + this.sharpness = sharpness; + this.halfSharp = halfSharp; + dummy0 = dummy1 = 0; + } + + public readonly uint sharpness; + public readonly uint halfSharp; + public readonly uint dummy0; + public readonly uint dummy1; + } + } +} diff --git a/Assets/Scripts/Core/Fsr3Upscaler.cs.meta b/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2.cs.meta similarity index 100% rename from Assets/Scripts/Core/Fsr3Upscaler.cs.meta rename to Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2.cs.meta diff --git a/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Assets.cs b/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Assets.cs new file mode 100644 index 0000000..c30a8f6 --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Assets.cs @@ -0,0 +1,152 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +using UnityEngine; + +namespace FidelityFX.FSR2 +{ + /// + /// Scriptable object containing all shader resources required by FidelityFX Super Resolution 2 (FSR2). + /// These can be stored in an asset file and referenced from a scene or prefab, avoiding the need to load the shaders from a Resources folder. + /// + [CreateAssetMenu(fileName = "FSR2 Assets", menuName = "FidelityFX/FSR2 Assets", order = 1102)] + public class Fsr2Assets : ScriptableObject + { + public Fsr2Shaders shaders; + +#if UNITY_EDITOR + private void Reset() + { + shaders = new Fsr2Shaders + { + computeLuminancePyramidPass = FindComputeShader("ffx_fsr2_compute_luminance_pyramid_pass"), + reconstructPreviousDepthPass = FindComputeShader("ffx_fsr2_reconstruct_previous_depth_pass"), + depthClipPass = FindComputeShader("ffx_fsr2_depth_clip_pass"), + lockPass = FindComputeShader("ffx_fsr2_lock_pass"), + accumulatePass = FindComputeShader("ffx_fsr2_accumulate_pass"), + sharpenPass = FindComputeShader("ffx_fsr2_rcas_pass"), + autoGenReactivePass = FindComputeShader("ffx_fsr2_autogen_reactive_pass"), + tcrAutoGenPass = FindComputeShader("ffx_fsr2_tcr_autogen_pass"), + }; + } + + private static ComputeShader FindComputeShader(string name) + { + string[] assetGuids = UnityEditor.AssetDatabase.FindAssets($"t:ComputeShader {name}"); + if (assetGuids == null || assetGuids.Length == 0) + return null; + + string assetPath = UnityEditor.AssetDatabase.GUIDToAssetPath(assetGuids[0]); + return UnityEditor.AssetDatabase.LoadAssetAtPath(assetPath); + } +#endif + } + + /// + /// All the compute shaders used by FSR2. + /// + [System.Serializable] + public class Fsr2Shaders + { + /// + /// The compute shader used by the luminance pyramid computation pass. + /// + public ComputeShader computeLuminancePyramidPass; + + /// + /// The compute shader used by the previous depth reconstruction pass. + /// + public ComputeShader reconstructPreviousDepthPass; + + /// + /// The compute shader used by the depth clip pass. + /// + public ComputeShader depthClipPass; + + /// + /// The compute shader used by the lock pass. + /// + public ComputeShader lockPass; + + /// + /// The compute shader used by the accumulation pass. + /// + public ComputeShader accumulatePass; + + /// + /// The compute shader used by the RCAS sharpening pass. + /// + public ComputeShader sharpenPass; + + /// + /// The compute shader used to auto-generate a reactive mask. + /// + public ComputeShader autoGenReactivePass; + + /// + /// The compute shader used to auto-generate a transparency & composition mask. + /// + public ComputeShader tcrAutoGenPass; + + /// + /// Returns a copy of this class and its contents. + /// + public Fsr2Shaders Clone() + { + return (Fsr2Shaders)MemberwiseClone(); + } + + /// + /// Returns a copy of this class with clones of all its shaders. + /// This can be useful if you're running multiple FSR2 instances with different shader configurations. + /// Be sure to clean up these clones through Dispose once you're done with them. + /// + public Fsr2Shaders DeepCopy() + { + return new Fsr2Shaders + { + computeLuminancePyramidPass = Object.Instantiate(computeLuminancePyramidPass), + reconstructPreviousDepthPass = Object.Instantiate(reconstructPreviousDepthPass), + depthClipPass = Object.Instantiate(depthClipPass), + lockPass = Object.Instantiate(lockPass), + accumulatePass = Object.Instantiate(accumulatePass), + sharpenPass = Object.Instantiate(sharpenPass), + autoGenReactivePass = Object.Instantiate(autoGenReactivePass), + tcrAutoGenPass = Object.Instantiate(tcrAutoGenPass), + }; + } + + /// + /// Destroy all the shaders within this instance. + /// Use this only on clones created through DeepCopy. + /// + public void Dispose() + { + Object.Destroy(computeLuminancePyramidPass); + Object.Destroy(reconstructPreviousDepthPass); + Object.Destroy(depthClipPass); + Object.Destroy(lockPass); + Object.Destroy(accumulatePass); + Object.Destroy(sharpenPass); + Object.Destroy(autoGenReactivePass); + Object.Destroy(tcrAutoGenPass); + } + } +} diff --git a/Assets/Scripts/Core/Fsr3UpscalerAssets.cs.meta b/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Assets.cs.meta similarity index 100% rename from Assets/Scripts/Core/Fsr3UpscalerAssets.cs.meta rename to Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Assets.cs.meta diff --git a/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Callbacks.cs b/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Callbacks.cs new file mode 100644 index 0000000..06151b5 --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Callbacks.cs @@ -0,0 +1,81 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +using UnityEngine; + +namespace FidelityFX.FSR2 +{ + /// + /// A collection of callbacks required by the FSR2 process. + /// This allows some customization by the game dev on how to integrate FSR2 upscaling into their own game setup. + /// + public interface IFsr2Callbacks + { + /// + /// Apply a mipmap bias to in-game textures to prevent them from becoming blurry as the internal rendering resolution lowers. + /// This will need to be customized on a per-game basis, as there is no clear universal way to determine what are "in-game" textures. + /// The default implementation will simply apply a mipmap bias to all 2D textures, which will include things like UI textures and which might miss things like terrain texture arrays. + /// + /// Depending on how your game organizes its assets, you will want to create a filter that more specifically selects the textures that need to have this mipmap bias applied. + /// You may also want to store the bias offset value and apply it to any assets that are loaded in on demand. + /// + void ApplyMipmapBias(float biasOffset); + + void UndoMipmapBias(); + } + + /// + /// Default implementation of IFsr2Callbacks. + /// These are fine for testing but a proper game will want to extend and override these methods. + /// + public class Fsr2CallbacksBase: IFsr2Callbacks + { + protected float CurrentBiasOffset = 0; + + public virtual void ApplyMipmapBias(float biasOffset) + { + if (float.IsNaN(biasOffset) || float.IsInfinity(biasOffset)) + return; + + CurrentBiasOffset += biasOffset; + + if (Mathf.Approximately(CurrentBiasOffset, 0f)) + { + CurrentBiasOffset = 0f; + } + + foreach (var texture in Resources.FindObjectsOfTypeAll()) + { + if (texture.mipmapCount <= 1) + continue; + + texture.mipMapBias += biasOffset; + } + } + + public virtual void UndoMipmapBias() + { + if (CurrentBiasOffset == 0f) + return; + + ApplyMipmapBias(-CurrentBiasOffset); + } + } +} diff --git a/Assets/Scripts/Core/Fsr3UpscalerCallbacks.cs.meta b/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Callbacks.cs.meta similarity index 100% rename from Assets/Scripts/Core/Fsr3UpscalerCallbacks.cs.meta rename to Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Callbacks.cs.meta diff --git a/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Context.cs b/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Context.cs new file mode 100644 index 0000000..38d380b --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Context.cs @@ -0,0 +1,618 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +using System; +using System.Runtime.InteropServices; +using UnityEngine; +using UnityEngine.Rendering; + +namespace FidelityFX.FSR2 +{ + /// + /// This class loosely matches the FfxFsr2Context struct from the original FSR2 codebase. + /// It manages the various resources and compute passes required by the FSR2 process. + /// Note that this class does not know anything about Unity render pipelines; all it knows is CommandBuffers and RenderTargetIdentifiers. + /// This should make it suitable for integration with any of the available Unity render pipelines. + /// + public class Fsr2Context + { + private const int MaxQueuedFrames = 16; + + private Fsr2.ContextDescription _contextDescription; + private CommandBuffer _commandBuffer; + + private Fsr2Pass _computeLuminancePyramidPass; + private Fsr2Pass _reconstructPreviousDepthPass; + private Fsr2Pass _depthClipPass; + private Fsr2Pass _lockPass; + private Fsr2Pass _accumulatePass; + private Fsr2Pass _sharpenPass; + private Fsr2Pass _generateReactivePass; + private Fsr2Pass _tcrAutogeneratePass; + + private readonly Fsr2Resources _resources = new Fsr2Resources(); + + private ComputeBuffer _upscalerConstantsBuffer; + private readonly Fsr2.UpscalerConstants[] _upscalerConstantsArray = { new Fsr2.UpscalerConstants() }; + private ref Fsr2.UpscalerConstants UpscalerConsts => ref _upscalerConstantsArray[0]; + + private ComputeBuffer _spdConstantsBuffer; + private readonly Fsr2.SpdConstants[] _spdConstantsArray = { new Fsr2.SpdConstants() }; + private ref Fsr2.SpdConstants SpdConsts => ref _spdConstantsArray[0]; + + private ComputeBuffer _rcasConstantsBuffer; + private readonly Fsr2.RcasConstants[] _rcasConstantsArray = new Fsr2.RcasConstants[1]; + private ref Fsr2.RcasConstants RcasConsts => ref _rcasConstantsArray[0]; + + private ComputeBuffer _generateReactiveConstantsBuffer; + private readonly Fsr2.GenerateReactiveConstants[] _generateReactiveConstantsArray = { new Fsr2.GenerateReactiveConstants() }; + private ref Fsr2.GenerateReactiveConstants GenReactiveConsts => ref _generateReactiveConstantsArray[0]; + + private ComputeBuffer _tcrAutogenerateConstantsBuffer; + private readonly Fsr2.GenerateReactiveConstants2[] _tcrAutogenerateConstantsArray = { new Fsr2.GenerateReactiveConstants2() }; + private ref Fsr2.GenerateReactiveConstants2 TcrAutoGenConsts => ref _tcrAutogenerateConstantsArray[0]; + + private bool _firstExecution; + private Vector2 _previousJitterOffset; + private int _resourceFrameIndex; + + public void Create(Fsr2.ContextDescription contextDescription) + { + _contextDescription = contextDescription; + _commandBuffer = new CommandBuffer { name = "FSR2" }; + + _upscalerConstantsBuffer = CreateConstantBuffer(); + _spdConstantsBuffer = CreateConstantBuffer(); + _rcasConstantsBuffer = CreateConstantBuffer(); + _generateReactiveConstantsBuffer = CreateConstantBuffer(); + _tcrAutogenerateConstantsBuffer = CreateConstantBuffer(); + + // Set defaults + _firstExecution = true; + _resourceFrameIndex = 0; + + UpscalerConsts.displaySize = _contextDescription.DisplaySize; + + _resources.Create(_contextDescription); + CreatePasses(); + } + + private void CreatePasses() + { + _computeLuminancePyramidPass = new Fsr2ComputeLuminancePyramidPass(_contextDescription, _resources, _upscalerConstantsBuffer, _spdConstantsBuffer); + _reconstructPreviousDepthPass = new Fsr2ReconstructPreviousDepthPass(_contextDescription, _resources, _upscalerConstantsBuffer); + _depthClipPass = new Fsr2DepthClipPass(_contextDescription, _resources, _upscalerConstantsBuffer); + _lockPass = new Fsr2LockPass(_contextDescription, _resources, _upscalerConstantsBuffer); + _accumulatePass = new Fsr2AccumulatePass(_contextDescription, _resources, _upscalerConstantsBuffer); + _sharpenPass = new Fsr2SharpenPass(_contextDescription, _resources, _upscalerConstantsBuffer, _rcasConstantsBuffer); + _generateReactivePass = new Fsr2GenerateReactivePass(_contextDescription, _resources, _generateReactiveConstantsBuffer); + _tcrAutogeneratePass = new Fsr2TcrAutogeneratePass(_contextDescription, _resources, _upscalerConstantsBuffer, _tcrAutogenerateConstantsBuffer); + } + + public void Destroy() + { + DestroyPass(ref _tcrAutogeneratePass); + DestroyPass(ref _generateReactivePass); + DestroyPass(ref _sharpenPass); + DestroyPass(ref _accumulatePass); + DestroyPass(ref _lockPass); + DestroyPass(ref _depthClipPass); + DestroyPass(ref _reconstructPreviousDepthPass); + DestroyPass(ref _computeLuminancePyramidPass); + + _resources.Destroy(); + + DestroyConstantBuffer(ref _tcrAutogenerateConstantsBuffer); + DestroyConstantBuffer(ref _generateReactiveConstantsBuffer); + DestroyConstantBuffer(ref _rcasConstantsBuffer); + DestroyConstantBuffer(ref _spdConstantsBuffer); + DestroyConstantBuffer(ref _upscalerConstantsBuffer); + + if (_commandBuffer != null) + { + _commandBuffer.Dispose(); + _commandBuffer = null; + } + } + + public void Dispatch(Fsr2.DispatchDescription dispatchParams) + { + _commandBuffer.Clear(); + Dispatch(dispatchParams, _commandBuffer); + Graphics.ExecuteCommandBuffer(_commandBuffer); + } + + public void Dispatch(Fsr2.DispatchDescription dispatchParams, CommandBuffer commandBuffer) + { + if ((_contextDescription.Flags & Fsr2.InitializationFlags.EnableDebugChecking) != 0) + { + DebugCheckDispatch(dispatchParams); + } + + if (dispatchParams.UseTextureArrays) + commandBuffer.EnableShaderKeyword("UNITY_FSR_TEXTURE2D_X_ARRAY"); + + if (_firstExecution) + { + commandBuffer.SetRenderTarget(_resources.LockStatus[0]); + commandBuffer.ClearRenderTarget(false, true, Color.clear); + commandBuffer.SetRenderTarget(_resources.LockStatus[1]); + commandBuffer.ClearRenderTarget(false, true, Color.clear); + } + + int frameIndex = _resourceFrameIndex % 2; + bool resetAccumulation = dispatchParams.Reset || _firstExecution; + _firstExecution = false; + + // If auto exposure is enabled use the auto exposure SRV, otherwise what the app sends + if ((_contextDescription.Flags & Fsr2.InitializationFlags.EnableAutoExposure) != 0) + dispatchParams.Exposure = new ResourceView(_resources.AutoExposure); + else if (!dispatchParams.Exposure.IsValid) + dispatchParams.Exposure = new ResourceView(_resources.DefaultExposure); + + if (dispatchParams.EnableAutoReactive) + { + // Create the auto-TCR resources only when we need them + if (_resources.AutoReactive == null) + _resources.CreateTcrAutogenResources(_contextDescription); + + if (resetAccumulation) + { + RenderTargetIdentifier opaqueOnly = dispatchParams.ColorOpaqueOnly.IsValid ? dispatchParams.ColorOpaqueOnly.RenderTarget : Fsr2ShaderIDs.SrvOpaqueOnly; + commandBuffer.Blit(_resources.PrevPreAlpha[frameIndex ^ 1], opaqueOnly); + } + } + else if (_resources.AutoReactive != null) + { + // Destroy the auto-TCR resources if we don't use the feature + _resources.DestroyTcrAutogenResources(); + } + + if (!dispatchParams.Reactive.IsValid) dispatchParams.Reactive = new ResourceView(_resources.DefaultReactive); + if (!dispatchParams.TransparencyAndComposition.IsValid) dispatchParams.TransparencyAndComposition = new ResourceView(_resources.DefaultReactive); + Fsr2Resources.CreateAliasableResources(commandBuffer, _contextDescription, dispatchParams); + + SetupConstants(dispatchParams, resetAccumulation); + + // Reactive mask bias + const int threadGroupWorkRegionDim = 8; + int dispatchSrcX = (UpscalerConsts.renderSize.x + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; + int dispatchSrcY = (UpscalerConsts.renderSize.y + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; + int dispatchDstX = (_contextDescription.DisplaySize.x + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; + int dispatchDstY = (_contextDescription.DisplaySize.y + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; + + // Clear reconstructed depth for max depth store + if (resetAccumulation) + { + commandBuffer.SetRenderTarget(_resources.LockStatus[frameIndex ^ 1]); + commandBuffer.ClearRenderTarget(false, true, Color.clear); + + commandBuffer.SetRenderTarget(_resources.InternalUpscaled[frameIndex ^ 1]); + commandBuffer.ClearRenderTarget(false, true, Color.clear); + + commandBuffer.SetRenderTarget(_resources.SceneLuminance); + commandBuffer.ClearRenderTarget(false, true, Color.clear); + + // Auto exposure always used to track luma changes in locking logic + commandBuffer.SetRenderTarget(_resources.AutoExposure); + commandBuffer.ClearRenderTarget(false, true, new Color(0f, 1e8f, 0f, 0f)); + + // Reset atomic counter to 0 + commandBuffer.SetRenderTarget(_resources.SpdAtomicCounter); + commandBuffer.ClearRenderTarget(false, true, Color.clear); + } + + // FSR3: need to clear here since we need the content of this surface for frame interpolation, so clearing in the lock pass is not an option + bool depthInverted = (_contextDescription.Flags & Fsr2.InitializationFlags.EnableDepthInverted) == Fsr2.InitializationFlags.EnableDepthInverted; + commandBuffer.SetRenderTarget(Fsr2ShaderIDs.UavReconstructedPrevNearestDepth); + commandBuffer.ClearRenderTarget(false, true, depthInverted ? Color.clear : Color.white); + + // Auto exposure + SetupSpdConstants(dispatchParams, out var dispatchThreadGroupCount); + + // Initialize constant buffers data + commandBuffer.SetBufferData(_upscalerConstantsBuffer, _upscalerConstantsArray); + commandBuffer.SetBufferData(_spdConstantsBuffer, _spdConstantsArray); + + // Auto reactive + if (dispatchParams.EnableAutoReactive) + { + GenerateTransparencyCompositionReactive(dispatchParams, commandBuffer, frameIndex); + dispatchParams.Reactive = new ResourceView(_resources.AutoReactive); + dispatchParams.TransparencyAndComposition = new ResourceView(_resources.AutoComposition); + } + + // Compute luminance pyramid + _computeLuminancePyramidPass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchThreadGroupCount.x, dispatchThreadGroupCount.y); + + // Reconstruct previous depth + _reconstructPreviousDepthPass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchSrcX, dispatchSrcY); + + // Depth clip + _depthClipPass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchSrcX, dispatchSrcY); + + // Create locks + _lockPass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchSrcX, dispatchSrcY); + + // Accumulate + _accumulatePass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchDstX, dispatchDstY); + + if (dispatchParams.EnableSharpening) + { + // Compute the constants + SetupRcasConstants(dispatchParams); + commandBuffer.SetBufferData(_rcasConstantsBuffer, _rcasConstantsArray); + + // Dispatch RCAS + const int threadGroupWorkRegionDimRcas = 16; + int threadGroupsX = (Screen.width + threadGroupWorkRegionDimRcas - 1) / threadGroupWorkRegionDimRcas; + int threadGroupsY = (Screen.height + threadGroupWorkRegionDimRcas - 1) / threadGroupWorkRegionDimRcas; + _sharpenPass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, threadGroupsX, threadGroupsY); + } + + _resourceFrameIndex = (_resourceFrameIndex + 1) % MaxQueuedFrames; + + Fsr2Resources.DestroyAliasableResources(commandBuffer); + + commandBuffer.DisableShaderKeyword("UNITY_FSR_TEXTURE2D_X_ARRAY"); + } + + public void GenerateReactiveMask(Fsr2.GenerateReactiveDescription dispatchParams) + { + _commandBuffer.Clear(); + GenerateReactiveMask(dispatchParams, _commandBuffer); + Graphics.ExecuteCommandBuffer(_commandBuffer); + } + + public void GenerateReactiveMask(Fsr2.GenerateReactiveDescription dispatchParams, CommandBuffer commandBuffer) + { + const int threadGroupWorkRegionDim = 8; + int dispatchSrcX = (dispatchParams.RenderSize.x + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; + int dispatchSrcY = (dispatchParams.RenderSize.y + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; + + GenReactiveConsts.scale = dispatchParams.Scale; + GenReactiveConsts.threshold = dispatchParams.CutoffThreshold; + GenReactiveConsts.binaryValue = dispatchParams.BinaryValue; + GenReactiveConsts.flags = (uint)dispatchParams.Flags; + commandBuffer.SetBufferData(_generateReactiveConstantsBuffer, _generateReactiveConstantsArray); + + ((Fsr2GenerateReactivePass)_generateReactivePass).ScheduleDispatch(commandBuffer, dispatchParams, dispatchSrcX, dispatchSrcY); + } + + private void GenerateTransparencyCompositionReactive(Fsr2.DispatchDescription dispatchParams, CommandBuffer commandBuffer, int frameIndex) + { + const int threadGroupWorkRegionDim = 8; + int dispatchSrcX = (dispatchParams.RenderSize.x + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; + int dispatchSrcY = (dispatchParams.RenderSize.y + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; + + TcrAutoGenConsts.autoTcThreshold = dispatchParams.AutoTcThreshold; + TcrAutoGenConsts.autoTcScale = dispatchParams.AutoTcScale; + TcrAutoGenConsts.autoReactiveScale = dispatchParams.AutoReactiveScale; + TcrAutoGenConsts.autoReactiveMax = dispatchParams.AutoReactiveMax; + commandBuffer.SetBufferData(_tcrAutogenerateConstantsBuffer, _tcrAutogenerateConstantsArray); + + _tcrAutogeneratePass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchSrcX, dispatchSrcY); + } + + private void SetupConstants(Fsr2.DispatchDescription dispatchParams, bool resetAccumulation) + { + ref Fsr2.UpscalerConstants constants = ref UpscalerConsts; + + constants.jitterOffset = dispatchParams.JitterOffset; + constants.renderSize = dispatchParams.RenderSize; + constants.maxRenderSize = _contextDescription.MaxRenderSize; + constants.inputColorResourceDimensions = dispatchParams.InputResourceSize; + + // Compute the horizontal FOV for the shader from the vertical one + float aspectRatio = (float)dispatchParams.RenderSize.x / dispatchParams.RenderSize.y; + float cameraAngleHorizontal = Mathf.Atan(Mathf.Tan(dispatchParams.CameraFovAngleVertical / 2.0f) * aspectRatio) * 2.0f; + constants.tanHalfFOV = Mathf.Tan(cameraAngleHorizontal * 0.5f); + constants.viewSpaceToMetersFactor = (dispatchParams.ViewSpaceToMetersFactor > 0.0f) ? dispatchParams.ViewSpaceToMetersFactor : 1.0f; + + // Compute params to enable device depth to view space depth computation in shader + constants.deviceToViewDepth = SetupDeviceDepthToViewSpaceDepthParams(dispatchParams); + + // To be updated if resource is larger than the actual image size + constants.downscaleFactor = new Vector2((float)constants.renderSize.x / _contextDescription.DisplaySize.x, (float)constants.renderSize.y / _contextDescription.DisplaySize.y); + constants.previousFramePreExposure = constants.preExposure; + constants.preExposure = (dispatchParams.PreExposure != 0) ? dispatchParams.PreExposure : 1.0f; + + // Motion vector data + Vector2Int motionVectorsTargetSize = (_contextDescription.Flags & Fsr2.InitializationFlags.EnableDisplayResolutionMotionVectors) != 0 ? constants.displaySize : constants.renderSize; + constants.motionVectorScale = dispatchParams.MotionVectorScale / motionVectorsTargetSize; + + // Compute jitter cancellation + if ((_contextDescription.Flags & Fsr2.InitializationFlags.EnableMotionVectorsJitterCancellation) != 0) + { + constants.motionVectorJitterCancellation = (_previousJitterOffset - constants.jitterOffset) / motionVectorsTargetSize; + _previousJitterOffset = constants.jitterOffset; + } + + int jitterPhaseCount = Fsr2.GetJitterPhaseCount(dispatchParams.RenderSize.x, _contextDescription.DisplaySize.x); + if (resetAccumulation || constants.jitterPhaseCount == 0) + { + constants.jitterPhaseCount = jitterPhaseCount; + } + else + { + int jitterPhaseCountDelta = (int)(jitterPhaseCount - constants.jitterPhaseCount); + if (jitterPhaseCountDelta > 0) + constants.jitterPhaseCount++; + else if (jitterPhaseCountDelta < 0) + constants.jitterPhaseCount--; + } + + // Convert delta time to seconds and clamp to [0, 1] + constants.deltaTime = Mathf.Clamp01(dispatchParams.FrameTimeDelta); + + if (resetAccumulation) + constants.frameIndex = 0; + else + constants.frameIndex++; + + // Shading change usage of the SPD mip levels + constants.lumaMipLevelToUse = Fsr2Pass.ShadingChangeMipLevel; + + float mipDiv = 2 << constants.lumaMipLevelToUse; + constants.lumaMipDimensions.x = (int)(constants.maxRenderSize.x / mipDiv); + constants.lumaMipDimensions.y = (int)(constants.maxRenderSize.y / mipDiv); + } + + private Vector4 SetupDeviceDepthToViewSpaceDepthParams(Fsr2.DispatchDescription dispatchParams) + { + bool inverted = (_contextDescription.Flags & Fsr2.InitializationFlags.EnableDepthInverted) != 0; + bool infinite = (_contextDescription.Flags & Fsr2.InitializationFlags.EnableDepthInfinite) != 0; + + // make sure it has no impact if near and far plane values are swapped in dispatch params + // the flags "inverted" and "infinite" will decide what transform to use + float min = Mathf.Min(dispatchParams.CameraNear, dispatchParams.CameraFar); + float max = Mathf.Max(dispatchParams.CameraNear, dispatchParams.CameraFar); + + if (inverted) + { + (min, max) = (max, min); + } + + float q = max / (min - max); + float d = -1.0f; + + Vector4 matrixElemC = new Vector4(q, -1.0f - Mathf.Epsilon, q, 0.0f + Mathf.Epsilon); + Vector4 matrixElemE = new Vector4(q * min, -min - Mathf.Epsilon, q * min, max); + + // Revert x and y coords + float aspect = (float)dispatchParams.RenderSize.x / dispatchParams.RenderSize.y; + float cotHalfFovY = Mathf.Cos(0.5f * dispatchParams.CameraFovAngleVertical) / Mathf.Sin(0.5f * dispatchParams.CameraFovAngleVertical); + + int matrixIndex = (inverted ? 2 : 0) + (infinite ? 1 : 0); + return new Vector4( + d * matrixElemC[matrixIndex], + matrixElemE[matrixIndex], + aspect / cotHalfFovY, + 1.0f / cotHalfFovY); + } + + private void SetupRcasConstants(Fsr2.DispatchDescription dispatchParams) + { + int sharpnessIndex = Mathf.RoundToInt(Mathf.Clamp01(dispatchParams.Sharpness) * (RcasConfigs.Length - 1)); + RcasConsts = RcasConfigs[sharpnessIndex]; + } + + private void SetupSpdConstants(Fsr2.DispatchDescription dispatchParams, out Vector2Int dispatchThreadGroupCount) + { + RectInt rectInfo = new RectInt(0, 0, dispatchParams.RenderSize.x, dispatchParams.RenderSize.y); + SpdSetup(rectInfo, out dispatchThreadGroupCount, out var workGroupOffset, out var numWorkGroupsAndMips); + + // Downsample + ref Fsr2.SpdConstants spdConstants = ref SpdConsts; + spdConstants.numWorkGroups = (uint)numWorkGroupsAndMips.x; + spdConstants.mips = (uint)numWorkGroupsAndMips.y; + spdConstants.workGroupOffsetX = (uint)workGroupOffset.x; + spdConstants.workGroupOffsetY = (uint)workGroupOffset.y; + spdConstants.renderSizeX = (uint)dispatchParams.RenderSize.x; + spdConstants.renderSizeY = (uint)dispatchParams.RenderSize.y; + } + + private static void SpdSetup(RectInt rectInfo, out Vector2Int dispatchThreadGroupCount, out Vector2Int workGroupOffset, out Vector2Int numWorkGroupsAndMips, int mips = -1) + { + workGroupOffset = new Vector2Int(rectInfo.x / 64, rectInfo.y / 64); + + int endIndexX = (rectInfo.x + rectInfo.width - 1) / 64; + int endIndexY = (rectInfo.y + rectInfo.height - 1) / 64; + + dispatchThreadGroupCount = new Vector2Int(endIndexX + 1 - workGroupOffset.x, endIndexY + 1 - workGroupOffset.y); + + numWorkGroupsAndMips = new Vector2Int(dispatchThreadGroupCount.x * dispatchThreadGroupCount.y, mips); + if (mips < 0) + { + float resolution = Math.Max(rectInfo.width, rectInfo.height); + numWorkGroupsAndMips.y = Math.Min(Mathf.FloorToInt(Mathf.Log(resolution, 2.0f)), 12); + } + } + + private void DebugCheckDispatch(Fsr2.DispatchDescription dispatchParams) + { + if (!dispatchParams.Color.IsValid) + { + Debug.LogError("Color resource is null"); + } + + if (!dispatchParams.Depth.IsValid) + { + Debug.LogError("Depth resource is null"); + } + + if (!dispatchParams.MotionVectors.IsValid) + { + Debug.LogError("MotionVectors resource is null"); + } + + if (!dispatchParams.Output.IsValid) + { + Debug.LogError("Output resource is null"); + } + + if (dispatchParams.Exposure.IsValid && (_contextDescription.Flags & Fsr2.InitializationFlags.EnableAutoExposure) != 0) + { + Debug.LogWarning("Exposure resource provided, however auto exposure flag is present"); + } + + if (Mathf.Abs(dispatchParams.JitterOffset.x) > 1.0f || Mathf.Abs(dispatchParams.JitterOffset.y) > 1.0f) + { + Debug.LogWarning("JitterOffset contains value outside of expected range [-1.0, 1.0]"); + } + + if (dispatchParams.MotionVectorScale.x > _contextDescription.MaxRenderSize.x || dispatchParams.MotionVectorScale.y > _contextDescription.MaxRenderSize.y) + { + Debug.LogWarning("MotionVectorScale contains scale value greater than MaxRenderSize"); + } + + if (dispatchParams.MotionVectorScale.x == 0.0f || dispatchParams.MotionVectorScale.y == 0.0f) + { + Debug.LogWarning("MotionVectorScale contains zero scale value"); + } + + if (dispatchParams.RenderSize.x > _contextDescription.MaxRenderSize.x || dispatchParams.RenderSize.y > _contextDescription.MaxRenderSize.y) + { + Debug.LogWarning("RenderSize is greater than context MaxRenderSize"); + } + + if (dispatchParams.RenderSize.x == 0 || dispatchParams.RenderSize.y == 0) + { + Debug.LogWarning("RenderSize contains zero dimension"); + } + + if (dispatchParams.FrameTimeDelta > 1.0f) + { + Debug.LogWarning("FrameTimeDelta is greater than 1.0f - this value should be seconds (~0.0166 for 60fps)"); + } + + if (dispatchParams.PreExposure == 0.0f) + { + Debug.LogError("PreExposure provided as 0.0f which is invalid"); + } + + bool infiniteDepth = (_contextDescription.Flags & Fsr2.InitializationFlags.EnableDepthInfinite) != 0; + bool inverseDepth = (_contextDescription.Flags & Fsr2.InitializationFlags.EnableDepthInverted) != 0; + + if (inverseDepth) + { + if (dispatchParams.CameraNear < dispatchParams.CameraFar) + { + Debug.LogWarning("EnableDepthInverted flag is present yet CameraNear is less than CameraFar"); + } + + if (infiniteDepth) + { + if (dispatchParams.CameraNear < float.MaxValue) + { + Debug.LogWarning("EnableDepthInfinite and EnableDepthInverted present, yet CameraNear != float.MaxValue"); + } + } + + if (dispatchParams.CameraFar < 0.075f) + { + Debug.LogWarning("EnableDepthInverted present, CameraFar value is very low which may result in depth separation artefacting"); + } + } + else + { + if (dispatchParams.CameraNear > dispatchParams.CameraFar) + { + Debug.LogWarning("CameraNear is greater than CameraFar in non-inverted-depth context"); + } + + if (infiniteDepth) + { + if (dispatchParams.CameraFar < float.MaxValue) + { + Debug.LogWarning("EnableDepthInfinite present, yet CameraFar != float.MaxValue"); + } + } + + if (dispatchParams.CameraNear < 0.075f) + { + Debug.LogWarning("CameraNear value is very low which may result in depth separation artefacting"); + } + } + + if (dispatchParams.CameraFovAngleVertical <= 0.0f) + { + Debug.LogError("CameraFovAngleVertical is 0.0f - this value should be > 0.0f"); + } + + if (dispatchParams.CameraFovAngleVertical > Mathf.PI) + { + Debug.LogError("CameraFovAngleVertical is greater than 180 degrees/PI"); + } + } + + /// + /// The FSR2 C++ codebase uses floats bitwise converted to ints to pass sharpness parameters to the RCAS shader. + /// This is not possible in C# without enabling unsafe code compilation, so to avoid that we instead use a table of precomputed values. + /// + private static readonly Fsr2.RcasConstants[] RcasConfigs = new [] + { + new Fsr2.RcasConstants(1048576000u, 872428544u), + new Fsr2.RcasConstants(1049178080u, 877212745u), + new Fsr2.RcasConstants(1049823372u, 882390168u), + new Fsr2.RcasConstants(1050514979u, 887895276u), + new Fsr2.RcasConstants(1051256227u, 893859143u), + new Fsr2.RcasConstants(1052050675u, 900216232u), + new Fsr2.RcasConstants(1052902144u, 907032080u), + new Fsr2.RcasConstants(1053814727u, 914306687u), + new Fsr2.RcasConstants(1054792807u, 922105590u), + new Fsr2.RcasConstants(1055841087u, 930494326u), + new Fsr2.RcasConstants(1056964608u, 939538432u), + new Fsr2.RcasConstants(1057566688u, 944322633u), + new Fsr2.RcasConstants(1058211980u, 949500056u), + new Fsr2.RcasConstants(1058903587u, 955005164u), + new Fsr2.RcasConstants(1059644835u, 960969031u), + new Fsr2.RcasConstants(1060439283u, 967326120u), + new Fsr2.RcasConstants(1061290752u, 974141968u), + new Fsr2.RcasConstants(1062203335u, 981416575u), + new Fsr2.RcasConstants(1063181415u, 989215478u), + new Fsr2.RcasConstants(1064229695u, 997604214u), + new Fsr2.RcasConstants(1065353216u, 1006648320), + }; + + private static ComputeBuffer CreateConstantBuffer() where TConstants: struct + { + return new ComputeBuffer(1, Marshal.SizeOf(), ComputeBufferType.Constant); + } + + private static void DestroyConstantBuffer(ref ComputeBuffer bufferRef) + { + if (bufferRef == null) + return; + + bufferRef.Release(); + bufferRef = null; + } + + private static void DestroyPass(ref Fsr2Pass pass) + { + if (pass == null) + return; + + pass.Dispose(); + pass = null; + } + } +} diff --git a/Assets/Scripts/Core/Fsr3UpscalerContext.cs.meta b/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Context.cs.meta similarity index 100% rename from Assets/Scripts/Core/Fsr3UpscalerContext.cs.meta rename to Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Context.cs.meta diff --git a/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Pass.cs b/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Pass.cs new file mode 100644 index 0000000..bda5092 --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Pass.cs @@ -0,0 +1,380 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +using System; +using System.Runtime.InteropServices; +using UnityEngine; +using UnityEngine.Profiling; +using UnityEngine.Rendering; + +namespace FidelityFX.FSR2 +{ + /// + /// Base class for all of the compute passes that make up the FSR2 process. + /// This loosely matches the FfxPipelineState struct from the original FSR2 codebase, wrapped in an object-oriented blanket. + /// These classes are responsible for loading compute shaders, managing temporary resources, binding resources to shader kernels and dispatching said shaders. + /// + internal abstract class Fsr2Pass: IDisposable + { + internal const int ShadingChangeMipLevel = 4; // This matches the FFX_FSR2_SHADING_CHANGE_MIP_LEVEL define + + protected readonly Fsr2.ContextDescription ContextDescription; + protected readonly Fsr2Resources Resources; + protected readonly ComputeBuffer Constants; + + protected ComputeShader ComputeShader; + protected int KernelIndex; + + protected CustomSampler Sampler; + + protected Fsr2Pass(Fsr2.ContextDescription contextDescription, Fsr2Resources resources, ComputeBuffer constants) + { + ContextDescription = contextDescription; + Resources = resources; + Constants = constants; + } + + public virtual void Dispose() + { + } + + public void ScheduleDispatch(CommandBuffer commandBuffer, Fsr2.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) + { + commandBuffer.BeginSample(Sampler); + DoScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchX, dispatchY); + commandBuffer.EndSample(Sampler); + } + + protected abstract void DoScheduleDispatch(CommandBuffer commandBuffer, Fsr2.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY); + + protected void InitComputeShader(string passName, ComputeShader shader) + { + InitComputeShader(passName, shader, ContextDescription.Flags); + } + + private void InitComputeShader(string passName, ComputeShader shader, Fsr2.InitializationFlags flags) + { + if (shader == null) + { + throw new MissingReferenceException($"Shader for FSR2 pass '{passName}' could not be loaded! Please ensure it is included in the project correctly."); + } + + ComputeShader = shader; + KernelIndex = ComputeShader.FindKernel("CS"); + Sampler = CustomSampler.Create(passName); + + bool useLut = false; +#if UNITY_2022_1_OR_NEWER // This will also work in 2020.3.43+ and 2021.3.14+ + if (SystemInfo.computeSubGroupSize == 64) + { + useLut = true; + } +#endif + + // This matches the permutation rules from the CreatePipeline* functions + if ((flags & Fsr2.InitializationFlags.EnableHighDynamicRange) != 0) ComputeShader.EnableKeyword("FFX_FSR2_OPTION_HDR_COLOR_INPUT"); + if ((flags & Fsr2.InitializationFlags.EnableDisplayResolutionMotionVectors) == 0) ComputeShader.EnableKeyword("FFX_FSR2_OPTION_LOW_RESOLUTION_MOTION_VECTORS"); + if ((flags & Fsr2.InitializationFlags.EnableMotionVectorsJitterCancellation) != 0) ComputeShader.EnableKeyword("FFX_FSR2_OPTION_JITTERED_MOTION_VECTORS"); + if ((flags & Fsr2.InitializationFlags.EnableDepthInverted) != 0) ComputeShader.EnableKeyword("FFX_FSR2_OPTION_INVERTED_DEPTH"); + if (useLut) ComputeShader.EnableKeyword("FFX_FSR2_OPTION_REPROJECT_USE_LANCZOS_TYPE"); + if ((flags & Fsr2.InitializationFlags.EnableFP16Usage) != 0) ComputeShader.EnableKeyword("FFX_HALF"); + } + } + + internal class Fsr2ComputeLuminancePyramidPass : Fsr2Pass + { + private readonly ComputeBuffer _spdConstants; + + public Fsr2ComputeLuminancePyramidPass(Fsr2.ContextDescription contextDescription, Fsr2Resources resources, ComputeBuffer constants, ComputeBuffer spdConstants) + : base(contextDescription, resources, constants) + { + _spdConstants = spdConstants; + + InitComputeShader("Compute Luminance Pyramid", contextDescription.Shaders.computeLuminancePyramidPass); + } + + protected override void DoScheduleDispatch(CommandBuffer commandBuffer, Fsr2.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) + { + ref var color = ref dispatchParams.Color; + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvInputColor, color.RenderTarget, color.MipLevel, color.SubElement); + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.UavSpdAtomicCount, Resources.SpdAtomicCounter); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.UavExposureMipLumaChange, Resources.SceneLuminance, ShadingChangeMipLevel); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.UavExposureMip5, Resources.SceneLuminance, 5); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.UavAutoExposure, Resources.AutoExposure); + + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr2ShaderIDs.CbFsr2, Constants, 0, Marshal.SizeOf()); + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr2ShaderIDs.CbSpd, _spdConstants, 0, Marshal.SizeOf()); + + commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); + } + } + + internal class Fsr2ReconstructPreviousDepthPass : Fsr2Pass + { + public Fsr2ReconstructPreviousDepthPass(Fsr2.ContextDescription contextDescription, Fsr2Resources resources, ComputeBuffer constants) + : base(contextDescription, resources, constants) + { + InitComputeShader("Reconstruct & Dilate", contextDescription.Shaders.reconstructPreviousDepthPass); + } + + protected override void DoScheduleDispatch(CommandBuffer commandBuffer, Fsr2.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) + { + ref var color = ref dispatchParams.Color; + ref var depth = ref dispatchParams.Depth; + ref var motionVectors = ref dispatchParams.MotionVectors; + ref var exposure = ref dispatchParams.Exposure; + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvInputColor, color.RenderTarget, color.MipLevel, color.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvInputDepth, depth.RenderTarget, depth.MipLevel, depth.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvInputMotionVectors, motionVectors.RenderTarget, motionVectors.MipLevel, motionVectors.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvInputExposure, exposure.RenderTarget, exposure.MipLevel, exposure.SubElement); + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.UavDilatedMotionVectors, Resources.DilatedMotionVectors[frameIndex]); + + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr2ShaderIDs.CbFsr2, Constants, 0, Marshal.SizeOf()); + + commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); + } + } + + internal class Fsr2DepthClipPass : Fsr2Pass + { + public Fsr2DepthClipPass(Fsr2.ContextDescription contextDescription, Fsr2Resources resources, ComputeBuffer constants) + : base(contextDescription, resources, constants) + { + InitComputeShader("Depth Clip", contextDescription.Shaders.depthClipPass); + } + + protected override void DoScheduleDispatch(CommandBuffer commandBuffer, Fsr2.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) + { + ref var color = ref dispatchParams.Color; + ref var depth = ref dispatchParams.Depth; + ref var motionVectors = ref dispatchParams.MotionVectors; + ref var exposure = ref dispatchParams.Exposure; + ref var reactive = ref dispatchParams.Reactive; + ref var tac = ref dispatchParams.TransparencyAndComposition; + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvInputColor, color.RenderTarget, color.MipLevel, color.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvInputDepth, depth.RenderTarget, depth.MipLevel, depth.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvInputMotionVectors, motionVectors.RenderTarget, motionVectors.MipLevel, motionVectors.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvInputExposure, exposure.RenderTarget, exposure.MipLevel, exposure.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvReactiveMask, reactive.RenderTarget, reactive.MipLevel, reactive.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvTransparencyAndCompositionMask, tac.RenderTarget, tac.MipLevel, tac.SubElement); + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvReconstructedPrevNearestDepth, Fsr2ShaderIDs.UavReconstructedPrevNearestDepth); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvDilatedMotionVectors, Resources.DilatedMotionVectors[frameIndex]); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvDilatedDepth, Fsr2ShaderIDs.UavDilatedDepth); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvPrevDilatedMotionVectors, Resources.DilatedMotionVectors[frameIndex ^ 1]); + + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr2ShaderIDs.CbFsr2, Constants, 0, Marshal.SizeOf()); + + commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); + } + } + + internal class Fsr2LockPass : Fsr2Pass + { + public Fsr2LockPass(Fsr2.ContextDescription contextDescription, Fsr2Resources resources, ComputeBuffer constants) + : base(contextDescription, resources, constants) + { + InitComputeShader("Create Locks", contextDescription.Shaders.lockPass); + } + + protected override void DoScheduleDispatch(CommandBuffer commandBuffer, Fsr2.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) + { + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvLockInputLuma, Fsr2ShaderIDs.UavLockInputLuma); + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr2ShaderIDs.CbFsr2, Constants, 0, Marshal.SizeOf()); + + commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); + } + } + + internal class Fsr2AccumulatePass : Fsr2Pass + { + private const string SharpeningKeyword = "FFX_FSR2_OPTION_APPLY_SHARPENING"; + +#if UNITY_2021_2_OR_NEWER + private readonly LocalKeyword _sharpeningKeyword; +#endif + + public Fsr2AccumulatePass(Fsr2.ContextDescription contextDescription, Fsr2Resources resources, ComputeBuffer constants) + : base(contextDescription, resources, constants) + { + InitComputeShader("Reproject & Accumulate", contextDescription.Shaders.accumulatePass); +#if UNITY_2021_2_OR_NEWER + _sharpeningKeyword = new LocalKeyword(ComputeShader, SharpeningKeyword); +#endif + } + + protected override void DoScheduleDispatch(CommandBuffer commandBuffer, Fsr2.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) + { +#if UNITY_2021_2_OR_NEWER + if (dispatchParams.EnableSharpening) + commandBuffer.EnableKeyword(ComputeShader, _sharpeningKeyword); + else + commandBuffer.DisableKeyword(ComputeShader, _sharpeningKeyword); +#else + if (dispatchParams.EnableSharpening) + commandBuffer.EnableShaderKeyword(SharpeningKeyword); + else + commandBuffer.DisableShaderKeyword(SharpeningKeyword); +#endif + + if ((ContextDescription.Flags & Fsr2.InitializationFlags.EnableDisplayResolutionMotionVectors) == 0) + { + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvDilatedMotionVectors, Resources.DilatedMotionVectors[frameIndex]); + } + else + { + ref var motionVectors = ref dispatchParams.MotionVectors; + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvInputMotionVectors, motionVectors.RenderTarget, motionVectors.MipLevel, motionVectors.SubElement); + } + + ref var exposure = ref dispatchParams.Exposure; + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvInputExposure, exposure.RenderTarget, exposure.MipLevel, exposure.SubElement); + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvDilatedReactiveMasks, Fsr2ShaderIDs.UavDilatedReactiveMasks); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvInternalUpscaled, Resources.InternalUpscaled[frameIndex ^ 1]); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvLockStatus, Resources.LockStatus[frameIndex ^ 1]); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvPreparedInputColor, Fsr2ShaderIDs.UavPreparedInputColor); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvLanczosLut, Resources.LanczosLut); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvUpscaleMaximumBiasLut, Resources.MaximumBiasLut); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvSceneLuminanceMips, Resources.SceneLuminance); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvAutoExposure, Resources.AutoExposure); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvLumaHistory, Resources.LumaHistory[frameIndex ^ 1]); + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.UavInternalUpscaled, Resources.InternalUpscaled[frameIndex]); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.UavLockStatus, Resources.LockStatus[frameIndex]); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.UavLumaHistory, Resources.LumaHistory[frameIndex]); + + ref var output = ref dispatchParams.Output; + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.UavUpscaledOutput, output.RenderTarget, output.MipLevel, output.SubElement); + + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr2ShaderIDs.CbFsr2, Constants, 0, Marshal.SizeOf()); + + commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); + } + } + + internal class Fsr2SharpenPass : Fsr2Pass + { + private readonly ComputeBuffer _rcasConstants; + + public Fsr2SharpenPass(Fsr2.ContextDescription contextDescription, Fsr2Resources resources, ComputeBuffer constants, ComputeBuffer rcasConstants) + : base(contextDescription, resources, constants) + { + _rcasConstants = rcasConstants; + + InitComputeShader("RCAS Sharpening", contextDescription.Shaders.sharpenPass); + } + + protected override void DoScheduleDispatch(CommandBuffer commandBuffer, Fsr2.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) + { + ref var exposure = ref dispatchParams.Exposure; + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvInputExposure, exposure.RenderTarget, exposure.MipLevel, exposure.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvRcasInput, Resources.InternalUpscaled[frameIndex]); + + ref var output = ref dispatchParams.Output; + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.UavUpscaledOutput, output.RenderTarget, output.MipLevel, output.SubElement); + + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr2ShaderIDs.CbFsr2, Constants, 0, Marshal.SizeOf()); + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr2ShaderIDs.CbRcas, _rcasConstants, 0, Marshal.SizeOf()); + + commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); + } + } + + internal class Fsr2GenerateReactivePass : Fsr2Pass + { + private readonly ComputeBuffer _generateReactiveConstants; + + public Fsr2GenerateReactivePass(Fsr2.ContextDescription contextDescription, Fsr2Resources resources, ComputeBuffer generateReactiveConstants) + : base(contextDescription, resources, null) + { + _generateReactiveConstants = generateReactiveConstants; + + InitComputeShader("Auto-Generate Reactive Mask", contextDescription.Shaders.autoGenReactivePass); + } + + protected override void DoScheduleDispatch(CommandBuffer commandBuffer, Fsr2.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) + { + } + + public void ScheduleDispatch(CommandBuffer commandBuffer, Fsr2.GenerateReactiveDescription dispatchParams, int dispatchX, int dispatchY) + { + commandBuffer.BeginSample(Sampler); + + ref var opaqueOnly = ref dispatchParams.ColorOpaqueOnly; + ref var color = ref dispatchParams.ColorPreUpscale; + ref var reactive = ref dispatchParams.OutReactive; + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvOpaqueOnly, opaqueOnly.RenderTarget, opaqueOnly.MipLevel, opaqueOnly.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvInputColor, color.RenderTarget, color.MipLevel, color.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.UavAutoReactive, reactive.RenderTarget, reactive.MipLevel, reactive.SubElement); + + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr2ShaderIDs.CbGenReactive, _generateReactiveConstants, 0, Marshal.SizeOf()); + + commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); + + commandBuffer.EndSample(Sampler); + } + } + + internal class Fsr2TcrAutogeneratePass : Fsr2Pass + { + private readonly ComputeBuffer _tcrAutogenerateConstants; + + public Fsr2TcrAutogeneratePass(Fsr2.ContextDescription contextDescription, Fsr2Resources resources, ComputeBuffer constants, ComputeBuffer tcrAutogenerateConstants) + : base(contextDescription, resources, constants) + { + _tcrAutogenerateConstants = tcrAutogenerateConstants; + + InitComputeShader("Auto-Generate Transparency & Composition Mask", contextDescription.Shaders.tcrAutoGenPass); + } + + protected override void DoScheduleDispatch(CommandBuffer commandBuffer, Fsr2.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) + { + ref var color = ref dispatchParams.Color; + ref var motionVectors = ref dispatchParams.MotionVectors; + ref var opaqueOnly = ref dispatchParams.ColorOpaqueOnly; + ref var reactive = ref dispatchParams.Reactive; + ref var tac = ref dispatchParams.TransparencyAndComposition; + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvOpaqueOnly, opaqueOnly.RenderTarget, opaqueOnly.MipLevel, opaqueOnly.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvInputColor, color.RenderTarget, color.MipLevel, color.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvInputMotionVectors, motionVectors.RenderTarget, motionVectors.MipLevel, motionVectors.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvPrevColorPreAlpha, Resources.PrevPreAlpha[frameIndex ^ 1]); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvPrevColorPostAlpha, Resources.PrevPostAlpha[frameIndex ^ 1]); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvReactiveMask, reactive.RenderTarget, reactive.MipLevel, reactive.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.SrvTransparencyAndCompositionMask, tac.RenderTarget, tac.MipLevel, tac.SubElement); + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.UavAutoReactive, Resources.AutoReactive); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.UavAutoComposition, Resources.AutoComposition); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.UavPrevColorPreAlpha, Resources.PrevPreAlpha[frameIndex]); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr2ShaderIDs.UavPrevColorPostAlpha, Resources.PrevPostAlpha[frameIndex]); + + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr2ShaderIDs.CbFsr2, Constants, 0, Marshal.SizeOf()); + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr2ShaderIDs.CbGenReactive, _tcrAutogenerateConstants, 0, Marshal.SizeOf()); + + commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); + } + } +} diff --git a/Assets/Scripts/Core/Fsr3UpscalerPass.cs.meta b/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Pass.cs.meta similarity index 100% rename from Assets/Scripts/Core/Fsr3UpscalerPass.cs.meta rename to Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Pass.cs.meta diff --git a/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Resources.cs b/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Resources.cs new file mode 100644 index 0000000..4c1c50e --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Resources.cs @@ -0,0 +1,258 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +using System; +using UnityEngine; +using UnityEngine.Experimental.Rendering; +using UnityEngine.Rendering; + +namespace FidelityFX.FSR2 +{ + /// + /// Helper class for bundling and managing persistent resources required by the FSR2 process. + /// This includes lookup tables, default fallback resources and double-buffered resources that get swapped between frames. + /// + internal class Fsr2Resources + { + public Texture2D DefaultExposure; + public Texture2D DefaultReactive; + public Texture2D LanczosLut; + public Texture2D MaximumBiasLut; + public RenderTexture SpdAtomicCounter; + public RenderTexture AutoExposure; + public RenderTexture SceneLuminance; + public RenderTexture AutoReactive; + public RenderTexture AutoComposition; + public readonly RenderTexture[] DilatedMotionVectors = new RenderTexture[2]; + public readonly RenderTexture[] LockStatus = new RenderTexture[2]; + public readonly RenderTexture[] InternalUpscaled = new RenderTexture[2]; + public readonly RenderTexture[] LumaHistory = new RenderTexture[2]; + public readonly RenderTexture[] PrevPreAlpha = new RenderTexture[2]; + public readonly RenderTexture[] PrevPostAlpha = new RenderTexture[2]; + + public void Create(Fsr2.ContextDescription contextDescription) + { + // Generate the data for the LUT + const int lanczos2LutWidth = 128; + float[] lanczos2Weights = new float[lanczos2LutWidth]; + for (int currentLanczosWidthIndex = 0; currentLanczosWidthIndex < lanczos2LutWidth; ++currentLanczosWidthIndex) + { + float x = 2.0f * currentLanczosWidthIndex / (lanczos2LutWidth - 1); + float y = Fsr2.Lanczos2(x); + lanczos2Weights[currentLanczosWidthIndex] = y; + } + + float[] maximumBias = new float[MaximumBiasTextureWidth * MaximumBiasTextureHeight]; + for (int i = 0; i < maximumBias.Length; ++i) + { + maximumBias[i] = MaximumBias[i] / 2.0f; + } + + // Resource FSR2_LanczosLutData: FFX_RESOURCE_USAGE_READ_ONLY, FFX_SURFACE_FORMAT_R16_SNORM, FFX_RESOURCE_FLAGS_NONE + // R16_SNorm textures are not supported by Unity on most platforms, strangely enough. So instead we use R32_SFloat and upload pre-normalized float data. + LanczosLut = new Texture2D(lanczos2LutWidth, 1, GraphicsFormat.R32_SFloat, TextureCreationFlags.None) { name = "FSR2_LanczosLutData" }; + LanczosLut.SetPixelData(lanczos2Weights, 0); + LanczosLut.Apply(); + + // Resource FSR2_MaximumUpsampleBias: FFX_RESOURCE_USAGE_READ_ONLY, FFX_SURFACE_FORMAT_R16_SNORM, FFX_RESOURCE_FLAGS_NONE + MaximumBiasLut = new Texture2D(MaximumBiasTextureWidth, MaximumBiasTextureHeight, GraphicsFormat.R32_SFloat, TextureCreationFlags.None) { name = "FSR2_MaximumUpsampleBias" }; + MaximumBiasLut.SetPixelData(maximumBias, 0); + MaximumBiasLut.Apply(); + + // Resource FSR2_DefaultExposure: FFX_RESOURCE_USAGE_READ_ONLY, FFX_SURFACE_FORMAT_R32G32_FLOAT, FFX_RESOURCE_FLAGS_NONE + DefaultExposure = new Texture2D(1, 1, GraphicsFormat.R32G32_SFloat, TextureCreationFlags.None) { name = "FSR2_DefaultExposure" }; + DefaultExposure.SetPixel(0, 0, Color.clear); + DefaultExposure.Apply(); + + // Resource FSR2_DefaultReactivityMask: FFX_RESOURCE_USAGE_READ_ONLY, FFX_SURFACE_FORMAT_R8_UNORM, FFX_RESOURCE_FLAGS_NONE + DefaultReactive = new Texture2D(1, 1, GraphicsFormat.R8_UNorm, TextureCreationFlags.None) { name = "FSR2_DefaultReactivityMask" }; + DefaultReactive.SetPixel(0, 0, Color.clear); + DefaultReactive.Apply(); + + // Resource FSR2_SpdAtomicCounter: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R32_UINT, FFX_RESOURCE_FLAGS_ALIASABLE + // Despite what the original FSR2 codebase says, this resource really isn't aliasable. Resetting this counter to 0 every frame breaks auto-exposure on MacOS Metal. + SpdAtomicCounter = new RenderTexture(1, 1, 0, GraphicsFormat.R32_UInt) { name = "FSR2_SpdAtomicCounter", enableRandomWrite = true }; + SpdAtomicCounter.Create(); + + // Resource FSR2_AutoExposure: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R32G32_FLOAT, FFX_RESOURCE_FLAGS_NONE + AutoExposure = new RenderTexture(1, 1, 0, GraphicsFormat.R32G32_SFloat) { name = "FSR2_AutoExposure", enableRandomWrite = true }; + AutoExposure.Create(); + + // Resource FSR2_ExposureMips: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16_FLOAT, FFX_RESOURCE_FLAGS_ALIASABLE + // This is a rather special case: it's an aliasable resource, but because we require a mipmap chain and bind specific mip levels per shader, we can't easily use temporary RTs for this. + int w = contextDescription.MaxRenderSize.x / 2, h = contextDescription.MaxRenderSize.y / 2; + int mipCount = 1 + Mathf.FloorToInt(Mathf.Log(Math.Max(w, h), 2.0f)); + SceneLuminance = new RenderTexture(w, h, 0, GraphicsFormat.R16_SFloat, mipCount) { name = "FSR2_ExposureMips", enableRandomWrite = true, useMipMap = true, autoGenerateMips = false }; + SceneLuminance.Create(); + + // Resources FSR2_InternalDilatedVelocity1/2: FFX_RESOURCE_USAGE_RENDERTARGET | FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16G16_FLOAT, FFX_RESOURCE_FLAGS_NONE + CreateDoubleBufferedResource(DilatedMotionVectors, "FSR2_InternalDilatedVelocity", contextDescription.MaxRenderSize, GraphicsFormat.R16G16_SFloat); + + // Resources FSR2_LockStatus1/2: FFX_RESOURCE_USAGE_RENDERTARGET | FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16G16_FLOAT, FFX_RESOURCE_FLAGS_NONE + CreateDoubleBufferedResource(LockStatus, "FSR2_LockStatus", contextDescription.DisplaySize, GraphicsFormat.R16G16_SFloat); + + // Resources FSR2_InternalUpscaled1/2: FFX_RESOURCE_USAGE_RENDERTARGET | FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16G16B16A16_FLOAT, FFX_RESOURCE_FLAGS_NONE + CreateDoubleBufferedResource(InternalUpscaled, "FSR2_InternalUpscaled", contextDescription.DisplaySize, GraphicsFormat.R16G16B16A16_SFloat); + + // Resources FSR2_LumaHistory1/2: FFX_RESOURCE_USAGE_RENDERTARGET | FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R8G8B8A8_UNORM, FFX_RESOURCE_FLAGS_NONE + CreateDoubleBufferedResource(LumaHistory, "FSR2_LumaHistory", contextDescription.DisplaySize, GraphicsFormat.R8G8B8A8_UNorm); + } + + public void CreateTcrAutogenResources(Fsr2.ContextDescription contextDescription) + { + // Resource FSR2_AutoReactive: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R8_UNORM, FFX_RESOURCE_FLAGS_NONE + AutoReactive = new RenderTexture(contextDescription.MaxRenderSize.x, contextDescription.MaxRenderSize.y, 0, GraphicsFormat.R8_UNorm) { name = "FSR2_AutoReactive", enableRandomWrite = true }; + AutoReactive.Create(); + + // Resource FSR2_AutoComposition: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R8_UNORM, FFX_RESOURCE_FLAGS_NONE + AutoComposition = new RenderTexture(contextDescription.MaxRenderSize.x, contextDescription.MaxRenderSize.y, 0, GraphicsFormat.R8_UNorm) { name = "FSR2_AutoComposition", enableRandomWrite = true }; + AutoComposition.Create(); + + // Resources FSR2_PrevPreAlpha0/1: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R11G11B10_FLOAT, FFX_RESOURCE_FLAGS_NONE + CreateDoubleBufferedResource(PrevPreAlpha, "FSR2_PrevPreAlpha", contextDescription.MaxRenderSize, GraphicsFormat.B10G11R11_UFloatPack32); + + // Resources FSR2_PrevPostAlpha0/1: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R11G11B10_FLOAT, FFX_RESOURCE_FLAGS_NONE + CreateDoubleBufferedResource(PrevPostAlpha, "FSR2_PrevPostAlpha", contextDescription.MaxRenderSize, GraphicsFormat.B10G11R11_UFloatPack32); + } + + // Set up shared aliasable resources, i.e. temporary render textures + // These do not need to persist between frames, but they do need to be available between passes + public static void CreateAliasableResources(CommandBuffer commandBuffer, Fsr2.ContextDescription contextDescription, Fsr2.DispatchDescription dispatchParams) + { + Vector2Int displaySize = contextDescription.DisplaySize; + Vector2Int maxRenderSize = contextDescription.MaxRenderSize; + + // FSR2_ReconstructedPrevNearestDepth: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R32_UINT, FFX_RESOURCE_FLAGS_ALIASABLE + commandBuffer.GetTemporaryRT(Fsr2ShaderIDs.UavReconstructedPrevNearestDepth, maxRenderSize.x, maxRenderSize.y, 0, default, GraphicsFormat.R32_UInt, 1, true); + + // FSR2_DilatedDepth: FFX_RESOURCE_USAGE_RENDERTARGET | FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R32_FLOAT, FFX_RESOURCE_FLAGS_ALIASABLE + commandBuffer.GetTemporaryRT(Fsr2ShaderIDs.UavDilatedDepth, maxRenderSize.x, maxRenderSize.y, 0, default, GraphicsFormat.R32_SFloat, 1, true); + + // FSR2_LockInputLuma: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16_FLOAT, FFX_RESOURCE_FLAGS_ALIASABLE + commandBuffer.GetTemporaryRT(Fsr2ShaderIDs.UavLockInputLuma, maxRenderSize.x, maxRenderSize.y, 0, default, GraphicsFormat.R16_SFloat, 1, true); + + // FSR2_DilatedReactiveMasks: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R8G8_UNORM, FFX_RESOURCE_FLAGS_ALIASABLE + commandBuffer.GetTemporaryRT(Fsr2ShaderIDs.UavDilatedReactiveMasks, maxRenderSize.x, maxRenderSize.y, 0, default, GraphicsFormat.R8G8_UNorm, 1, true); + + // FSR2_PreparedInputColor: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16G16B16A16_FLOAT, FFX_RESOURCE_FLAGS_ALIASABLE + commandBuffer.GetTemporaryRT(Fsr2ShaderIDs.UavPreparedInputColor, maxRenderSize.x, maxRenderSize.y, 0, default, GraphicsFormat.R16G16B16A16_SFloat, 1, true); + + // FSR2_NewLocks: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R8_UNORM, FFX_RESOURCE_FLAGS_ALIASABLE + commandBuffer.GetTemporaryRT(Fsr2ShaderIDs.UavNewLocks, displaySize.x, displaySize.y, 0, default, GraphicsFormat.R8_UNorm, 1, true); + } + + public static void DestroyAliasableResources(CommandBuffer commandBuffer) + { + // Release all of the aliasable resources used this frame + commandBuffer.ReleaseTemporaryRT(Fsr2ShaderIDs.UavReconstructedPrevNearestDepth); + commandBuffer.ReleaseTemporaryRT(Fsr2ShaderIDs.UavDilatedDepth); + commandBuffer.ReleaseTemporaryRT(Fsr2ShaderIDs.UavLockInputLuma); + commandBuffer.ReleaseTemporaryRT(Fsr2ShaderIDs.UavDilatedReactiveMasks); + commandBuffer.ReleaseTemporaryRT(Fsr2ShaderIDs.UavPreparedInputColor); + commandBuffer.ReleaseTemporaryRT(Fsr2ShaderIDs.UavNewLocks); + } + + private static void CreateDoubleBufferedResource(RenderTexture[] resource, string name, Vector2Int size, GraphicsFormat format) + { + for (int i = 0; i < 2; ++i) + { + resource[i] = new RenderTexture(size.x, size.y, 0, format) { name = name + (i + 1), enableRandomWrite = true }; + resource[i].Create(); + } + } + + public void Destroy() + { + DestroyTcrAutogenResources(); + + DestroyResource(LumaHistory); + DestroyResource(InternalUpscaled); + DestroyResource(LockStatus); + DestroyResource(DilatedMotionVectors); + DestroyResource(ref SceneLuminance); + DestroyResource(ref AutoExposure); + DestroyResource(ref DefaultReactive); + DestroyResource(ref DefaultExposure); + DestroyResource(ref MaximumBiasLut); + DestroyResource(ref LanczosLut); + } + + public void DestroyTcrAutogenResources() + { + DestroyResource(PrevPostAlpha); + DestroyResource(PrevPreAlpha); + DestroyResource(ref AutoComposition); + DestroyResource(ref AutoReactive); + } + + private static void DestroyResource(ref Texture2D resource) + { + if (resource == null) + return; + +#if UNITY_EDITOR + if (Application.isPlaying && !UnityEditor.EditorApplication.isPaused) + UnityEngine.Object.Destroy(resource); + else + UnityEngine.Object.DestroyImmediate(resource); +#else + UnityEngine.Object.Destroy(resource); +#endif + resource = null; + } + + private static void DestroyResource(ref RenderTexture resource) + { + if (resource == null) + return; + + resource.Release(); + resource = null; + } + + private static void DestroyResource(RenderTexture[] resource) + { + for (int i = 0; i < resource.Length; ++i) + DestroyResource(ref resource[i]); + } + + private const int MaximumBiasTextureWidth = 16; + private const int MaximumBiasTextureHeight = 16; + private static readonly float[] MaximumBias = + { + 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.876f, 1.809f, 1.772f, 1.753f, 1.748f, + 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.869f, 1.801f, 1.764f, 1.745f, 1.739f, + 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.976f, 1.841f, 1.774f, 1.737f, 1.716f, 1.71f, + 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.914f, 1.784f, 1.716f, 1.673f, 1.649f, 1.641f, + 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.793f, 1.676f, 1.604f, 1.562f, 1.54f, 1.533f, + 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.802f, 1.619f, 1.536f, 1.492f, 1.467f, 1.454f, 1.449f, + 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.812f, 1.575f, 1.496f, 1.456f, 1.432f, 1.416f, 1.408f, 1.405f, + 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.555f, 1.479f, 1.438f, 1.413f, 1.398f, 1.387f, 1.381f, 1.379f, + 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.812f, 1.555f, 1.474f, 1.43f, 1.404f, 1.387f, 1.376f, 1.368f, 1.363f, 1.362f, + 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.802f, 1.575f, 1.479f, 1.43f, 1.401f, 1.382f, 1.369f, 1.36f, 1.354f, 1.351f, 1.35f, + 2.0f, 2.0f, 1.976f, 1.914f, 1.793f, 1.619f, 1.496f, 1.438f, 1.404f, 1.382f, 1.367f, 1.357f, 1.349f, 1.344f, 1.341f, 1.34f, + 1.876f, 1.869f, 1.841f, 1.784f, 1.676f, 1.536f, 1.456f, 1.413f, 1.387f, 1.369f, 1.357f, 1.347f, 1.341f, 1.336f, 1.333f, 1.332f, + 1.809f, 1.801f, 1.774f, 1.716f, 1.604f, 1.492f, 1.432f, 1.398f, 1.376f, 1.36f, 1.349f, 1.341f, 1.335f, 1.33f, 1.328f, 1.327f, + 1.772f, 1.764f, 1.737f, 1.673f, 1.562f, 1.467f, 1.416f, 1.387f, 1.368f, 1.354f, 1.344f, 1.336f, 1.33f, 1.326f, 1.323f, 1.323f, + 1.753f, 1.745f, 1.716f, 1.649f, 1.54f, 1.454f, 1.408f, 1.381f, 1.363f, 1.351f, 1.341f, 1.333f, 1.328f, 1.323f, 1.321f, 1.32f, + 1.748f, 1.739f, 1.71f, 1.641f, 1.533f, 1.449f, 1.405f, 1.379f, 1.362f, 1.35f, 1.34f, 1.332f, 1.327f, 1.323f, 1.32f, 1.319f, + }; + } +} diff --git a/Assets/Scripts/Core/Fsr3UpscalerResources.cs.meta b/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Resources.cs.meta similarity index 100% rename from Assets/Scripts/Core/Fsr3UpscalerResources.cs.meta rename to Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2Resources.cs.meta diff --git a/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2ShaderIDs.cs b/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2ShaderIDs.cs new file mode 100644 index 0000000..0adbf6b --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2ShaderIDs.cs @@ -0,0 +1,80 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +using UnityEngine; + +namespace FidelityFX.FSR2 +{ + public static class Fsr2ShaderIDs + { + // Shader resource views, i.e. read-only bindings + public static readonly int SrvInputColor = Shader.PropertyToID("r_input_color_jittered"); + public static readonly int SrvOpaqueOnly = Shader.PropertyToID("r_input_opaque_only"); + public static readonly int SrvInputMotionVectors = Shader.PropertyToID("r_input_motion_vectors"); + public static readonly int SrvInputDepth = Shader.PropertyToID("r_input_depth"); + public static readonly int SrvInputExposure = Shader.PropertyToID("r_input_exposure"); + public static readonly int SrvAutoExposure = Shader.PropertyToID("r_auto_exposure"); + public static readonly int SrvReactiveMask = Shader.PropertyToID("r_reactive_mask"); + public static readonly int SrvTransparencyAndCompositionMask = Shader.PropertyToID("r_transparency_and_composition_mask"); + public static readonly int SrvReconstructedPrevNearestDepth = Shader.PropertyToID("r_reconstructed_previous_nearest_depth"); + public static readonly int SrvDilatedMotionVectors = Shader.PropertyToID("r_dilated_motion_vectors"); + public static readonly int SrvPrevDilatedMotionVectors = Shader.PropertyToID("r_previous_dilated_motion_vectors"); + public static readonly int SrvDilatedDepth = Shader.PropertyToID("r_dilatedDepth"); + public static readonly int SrvInternalUpscaled = Shader.PropertyToID("r_internal_upscaled_color"); + public static readonly int SrvLockStatus = Shader.PropertyToID("r_lock_status"); + public static readonly int SrvLockInputLuma = Shader.PropertyToID("r_lock_input_luma"); + public static readonly int SrvPreparedInputColor = Shader.PropertyToID("r_prepared_input_color"); + public static readonly int SrvLumaHistory = Shader.PropertyToID("r_luma_history"); + public static readonly int SrvRcasInput = Shader.PropertyToID("r_rcas_input"); + public static readonly int SrvLanczosLut = Shader.PropertyToID("r_lanczos_lut"); + public static readonly int SrvSceneLuminanceMips = Shader.PropertyToID("r_imgMips"); + public static readonly int SrvUpscaleMaximumBiasLut = Shader.PropertyToID("r_upsample_maximum_bias_lut"); + public static readonly int SrvDilatedReactiveMasks = Shader.PropertyToID("r_dilated_reactive_masks"); + public static readonly int SrvPrevColorPreAlpha = Shader.PropertyToID("r_input_prev_color_pre_alpha"); + public static readonly int SrvPrevColorPostAlpha = Shader.PropertyToID("r_input_prev_color_post_alpha"); + + // Unordered access views, i.e. random read/write bindings + public static readonly int UavReconstructedPrevNearestDepth = Shader.PropertyToID("rw_reconstructed_previous_nearest_depth"); + public static readonly int UavDilatedMotionVectors = Shader.PropertyToID("rw_dilated_motion_vectors"); + public static readonly int UavDilatedDepth = Shader.PropertyToID("rw_dilatedDepth"); + public static readonly int UavInternalUpscaled = Shader.PropertyToID("rw_internal_upscaled_color"); + public static readonly int UavLockStatus = Shader.PropertyToID("rw_lock_status"); + public static readonly int UavLockInputLuma = Shader.PropertyToID("rw_lock_input_luma"); + public static readonly int UavNewLocks = Shader.PropertyToID("rw_new_locks"); + public static readonly int UavPreparedInputColor = Shader.PropertyToID("rw_prepared_input_color"); + public static readonly int UavLumaHistory = Shader.PropertyToID("rw_luma_history"); + public static readonly int UavUpscaledOutput = Shader.PropertyToID("rw_upscaled_output"); + public static readonly int UavExposureMipLumaChange = Shader.PropertyToID("rw_img_mip_shading_change"); + public static readonly int UavExposureMip5 = Shader.PropertyToID("rw_img_mip_5"); + public static readonly int UavDilatedReactiveMasks = Shader.PropertyToID("rw_dilated_reactive_masks"); + public static readonly int UavAutoExposure = Shader.PropertyToID("rw_auto_exposure"); + public static readonly int UavSpdAtomicCount = Shader.PropertyToID("rw_spd_global_atomic"); + public static readonly int UavAutoReactive = Shader.PropertyToID("rw_output_autoreactive"); + public static readonly int UavAutoComposition = Shader.PropertyToID("rw_output_autocomposition"); + public static readonly int UavPrevColorPreAlpha = Shader.PropertyToID("rw_output_prev_color_pre_alpha"); + public static readonly int UavPrevColorPostAlpha = Shader.PropertyToID("rw_output_prev_color_post_alpha"); + + // Constant buffer bindings + public static readonly int CbFsr2 = Shader.PropertyToID("cbFSR2"); + public static readonly int CbSpd = Shader.PropertyToID("cbSPD"); + public static readonly int CbRcas = Shader.PropertyToID("cbRCAS"); + public static readonly int CbGenReactive = Shader.PropertyToID("cbGenerateReactive"); + } +} diff --git a/Assets/Scripts/Core/Fsr3ShaderIDs.cs.meta b/Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2ShaderIDs.cs.meta similarity index 100% rename from Assets/Scripts/Core/Fsr3ShaderIDs.cs.meta rename to Packages/fidelityfx.fsr/Runtime/FSR2/Fsr2ShaderIDs.cs.meta diff --git a/Packages/fidelityfx.fsr/Runtime/FSR3.meta b/Packages/fidelityfx.fsr/Runtime/FSR3.meta new file mode 100644 index 0000000..5399ea1 --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR3.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: 16d16ad51c3bc2a429d3da6788f65e07 +folderAsset: yes +DefaultImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3ShaderIDs.cs b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3ShaderIDs.cs new file mode 100644 index 0000000..6e543fe --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3ShaderIDs.cs @@ -0,0 +1,91 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +using UnityEngine; + +namespace FidelityFX.FSR3 +{ + public static class Fsr3ShaderIDs + { + // Shader resource views, i.e. read-only bindings + public static readonly int SrvInputColor = Shader.PropertyToID("r_input_color_jittered"); + public static readonly int SrvOpaqueOnly = Shader.PropertyToID("r_input_opaque_only"); + public static readonly int SrvInputMotionVectors = Shader.PropertyToID("r_input_motion_vectors"); + public static readonly int SrvInputDepth = Shader.PropertyToID("r_input_depth"); + public static readonly int SrvInputExposure = Shader.PropertyToID("r_input_exposure"); + public static readonly int SrvFrameInfo = Shader.PropertyToID("r_frame_info"); + public static readonly int SrvReactiveMask = Shader.PropertyToID("r_reactive_mask"); + public static readonly int SrvTransparencyAndCompositionMask = Shader.PropertyToID("r_transparency_and_composition_mask"); + public static readonly int SrvReconstructedPrevNearestDepth = Shader.PropertyToID("r_reconstructed_previous_nearest_depth"); + public static readonly int SrvDilatedMotionVectors = Shader.PropertyToID("r_dilated_motion_vectors"); + public static readonly int SrvDilatedDepth = Shader.PropertyToID("r_dilated_depth"); + public static readonly int SrvInternalUpscaled = Shader.PropertyToID("r_internal_upscaled_color"); + public static readonly int SrvAccumulation = Shader.PropertyToID("r_accumulation"); + public static readonly int SrvLumaHistory = Shader.PropertyToID("r_luma_history"); + public static readonly int SrvRcasInput = Shader.PropertyToID("r_rcas_input"); + public static readonly int SrvLanczosLut = Shader.PropertyToID("r_lanczos_lut"); + public static readonly int SrvSpdMips = Shader.PropertyToID("r_spd_mips"); + public static readonly int SrvDilatedReactiveMasks = Shader.PropertyToID("r_dilated_reactive_masks"); + public static readonly int SrvNewLocks = Shader.PropertyToID("r_new_locks"); + public static readonly int SrvFarthestDepth = Shader.PropertyToID("r_farthest_depth"); + public static readonly int SrvFarthestDepthMip1 = Shader.PropertyToID("r_farthest_depth_mip1"); + public static readonly int SrvShadingChange = Shader.PropertyToID("r_shading_change"); + public static readonly int SrvCurrentLuma = Shader.PropertyToID("r_current_luma"); + public static readonly int SrvPreviousLuma = Shader.PropertyToID("r_previous_luma"); + public static readonly int SrvLumaInstability = Shader.PropertyToID("r_luma_instability"); + public static readonly int SrvPrevColorPreAlpha = Shader.PropertyToID("r_input_prev_color_pre_alpha"); + public static readonly int SrvPrevColorPostAlpha = Shader.PropertyToID("r_input_prev_color_post_alpha"); + + // Unordered access views, i.e. random read/write bindings + public static readonly int UavReconstructedPrevNearestDepth = Shader.PropertyToID("rw_reconstructed_previous_nearest_depth"); + public static readonly int UavDilatedMotionVectors = Shader.PropertyToID("rw_dilated_motion_vectors"); + public static readonly int UavDilatedDepth = Shader.PropertyToID("rw_dilated_depth"); + public static readonly int UavInternalUpscaled = Shader.PropertyToID("rw_internal_upscaled_color"); + public static readonly int UavAccumulation = Shader.PropertyToID("rw_accumulation"); + public static readonly int UavLumaHistory = Shader.PropertyToID("rw_luma_history"); + public static readonly int UavUpscaledOutput = Shader.PropertyToID("rw_upscaled_output"); + public static readonly int UavDilatedReactiveMasks = Shader.PropertyToID("rw_dilated_reactive_masks"); + public static readonly int UavFrameInfo = Shader.PropertyToID("rw_frame_info"); + public static readonly int UavSpdAtomicCount = Shader.PropertyToID("rw_spd_global_atomic"); + public static readonly int UavNewLocks = Shader.PropertyToID("rw_new_locks"); + public static readonly int UavAutoReactive = Shader.PropertyToID("rw_output_autoreactive"); + public static readonly int UavShadingChange = Shader.PropertyToID("rw_shading_change"); + public static readonly int UavFarthestDepth = Shader.PropertyToID("rw_farthest_depth"); + public static readonly int UavFarthestDepthMip1 = Shader.PropertyToID("rw_farthest_depth_mip1"); + public static readonly int UavCurrentLuma = Shader.PropertyToID("rw_current_luma"); + public static readonly int UavLumaInstability = Shader.PropertyToID("rw_luma_instability"); + public static readonly int UavIntermediate = Shader.PropertyToID("rw_intermediate_fp16x1"); + public static readonly int UavSpdMip0 = Shader.PropertyToID("rw_spd_mip0"); + public static readonly int UavSpdMip1 = Shader.PropertyToID("rw_spd_mip1"); + public static readonly int UavSpdMip2 = Shader.PropertyToID("rw_spd_mip2"); + public static readonly int UavSpdMip3 = Shader.PropertyToID("rw_spd_mip3"); + public static readonly int UavSpdMip4 = Shader.PropertyToID("rw_spd_mip4"); + public static readonly int UavSpdMip5 = Shader.PropertyToID("rw_spd_mip5"); + public static readonly int UavAutoComposition = Shader.PropertyToID("rw_output_autocomposition"); + public static readonly int UavPrevColorPreAlpha = Shader.PropertyToID("rw_output_prev_color_pre_alpha"); + public static readonly int UavPrevColorPostAlpha = Shader.PropertyToID("rw_output_prev_color_post_alpha"); + + // Constant buffer bindings + public static readonly int CbFsr3Upscaler = Shader.PropertyToID("cbFSR3Upscaler"); + public static readonly int CbSpd = Shader.PropertyToID("cbSPD"); + public static readonly int CbRcas = Shader.PropertyToID("cbRCAS"); + public static readonly int CbGenReactive = Shader.PropertyToID("cbGenerateReactive"); + } +} diff --git a/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3ShaderIDs.cs.meta b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3ShaderIDs.cs.meta new file mode 100644 index 0000000..734aedb --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3ShaderIDs.cs.meta @@ -0,0 +1,3 @@ +fileFormatVersion: 2 +guid: a1e3e7c189594b94897510b4a20b8a00 +timeCreated: 1679060863 \ No newline at end of file diff --git a/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3Upscaler.cs b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3Upscaler.cs new file mode 100644 index 0000000..d69aa29 --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3Upscaler.cs @@ -0,0 +1,313 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +using System; +using System.Runtime.InteropServices; +using UnityEngine; +using UnityEngine.Rendering; + +namespace FidelityFX.FSR3 +{ + /// + /// A collection of helper functions and data structures required by the FSR3 Upscaler process. + /// + public static class Fsr3Upscaler + { + /// + /// Creates a new FSR3 Upscaler context with standard parameters that are appropriate for the current platform. + /// + public static Fsr3UpscalerContext CreateContext(Vector2Int displaySize, Vector2Int maxRenderSize, Fsr3UpscalerShaders shaders, InitializationFlags flags = 0) + { + if (SystemInfo.usesReversedZBuffer) + flags |= InitializationFlags.EnableDepthInverted; + else + flags &= ~InitializationFlags.EnableDepthInverted; + +#if UNITY_EDITOR || DEVELOPMENT_BUILD + flags |= InitializationFlags.EnableDebugChecking; +#endif + + Debug.Log($"Setting up FSR3 Upscaler with render size: {maxRenderSize.x}x{maxRenderSize.y}, display size: {displaySize.x}x{displaySize.y}, flags: {flags}"); + + var contextDescription = new ContextDescription + { + Flags = flags, + MaxUpscaleSize = displaySize, + MaxRenderSize = maxRenderSize, + Shaders = shaders, + }; + + var context = new Fsr3UpscalerContext(); + context.Create(contextDescription); + return context; + } + + public static float GetUpscaleRatioFromQualityMode(QualityMode qualityMode) + { + switch (qualityMode) + { + case QualityMode.NativeAA: + return 1.0f; + case QualityMode.UltraQuality: + return 1.2f; + case QualityMode.Quality: + return 1.5f; + case QualityMode.Balanced: + return 1.7f; + case QualityMode.Performance: + return 2.0f; + case QualityMode.UltraPerformance: + return 3.0f; + default: + return 1.0f; + } + } + + public static void GetRenderResolutionFromQualityMode( + out int renderWidth, out int renderHeight, + int displayWidth, int displayHeight, QualityMode qualityMode) + { + float ratio = GetUpscaleRatioFromQualityMode(qualityMode); + renderWidth = Mathf.RoundToInt(displayWidth / ratio); + renderHeight = Mathf.RoundToInt(displayHeight / ratio); + } + + public static float GetMipmapBiasOffset(int renderWidth, int displayWidth) + { + return Mathf.Log((float)renderWidth / displayWidth, 2.0f) - 1.0f; + } + + public static int GetJitterPhaseCount(int renderWidth, int displayWidth) + { + const float basePhaseCount = 8.0f; + int jitterPhaseCount = (int)(basePhaseCount * Mathf.Pow((float)displayWidth / renderWidth, 2.0f)); + return jitterPhaseCount; + } + + public static void GetJitterOffset(out float outX, out float outY, int index, int phaseCount) + { + outX = Halton((index % phaseCount) + 1, 2) - 0.5f; + outY = Halton((index % phaseCount) + 1, 3) - 0.5f; + } + + // Calculate halton number for index and base. + private static float Halton(int index, int @base) + { + float f = 1.0f, result = 0.0f; + + for (int currentIndex = index; currentIndex > 0;) { + + f /= @base; + result += f * (currentIndex % @base); + currentIndex = (int)Mathf.Floor((float)currentIndex / @base); + } + + return result; + } + + public static float Lanczos2(float value) + { + return Mathf.Abs(value) < Mathf.Epsilon ? 1.0f : Mathf.Sin(Mathf.PI * value) / (Mathf.PI * value) * (Mathf.Sin(0.5f * Mathf.PI * value) / (0.5f * Mathf.PI * value)); + } + +#if !UNITY_2021_1_OR_NEWER + internal static void SetBufferData(this CommandBuffer commandBuffer, ComputeBuffer computeBuffer, Array data) + { + commandBuffer.SetComputeBufferData(computeBuffer, data); + } +#endif + + public enum QualityMode + { + NativeAA = 0, + UltraQuality = 1, + Quality = 2, + Balanced = 3, + Performance = 4, + UltraPerformance = 5, + } + + [Flags] + public enum InitializationFlags + { + EnableHighDynamicRange = 1 << 0, + EnableDisplayResolutionMotionVectors = 1 << 1, + EnableMotionVectorsJitterCancellation = 1 << 2, + EnableDepthInverted = 1 << 3, + EnableDepthInfinite = 1 << 4, + EnableAutoExposure = 1 << 5, + EnableDynamicResolution = 1 << 6, + EnableFP16Usage = 1 << 7, + EnableDebugChecking = 1 << 8, + } + + [Flags] + public enum DispatchFlags + { + DrawDebugView = 1 << 0, + } + + /// + /// A structure encapsulating the parameters required to initialize FidelityFX Super Resolution 3 upscaling. + /// + public struct ContextDescription + { + public InitializationFlags Flags; + public Vector2Int MaxRenderSize; + public Vector2Int MaxUpscaleSize; + public Fsr3UpscalerShaders Shaders; + } + + /// + /// A structure encapsulating the parameters for dispatching the various passes of FidelityFX Super Resolution 3. + /// + public class DispatchDescription + { + public ResourceView Color; + public ResourceView Depth; + public ResourceView MotionVectors; + public ResourceView Exposure; // optional + public ResourceView Reactive; // optional + public ResourceView TransparencyAndComposition; // optional + public ResourceView Output; + public Vector2 JitterOffset; + public Vector2 MotionVectorScale; + public Vector2Int RenderSize; + public Vector2Int UpscaleSize; + public bool EnableSharpening; + public float Sharpness; + public float FrameTimeDelta; // in seconds + public float PreExposure; + public bool Reset; + public float CameraNear; + public float CameraFar; + public float CameraFovAngleVertical; + public float ViewSpaceToMetersFactor; + public DispatchFlags Flags; + public bool UseTextureArrays; // Enable texture array bindings, primarily used for HDRP and XR + + // EXPERIMENTAL reactive mask generation parameters + public bool EnableAutoReactive; + public ResourceView ColorOpaqueOnly; + public float AutoTcThreshold = 0.05f; + public float AutoTcScale = 1.0f; + public float AutoReactiveScale = 5.0f; + public float AutoReactiveMax = 0.9f; + } + + /// + /// A structure encapsulating the parameters for automatic generation of a reactive mask. + /// The default values for Scale, CutoffThreshold, BinaryValue and Flags were taken from the FSR3 demo project. + /// + public class GenerateReactiveDescription + { + public ResourceView ColorOpaqueOnly; + public ResourceView ColorPreUpscale; + public ResourceView OutReactive; + public Vector2Int RenderSize; + public float Scale = 0.5f; + public float CutoffThreshold = 0.2f; + public float BinaryValue = 0.9f; + public GenerateReactiveFlags Flags = GenerateReactiveFlags.ApplyTonemap | GenerateReactiveFlags.ApplyThreshold | GenerateReactiveFlags.UseComponentsMax; + } + + [Flags] + public enum GenerateReactiveFlags + { + ApplyTonemap = 1 << 0, + ApplyInverseTonemap = 1 << 1, + ApplyThreshold = 1 << 2, + UseComponentsMax = 1 << 3, + } + + [Serializable, StructLayout(LayoutKind.Sequential)] + internal struct UpscalerConstants + { + public Vector2Int renderSize; + public Vector2Int previousFrameRenderSize; + + public Vector2Int upscaleSize; + public Vector2Int previousFrameUpscaleSize; + + public Vector2Int maxRenderSize; + public Vector2Int maxUpscaleSize; + + public Vector4 deviceToViewDepth; + + public Vector2 jitterOffset; + public Vector2 previousFrameJitterOffset; + + public Vector2 motionVectorScale; + public Vector2 downscaleFactor; + + public Vector2 motionVectorJitterCancellation; + public float tanHalfFOV; + public float jitterPhaseCount; + + public float deltaTime; + public float deltaPreExposure; + public float viewSpaceToMetersFactor; + public float frameIndex; + } + + [Serializable, StructLayout(LayoutKind.Sequential)] + internal struct SpdConstants + { + public uint mips; + public uint numWorkGroups; + public uint workGroupOffsetX, workGroupOffsetY; + public uint renderSizeX, renderSizeY; + } + + [Serializable, StructLayout(LayoutKind.Sequential)] + internal struct GenerateReactiveConstants + { + public float scale; + public float threshold; + public float binaryValue; + public uint flags; + } + + [Serializable, StructLayout(LayoutKind.Sequential)] + internal struct GenerateReactiveConstants2 + { + public float autoTcThreshold; + public float autoTcScale; + public float autoReactiveScale; + public float autoReactiveMax; + } + + [Serializable, StructLayout(LayoutKind.Sequential)] + internal struct RcasConstants + { + public RcasConstants(uint sharpness, uint halfSharp) + { + this.sharpness = sharpness; + this.halfSharp = halfSharp; + dummy0 = dummy1 = 0; + } + + public readonly uint sharpness; + public readonly uint halfSharp; + public readonly uint dummy0; + public readonly uint dummy1; + } + } +} diff --git a/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3Upscaler.cs.meta b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3Upscaler.cs.meta new file mode 100644 index 0000000..1c62899 --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3Upscaler.cs.meta @@ -0,0 +1,3 @@ +fileFormatVersion: 2 +guid: 7ee72b891c35d614eac306ca6154b66d +timeCreated: 1673441954 \ No newline at end of file diff --git a/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerAssets.cs b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerAssets.cs new file mode 100644 index 0000000..84c4474 --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerAssets.cs @@ -0,0 +1,177 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +using UnityEngine; +using UnityEngine.Serialization; + +namespace FidelityFX.FSR3 +{ + /// + /// Scriptable object containing all shader resources required by FidelityFX Super Resolution 3 (FSR3) Upscaler. + /// These can be stored in an asset file and referenced from a scene or prefab, avoiding the need to load the shaders from a Resources folder. + /// + [CreateAssetMenu(fileName = "FSR3 Upscaler Assets", menuName = "FidelityFX/FSR3 Upscaler Assets", order = 1103)] + public class Fsr3UpscalerAssets : ScriptableObject + { + public Fsr3UpscalerShaders shaders; + +#if UNITY_EDITOR + private void Reset() + { + shaders = new Fsr3UpscalerShaders + { + prepareInputsPass = FindComputeShader("ffx_fsr3upscaler_prepare_inputs_pass"), + lumaPyramidPass = FindComputeShader("ffx_fsr3upscaler_luma_pyramid_pass"), + shadingChangePyramidPass = FindComputeShader("ffx_fsr3upscaler_shading_change_pyramid_pass"), + shadingChangePass = FindComputeShader("ffx_fsr3upscaler_shading_change_pass"), + prepareReactivityPass = FindComputeShader("ffx_fsr3upscaler_prepare_reactivity_pass"), + lumaInstabilityPass = FindComputeShader("ffx_fsr3upscaler_luma_instability_pass"), + accumulatePass = FindComputeShader("ffx_fsr3upscaler_accumulate_pass"), + sharpenPass = FindComputeShader("ffx_fsr3upscaler_rcas_pass"), + autoGenReactivePass = FindComputeShader("ffx_fsr3upscaler_autogen_reactive_pass"), + tcrAutoGenPass = FindComputeShader("ffx_fsr3upscaler_tcr_autogen_pass"), + debugViewPass = FindComputeShader("ffx_fsr3upscaler_debug_view_pass"), + }; + } + + private static ComputeShader FindComputeShader(string name) + { + string[] assetGuids = UnityEditor.AssetDatabase.FindAssets($"t:ComputeShader {name}"); + if (assetGuids == null || assetGuids.Length == 0) + return null; + + string assetPath = UnityEditor.AssetDatabase.GUIDToAssetPath(assetGuids[0]); + return UnityEditor.AssetDatabase.LoadAssetAtPath(assetPath); + } +#endif + } + + /// + /// All the compute shaders used by the FSR3 Upscaler. + /// + [System.Serializable] + public class Fsr3UpscalerShaders + { + /// + /// The compute shader used by the prepare inputs pass. + /// + public ComputeShader prepareInputsPass; + + /// + /// The compute shader used by the luminance pyramid computation pass. + /// + public ComputeShader lumaPyramidPass; + + /// + /// The compute shader used by the shading change pyramid pass. + /// + public ComputeShader shadingChangePyramidPass; + + /// + /// The compute shader used by the shading change pass. + /// + public ComputeShader shadingChangePass; + + /// + /// The compute shader used by the prepare reactivity pass. + /// + public ComputeShader prepareReactivityPass; + + /// + /// The compute shader used by the luma instability pass. + /// + public ComputeShader lumaInstabilityPass; + + /// + /// The compute shader used by the accumulation pass. + /// + public ComputeShader accumulatePass; + + /// + /// The compute shader used by the RCAS sharpening pass. + /// + public ComputeShader sharpenPass; + + /// + /// The compute shader used to auto-generate a reactive mask. + /// + public ComputeShader autoGenReactivePass; + + /// + /// The compute shader used to auto-generate a transparency & composition mask. + /// + public ComputeShader tcrAutoGenPass; + + /// + /// The compute shader used to display a debug view. + /// + public ComputeShader debugViewPass; + + /// + /// Returns a copy of this class and its contents. + /// + public Fsr3UpscalerShaders Clone() + { + return (Fsr3UpscalerShaders)MemberwiseClone(); + } + + /// + /// Returns a copy of this class with clones of all its shaders. + /// This can be useful if you're running multiple FSR3 Upscaler instances with different shader configurations. + /// Be sure to clean up these clones through Dispose once you're done with them. + /// + public Fsr3UpscalerShaders DeepCopy() + { + return new Fsr3UpscalerShaders + { + prepareInputsPass = Object.Instantiate(prepareInputsPass), + lumaPyramidPass = Object.Instantiate(lumaPyramidPass), + shadingChangePyramidPass = Object.Instantiate(shadingChangePyramidPass), + shadingChangePass = Object.Instantiate(shadingChangePass), + prepareReactivityPass = Object.Instantiate(prepareReactivityPass), + lumaInstabilityPass = Object.Instantiate(lumaInstabilityPass), + accumulatePass = Object.Instantiate(accumulatePass), + sharpenPass = Object.Instantiate(sharpenPass), + autoGenReactivePass = Object.Instantiate(autoGenReactivePass), + tcrAutoGenPass = Object.Instantiate(tcrAutoGenPass), + debugViewPass = Object.Instantiate(debugViewPass), + }; + } + + /// + /// Destroy all the shaders within this instance. + /// Use this only on clones created through DeepCopy. + /// + public void Dispose() + { + Object.Destroy(prepareInputsPass); + Object.Destroy(lumaPyramidPass); + Object.Destroy(shadingChangePyramidPass); + Object.Destroy(shadingChangePass); + Object.Destroy(prepareReactivityPass); + Object.Destroy(lumaInstabilityPass); + Object.Destroy(accumulatePass); + Object.Destroy(sharpenPass); + Object.Destroy(autoGenReactivePass); + Object.Destroy(tcrAutoGenPass); + Object.Destroy(debugViewPass); + } + } +} diff --git a/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerAssets.cs.meta b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerAssets.cs.meta new file mode 100644 index 0000000..7c8565b --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerAssets.cs.meta @@ -0,0 +1,11 @@ +fileFormatVersion: 2 +guid: 2dbcc608a4754d049a14a0bcce2eb40b +MonoImporter: + externalObjects: {} + serializedVersion: 2 + defaultReferences: [] + executionOrder: 0 + icon: {instanceID: 0} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerCallbacks.cs b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerCallbacks.cs new file mode 100644 index 0000000..2cb48f2 --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerCallbacks.cs @@ -0,0 +1,81 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +using UnityEngine; + +namespace FidelityFX.FSR3 +{ + /// + /// A collection of callbacks required by the FSR3 Upscaler process. + /// This allows some customization by the game dev on how to integrate FSR3 upscaling into their own game setup. + /// + public interface IFsr3UpscalerCallbacks + { + /// + /// Apply a mipmap bias to in-game textures to prevent them from becoming blurry as the internal rendering resolution lowers. + /// This will need to be customized on a per-game basis, as there is no clear universal way to determine what are "in-game" textures. + /// The default implementation will simply apply a mipmap bias to all 2D textures, which will include things like UI textures and which might miss things like terrain texture arrays. + /// + /// Depending on how your game organizes its assets, you will want to create a filter that more specifically selects the textures that need to have this mipmap bias applied. + /// You may also want to store the bias offset value and apply it to any assets that are loaded in on demand. + /// + void ApplyMipmapBias(float biasOffset); + + void UndoMipmapBias(); + } + + /// + /// Default implementation of IFsr3UpscalerCallbacks. + /// These are fine for testing but a proper game will want to extend and override these methods. + /// + public class Fsr3UpscalerCallbacksBase: IFsr3UpscalerCallbacks + { + protected float CurrentBiasOffset = 0; + + public virtual void ApplyMipmapBias(float biasOffset) + { + if (float.IsNaN(biasOffset) || float.IsInfinity(biasOffset)) + return; + + CurrentBiasOffset += biasOffset; + + if (Mathf.Approximately(CurrentBiasOffset, 0f)) + { + CurrentBiasOffset = 0f; + } + + foreach (var texture in Resources.FindObjectsOfTypeAll()) + { + if (texture.mipmapCount <= 1) + continue; + + texture.mipMapBias += biasOffset; + } + } + + public virtual void UndoMipmapBias() + { + if (CurrentBiasOffset == 0f) + return; + + ApplyMipmapBias(-CurrentBiasOffset); + } + } +} diff --git a/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerCallbacks.cs.meta b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerCallbacks.cs.meta new file mode 100644 index 0000000..d68162b --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerCallbacks.cs.meta @@ -0,0 +1,11 @@ +fileFormatVersion: 2 +guid: d52448255f58a6a42bfcf1634f4dd1a8 +MonoImporter: + externalObjects: {} + serializedVersion: 2 + defaultReferences: [] + executionOrder: 0 + icon: {instanceID: 0} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerContext.cs b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerContext.cs new file mode 100644 index 0000000..2b7f2ef --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerContext.cs @@ -0,0 +1,657 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +using System; +using System.Runtime.InteropServices; +using UnityEngine; +using UnityEngine.Rendering; + +namespace FidelityFX.FSR3 +{ + /// + /// This class loosely matches the FfxFsr3UpscalerContext struct from the original FSR3 codebase. + /// It manages the various resources and compute passes required by the FSR3 Upscaler process. + /// Note that this class does not know anything about Unity render pipelines; all it knows is CommandBuffers and RenderTargetIdentifiers. + /// This should make it suitable for integration with any of the available Unity render pipelines. + /// + public class Fsr3UpscalerContext + { + private const int MaxQueuedFrames = 16; + + private Fsr3Upscaler.ContextDescription _contextDescription; + private CommandBuffer _commandBuffer; + + private Fsr3UpscalerPass _prepareInputsPass; + private Fsr3UpscalerPass _lumaPyramidPass; + private Fsr3UpscalerPass _shadingChangePyramidPass; + private Fsr3UpscalerPass _shadingChangePass; + private Fsr3UpscalerPass _prepareReactivityPass; + private Fsr3UpscalerPass _lumaInstabilityPass; + private Fsr3UpscalerPass _accumulatePass; + private Fsr3UpscalerPass _sharpenPass; + private Fsr3UpscalerPass _generateReactivePass; + private Fsr3UpscalerPass _tcrAutogeneratePass; +#if UNITY_EDITOR || DEVELOPMENT_BUILD + private Fsr3UpscalerPass _debugViewPass; +#endif + + private readonly Fsr3UpscalerResources _resources = new Fsr3UpscalerResources(); + + private ComputeBuffer _upscalerConstantsBuffer; + private readonly Fsr3Upscaler.UpscalerConstants[] _upscalerConstantsArray = { new Fsr3Upscaler.UpscalerConstants() }; + private ref Fsr3Upscaler.UpscalerConstants UpscalerConsts => ref _upscalerConstantsArray[0]; + + private ComputeBuffer _spdConstantsBuffer; + private readonly Fsr3Upscaler.SpdConstants[] _spdConstantsArray = { new Fsr3Upscaler.SpdConstants() }; + private ref Fsr3Upscaler.SpdConstants SpdConsts => ref _spdConstantsArray[0]; + + private ComputeBuffer _rcasConstantsBuffer; + private readonly Fsr3Upscaler.RcasConstants[] _rcasConstantsArray = new Fsr3Upscaler.RcasConstants[1]; + private ref Fsr3Upscaler.RcasConstants RcasConsts => ref _rcasConstantsArray[0]; + + private ComputeBuffer _generateReactiveConstantsBuffer; + private readonly Fsr3Upscaler.GenerateReactiveConstants[] _generateReactiveConstantsArray = { new Fsr3Upscaler.GenerateReactiveConstants() }; + private ref Fsr3Upscaler.GenerateReactiveConstants GenReactiveConsts => ref _generateReactiveConstantsArray[0]; + + private ComputeBuffer _tcrAutogenerateConstantsBuffer; + private readonly Fsr3Upscaler.GenerateReactiveConstants2[] _tcrAutogenerateConstantsArray = { new Fsr3Upscaler.GenerateReactiveConstants2() }; + private ref Fsr3Upscaler.GenerateReactiveConstants2 TcrAutoGenConsts => ref _tcrAutogenerateConstantsArray[0]; + + private bool _firstExecution; + private int _resourceFrameIndex; + private Vector2 _previousJitterOffset; + private float _preExposure; + private float _previousFramePreExposure; + + public void Create(Fsr3Upscaler.ContextDescription contextDescription) + { + _contextDescription = contextDescription; + _commandBuffer = new CommandBuffer { name = "FSR3 Upscaler" }; + + _upscalerConstantsBuffer = CreateConstantBuffer(); + _spdConstantsBuffer = CreateConstantBuffer(); + _rcasConstantsBuffer = CreateConstantBuffer(); + _generateReactiveConstantsBuffer = CreateConstantBuffer(); + _tcrAutogenerateConstantsBuffer = CreateConstantBuffer(); + + // Set defaults + _firstExecution = true; + _resourceFrameIndex = 0; + + UpscalerConsts.maxUpscaleSize = _contextDescription.MaxUpscaleSize; + + _resources.Create(_contextDescription); + CreatePasses(); + } + + private void CreatePasses() + { + _prepareInputsPass = new Fsr3UpscalerPrepareInputsPass(_contextDescription, _resources, _upscalerConstantsBuffer); + _lumaPyramidPass = new Fsr3UpscalerLumaPyramidPass(_contextDescription, _resources, _upscalerConstantsBuffer, _spdConstantsBuffer); + _shadingChangePyramidPass = new Fsr3UpscalerShadingChangePyramidPass(_contextDescription, _resources, _upscalerConstantsBuffer, _spdConstantsBuffer); + _shadingChangePass = new Fsr3UpscalerShadingChangePass(_contextDescription, _resources, _upscalerConstantsBuffer); + _prepareReactivityPass = new Fsr3UpscalerPrepareReactivityPass(_contextDescription, _resources, _upscalerConstantsBuffer); + _lumaInstabilityPass = new Fsr3UpscalerLumaInstabilityPass(_contextDescription, _resources, _upscalerConstantsBuffer); + _accumulatePass = new Fsr3UpscalerAccumulatePass(_contextDescription, _resources, _upscalerConstantsBuffer); + _sharpenPass = new Fsr3UpscalerSharpenPass(_contextDescription, _resources, _upscalerConstantsBuffer, _rcasConstantsBuffer); + _generateReactivePass = new Fsr3UpscalerGenerateReactivePass(_contextDescription, _resources, _generateReactiveConstantsBuffer); + _tcrAutogeneratePass = new Fsr3UpscalerTcrAutogeneratePass(_contextDescription, _resources, _upscalerConstantsBuffer, _tcrAutogenerateConstantsBuffer); +#if UNITY_EDITOR || DEVELOPMENT_BUILD + _debugViewPass = new Fsr3UpscalerDebugViewPass(_contextDescription, _resources, _upscalerConstantsBuffer); +#endif + } + + public void Destroy() + { +#if UNITY_EDITOR || DEVELOPMENT_BUILD + DestroyPass(ref _debugViewPass); +#endif + DestroyPass(ref _tcrAutogeneratePass); + DestroyPass(ref _generateReactivePass); + DestroyPass(ref _lumaPyramidPass); + DestroyPass(ref _sharpenPass); + DestroyPass(ref _accumulatePass); + DestroyPass(ref _prepareReactivityPass); + DestroyPass(ref _shadingChangePass); + DestroyPass(ref _shadingChangePyramidPass); + DestroyPass(ref _lumaInstabilityPass); + DestroyPass(ref _prepareInputsPass); + + _resources.Destroy(); + + DestroyConstantBuffer(ref _tcrAutogenerateConstantsBuffer); + DestroyConstantBuffer(ref _generateReactiveConstantsBuffer); + DestroyConstantBuffer(ref _rcasConstantsBuffer); + DestroyConstantBuffer(ref _spdConstantsBuffer); + DestroyConstantBuffer(ref _upscalerConstantsBuffer); + + if (_commandBuffer != null) + { + _commandBuffer.Dispose(); + _commandBuffer = null; + } + } + + public void Dispatch(Fsr3Upscaler.DispatchDescription dispatchParams) + { + _commandBuffer.Clear(); + Dispatch(dispatchParams, _commandBuffer); + Graphics.ExecuteCommandBuffer(_commandBuffer); + } + + public void Dispatch(Fsr3Upscaler.DispatchDescription dispatchParams, CommandBuffer commandBuffer) + { + if ((_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDebugChecking) != 0) + { + DebugCheckDispatch(dispatchParams); + } + + if (dispatchParams.UseTextureArrays) + commandBuffer.EnableShaderKeyword("UNITY_FSR_TEXTURE2D_X_ARRAY"); + + if (_firstExecution) + { + commandBuffer.SetRenderTarget(_resources.Accumulation[0]); + commandBuffer.ClearRenderTarget(false, true, Color.clear); + commandBuffer.SetRenderTarget(_resources.Accumulation[1]); + commandBuffer.ClearRenderTarget(false, true, Color.clear); + commandBuffer.SetRenderTarget(_resources.Luma[0]); + commandBuffer.ClearRenderTarget(false, true, Color.clear); + commandBuffer.SetRenderTarget(_resources.Luma[1]); + commandBuffer.ClearRenderTarget(false, true, Color.clear); + } + + int frameIndex = _resourceFrameIndex % 2; + bool resetAccumulation = dispatchParams.Reset || _firstExecution; + _firstExecution = false; + + // If auto exposure is enabled use the auto exposure SRV, otherwise what the app sends + if ((_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableAutoExposure) != 0) + dispatchParams.Exposure = new ResourceView(_resources.FrameInfo); + else if (!dispatchParams.Exposure.IsValid) + dispatchParams.Exposure = new ResourceView(_resources.DefaultExposure); + + if (dispatchParams.EnableAutoReactive) + { + // Create the auto-TCR resources only when we need them + if (_resources.AutoReactive == null) + _resources.CreateTcrAutogenResources(_contextDescription); + + if (resetAccumulation) + { + RenderTargetIdentifier opaqueOnly = dispatchParams.ColorOpaqueOnly.IsValid ? dispatchParams.ColorOpaqueOnly.RenderTarget : Fsr3ShaderIDs.SrvOpaqueOnly; + commandBuffer.Blit(_resources.PrevPreAlpha[frameIndex ^ 1], opaqueOnly); + } + } + else if (_resources.AutoReactive != null) + { + // Destroy the auto-TCR resources if we don't use the feature + _resources.DestroyTcrAutogenResources(); + } + + if (!dispatchParams.Reactive.IsValid) dispatchParams.Reactive = new ResourceView(_resources.DefaultReactive); + if (!dispatchParams.TransparencyAndComposition.IsValid) dispatchParams.TransparencyAndComposition = new ResourceView(_resources.DefaultReactive); + Fsr3UpscalerResources.CreateAliasableResources(commandBuffer, _contextDescription, dispatchParams); + + SetupConstants(dispatchParams, resetAccumulation); + + // Reactive mask bias + const int threadGroupWorkRegionDim = 8; + int dispatchSrcX = (UpscalerConsts.renderSize.x + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; + int dispatchSrcY = (UpscalerConsts.renderSize.y + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; + int dispatchDstX = (UpscalerConsts.upscaleSize.x + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; + int dispatchDstY = (UpscalerConsts.upscaleSize.y + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; + int dispatchShadingChangePassX = ((UpscalerConsts.renderSize.x / 2) + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; + int dispatchShadingChangePassY = ((UpscalerConsts.renderSize.y / 2) + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; + + // Clear reconstructed depth for max depth store + if (resetAccumulation) + { + commandBuffer.SetRenderTarget(_resources.Accumulation[frameIndex ^ 1]); + commandBuffer.ClearRenderTarget(false, true, Color.clear); + + commandBuffer.SetRenderTarget(_resources.SpdMips); + commandBuffer.ClearRenderTarget(false, true, Color.clear); + + // Auto exposure always used to track luma changes in locking logic + commandBuffer.SetRenderTarget(_resources.FrameInfo); + commandBuffer.ClearRenderTarget(false, true, new Color(0f, 1e8f, 0f, 0f)); + + // Reset atomic counter to 0 + commandBuffer.SetRenderTarget(_resources.SpdAtomicCounter); + commandBuffer.ClearRenderTarget(false, true, Color.clear); + } + + // FSR3: need to clear here since we need the content of this surface for frame interpolation, so clearing in the lock pass is not an option + bool depthInverted = (_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDepthInverted) == Fsr3Upscaler.InitializationFlags.EnableDepthInverted; + commandBuffer.SetRenderTarget(_resources.ReconstructedPrevNearestDepth); + commandBuffer.ClearRenderTarget(false, true, depthInverted ? Color.clear : Color.white); + + // Auto exposure + SetupSpdConstants(dispatchParams, out var dispatchThreadGroupCount); + + // Initialize constant buffers data + commandBuffer.SetBufferData(_upscalerConstantsBuffer, _upscalerConstantsArray); + commandBuffer.SetBufferData(_spdConstantsBuffer, _spdConstantsArray); + + // Auto reactive + if (dispatchParams.EnableAutoReactive) + { + GenerateTransparencyCompositionReactive(dispatchParams, commandBuffer, frameIndex); + dispatchParams.Reactive = new ResourceView(_resources.AutoReactive); + dispatchParams.TransparencyAndComposition = new ResourceView(_resources.AutoComposition); + } + + _prepareInputsPass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchSrcX, dispatchSrcY); + _lumaPyramidPass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchThreadGroupCount.x, dispatchThreadGroupCount.y); + _shadingChangePyramidPass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchThreadGroupCount.x, dispatchThreadGroupCount.y); + _shadingChangePass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchShadingChangePassX, dispatchShadingChangePassY); + _prepareReactivityPass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchSrcX, dispatchSrcY); + _lumaInstabilityPass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchSrcX, dispatchSrcY); + + _accumulatePass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchDstX, dispatchDstY); + + if (dispatchParams.EnableSharpening) + { + // Compute the constants + SetupRcasConstants(dispatchParams); + commandBuffer.SetBufferData(_rcasConstantsBuffer, _rcasConstantsArray); + + // Dispatch RCAS + const int threadGroupWorkRegionDimRcas = 16; + int threadGroupsX = (UpscalerConsts.upscaleSize.x + threadGroupWorkRegionDimRcas - 1) / threadGroupWorkRegionDimRcas; + int threadGroupsY = (UpscalerConsts.upscaleSize.y + threadGroupWorkRegionDimRcas - 1) / threadGroupWorkRegionDimRcas; + _sharpenPass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, threadGroupsX, threadGroupsY); + } + +#if UNITY_EDITOR || DEVELOPMENT_BUILD + if ((dispatchParams.Flags & Fsr3Upscaler.DispatchFlags.DrawDebugView) != 0) + { + _debugViewPass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchDstX, dispatchDstY); + } +#endif + + _resourceFrameIndex = (_resourceFrameIndex + 1) % MaxQueuedFrames; + + Fsr3UpscalerResources.DestroyAliasableResources(commandBuffer); + + commandBuffer.DisableShaderKeyword("UNITY_FSR_TEXTURE2D_X_ARRAY"); + } + + public void GenerateReactiveMask(Fsr3Upscaler.GenerateReactiveDescription dispatchParams) + { + _commandBuffer.Clear(); + GenerateReactiveMask(dispatchParams, _commandBuffer); + Graphics.ExecuteCommandBuffer(_commandBuffer); + } + + public void GenerateReactiveMask(Fsr3Upscaler.GenerateReactiveDescription dispatchParams, CommandBuffer commandBuffer) + { + const int threadGroupWorkRegionDim = 8; + int dispatchSrcX = (dispatchParams.RenderSize.x + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; + int dispatchSrcY = (dispatchParams.RenderSize.y + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; + + GenReactiveConsts.scale = dispatchParams.Scale; + GenReactiveConsts.threshold = dispatchParams.CutoffThreshold; + GenReactiveConsts.binaryValue = dispatchParams.BinaryValue; + GenReactiveConsts.flags = (uint)dispatchParams.Flags; + commandBuffer.SetBufferData(_generateReactiveConstantsBuffer, _generateReactiveConstantsArray); + + ((Fsr3UpscalerGenerateReactivePass)_generateReactivePass).ScheduleDispatch(commandBuffer, dispatchParams, dispatchSrcX, dispatchSrcY); + } + + private void GenerateTransparencyCompositionReactive(Fsr3Upscaler.DispatchDescription dispatchParams, CommandBuffer commandBuffer, int frameIndex) + { + const int threadGroupWorkRegionDim = 8; + int dispatchSrcX = (dispatchParams.RenderSize.x + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; + int dispatchSrcY = (dispatchParams.RenderSize.y + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; + + TcrAutoGenConsts.autoTcThreshold = dispatchParams.AutoTcThreshold; + TcrAutoGenConsts.autoTcScale = dispatchParams.AutoTcScale; + TcrAutoGenConsts.autoReactiveScale = dispatchParams.AutoReactiveScale; + TcrAutoGenConsts.autoReactiveMax = dispatchParams.AutoReactiveMax; + commandBuffer.SetBufferData(_tcrAutogenerateConstantsBuffer, _tcrAutogenerateConstantsArray); + + _tcrAutogeneratePass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchSrcX, dispatchSrcY); + } + + private void SetupConstants(Fsr3Upscaler.DispatchDescription dispatchParams, bool resetAccumulation) + { + ref Fsr3Upscaler.UpscalerConstants constants = ref UpscalerConsts; + + constants.previousFrameJitterOffset = constants.jitterOffset; + constants.jitterOffset = dispatchParams.JitterOffset; + + constants.previousFrameRenderSize = constants.renderSize; + constants.renderSize = dispatchParams.RenderSize; + constants.maxRenderSize = _contextDescription.MaxRenderSize; + + // Compute the horizontal FOV for the shader from the vertical one + float aspectRatio = (float)dispatchParams.RenderSize.x / dispatchParams.RenderSize.y; + float cameraAngleHorizontal = Mathf.Atan(Mathf.Tan(dispatchParams.CameraFovAngleVertical / 2.0f) * aspectRatio) * 2.0f; + constants.tanHalfFOV = Mathf.Tan(cameraAngleHorizontal * 0.5f); + constants.viewSpaceToMetersFactor = (dispatchParams.ViewSpaceToMetersFactor > 0.0f) ? dispatchParams.ViewSpaceToMetersFactor : 1.0f; + + // Compute params to enable device depth to view space depth computation in shader + constants.deviceToViewDepth = SetupDeviceDepthToViewSpaceDepthParams(dispatchParams); + + constants.previousFrameUpscaleSize = constants.upscaleSize; + if (dispatchParams.UpscaleSize.x == 0 && dispatchParams.UpscaleSize.y == 0) + { + constants.upscaleSize = _contextDescription.MaxUpscaleSize; + } + else + { + constants.upscaleSize = dispatchParams.UpscaleSize; + } + + // To be updated if resource is larger than the actual image size + constants.downscaleFactor = new Vector2((float)constants.renderSize.x / constants.upscaleSize.x, (float)constants.renderSize.y / constants.upscaleSize.y); + + // Calculate pre-exposure relevant factors + constants.deltaPreExposure = 1.0f; + _previousFramePreExposure = _preExposure; + _preExposure = dispatchParams.PreExposure != 0.0f ? dispatchParams.PreExposure : 1.0f; + + if (_previousFramePreExposure > 0.0f) + { + constants.deltaPreExposure = _preExposure / _previousFramePreExposure; + } + + // Motion vector data + Vector2Int motionVectorsTargetSize = (_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDisplayResolutionMotionVectors) != 0 ? constants.upscaleSize : constants.renderSize; + constants.motionVectorScale = dispatchParams.MotionVectorScale / motionVectorsTargetSize; + + // Compute jitter cancellation + if ((_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableMotionVectorsJitterCancellation) != 0) + { + constants.motionVectorJitterCancellation = (_previousJitterOffset - constants.jitterOffset) / motionVectorsTargetSize; + _previousJitterOffset = constants.jitterOffset; + } + + int jitterPhaseCount = Fsr3Upscaler.GetJitterPhaseCount(dispatchParams.RenderSize.x, _contextDescription.MaxUpscaleSize.x); + if (resetAccumulation || constants.jitterPhaseCount == 0) + { + constants.jitterPhaseCount = jitterPhaseCount; + } + else + { + int jitterPhaseCountDelta = (int)(jitterPhaseCount - constants.jitterPhaseCount); + if (jitterPhaseCountDelta > 0) + constants.jitterPhaseCount++; + else if (jitterPhaseCountDelta < 0) + constants.jitterPhaseCount--; + } + + // Convert delta time to seconds and clamp to [0, 1] + constants.deltaTime = Mathf.Clamp01(dispatchParams.FrameTimeDelta); + + if (resetAccumulation) + constants.frameIndex = 0; + else + constants.frameIndex += 1.0f; + } + + private Vector4 SetupDeviceDepthToViewSpaceDepthParams(Fsr3Upscaler.DispatchDescription dispatchParams) + { + bool inverted = (_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDepthInverted) != 0; + bool infinite = (_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDepthInfinite) != 0; + + // make sure it has no impact if near and far plane values are swapped in dispatch params + // the flags "inverted" and "infinite" will decide what transform to use + float min = Mathf.Min(dispatchParams.CameraNear, dispatchParams.CameraFar); + float max = Mathf.Max(dispatchParams.CameraNear, dispatchParams.CameraFar); + + if (inverted) + { + (min, max) = (max, min); + } + + float q = max / (min - max); + float d = -1.0f; + + Vector4 matrixElemC = new Vector4(q, -1.0f - Mathf.Epsilon, q, 0.0f + Mathf.Epsilon); + Vector4 matrixElemE = new Vector4(q * min, -min - Mathf.Epsilon, q * min, max); + + // Revert x and y coords + float aspect = (float)dispatchParams.RenderSize.x / dispatchParams.RenderSize.y; + float cotHalfFovY = Mathf.Cos(0.5f * dispatchParams.CameraFovAngleVertical) / Mathf.Sin(0.5f * dispatchParams.CameraFovAngleVertical); + + int matrixIndex = (inverted ? 2 : 0) + (infinite ? 1 : 0); + return new Vector4( + d * matrixElemC[matrixIndex], + matrixElemE[matrixIndex], + aspect / cotHalfFovY, + 1.0f / cotHalfFovY); + } + + private void SetupRcasConstants(Fsr3Upscaler.DispatchDescription dispatchParams) + { + int sharpnessIndex = Mathf.RoundToInt(Mathf.Clamp01(dispatchParams.Sharpness) * (RcasConfigs.Length - 1)); + RcasConsts = RcasConfigs[sharpnessIndex]; + } + + private void SetupSpdConstants(Fsr3Upscaler.DispatchDescription dispatchParams, out Vector2Int dispatchThreadGroupCount) + { + RectInt rectInfo = new RectInt(0, 0, dispatchParams.RenderSize.x, dispatchParams.RenderSize.y); + SpdSetup(rectInfo, out dispatchThreadGroupCount, out var workGroupOffset, out var numWorkGroupsAndMips); + + // Downsample + ref Fsr3Upscaler.SpdConstants spdConstants = ref SpdConsts; + spdConstants.numWorkGroups = (uint)numWorkGroupsAndMips.x; + spdConstants.mips = (uint)numWorkGroupsAndMips.y; + spdConstants.workGroupOffsetX = (uint)workGroupOffset.x; + spdConstants.workGroupOffsetY = (uint)workGroupOffset.y; + spdConstants.renderSizeX = (uint)dispatchParams.RenderSize.x; + spdConstants.renderSizeY = (uint)dispatchParams.RenderSize.y; + } + + private static void SpdSetup(RectInt rectInfo, out Vector2Int dispatchThreadGroupCount, out Vector2Int workGroupOffset, out Vector2Int numWorkGroupsAndMips, int mips = -1) + { + workGroupOffset = new Vector2Int(rectInfo.x / 64, rectInfo.y / 64); + + int endIndexX = (rectInfo.x + rectInfo.width - 1) / 64; + int endIndexY = (rectInfo.y + rectInfo.height - 1) / 64; + + dispatchThreadGroupCount = new Vector2Int(endIndexX + 1 - workGroupOffset.x, endIndexY + 1 - workGroupOffset.y); + + numWorkGroupsAndMips = new Vector2Int(dispatchThreadGroupCount.x * dispatchThreadGroupCount.y, mips); + if (mips < 0) + { + float resolution = Math.Max(rectInfo.width, rectInfo.height); + numWorkGroupsAndMips.y = Math.Min(Mathf.FloorToInt(Mathf.Log(resolution, 2.0f)), 12); + } + } + + private void DebugCheckDispatch(Fsr3Upscaler.DispatchDescription dispatchParams) + { + if (!dispatchParams.Color.IsValid) + { + Debug.LogError("Color resource is null"); + } + + if (!dispatchParams.Depth.IsValid) + { + Debug.LogError("Depth resource is null"); + } + + if (!dispatchParams.MotionVectors.IsValid) + { + Debug.LogError("MotionVectors resource is null"); + } + + if (dispatchParams.Exposure.IsValid && (_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableAutoExposure) != 0) + { + Debug.LogWarning("Exposure resource provided, however auto exposure flag is present"); + } + + if (!dispatchParams.Output.IsValid) + { + Debug.LogError("Output resource is null"); + } + + if (Mathf.Abs(dispatchParams.JitterOffset.x) > 1.0f || Mathf.Abs(dispatchParams.JitterOffset.y) > 1.0f) + { + Debug.LogWarning("JitterOffset contains value outside of expected range [-1.0, 1.0]"); + } + + if (dispatchParams.MotionVectorScale.x > _contextDescription.MaxRenderSize.x || dispatchParams.MotionVectorScale.y > _contextDescription.MaxRenderSize.y) + { + Debug.LogWarning("MotionVectorScale contains scale value greater than MaxRenderSize"); + } + + if (dispatchParams.MotionVectorScale.x == 0.0f || dispatchParams.MotionVectorScale.y == 0.0f) + { + Debug.LogWarning("MotionVectorScale contains zero scale value"); + } + + if (dispatchParams.RenderSize.x > _contextDescription.MaxRenderSize.x || dispatchParams.RenderSize.y > _contextDescription.MaxRenderSize.y) + { + Debug.LogWarning("RenderSize is greater than context MaxRenderSize"); + } + + if (dispatchParams.RenderSize.x == 0 || dispatchParams.RenderSize.y == 0) + { + Debug.LogWarning("RenderSize contains zero dimension"); + } + + if (dispatchParams.Sharpness < 0.0f || dispatchParams.Sharpness > 1.0f) + { + Debug.LogWarning("Sharpness contains value outside of expected range [0.0, 1.0]"); + } + + if (dispatchParams.FrameTimeDelta > 1.0f) + { + Debug.LogWarning("FrameTimeDelta is greater than 1.0f - this value should be seconds (~0.0166 for 60fps)"); + } + + if (dispatchParams.PreExposure == 0.0f) + { + Debug.LogError("PreExposure provided as 0.0f which is invalid"); + } + + bool infiniteDepth = (_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDepthInfinite) != 0; + bool inverseDepth = (_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDepthInverted) != 0; + + if (inverseDepth) + { + if (dispatchParams.CameraNear < dispatchParams.CameraFar) + { + Debug.LogWarning("EnableDepthInverted flag is present yet CameraNear is less than CameraFar"); + } + + if (infiniteDepth) + { + if (dispatchParams.CameraNear < float.MaxValue) + { + Debug.LogWarning("EnableDepthInfinite and EnableDepthInverted present, yet CameraNear != float.MaxValue"); + } + } + + if (dispatchParams.CameraFar < 0.075f) + { + Debug.LogWarning("EnableDepthInverted present, CameraFar value is very low which may result in depth separation artefacting"); + } + } + else + { + if (dispatchParams.CameraNear > dispatchParams.CameraFar) + { + Debug.LogWarning("CameraNear is greater than CameraFar in non-inverted-depth context"); + } + + if (infiniteDepth) + { + if (dispatchParams.CameraFar < float.MaxValue) + { + Debug.LogWarning("EnableDepthInfinite present, yet CameraFar != float.MaxValue"); + } + } + + if (dispatchParams.CameraNear < 0.075f) + { + Debug.LogWarning("CameraNear value is very low which may result in depth separation artefacting"); + } + } + + if (dispatchParams.CameraFovAngleVertical <= 0.0f) + { + Debug.LogError("CameraFovAngleVertical is 0.0f - this value should be > 0.0f"); + } + + if (dispatchParams.CameraFovAngleVertical > Mathf.PI) + { + Debug.LogError("CameraFovAngleVertical is greater than 180 degrees/PI"); + } + } + + /// + /// The FSR3 C++ codebase uses floats bitwise converted to ints to pass sharpness parameters to the RCAS shader. + /// This is not possible in C# without enabling unsafe code compilation, so to avoid that we instead use a table of precomputed values. + /// + private static readonly Fsr3Upscaler.RcasConstants[] RcasConfigs = new [] + { + new Fsr3Upscaler.RcasConstants(1048576000u, 872428544u), + new Fsr3Upscaler.RcasConstants(1049178080u, 877212745u), + new Fsr3Upscaler.RcasConstants(1049823372u, 882390168u), + new Fsr3Upscaler.RcasConstants(1050514979u, 887895276u), + new Fsr3Upscaler.RcasConstants(1051256227u, 893859143u), + new Fsr3Upscaler.RcasConstants(1052050675u, 900216232u), + new Fsr3Upscaler.RcasConstants(1052902144u, 907032080u), + new Fsr3Upscaler.RcasConstants(1053814727u, 914306687u), + new Fsr3Upscaler.RcasConstants(1054792807u, 922105590u), + new Fsr3Upscaler.RcasConstants(1055841087u, 930494326u), + new Fsr3Upscaler.RcasConstants(1056964608u, 939538432u), + new Fsr3Upscaler.RcasConstants(1057566688u, 944322633u), + new Fsr3Upscaler.RcasConstants(1058211980u, 949500056u), + new Fsr3Upscaler.RcasConstants(1058903587u, 955005164u), + new Fsr3Upscaler.RcasConstants(1059644835u, 960969031u), + new Fsr3Upscaler.RcasConstants(1060439283u, 967326120u), + new Fsr3Upscaler.RcasConstants(1061290752u, 974141968u), + new Fsr3Upscaler.RcasConstants(1062203335u, 981416575u), + new Fsr3Upscaler.RcasConstants(1063181415u, 989215478u), + new Fsr3Upscaler.RcasConstants(1064229695u, 997604214u), + new Fsr3Upscaler.RcasConstants(1065353216u, 1006648320), + }; + + private static ComputeBuffer CreateConstantBuffer() where TConstants: struct + { + return new ComputeBuffer(1, Marshal.SizeOf(), ComputeBufferType.Constant); + } + + private static void DestroyConstantBuffer(ref ComputeBuffer bufferRef) + { + if (bufferRef == null) + return; + + bufferRef.Release(); + bufferRef = null; + } + + private static void DestroyPass(ref Fsr3UpscalerPass pass) + { + if (pass == null) + return; + + pass.Dispose(); + pass = null; + } + } +} diff --git a/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerContext.cs.meta b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerContext.cs.meta new file mode 100644 index 0000000..0b27f6d --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerContext.cs.meta @@ -0,0 +1,3 @@ +fileFormatVersion: 2 +guid: 742dda30b4604d5488e7eeda3cf04d56 +timeCreated: 1673442225 \ No newline at end of file diff --git a/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerPass.cs b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerPass.cs new file mode 100644 index 0000000..3b76eb4 --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerPass.cs @@ -0,0 +1,472 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +using System; +using System.Runtime.InteropServices; +using UnityEngine; +using UnityEngine.Profiling; +using UnityEngine.Rendering; + +namespace FidelityFX.FSR3 +{ + /// + /// Base class for all of the compute passes that make up the FSR3 Upscaler process. + /// This loosely matches the FfxPipelineState struct from the original FSR3 codebase, wrapped in an object-oriented blanket. + /// These classes are responsible for loading compute shaders, managing temporary resources, binding resources to shader kernels and dispatching said shaders. + /// + internal abstract class Fsr3UpscalerPass: IDisposable + { + protected readonly Fsr3Upscaler.ContextDescription ContextDescription; + protected readonly Fsr3UpscalerResources Resources; + protected readonly ComputeBuffer Constants; + + protected ComputeShader ComputeShader; + protected int KernelIndex; + + protected CustomSampler Sampler; + + protected Fsr3UpscalerPass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants) + { + ContextDescription = contextDescription; + Resources = resources; + Constants = constants; + } + + public virtual void Dispose() + { + } + + public void ScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) + { + commandBuffer.BeginSample(Sampler); + DoScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchX, dispatchY); + commandBuffer.EndSample(Sampler); + } + + protected abstract void DoScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY); + + protected void InitComputeShader(string passName, ComputeShader shader) + { + InitComputeShader(passName, shader, ContextDescription.Flags); + } + + private void InitComputeShader(string passName, ComputeShader shader, Fsr3Upscaler.InitializationFlags flags) + { + if (shader == null) + { + throw new MissingReferenceException($"Shader for FSR3 Upscaler pass '{passName}' could not be loaded! Please ensure it is included in the project correctly."); + } + + ComputeShader = shader; + KernelIndex = ComputeShader.FindKernel("CS"); + Sampler = CustomSampler.Create(passName); + + bool useLut = false; +#if UNITY_2022_1_OR_NEWER // This will also work in 2020.3.43+ and 2021.3.14+ + if (SystemInfo.computeSubGroupSize == 64) + { + useLut = true; + } +#endif + + // This matches the permutation rules from the CreatePipeline* functions + if ((flags & Fsr3Upscaler.InitializationFlags.EnableHighDynamicRange) != 0) ComputeShader.EnableKeyword("FFX_FSR3UPSCALER_OPTION_HDR_COLOR_INPUT"); + if ((flags & Fsr3Upscaler.InitializationFlags.EnableDisplayResolutionMotionVectors) == 0) ComputeShader.EnableKeyword("FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS"); + if ((flags & Fsr3Upscaler.InitializationFlags.EnableMotionVectorsJitterCancellation) != 0) ComputeShader.EnableKeyword("FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS"); + if ((flags & Fsr3Upscaler.InitializationFlags.EnableDepthInverted) != 0) ComputeShader.EnableKeyword("FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH"); + if (useLut) ComputeShader.EnableKeyword("FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE"); + if ((flags & Fsr3Upscaler.InitializationFlags.EnableFP16Usage) != 0) ComputeShader.EnableKeyword("FFX_HALF"); + } + } + + internal class Fsr3UpscalerPrepareInputsPass : Fsr3UpscalerPass + { + public Fsr3UpscalerPrepareInputsPass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants) + : base(contextDescription, resources, constants) + { + InitComputeShader("Prepare Inputs", contextDescription.Shaders.prepareInputsPass); + } + + protected override void DoScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) + { + ref var color = ref dispatchParams.Color; + ref var depth = ref dispatchParams.Depth; + ref var motionVectors = ref dispatchParams.MotionVectors; + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputColor, color.RenderTarget, color.MipLevel, color.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputDepth, depth.RenderTarget, depth.MipLevel, depth.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputMotionVectors, motionVectors.RenderTarget, motionVectors.MipLevel, motionVectors.SubElement); + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavDilatedMotionVectors, Resources.DilatedVelocity); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavDilatedDepth, Resources.DilatedDepth); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavReconstructedPrevNearestDepth, Resources.ReconstructedPrevNearestDepth); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavFarthestDepth, Fsr3ShaderIDs.UavIntermediate); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavCurrentLuma, Resources.Luma[frameIndex]); + + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); + + commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); + } + } + + internal class Fsr3UpscalerLumaPyramidPass : Fsr3UpscalerPass + { + private readonly ComputeBuffer _spdConstants; + + public Fsr3UpscalerLumaPyramidPass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants, ComputeBuffer spdConstants) + : base(contextDescription, resources, constants) + { + _spdConstants = spdConstants; + + InitComputeShader("Compute Luminance Pyramid", contextDescription.Shaders.lumaPyramidPass); + } + + protected override void DoScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) + { + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvCurrentLuma, Resources.Luma[frameIndex]); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvFarthestDepth, Fsr3ShaderIDs.UavIntermediate); + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavSpdAtomicCount, Resources.SpdAtomicCounter); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavFrameInfo, Resources.FrameInfo); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavSpdMip0, Resources.SpdMips, 0); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavSpdMip1, Resources.SpdMips, 1); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavSpdMip2, Resources.SpdMips, 2); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavSpdMip3, Resources.SpdMips, 3); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavSpdMip4, Resources.SpdMips, 4); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavSpdMip5, Resources.SpdMips, 5); + + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbSpd, _spdConstants, 0, Marshal.SizeOf()); + + commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); + } + } + + internal class Fsr3UpscalerShadingChangePyramidPass : Fsr3UpscalerPass + { + private readonly ComputeBuffer _spdConstants; + + public Fsr3UpscalerShadingChangePyramidPass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants, ComputeBuffer spdConstants) + : base(contextDescription, resources, constants) + { + _spdConstants = spdConstants; + + InitComputeShader("Compute Shading Change Pyramid", contextDescription.Shaders.shadingChangePyramidPass); + } + + protected override void DoScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) + { + ref var exposure = ref dispatchParams.Exposure; + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvCurrentLuma, Resources.Luma[frameIndex]); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvPreviousLuma, Resources.Luma[frameIndex ^ 1]); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvDilatedMotionVectors, Resources.DilatedVelocity); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputExposure, exposure.RenderTarget, exposure.MipLevel, exposure.SubElement); + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavSpdAtomicCount, Resources.SpdAtomicCounter); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavSpdMip0, Resources.SpdMips, 0); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavSpdMip1, Resources.SpdMips, 1); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavSpdMip2, Resources.SpdMips, 2); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavSpdMip3, Resources.SpdMips, 3); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavSpdMip4, Resources.SpdMips, 4); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavSpdMip5, Resources.SpdMips, 5); + + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbSpd, _spdConstants, 0, Marshal.SizeOf()); + + commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); + } + } + + internal class Fsr3UpscalerShadingChangePass : Fsr3UpscalerPass + { + public Fsr3UpscalerShadingChangePass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants) + : base(contextDescription, resources, constants) + { + InitComputeShader("Compute Shading Change", contextDescription.Shaders.shadingChangePass); + } + + protected override void DoScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) + { + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvSpdMips, Resources.SpdMips); + + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); + + commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); + } + } + + internal class Fsr3UpscalerPrepareReactivityPass : Fsr3UpscalerPass + { + public Fsr3UpscalerPrepareReactivityPass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants) + : base(contextDescription, resources, constants) + { + InitComputeShader("Prepare Reactivity", contextDescription.Shaders.prepareReactivityPass); + } + + protected override void DoScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) + { + ref var exposure = ref dispatchParams.Exposure; + ref var reactive = ref dispatchParams.Reactive; + ref var tac = ref dispatchParams.TransparencyAndComposition; + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvReconstructedPrevNearestDepth, Resources.ReconstructedPrevNearestDepth); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvDilatedMotionVectors, Resources.DilatedVelocity); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvDilatedDepth, Resources.DilatedDepth); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvReactiveMask, reactive.RenderTarget, reactive.MipLevel, reactive.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvTransparencyAndCompositionMask, tac.RenderTarget, tac.MipLevel, tac.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvAccumulation, Resources.Accumulation[frameIndex ^ 1]); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvShadingChange, Fsr3ShaderIDs.UavShadingChange); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvCurrentLuma, Resources.Luma[frameIndex]); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputExposure, exposure.RenderTarget, exposure.MipLevel, exposure.SubElement); + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavAccumulation, Resources.Accumulation[frameIndex]); + + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); + + commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); + } + } + + internal class Fsr3UpscalerLumaInstabilityPass : Fsr3UpscalerPass + { + public Fsr3UpscalerLumaInstabilityPass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants) + : base(contextDescription, resources, constants) + { + InitComputeShader("Compute Luminance Instability", contextDescription.Shaders.lumaInstabilityPass); + } + + protected override void DoScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) + { + ref var exposure = ref dispatchParams.Exposure; + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputExposure, exposure.RenderTarget, exposure.MipLevel, exposure.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvDilatedReactiveMasks, Fsr3ShaderIDs.UavDilatedReactiveMasks); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvDilatedMotionVectors, Resources.DilatedVelocity); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvFrameInfo, Resources.FrameInfo); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvLumaHistory, Resources.LumaHistory[frameIndex ^ 1]); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvFarthestDepthMip1, Fsr3ShaderIDs.UavFarthestDepthMip1); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvCurrentLuma, Resources.Luma[frameIndex]); + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavLumaHistory, Resources.LumaHistory[frameIndex]); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavLumaInstability, Fsr3ShaderIDs.UavIntermediate); + + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); + + commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); + } + } + + internal class Fsr3UpscalerAccumulatePass : Fsr3UpscalerPass + { + private const string SharpeningKeyword = "FFX_FSR3UPSCALER_OPTION_APPLY_SHARPENING"; + +#if UNITY_2021_2_OR_NEWER + private readonly LocalKeyword _sharpeningKeyword; +#endif + + public Fsr3UpscalerAccumulatePass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants) + : base(contextDescription, resources, constants) + { + InitComputeShader("Accumulate", contextDescription.Shaders.accumulatePass); +#if UNITY_2021_2_OR_NEWER + _sharpeningKeyword = new LocalKeyword(ComputeShader, SharpeningKeyword); +#endif + } + + protected override void DoScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) + { +#if UNITY_2021_2_OR_NEWER + if (dispatchParams.EnableSharpening) + commandBuffer.EnableKeyword(ComputeShader, _sharpeningKeyword); + else + commandBuffer.DisableKeyword(ComputeShader, _sharpeningKeyword); +#else + if (dispatchParams.EnableSharpening) + commandBuffer.EnableShaderKeyword(SharpeningKeyword); + else + commandBuffer.DisableShaderKeyword(SharpeningKeyword); +#endif + + ref var color = ref dispatchParams.Color; + ref var exposure = ref dispatchParams.Exposure; + ref var output = ref dispatchParams.Output; + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputExposure, exposure.RenderTarget, exposure.MipLevel, exposure.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvDilatedReactiveMasks, Fsr3ShaderIDs.UavDilatedReactiveMasks); + + if ((ContextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDisplayResolutionMotionVectors) == 0) + { + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvDilatedMotionVectors, Resources.DilatedVelocity); + } + else + { + ref var motionVectors = ref dispatchParams.MotionVectors; + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputMotionVectors, motionVectors.RenderTarget, motionVectors.MipLevel, motionVectors.SubElement); + } + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInternalUpscaled, Resources.InternalUpscaled[frameIndex ^ 1]); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvLanczosLut, Resources.LanczosLut); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvFarthestDepthMip1, Fsr3ShaderIDs.UavFarthestDepthMip1); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvCurrentLuma, Resources.Luma[frameIndex]); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvLumaInstability, Fsr3ShaderIDs.UavIntermediate); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputColor, color.RenderTarget, color.MipLevel, color.SubElement); + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavInternalUpscaled, Resources.InternalUpscaled[frameIndex]); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavUpscaledOutput, output.RenderTarget, output.MipLevel, output.SubElement); + + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); + + commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); + } + } + + internal class Fsr3UpscalerSharpenPass : Fsr3UpscalerPass + { + private readonly ComputeBuffer _rcasConstants; + + public Fsr3UpscalerSharpenPass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants, ComputeBuffer rcasConstants) + : base(contextDescription, resources, constants) + { + _rcasConstants = rcasConstants; + + InitComputeShader("RCAS Sharpening", contextDescription.Shaders.sharpenPass); + } + + protected override void DoScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) + { + ref var exposure = ref dispatchParams.Exposure; + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputExposure, exposure.RenderTarget, exposure.MipLevel, exposure.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvRcasInput, Resources.InternalUpscaled[frameIndex]); + + ref var output = ref dispatchParams.Output; + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavUpscaledOutput, output.RenderTarget, output.MipLevel, output.SubElement); + + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbRcas, _rcasConstants, 0, Marshal.SizeOf()); + + commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); + } + } + + internal class Fsr3UpscalerGenerateReactivePass : Fsr3UpscalerPass + { + private readonly ComputeBuffer _generateReactiveConstants; + + public Fsr3UpscalerGenerateReactivePass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer generateReactiveConstants) + : base(contextDescription, resources, null) + { + _generateReactiveConstants = generateReactiveConstants; + + InitComputeShader("Auto-Generate Reactive Mask", contextDescription.Shaders.autoGenReactivePass); + } + + protected override void DoScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) + { + } + + public void ScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.GenerateReactiveDescription dispatchParams, int dispatchX, int dispatchY) + { + commandBuffer.BeginSample(Sampler); + + ref var opaqueOnly = ref dispatchParams.ColorOpaqueOnly; + ref var color = ref dispatchParams.ColorPreUpscale; + ref var reactive = ref dispatchParams.OutReactive; + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvOpaqueOnly, opaqueOnly.RenderTarget, opaqueOnly.MipLevel, opaqueOnly.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputColor, color.RenderTarget, color.MipLevel, color.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavAutoReactive, reactive.RenderTarget, reactive.MipLevel, reactive.SubElement); + + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbGenReactive, _generateReactiveConstants, 0, Marshal.SizeOf()); + + commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); + + commandBuffer.EndSample(Sampler); + } + } + + internal class Fsr3UpscalerTcrAutogeneratePass : Fsr3UpscalerPass + { + private readonly ComputeBuffer _tcrAutogenerateConstants; + + public Fsr3UpscalerTcrAutogeneratePass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants, ComputeBuffer tcrAutogenerateConstants) + : base(contextDescription, resources, constants) + { + _tcrAutogenerateConstants = tcrAutogenerateConstants; + + InitComputeShader("Auto-Generate Transparency & Composition Mask", contextDescription.Shaders.tcrAutoGenPass); + } + + protected override void DoScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) + { + ref var color = ref dispatchParams.Color; + ref var motionVectors = ref dispatchParams.MotionVectors; + ref var opaqueOnly = ref dispatchParams.ColorOpaqueOnly; + ref var reactive = ref dispatchParams.Reactive; + ref var tac = ref dispatchParams.TransparencyAndComposition; + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvOpaqueOnly, opaqueOnly.RenderTarget, opaqueOnly.MipLevel, opaqueOnly.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputColor, color.RenderTarget, color.MipLevel, color.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputMotionVectors, motionVectors.RenderTarget, motionVectors.MipLevel, motionVectors.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvPrevColorPreAlpha, Resources.PrevPreAlpha[frameIndex ^ 1]); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvPrevColorPostAlpha, Resources.PrevPostAlpha[frameIndex ^ 1]); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvReactiveMask, reactive.RenderTarget, reactive.MipLevel, reactive.SubElement); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvTransparencyAndCompositionMask, tac.RenderTarget, tac.MipLevel, tac.SubElement); + + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavAutoReactive, Resources.AutoReactive); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavAutoComposition, Resources.AutoComposition); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavPrevColorPreAlpha, Resources.PrevPreAlpha[frameIndex]); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavPrevColorPostAlpha, Resources.PrevPostAlpha[frameIndex]); + + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbGenReactive, _tcrAutogenerateConstants, 0, Marshal.SizeOf()); + + commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); + } + } + +#if UNITY_EDITOR || DEVELOPMENT_BUILD + internal class Fsr3UpscalerDebugViewPass : Fsr3UpscalerPass + { + public Fsr3UpscalerDebugViewPass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants) + : base(contextDescription, resources, constants) + { + InitComputeShader("Debug View", contextDescription.Shaders.debugViewPass); + } + + protected override void DoScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) + { + ref var exposure = ref dispatchParams.Exposure; + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvDilatedReactiveMasks, Fsr3ShaderIDs.UavDilatedReactiveMasks); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvDilatedMotionVectors, Resources.DilatedVelocity); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvDilatedDepth, Resources.DilatedDepth); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInternalUpscaled, Resources.InternalUpscaled[frameIndex]); + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputExposure, exposure.RenderTarget, exposure.MipLevel, exposure.SubElement); + + ref var output = ref dispatchParams.Output; + commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavUpscaledOutput, output.RenderTarget, output.MipLevel, output.SubElement); + + commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); + + commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); + } + } +#endif +} diff --git a/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerPass.cs.meta b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerPass.cs.meta new file mode 100644 index 0000000..424e36c --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerPass.cs.meta @@ -0,0 +1,3 @@ +fileFormatVersion: 2 +guid: 9e0cf27d80d1d4d4c81450791a411ead +timeCreated: 1676885169 \ No newline at end of file diff --git a/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerResources.cs b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerResources.cs new file mode 100644 index 0000000..112b98d --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerResources.cs @@ -0,0 +1,245 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +using System; +using UnityEngine; +using UnityEngine.Experimental.Rendering; +using UnityEngine.Rendering; + +namespace FidelityFX.FSR3 +{ + /// + /// Helper class for bundling and managing persistent resources required by the FSR3 Upscaler process. + /// This includes lookup tables, default fallback resources and double-buffered resources that get swapped between frames. + /// + internal class Fsr3UpscalerResources + { + public Texture2D LanczosLut; + public Texture2D DefaultExposure; + public Texture2D DefaultReactive; + + public RenderTexture SpdAtomicCounter; + public RenderTexture SpdMips; + public RenderTexture DilatedVelocity; + public RenderTexture DilatedDepth; + public RenderTexture ReconstructedPrevNearestDepth; + public RenderTexture FrameInfo; + public RenderTexture AutoReactive; + public RenderTexture AutoComposition; + + public readonly RenderTexture[] Accumulation = new RenderTexture[2]; + public readonly RenderTexture[] Luma = new RenderTexture[2]; + public readonly RenderTexture[] InternalUpscaled = new RenderTexture[2]; + public readonly RenderTexture[] LumaHistory = new RenderTexture[2]; + public readonly RenderTexture[] PrevPreAlpha = new RenderTexture[2]; + public readonly RenderTexture[] PrevPostAlpha = new RenderTexture[2]; + + public void Create(Fsr3Upscaler.ContextDescription contextDescription) + { + // Generate the data for the LUT + const int lanczos2LutWidth = 128; + float[] lanczos2Weights = new float[lanczos2LutWidth]; + for (int currentLanczosWidthIndex = 0; currentLanczosWidthIndex < lanczos2LutWidth; ++currentLanczosWidthIndex) + { + float x = 2.0f * currentLanczosWidthIndex / (lanczos2LutWidth - 1); + float y = Fsr3Upscaler.Lanczos2(x); + lanczos2Weights[currentLanczosWidthIndex] = y; + } + + Vector2Int maxRenderSize = contextDescription.MaxRenderSize; + Vector2Int maxRenderSizeDiv2 = maxRenderSize / 2; + + // Resource FSR3UPSCALER_LanczosLutData: FFX_RESOURCE_USAGE_READ_ONLY, FFX_SURFACE_FORMAT_R16_SNORM, FFX_RESOURCE_FLAGS_NONE + // R16_SNorm textures are not supported by Unity on most platforms, strangely enough. So instead we use R32_SFloat and upload pre-normalized float data. + LanczosLut = new Texture2D(lanczos2LutWidth, 1, GraphicsFormat.R32_SFloat, TextureCreationFlags.None) { name = "FSR3UPSCALER_LanczosLutData" }; + LanczosLut.SetPixelData(lanczos2Weights, 0); + LanczosLut.Apply(); + + // Resource FSR3UPSCALER_DefaultReactivityMask: FFX_RESOURCE_USAGE_READ_ONLY, FFX_SURFACE_FORMAT_R8_UNORM, FFX_RESOURCE_FLAGS_NONE + DefaultReactive = new Texture2D(1, 1, GraphicsFormat.R8_UNorm, TextureCreationFlags.None) { name = "FSR3UPSCALER_DefaultReactivityMask" }; + DefaultReactive.SetPixel(0, 0, Color.clear); + DefaultReactive.Apply(); + + // Resource FSR3UPSCALER_DefaultExposure: FFX_RESOURCE_USAGE_READ_ONLY, FFX_SURFACE_FORMAT_R32G32_FLOAT, FFX_RESOURCE_FLAGS_NONE + DefaultExposure = new Texture2D(1, 1, GraphicsFormat.R32G32_SFloat, TextureCreationFlags.None) { name = "FSR3UPSCALER_DefaultExposure" }; + DefaultExposure.SetPixel(0, 0, Color.clear); + DefaultExposure.Apply(); + + // Resource FSR3UPSCALER_SpdAtomicCounter: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R32_UINT, FFX_RESOURCE_FLAGS_ALIASABLE + // Despite what the original FSR3 codebase says, this resource really isn't aliasable. Resetting this counter to 0 every frame breaks auto-exposure on MacOS Metal. + SpdAtomicCounter = new RenderTexture(1, 1, 0, GraphicsFormat.R32_UInt) { name = "FSR3UPSCALER_SpdAtomicCounter", enableRandomWrite = true }; + SpdAtomicCounter.Create(); + + // Resource FSR3UPSCALER_SpdMips: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16G16_FLOAT, FFX_RESOURCE_FLAGS_ALIASABLE + // This is a rather special case: it's an aliasable resource, but because we require a mipmap chain and bind specific mip levels per shader, we can't easily use temporary RTs for this. + int mipCount = 1 + Mathf.FloorToInt(Mathf.Log(Math.Max(maxRenderSizeDiv2.x, maxRenderSizeDiv2.y), 2.0f)); + SpdMips = new RenderTexture(maxRenderSizeDiv2.x, maxRenderSizeDiv2.y, 0, GraphicsFormat.R16G16_SFloat, mipCount) { name = "FSR3UPSCALER_SpdMips", enableRandomWrite = true, useMipMap = true, autoGenerateMips = false }; + SpdMips.Create(); + + // Resource FSR3UPSCALER_DilatedVelocity: FFX_RESOURCE_USAGE_RENDERTARGET | FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16G16_FLOAT, FFX_RESOURCE_FLAGS_NONE + DilatedVelocity = new RenderTexture(maxRenderSize.x, maxRenderSize.y, 0, GraphicsFormat.R16G16_SFloat) { name = "FSR3UPSCALER_DilatedVelocity", enableRandomWrite = true }; + DilatedVelocity.Create(); + + // Resource FSR3UPSCALER_DilatedDepth: FFX_RESOURCE_USAGE_RENDERTARGET | FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R32_FLOAT, FFX_RESOURCE_FLAGS_NONE + DilatedDepth = new RenderTexture(maxRenderSize.x, maxRenderSize.y, 0, GraphicsFormat.R32_SFloat) { name = "FSR3UPSCALER_DilatedDepth", enableRandomWrite = true }; + DilatedDepth.Create(); + + // Resource FSR3UPSCALER_ReconstructedPrevNearestDepth: FFX_RESOURCE_USAGE_RENDERTARGET | FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R32_UINT, FFX_RESOURCE_FLAGS_NONE + ReconstructedPrevNearestDepth = new RenderTexture(maxRenderSize.x, maxRenderSize.y, 0, GraphicsFormat.R32_UInt) { name = "FSR3UPSCALER_ReconstructedPrevNearestDepth", enableRandomWrite = true }; + ReconstructedPrevNearestDepth.Create(); + + // Resource FSR3UPSCALER_FrameInfo: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R32G32B32A32_FLOAT, FFX_RESOURCE_FLAGS_NONE + FrameInfo = new RenderTexture(1, 1, 0, GraphicsFormat.R32G32B32A32_SFloat) { name = "FSR3UPSCALER_FrameInfo", enableRandomWrite = true }; + FrameInfo.Create(); + + // Resources FSR3UPSCALER_Accumulation1/2: FFX_RESOURCE_USAGE_RENDERTARGET | FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R8_UNORM, FFX_RESOURCE_FLAGS_NONE + CreateDoubleBufferedResource(Accumulation, "FSR3UPSCALER_Accumulation", maxRenderSize, GraphicsFormat.R8_UNorm); + + // Resources FSR3UPSCALER_Luma1/2: FFX_RESOURCE_USAGE_RENDERTARGET | FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16_FLOAT, FFX_RESOURCE_FLAGS_NONE + CreateDoubleBufferedResource(Luma, "FSR3UPSCALER_Luma", maxRenderSize, GraphicsFormat.R16_SFloat); + + // Resources FSR3UPSCALER_InternalUpscaled1/2: FFX_RESOURCE_USAGE_RENDERTARGET | FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16G16B16A16_FLOAT, FFX_RESOURCE_FLAGS_NONE + CreateDoubleBufferedResource(InternalUpscaled, "FSR3UPSCALER_InternalUpscaled", contextDescription.MaxUpscaleSize, GraphicsFormat.R16G16B16A16_SFloat); + + // Resources FSR3UPSCALER_LumaHistory1/2: FFX_RESOURCE_USAGE_RENDERTARGET | FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16G16B16A16_FLOAT, FFX_RESOURCE_FLAGS_NONE + CreateDoubleBufferedResource(LumaHistory, "FSR3UPSCALER_LumaHistory", maxRenderSize, GraphicsFormat.R16G16B16A16_SFloat); + } + + public void CreateTcrAutogenResources(Fsr3Upscaler.ContextDescription contextDescription) + { + // Resource FSR3UPSCALER_AutoReactive: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R8_UNORM, FFX_RESOURCE_FLAGS_NONE + AutoReactive = new RenderTexture(contextDescription.MaxRenderSize.x, contextDescription.MaxRenderSize.y, 0, GraphicsFormat.R8_UNorm) { name = "FSR3UPSCALER_AutoReactive", enableRandomWrite = true }; + AutoReactive.Create(); + + // Resource FSR3UPSCALER_AutoComposition: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R8_UNORM, FFX_RESOURCE_FLAGS_NONE + AutoComposition = new RenderTexture(contextDescription.MaxRenderSize.x, contextDescription.MaxRenderSize.y, 0, GraphicsFormat.R8_UNorm) { name = "FSR3UPSCALER_AutoComposition", enableRandomWrite = true }; + AutoComposition.Create(); + + // Resources FSR3UPSCALER_PrevPreAlpha0/1: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R11G11B10_FLOAT, FFX_RESOURCE_FLAGS_NONE + CreateDoubleBufferedResource(PrevPreAlpha, "FSR3UPSCALER_PrevPreAlpha", contextDescription.MaxRenderSize, GraphicsFormat.B10G11R11_UFloatPack32); + + // Resources FSR3UPSCALER_PrevPostAlpha0/1: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R11G11B10_FLOAT, FFX_RESOURCE_FLAGS_NONE + CreateDoubleBufferedResource(PrevPostAlpha, "FSR3UPSCALER_PrevPostAlpha", contextDescription.MaxRenderSize, GraphicsFormat.B10G11R11_UFloatPack32); + } + + // Set up shared aliasable resources, i.e. temporary render textures + // These do not need to persist between frames, but they do need to be available between passes + public static void CreateAliasableResources(CommandBuffer commandBuffer, Fsr3Upscaler.ContextDescription contextDescription, Fsr3Upscaler.DispatchDescription dispatchParams) + { + Vector2Int maxUpscaleSize = contextDescription.MaxUpscaleSize; + Vector2Int maxRenderSize = contextDescription.MaxRenderSize; + Vector2Int maxRenderSizeDiv2 = maxRenderSize / 2; + + // FSR3UPSCALER_IntermediateFp16x1: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16_FLOAT, FFX_RESOURCE_FLAGS_ALIASABLE + commandBuffer.GetTemporaryRT(Fsr3ShaderIDs.UavIntermediate, maxRenderSize.x, maxRenderSize.y, 0, default, GraphicsFormat.R16_SFloat, 1, true); + + // FSR3UPSCALER_ShadingChange: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R8_UNORM, FFX_RESOURCE_FLAGS_ALIASABLE + commandBuffer.GetTemporaryRT(Fsr3ShaderIDs.UavShadingChange, maxRenderSizeDiv2.x, maxRenderSizeDiv2.y, 0, default, GraphicsFormat.R8_UNorm, 1, true); + + // FSR3UPSCALER_NewLocks: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R8_UNORM, FFX_RESOURCE_FLAGS_ALIASABLE + commandBuffer.GetTemporaryRT(Fsr3ShaderIDs.UavNewLocks, maxUpscaleSize.x, maxUpscaleSize.y, 0, default, GraphicsFormat.R8_UNorm, 1, true); + + // FSR3UPSCALER_FarthestDepthMip1: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16_FLOAT, FFX_RESOURCE_FLAGS_ALIASABLE + commandBuffer.GetTemporaryRT(Fsr3ShaderIDs.UavFarthestDepthMip1, maxRenderSizeDiv2.x, maxRenderSizeDiv2.y, 0, default, GraphicsFormat.R16_SFloat, 1, true); + + // FSR3UPSCALER_DilatedReactiveMasks: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R8G8B8A8_UNORM, FFX_RESOURCE_FLAGS_ALIASABLE + commandBuffer.GetTemporaryRT(Fsr3ShaderIDs.UavDilatedReactiveMasks, maxRenderSize.x, maxRenderSize.y, 0, default, GraphicsFormat.R8G8B8A8_UNorm, 1, true); + } + + public static void DestroyAliasableResources(CommandBuffer commandBuffer) + { + // Release all of the aliasable resources used this frame + commandBuffer.ReleaseTemporaryRT(Fsr3ShaderIDs.UavDilatedReactiveMasks); + commandBuffer.ReleaseTemporaryRT(Fsr3ShaderIDs.UavFarthestDepthMip1); + commandBuffer.ReleaseTemporaryRT(Fsr3ShaderIDs.UavNewLocks); + commandBuffer.ReleaseTemporaryRT(Fsr3ShaderIDs.UavShadingChange); + commandBuffer.ReleaseTemporaryRT(Fsr3ShaderIDs.UavIntermediate); + } + + private static void CreateDoubleBufferedResource(RenderTexture[] resource, string name, Vector2Int size, GraphicsFormat format) + { + for (int i = 0; i < 2; ++i) + { + resource[i] = new RenderTexture(size.x, size.y, 0, format) { name = name + (i + 1), enableRandomWrite = true }; + resource[i].Create(); + } + } + + public void Destroy() + { + DestroyTcrAutogenResources(); + + DestroyResource(LumaHistory); + DestroyResource(InternalUpscaled); + DestroyResource(Luma); + DestroyResource(Accumulation); + + DestroyResource(ref FrameInfo); + DestroyResource(ref ReconstructedPrevNearestDepth); + DestroyResource(ref DilatedDepth); + DestroyResource(ref DilatedVelocity); + DestroyResource(ref SpdMips); + DestroyResource(ref SpdAtomicCounter); + + DestroyResource(ref DefaultReactive); + DestroyResource(ref DefaultExposure); + DestroyResource(ref LanczosLut); + } + + public void DestroyTcrAutogenResources() + { + DestroyResource(PrevPostAlpha); + DestroyResource(PrevPreAlpha); + DestroyResource(ref AutoComposition); + DestroyResource(ref AutoReactive); + } + + private static void DestroyResource(ref Texture2D resource) + { + if (resource == null) + return; + +#if UNITY_EDITOR + if (Application.isPlaying && !UnityEditor.EditorApplication.isPaused) + UnityEngine.Object.Destroy(resource); + else + UnityEngine.Object.DestroyImmediate(resource); +#else + UnityEngine.Object.Destroy(resource); +#endif + resource = null; + } + + private static void DestroyResource(ref RenderTexture resource) + { + if (resource == null) + return; + + resource.Release(); + resource = null; + } + + private static void DestroyResource(RenderTexture[] resource) + { + for (int i = 0; i < resource.Length; ++i) + DestroyResource(ref resource[i]); + } + } +} diff --git a/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerResources.cs.meta b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerResources.cs.meta new file mode 100644 index 0000000..62d6dc1 --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FSR3/Fsr3UpscalerResources.cs.meta @@ -0,0 +1,3 @@ +fileFormatVersion: 2 +guid: 6749e0e55c9ac5a4f869bc73dfbb7163 +timeCreated: 1677236102 \ No newline at end of file diff --git a/Packages/fidelityfx.fsr/Runtime/FidelityFX.FSR.asmdef b/Packages/fidelityfx.fsr/Runtime/FidelityFX.FSR.asmdef new file mode 100644 index 0000000..4d4ce4c --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FidelityFX.FSR.asmdef @@ -0,0 +1,14 @@ +{ + "name": "FidelityFX.FSR", + "rootNamespace": "FidelityFX", + "references": [], + "includePlatforms": [], + "excludePlatforms": [], + "allowUnsafeCode": false, + "overrideReferences": false, + "precompiledReferences": [], + "autoReferenced": true, + "defineConstraints": [], + "versionDefines": [], + "noEngineReferences": false +} \ No newline at end of file diff --git a/Packages/fidelityfx.fsr/Runtime/FidelityFX.FSR.asmdef.meta b/Packages/fidelityfx.fsr/Runtime/FidelityFX.FSR.asmdef.meta new file mode 100644 index 0000000..7cae25d --- /dev/null +++ b/Packages/fidelityfx.fsr/Runtime/FidelityFX.FSR.asmdef.meta @@ -0,0 +1,7 @@ +fileFormatVersion: 2 +guid: eb1f729ceec147d4a813e8e75b65dc12 +AssemblyDefinitionImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders.meta b/Packages/fidelityfx.fsr/Shaders.meta new file mode 100644 index 0000000..cf5a649 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: af3a6660394230a4bb247808bbefe9f4 +folderAsset: yes +DefaultImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_accumulate_pass.compute b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_accumulate_pass.compute new file mode 100644 index 0000000..63532d7 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_accumulate_pass.compute @@ -0,0 +1,41 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#pragma kernel CS + +#pragma multi_compile_local __ FFX_HALF +#pragma multi_compile_local __ FFX_FSR2_OPTION_REPROJECT_USE_LANCZOS_TYPE +#pragma multi_compile_local __ FFX_FSR2_OPTION_HDR_COLOR_INPUT +#pragma multi_compile_local __ FFX_FSR2_OPTION_LOW_RESOLUTION_MOTION_VECTORS +#pragma multi_compile_local __ FFX_FSR2_OPTION_JITTERED_MOTION_VECTORS +#pragma multi_compile_local __ FFX_FSR2_OPTION_INVERTED_DEPTH +#pragma multi_compile_local __ FFX_FSR2_OPTION_APPLY_SHARPENING + +#pragma multi_compile __ UNITY_FSR_TEXTURE2D_X_ARRAY + +#include "ffx_fsr_unity_common.cginc" + +// Ensure the correct value is defined for this keyword, as it is used to select one of multiple sampler functions +#ifdef FFX_FSR2_OPTION_REPROJECT_USE_LANCZOS_TYPE +#undef FFX_FSR2_OPTION_REPROJECT_USE_LANCZOS_TYPE +#define FFX_FSR2_OPTION_REPROJECT_USE_LANCZOS_TYPE 1 +#endif + +#include "shaders/ffx_fsr2_accumulate_pass.hlsl" diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_accumulate_pass.compute.meta b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_accumulate_pass.compute.meta new file mode 100644 index 0000000..5936e95 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_accumulate_pass.compute.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: 7e791d69a5be98247a93b63897bc64df +ComputeShaderImporter: + externalObjects: {} + preprocessorOverride: 0 + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_autogen_reactive_pass.compute b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_autogen_reactive_pass.compute new file mode 100644 index 0000000..6bc2301 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_autogen_reactive_pass.compute @@ -0,0 +1,32 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#pragma kernel CS + +#pragma multi_compile_local __ FFX_HALF +#pragma multi_compile_local __ FFX_FSR2_OPTION_LOW_RESOLUTION_MOTION_VECTORS +#pragma multi_compile_local __ FFX_FSR2_OPTION_JITTERED_MOTION_VECTORS +#pragma multi_compile_local __ FFX_FSR2_OPTION_INVERTED_DEPTH + +#pragma multi_compile __ UNITY_FSR_TEXTURE2D_X_ARRAY + +#include "ffx_fsr_unity_common.cginc" + +#include "shaders/ffx_fsr2_autogen_reactive_pass.hlsl" diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_autogen_reactive_pass.compute.meta b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_autogen_reactive_pass.compute.meta new file mode 100644 index 0000000..69662d6 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_autogen_reactive_pass.compute.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: 67ee1b32ca5e4234db9f06984c783dee +ComputeShaderImporter: + externalObjects: {} + preprocessorOverride: 0 + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_compute_luminance_pyramid_pass.compute b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_compute_luminance_pyramid_pass.compute new file mode 100644 index 0000000..7058cc5 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_compute_luminance_pyramid_pass.compute @@ -0,0 +1,42 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#pragma kernel CS + +#pragma multi_compile_local __ FFX_HALF +#pragma multi_compile_local __ FFX_FSR2_OPTION_LOW_RESOLUTION_MOTION_VECTORS +#pragma multi_compile_local __ FFX_FSR2_OPTION_JITTERED_MOTION_VECTORS +#pragma multi_compile_local __ FFX_FSR2_OPTION_INVERTED_DEPTH + +#pragma multi_compile __ UNITY_FSR_TEXTURE2D_X_ARRAY + +#include "ffx_fsr_unity_common.cginc" + +// Wave operations require shader model 6.0; this can only be enabled when using DXC on D3D12 +// These pragmas are commented out by default as Unity will sometimes ignore the #if's and try to enable these features anyway. +// Uncomment the below lines if you intend to try wave operations on DX12 with the DXC compiler. +//#if defined(UNITY_COMPILER_DXC) && defined(SHADER_API_D3D12) +//#pragma require WaveBasic // Required for WaveGetLaneIndex +//#pragma require WaveBallot // Required for WaveReadLaneAt +//#else +#define FFX_SPD_NO_WAVE_OPERATIONS +//#endif + +#include "shaders/ffx_fsr2_compute_luminance_pyramid_pass.hlsl" diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_compute_luminance_pyramid_pass.compute.meta b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_compute_luminance_pyramid_pass.compute.meta new file mode 100644 index 0000000..91fe4a2 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_compute_luminance_pyramid_pass.compute.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: 04c3480675e29a340808141e68d4cc8b +ComputeShaderImporter: + externalObjects: {} + preprocessorOverride: 0 + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_depth_clip_pass.compute b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_depth_clip_pass.compute new file mode 100644 index 0000000..582ca6b --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_depth_clip_pass.compute @@ -0,0 +1,32 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#pragma kernel CS + +#pragma multi_compile_local __ FFX_HALF +#pragma multi_compile_local __ FFX_FSR2_OPTION_LOW_RESOLUTION_MOTION_VECTORS +#pragma multi_compile_local __ FFX_FSR2_OPTION_JITTERED_MOTION_VECTORS +#pragma multi_compile_local __ FFX_FSR2_OPTION_INVERTED_DEPTH + +#pragma multi_compile __ UNITY_FSR_TEXTURE2D_X_ARRAY + +#include "ffx_fsr_unity_common.cginc" + +#include "shaders/ffx_fsr2_depth_clip_pass.hlsl" diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_depth_clip_pass.compute.meta b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_depth_clip_pass.compute.meta new file mode 100644 index 0000000..72d9877 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_depth_clip_pass.compute.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: b207de122e2c4b844b89dcd7c5c77c80 +ComputeShaderImporter: + externalObjects: {} + preprocessorOverride: 0 + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_lock_pass.compute b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_lock_pass.compute new file mode 100644 index 0000000..fb12d2c --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_lock_pass.compute @@ -0,0 +1,32 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#pragma kernel CS + +#pragma multi_compile_local __ FFX_HALF +#pragma multi_compile_local __ FFX_FSR2_OPTION_LOW_RESOLUTION_MOTION_VECTORS +#pragma multi_compile_local __ FFX_FSR2_OPTION_JITTERED_MOTION_VECTORS +#pragma multi_compile_local __ FFX_FSR2_OPTION_INVERTED_DEPTH + +#pragma multi_compile __ UNITY_FSR_TEXTURE2D_X_ARRAY + +#include "ffx_fsr_unity_common.cginc" + +#include "shaders/ffx_fsr2_lock_pass.hlsl" diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_lock_pass.compute.meta b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_lock_pass.compute.meta new file mode 100644 index 0000000..3503b36 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_lock_pass.compute.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: 20b7864a7e7258946aaf0f1996febad3 +ComputeShaderImporter: + externalObjects: {} + preprocessorOverride: 0 + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_rcas_pass.compute b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_rcas_pass.compute new file mode 100644 index 0000000..e6ac7df --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_rcas_pass.compute @@ -0,0 +1,31 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#pragma kernel CS + +#pragma multi_compile_local __ FFX_FSR2_OPTION_LOW_RESOLUTION_MOTION_VECTORS +#pragma multi_compile_local __ FFX_FSR2_OPTION_JITTERED_MOTION_VECTORS +#pragma multi_compile_local __ FFX_FSR2_OPTION_INVERTED_DEPTH + +#pragma multi_compile __ UNITY_FSR_TEXTURE2D_X_ARRAY + +#include "ffx_fsr_unity_common.cginc" + +#include "shaders/ffx_fsr2_rcas_pass.hlsl" diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_rcas_pass.compute.meta b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_rcas_pass.compute.meta new file mode 100644 index 0000000..ecbdb42 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_rcas_pass.compute.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: 40815651f0f5d994cb73da9816a7ff9b +ComputeShaderImporter: + externalObjects: {} + preprocessorOverride: 0 + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_reconstruct_previous_depth_pass.compute b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_reconstruct_previous_depth_pass.compute new file mode 100644 index 0000000..12f2ec3 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_reconstruct_previous_depth_pass.compute @@ -0,0 +1,33 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#pragma kernel CS + +#pragma multi_compile_local __ FFX_HALF +#pragma multi_compile_local __ FFX_FSR2_OPTION_HDR_COLOR_INPUT +#pragma multi_compile_local __ FFX_FSR2_OPTION_LOW_RESOLUTION_MOTION_VECTORS +#pragma multi_compile_local __ FFX_FSR2_OPTION_JITTERED_MOTION_VECTORS +#pragma multi_compile_local __ FFX_FSR2_OPTION_INVERTED_DEPTH + +#pragma multi_compile __ UNITY_FSR_TEXTURE2D_X_ARRAY + +#include "ffx_fsr_unity_common.cginc" + +#include "shaders/ffx_fsr2_reconstruct_previous_depth_pass.hlsl" diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_reconstruct_previous_depth_pass.compute.meta b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_reconstruct_previous_depth_pass.compute.meta new file mode 100644 index 0000000..92da2a6 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_reconstruct_previous_depth_pass.compute.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: 5060dfafe45aa67459629186ceb7464e +ComputeShaderImporter: + externalObjects: {} + preprocessorOverride: 0 + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_tcr_autogen_pass.compute b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_tcr_autogen_pass.compute new file mode 100644 index 0000000..5d2668a --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_tcr_autogen_pass.compute @@ -0,0 +1,32 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#pragma kernel CS + +#pragma multi_compile_local __ FFX_HALF +#pragma multi_compile_local __ FFX_FSR2_OPTION_LOW_RESOLUTION_MOTION_VECTORS +#pragma multi_compile_local __ FFX_FSR2_OPTION_JITTERED_MOTION_VECTORS +#pragma multi_compile_local __ FFX_FSR2_OPTION_INVERTED_DEPTH + +#pragma multi_compile __ UNITY_FSR_TEXTURE2D_X_ARRAY + +#include "ffx_fsr_unity_common.cginc" + +#include "shaders/ffx_fsr2_tcr_autogen_pass.hlsl" diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_tcr_autogen_pass.compute.meta b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_tcr_autogen_pass.compute.meta new file mode 100644 index 0000000..fb8a28a --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr2_tcr_autogen_pass.compute.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: f8b1c27fb6a544b43b38903592240500 +ComputeShaderImporter: + externalObjects: {} + preprocessorOverride: 0 + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_accumulate_pass.compute b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_accumulate_pass.compute new file mode 100644 index 0000000..a8d6864 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_accumulate_pass.compute @@ -0,0 +1,39 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#pragma kernel CS + +#pragma multi_compile_local __ FFX_HALF +#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE +#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_HDR_COLOR_INPUT +#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS +#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_APPLY_SHARPENING + +#pragma multi_compile __ UNITY_FSR_TEXTURE2D_X_ARRAY + +#include "ffx_fsr_unity_common.cginc" + +// Ensure the correct value is defined for this keyword, as it is used to select one of multiple sampler functions +#ifdef FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE +#undef FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE +#define FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE 1 +#endif + +#include "shaders/ffx_fsr3upscaler_accumulate_pass.hlsl" diff --git a/Assets/Shaders/FSR3/ffx_fsr3upscaler_accumulate_pass.compute.meta b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_accumulate_pass.compute.meta similarity index 100% rename from Assets/Shaders/FSR3/ffx_fsr3upscaler_accumulate_pass.compute.meta rename to Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_accumulate_pass.compute.meta diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_autogen_reactive_pass.compute b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_autogen_reactive_pass.compute new file mode 100644 index 0000000..500352a --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_autogen_reactive_pass.compute @@ -0,0 +1,29 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#pragma kernel CS + +#pragma multi_compile_local __ FFX_HALF + +#pragma multi_compile __ UNITY_FSR_TEXTURE2D_X_ARRAY + +#include "ffx_fsr_unity_common.cginc" + +#include "shaders/ffx_fsr3upscaler_autogen_reactive_pass.hlsl" diff --git a/Assets/Shaders/FSR3/ffx_fsr3upscaler_autogen_reactive_pass.compute.meta b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_autogen_reactive_pass.compute.meta similarity index 100% rename from Assets/Shaders/FSR3/ffx_fsr3upscaler_autogen_reactive_pass.compute.meta rename to Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_autogen_reactive_pass.compute.meta diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_debug_view_pass.compute b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_debug_view_pass.compute new file mode 100644 index 0000000..027e414 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_debug_view_pass.compute @@ -0,0 +1,29 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#pragma kernel CS + +#pragma multi_compile_local __ FFX_HALF + +#pragma multi_compile __ UNITY_FSR_TEXTURE2D_X_ARRAY + +#include "ffx_fsr_unity_common.cginc" + +#include "shaders/ffx_fsr3upscaler_debug_view_pass.hlsl" diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_debug_view_pass.compute.meta b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_debug_view_pass.compute.meta new file mode 100644 index 0000000..f90cac3 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_debug_view_pass.compute.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: cb24a71d54164c54eb5e86839acd48c5 +ComputeShaderImporter: + externalObjects: {} + preprocessorOverride: 0 + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_luma_instability_pass.compute b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_luma_instability_pass.compute new file mode 100644 index 0000000..5cffc66 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_luma_instability_pass.compute @@ -0,0 +1,29 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#pragma kernel CS + +#pragma multi_compile_local __ FFX_HALF + +#pragma multi_compile __ UNITY_FSR_TEXTURE2D_X_ARRAY + +#include "ffx_fsr_unity_common.cginc" + +#include "shaders/ffx_fsr3upscaler_luma_instability_pass.hlsl" diff --git a/Assets/Shaders/FSR3/ffx_fsr3upscaler_lock_pass.compute.meta b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_luma_instability_pass.compute.meta similarity index 100% rename from Assets/Shaders/FSR3/ffx_fsr3upscaler_lock_pass.compute.meta rename to Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_luma_instability_pass.compute.meta diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_luma_pyramid_pass.compute b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_luma_pyramid_pass.compute new file mode 100644 index 0000000..347b0d6 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_luma_pyramid_pass.compute @@ -0,0 +1,39 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#pragma kernel CS + +#pragma multi_compile_local __ FFX_HALF + +#pragma multi_compile __ UNITY_FSR_TEXTURE2D_X_ARRAY + +#include "ffx_fsr_unity_common.cginc" + +// Wave operations require shader model 6.0; this can only be enabled when using DXC on D3D12 +// These pragmas are commented out by default as Unity will sometimes ignore the #if's and try to enable these features anyway. +// Uncomment the below lines if you intend to try wave operations on DX12 with the DXC compiler. +//#if defined(UNITY_COMPILER_DXC) && defined(SHADER_API_D3D12) +//#pragma require WaveBasic // Required for WaveGetLaneIndex +//#pragma require WaveBallot // Required for WaveReadLaneAt +//#else +#define FFX_SPD_NO_WAVE_OPERATIONS +//#endif + +#include "shaders/ffx_fsr3upscaler_luma_pyramid_pass.hlsl" diff --git a/Assets/Shaders/FSR3/ffx_fsr3upscaler_compute_luminance_pyramid_pass.compute.meta b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_luma_pyramid_pass.compute.meta similarity index 100% rename from Assets/Shaders/FSR3/ffx_fsr3upscaler_compute_luminance_pyramid_pass.compute.meta rename to Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_luma_pyramid_pass.compute.meta diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_prepare_inputs_pass.compute b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_prepare_inputs_pass.compute new file mode 100644 index 0000000..b71ea5d --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_prepare_inputs_pass.compute @@ -0,0 +1,31 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#pragma kernel CS + +#pragma multi_compile_local __ FFX_HALF +#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS +#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH + +#pragma multi_compile __ UNITY_FSR_TEXTURE2D_X_ARRAY + +#include "ffx_fsr_unity_common.cginc" + +#include "shaders/ffx_fsr3upscaler_prepare_inputs_pass.hlsl" diff --git a/Assets/Shaders/FSR3/ffx_fsr3upscaler_reconstruct_previous_depth_pass.compute.meta b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_prepare_inputs_pass.compute.meta similarity index 100% rename from Assets/Shaders/FSR3/ffx_fsr3upscaler_reconstruct_previous_depth_pass.compute.meta rename to Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_prepare_inputs_pass.compute.meta diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_prepare_reactivity_pass.compute b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_prepare_reactivity_pass.compute new file mode 100644 index 0000000..ebf4e91 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_prepare_reactivity_pass.compute @@ -0,0 +1,29 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#pragma kernel CS + +#pragma multi_compile_local __ FFX_HALF + +#pragma multi_compile __ UNITY_FSR_TEXTURE2D_X_ARRAY + +#include "ffx_fsr_unity_common.cginc" + +#include "shaders/ffx_fsr3upscaler_prepare_reactivity_pass.hlsl" diff --git a/Assets/Shaders/FSR3/ffx_fsr3upscaler_depth_clip_pass.compute.meta b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_prepare_reactivity_pass.compute.meta similarity index 100% rename from Assets/Shaders/FSR3/ffx_fsr3upscaler_depth_clip_pass.compute.meta rename to Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_prepare_reactivity_pass.compute.meta diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_rcas_pass.compute b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_rcas_pass.compute new file mode 100644 index 0000000..282a9ae --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_rcas_pass.compute @@ -0,0 +1,27 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#pragma kernel CS + +#pragma multi_compile __ UNITY_FSR_TEXTURE2D_X_ARRAY + +#include "ffx_fsr_unity_common.cginc" + +#include "shaders/ffx_fsr3upscaler_rcas_pass.hlsl" diff --git a/Assets/Shaders/FSR3/ffx_fsr3upscaler_rcas_pass.compute.meta b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_rcas_pass.compute.meta similarity index 100% rename from Assets/Shaders/FSR3/ffx_fsr3upscaler_rcas_pass.compute.meta rename to Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_rcas_pass.compute.meta diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_shading_change_pass.compute b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_shading_change_pass.compute new file mode 100644 index 0000000..34c3ade --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_shading_change_pass.compute @@ -0,0 +1,29 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#pragma kernel CS + +#pragma multi_compile_local __ FFX_HALF + +#pragma multi_compile __ UNITY_FSR_TEXTURE2D_X_ARRAY + +#include "ffx_fsr_unity_common.cginc" + +#include "shaders/ffx_fsr3upscaler_shading_change_pass.hlsl" diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_shading_change_pass.compute.meta b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_shading_change_pass.compute.meta new file mode 100644 index 0000000..c9ae3eb --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_shading_change_pass.compute.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: 9a2bff2f97619ed4989d9b0577ba0641 +ComputeShaderImporter: + externalObjects: {} + preprocessorOverride: 0 + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_shading_change_pyramid_pass.compute b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_shading_change_pyramid_pass.compute new file mode 100644 index 0000000..ffe5bfe --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_shading_change_pyramid_pass.compute @@ -0,0 +1,32 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#pragma kernel CS + +#pragma multi_compile_local __ FFX_HALF + +#pragma multi_compile __ UNITY_FSR_TEXTURE2D_X_ARRAY + +#include "ffx_fsr_unity_common.cginc" + +// Wave operations require shader model 6.0; this can only be enabled when using DXC on D3D12 +#define FFX_SPD_NO_WAVE_OPERATIONS + +#include "shaders/ffx_fsr3upscaler_shading_change_pyramid_pass.hlsl" diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_shading_change_pyramid_pass.compute.meta b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_shading_change_pyramid_pass.compute.meta new file mode 100644 index 0000000..889be6d --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_shading_change_pyramid_pass.compute.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: 251e663738905fa4d8817001682d802f +ComputeShaderImporter: + externalObjects: {} + preprocessorOverride: 0 + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_tcr_autogen_pass.compute b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_tcr_autogen_pass.compute new file mode 100644 index 0000000..952c052 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_tcr_autogen_pass.compute @@ -0,0 +1,30 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#pragma kernel CS + +#pragma multi_compile_local __ FFX_HALF +#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS + +#pragma multi_compile __ UNITY_FSR_TEXTURE2D_X_ARRAY + +#include "ffx_fsr_unity_common.cginc" + +#include "shaders/ffx_fsr3upscaler_tcr_autogen_pass.hlsl" diff --git a/Assets/Shaders/FSR3/ffx_fsr3upscaler_tcr_autogen_pass.compute.meta b/Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_tcr_autogen_pass.compute.meta similarity index 100% rename from Assets/Shaders/FSR3/ffx_fsr3upscaler_tcr_autogen_pass.compute.meta rename to Packages/fidelityfx.fsr/Shaders/ffx_fsr3upscaler_tcr_autogen_pass.compute.meta diff --git a/Packages/fidelityfx.fsr/Shaders/ffx_fsr_unity_common.cginc b/Packages/fidelityfx.fsr/Shaders/ffx_fsr_unity_common.cginc new file mode 100644 index 0000000..b1bcb5c --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/ffx_fsr_unity_common.cginc @@ -0,0 +1,82 @@ +// Copyright (c) 2024 Nico de Poel +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// +// The above copyright notice and this permission notice shall be included in all +// copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +// Suppress a few warnings produced by FFX's HLSL code +#pragma warning(disable: 3078) // Loop control variable conflicts +#pragma warning(disable: 3203) // Signed/unsigned mismatch +#pragma warning(disable: 3556) // Integer divides might be much slower, try using uints if possible + +#define FFX_GPU // Compiling for GPU +#define FFX_HLSL // Compile for plain HLSL + +// Use the DXC shader compiler on modern graphics APIs to enable a few advanced features +// The DXC-related pragmas are disabled by default, as DXC doesn't support all platforms yet and will break on some platforms when enabled. +// Consider this to be an experimental feature. If you want to benefit from 16-bit floating point and wave operations, and don't care about supporting older graphics APIs, then it's worth a try. +//#if defined(SHADER_API_D3D12) || defined(SHADER_API_VULKAN) || defined(SHADER_API_METAL) +//#pragma use_dxc // Using DXC will currently break DX11 support since DX11 and DX12 share the same shader bytecode in Unity. +//#endif + +// Enable half precision data types on platforms that support it +//#if defined(UNITY_COMPILER_DXC) && defined(FFX_HALF) +//#pragma require Native16Bit +//#endif + +// Hack to work around the lack of texture atomics on Metal +#if defined(SHADER_API_METAL) +#define InterlockedAdd(dest, val, orig) { (orig) = (dest); (dest) += (val); } +#define InterlockedMin(dest, val) { (dest) = min((dest), (val)); } +#define InterlockedMax(dest, val) { (dest) = max((dest), (val)); } +#endif + +// Workaround for HDRP using texture arrays for its camera buffers on some platforms +// The below defines are adapted from: Packages/com.unity.render-pipelines.core/ShaderLibrary/TextureXR.hlsl +#if ((defined(SHADER_API_D3D11) || defined(SHADER_API_D3D12)) && !defined(SHADER_API_XBOXONE) && !defined(SHADER_API_GAMECORE)) || defined(SHADER_API_PSSL) || defined(SHADER_API_VULKAN) + #define UNITY_TEXTURE2D_X_ARRAY_SUPPORTED +#endif + +// Control if TEXTURE2D_X macros will expand to texture arrays +#if defined(UNITY_TEXTURE2D_X_ARRAY_SUPPORTED) && defined(UNITY_FSR_TEXTURE2D_X_ARRAY) + #define USE_TEXTURE2D_X_AS_ARRAY +#endif + +// Early defines for single-pass instancing +#if defined(STEREO_INSTANCING_ON) && defined(UNITY_TEXTURE2D_X_ARRAY_SUPPORTED) + #define UNITY_STEREO_INSTANCING_ENABLED +#endif + +// Helper macros to handle XR single-pass with Texture2DArray +#if defined(USE_TEXTURE2D_X_AS_ARRAY) + + // Only single-pass stereo instancing used array indexing + #if defined(UNITY_STEREO_INSTANCING_ENABLED) + static uint unity_StereoEyeIndex; + #define SLICE_ARRAY_INDEX unity_StereoEyeIndex + #else + #define SLICE_ARRAY_INDEX 0 + #endif + + // Declare and sample camera buffers as texture arrays + #define UNITY_FSR_TEX2D(type) Texture2DArray + #define UNITY_FSR_RWTEX2D(type) RWTexture2DArray + #define UNITY_FSR_POS(pxPos) FfxUInt32x3(pxPos, SLICE_ARRAY_INDEX) + #define UNITY_FSR_UV(uv) FfxFloat32x3(uv, SLICE_ARRAY_INDEX) + #define UNITY_FSR_GETDIMS(tex, w, h) { FfxUInt32 uElements; (tex).GetDimensions((w), (h), uElements); } + +#endif diff --git a/Assets/Shaders/FSR3/ffx_fsr3upscaler_unity_common.cginc.meta b/Packages/fidelityfx.fsr/Shaders/ffx_fsr_unity_common.cginc.meta similarity index 100% rename from Assets/Shaders/FSR3/ffx_fsr3upscaler_unity_common.cginc.meta rename to Packages/fidelityfx.fsr/Shaders/ffx_fsr_unity_common.cginc.meta diff --git a/Packages/fidelityfx.fsr/Shaders/shaders.meta b/Packages/fidelityfx.fsr/Shaders/shaders.meta new file mode 100644 index 0000000..8e1a170 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: c3f8af1cab72f0e46acba11c5820d923 +folderAsset: yes +DefaultImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_common_types.h b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_common_types.h new file mode 100644 index 0000000..09d4502 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_common_types.h @@ -0,0 +1,654 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#ifndef FFX_COMMON_TYPES_H +#define FFX_COMMON_TYPES_H + +#if defined(FFX_CPU) +#define FFX_PARAMETER_IN +#define FFX_PARAMETER_OUT +#define FFX_PARAMETER_INOUT +#define FFX_PARAMETER_UNIFORM +#elif defined(FFX_HLSL) +#define FFX_PARAMETER_IN in +#define FFX_PARAMETER_OUT out +#define FFX_PARAMETER_INOUT inout +#define FFX_PARAMETER_UNIFORM uniform +#elif defined(FFX_GLSL) +#define FFX_PARAMETER_IN in +#define FFX_PARAMETER_OUT out +#define FFX_PARAMETER_INOUT inout +#define FFX_PARAMETER_UNIFORM const //[cacao_placeholder] until a better fit is found! +#endif // #if defined(FFX_CPU) + +#if defined(FFX_CPU) +/// A typedef for a boolean value. +/// +/// @ingroup CPUTypes +typedef bool FfxBoolean; + +/// A typedef for a unsigned 8bit integer. +/// +/// @ingroup CPUTypes +typedef uint8_t FfxUInt8; + +/// A typedef for a unsigned 16bit integer. +/// +/// @ingroup CPUTypes +typedef uint16_t FfxUInt16; + +/// A typedef for a unsigned 32bit integer. +/// +/// @ingroup CPUTypes +typedef uint32_t FfxUInt32; + +/// A typedef for a unsigned 64bit integer. +/// +/// @ingroup CPUTypes +typedef uint64_t FfxUInt64; + +/// A typedef for a signed 8bit integer. +/// +/// @ingroup CPUTypes +typedef int8_t FfxInt8; + +/// A typedef for a signed 16bit integer. +/// +/// @ingroup CPUTypes +typedef int16_t FfxInt16; + +/// A typedef for a signed 32bit integer. +/// +/// @ingroup CPUTypes +typedef int32_t FfxInt32; + +/// A typedef for a signed 64bit integer. +/// +/// @ingroup CPUTypes +typedef int64_t FfxInt64; + +/// A typedef for a floating point value. +/// +/// @ingroup CPUTypes +typedef float FfxFloat32; + +/// A typedef for a 2-dimensional floating point value. +/// +/// @ingroup CPUTypes +typedef float FfxFloat32x2[2]; + +/// A typedef for a 3-dimensional floating point value. +/// +/// @ingroup CPUTypes +typedef float FfxFloat32x3[3]; + +/// A typedef for a 4-dimensional floating point value. +/// +/// @ingroup CPUTypes +typedef float FfxFloat32x4[4]; + +/// A typedef for a 2x2 floating point matrix. +/// +/// @ingroup CPUTypes +typedef float FfxFloat32x2x2[4]; + +/// A typedef for a 3x3 floating point matrix. +/// +/// @ingroup CPUTypes +typedef float FfxFloat32x3x3[9]; + +/// A typedef for a 3x4 floating point matrix. +/// +/// @ingroup CPUTypes +typedef float FfxFloat32x3x4[12]; + +/// A typedef for a 4x4 floating point matrix. +/// +/// @ingroup CPUTypes +typedef float FfxFloat32x4x4[16]; + +/// A typedef for a 2-dimensional 32bit signed integer. +/// +/// @ingroup CPUTypes +typedef int32_t FfxInt32x2[2]; + +/// A typedef for a 3-dimensional 32bit signed integer. +/// +/// @ingroup CPUTypes +typedef int32_t FfxInt32x3[3]; + +/// A typedef for a 4-dimensional 32bit signed integer. +/// +/// @ingroup CPUTypes +typedef int32_t FfxInt32x4[4]; + +/// A typedef for a 2-dimensional 32bit usigned integer. +/// +/// @ingroup CPUTypes +typedef uint32_t FfxUInt32x2[2]; + +/// A typedef for a 3-dimensional 32bit unsigned integer. +/// +/// @ingroup CPUTypes +typedef uint32_t FfxUInt32x3[3]; + +/// A typedef for a 4-dimensional 32bit unsigned integer. +/// +/// @ingroup CPUTypes +typedef uint32_t FfxUInt32x4[4]; +#endif // #if defined(FFX_CPU) + +#if defined(FFX_HLSL) + +#define FfxFloat32Mat4 matrix +#define FfxFloat32Mat3 matrix + +/// A typedef for a boolean value. +/// +/// @ingroup HLSLTypes +typedef bool FfxBoolean; + +#if FFX_HLSL_SM>=62 + +/// @defgroup HLSL62Types HLSL 6.2 And Above Types +/// HLSL 6.2 and above type defines for all commonly used variables +/// +/// @ingroup HLSLTypes + +/// A typedef for a floating point value. +/// +/// @ingroup HLSL62Types +typedef float32_t FfxFloat32; + +/// A typedef for a 2-dimensional floating point value. +/// +/// @ingroup HLSL62Types +typedef float32_t2 FfxFloat32x2; + +/// A typedef for a 3-dimensional floating point value. +/// +/// @ingroup HLSL62Types +typedef float32_t3 FfxFloat32x3; + +/// A typedef for a 4-dimensional floating point value. +/// +/// @ingroup HLSL62Types +typedef float32_t4 FfxFloat32x4; + +/// A [cacao_placeholder] typedef for matrix type until confirmed. +typedef float4x4 FfxFloat32x4x4; +typedef float3x4 FfxFloat32x3x4; +typedef float3x3 FfxFloat32x3x3; +typedef float2x2 FfxFloat32x2x2; + +/// A typedef for a unsigned 32bit integer. +/// +/// @ingroup HLSL62Types +typedef uint32_t FfxUInt32; + +/// A typedef for a 2-dimensional 32bit unsigned integer. +/// +/// @ingroup HLSL62Types +typedef uint32_t2 FfxUInt32x2; + +/// A typedef for a 3-dimensional 32bit unsigned integer. +/// +/// @ingroup HLSL62Types +typedef uint32_t3 FfxUInt32x3; + +/// A typedef for a 4-dimensional 32bit unsigned integer. +/// +/// @ingroup HLSL62Types +typedef uint32_t4 FfxUInt32x4; + +/// A typedef for a signed 32bit integer. +/// +/// @ingroup HLSL62Types +typedef int32_t FfxInt32; + +/// A typedef for a 2-dimensional signed 32bit integer. +/// +/// @ingroup HLSL62Types +typedef int32_t2 FfxInt32x2; + +/// A typedef for a 3-dimensional signed 32bit integer. +/// +/// @ingroup HLSL62Types +typedef int32_t3 FfxInt32x3; + +/// A typedef for a 4-dimensional signed 32bit integer. +/// +/// @ingroup HLSL62Types +typedef int32_t4 FfxInt32x4; + +#else // #if FFX_HLSL_SM>=62 + +/// @defgroup HLSLBaseTypes HLSL 6.1 And Below Types +/// HLSL 6.1 and below type defines for all commonly used variables +/// +/// @ingroup HLSLTypes + +#define FfxFloat32 float +#define FfxFloat32x2 float2 +#define FfxFloat32x3 float3 +#define FfxFloat32x4 float4 + +/// A [cacao_placeholder] typedef for matrix type until confirmed. +#define FfxFloat32x4x4 float4x4 +#define FfxFloat32x3x4 float3x4 +#define FfxFloat32x3x3 float3x3 +#define FfxFloat32x2x2 float2x2 + +/// A typedef for a unsigned 32bit integer. +/// +/// @ingroup GPU +typedef uint FfxUInt32; +typedef uint2 FfxUInt32x2; +typedef uint3 FfxUInt32x3; +typedef uint4 FfxUInt32x4; + +typedef int FfxInt32; +typedef int2 FfxInt32x2; +typedef int3 FfxInt32x3; +typedef int4 FfxInt32x4; + +#endif // #if FFX_HLSL_SM>=62 + +#if FFX_HALF + +#if FFX_HLSL_SM >= 62 + +typedef float16_t FfxFloat16; +typedef float16_t2 FfxFloat16x2; +typedef float16_t3 FfxFloat16x3; +typedef float16_t4 FfxFloat16x4; + +/// A typedef for an unsigned 16bit integer. +/// +/// @ingroup HLSLTypes +typedef uint16_t FfxUInt16; +typedef uint16_t2 FfxUInt16x2; +typedef uint16_t3 FfxUInt16x3; +typedef uint16_t4 FfxUInt16x4; + +/// A typedef for a signed 16bit integer. +/// +/// @ingroup HLSLTypes +typedef int16_t FfxInt16; +typedef int16_t2 FfxInt16x2; +typedef int16_t3 FfxInt16x3; +typedef int16_t4 FfxInt16x4; +#elif SHADER_API_PSSL +#pragma argument(realtypes) // Enable true 16-bit types + +typedef half FfxFloat16; +typedef half2 FfxFloat16x2; +typedef half3 FfxFloat16x3; +typedef half4 FfxFloat16x4; + +/// A typedef for an unsigned 16bit integer. +/// +/// @ingroup GPU +typedef ushort FfxUInt16; +typedef ushort2 FfxUInt16x2; +typedef ushort3 FfxUInt16x3; +typedef ushort4 FfxUInt16x4; + +/// A typedef for a signed 16bit integer. +/// +/// @ingroup GPU +typedef short FfxInt16; +typedef short2 FfxInt16x2; +typedef short3 FfxInt16x3; +typedef short4 FfxInt16x4; +#else // #if FFX_HLSL_SM>=62 +typedef min16float FfxFloat16; +typedef min16float2 FfxFloat16x2; +typedef min16float3 FfxFloat16x3; +typedef min16float4 FfxFloat16x4; + +/// A typedef for an unsigned 16bit integer. +/// +/// @ingroup HLSLTypes +typedef min16uint FfxUInt16; +typedef min16uint2 FfxUInt16x2; +typedef min16uint3 FfxUInt16x3; +typedef min16uint4 FfxUInt16x4; + +/// A typedef for a signed 16bit integer. +/// +/// @ingroup HLSLTypes +typedef min16int FfxInt16; +typedef min16int2 FfxInt16x2; +typedef min16int3 FfxInt16x3; +typedef min16int4 FfxInt16x4; +#endif // #if FFX_HLSL_SM>=62 + +#endif // FFX_HALF + +#endif // #if defined(FFX_HLSL) + +#if defined(FFX_GLSL) + +#define FfxFloat32Mat4 mat4 +#define FfxFloat32Mat3 mat3 + +/// A typedef for a boolean value. +/// +/// @ingroup GLSLTypes +#define FfxBoolean bool +#define FfxFloat32 float +#define FfxFloat32x2 vec2 +#define FfxFloat32x3 vec3 +#define FfxFloat32x4 vec4 +#define FfxUInt32 uint +#define FfxUInt32x2 uvec2 +#define FfxUInt32x3 uvec3 +#define FfxUInt32x4 uvec4 +#define FfxInt32 int +#define FfxInt32x2 ivec2 +#define FfxInt32x3 ivec3 +#define FfxInt32x4 ivec4 + +/// A [cacao_placeholder] typedef for matrix type until confirmed. +#define FfxFloat32x4x4 mat4 +#define FfxFloat32x3x4 mat4x3 +#define FfxFloat32x3x3 mat3 +#define FfxFloat32x2x2 mat2 + +#if FFX_HALF +#define FfxFloat16 float16_t +#define FfxFloat16x2 f16vec2 +#define FfxFloat16x3 f16vec3 +#define FfxFloat16x4 f16vec4 +#define FfxUInt16 uint16_t +#define FfxUInt16x2 u16vec2 +#define FfxUInt16x3 u16vec3 +#define FfxUInt16x4 u16vec4 +#define FfxInt16 int16_t +#define FfxInt16x2 i16vec2 +#define FfxInt16x3 i16vec3 +#define FfxInt16x4 i16vec4 +#endif // FFX_HALF +#endif // #if defined(FFX_GLSL) + +// Global toggles: +// #define FFX_HALF (1) +// #define FFX_HLSL_SM (62) + +#if FFX_HALF && !defined(SHADER_API_PSSL) + +#if FFX_HLSL_SM >= 62 + +#define FFX_MIN16_SCALAR( TypeName, BaseComponentType ) typedef BaseComponentType##16_t TypeName; +#define FFX_MIN16_VECTOR( TypeName, BaseComponentType, COL ) typedef vector TypeName; +#define FFX_MIN16_MATRIX( TypeName, BaseComponentType, ROW, COL ) typedef matrix TypeName; + +#define FFX_16BIT_SCALAR( TypeName, BaseComponentType ) typedef BaseComponentType##16_t TypeName; +#define FFX_16BIT_VECTOR( TypeName, BaseComponentType, COL ) typedef vector TypeName; +#define FFX_16BIT_MATRIX( TypeName, BaseComponentType, ROW, COL ) typedef matrix TypeName; + +#else //FFX_HLSL_SM>=62 + +#define FFX_MIN16_SCALAR( TypeName, BaseComponentType ) typedef min16##BaseComponentType TypeName; +#define FFX_MIN16_VECTOR( TypeName, BaseComponentType, COL ) typedef vector TypeName; +#define FFX_MIN16_MATRIX( TypeName, BaseComponentType, ROW, COL ) typedef matrix TypeName; + +#define FFX_16BIT_SCALAR( TypeName, BaseComponentType ) FFX_MIN16_SCALAR( TypeName, BaseComponentType ); +#define FFX_16BIT_VECTOR( TypeName, BaseComponentType, COL ) FFX_MIN16_VECTOR( TypeName, BaseComponentType, COL ); +#define FFX_16BIT_MATRIX( TypeName, BaseComponentType, ROW, COL ) FFX_MIN16_MATRIX( TypeName, BaseComponentType, ROW, COL ); + +#endif //FFX_HLSL_SM>=62 + +#else //FFX_HALF + +#define FFX_MIN16_SCALAR( TypeName, BaseComponentType ) typedef BaseComponentType TypeName; +#define FFX_MIN16_VECTOR( TypeName, BaseComponentType, COL ) typedef vector TypeName; +#define FFX_MIN16_MATRIX( TypeName, BaseComponentType, ROW, COL ) typedef matrix TypeName; + +#define FFX_16BIT_SCALAR( TypeName, BaseComponentType ) typedef BaseComponentType TypeName; +#define FFX_16BIT_VECTOR( TypeName, BaseComponentType, COL ) typedef vector TypeName; +#define FFX_16BIT_MATRIX( TypeName, BaseComponentType, ROW, COL ) typedef matrix TypeName; + +#endif //FFX_HALF + +#if defined(FFX_GPU) +// Common typedefs: +#if defined(FFX_HLSL) && !defined(SHADER_API_PSSL) +FFX_MIN16_SCALAR( FFX_MIN16_F , float ); +FFX_MIN16_VECTOR( FFX_MIN16_F2, float, 2 ); +FFX_MIN16_VECTOR( FFX_MIN16_F3, float, 3 ); +FFX_MIN16_VECTOR( FFX_MIN16_F4, float, 4 ); + +FFX_MIN16_SCALAR( FFX_MIN16_I, int ); +FFX_MIN16_VECTOR( FFX_MIN16_I2, int, 2 ); +FFX_MIN16_VECTOR( FFX_MIN16_I3, int, 3 ); +FFX_MIN16_VECTOR( FFX_MIN16_I4, int, 4 ); + +FFX_MIN16_SCALAR( FFX_MIN16_U, uint ); +FFX_MIN16_VECTOR( FFX_MIN16_U2, uint, 2 ); +FFX_MIN16_VECTOR( FFX_MIN16_U3, uint, 3 ); +FFX_MIN16_VECTOR( FFX_MIN16_U4, uint, 4 ); + +FFX_16BIT_SCALAR( FFX_F16_t , float ); +FFX_16BIT_VECTOR( FFX_F16_t2, float, 2 ); +FFX_16BIT_VECTOR( FFX_F16_t3, float, 3 ); +FFX_16BIT_VECTOR( FFX_F16_t4, float, 4 ); + +FFX_16BIT_SCALAR( FFX_I16_t, int ); +FFX_16BIT_VECTOR( FFX_I16_t2, int, 2 ); +FFX_16BIT_VECTOR( FFX_I16_t3, int, 3 ); +FFX_16BIT_VECTOR( FFX_I16_t4, int, 4 ); + +FFX_16BIT_SCALAR( FFX_U16_t, uint ); +FFX_16BIT_VECTOR( FFX_U16_t2, uint, 2 ); +FFX_16BIT_VECTOR( FFX_U16_t3, uint, 3 ); +FFX_16BIT_VECTOR( FFX_U16_t4, uint, 4 ); + +#define TYPEDEF_MIN16_TYPES(Prefix) \ +typedef FFX_MIN16_F Prefix##_F; \ +typedef FFX_MIN16_F2 Prefix##_F2; \ +typedef FFX_MIN16_F3 Prefix##_F3; \ +typedef FFX_MIN16_F4 Prefix##_F4; \ +typedef FFX_MIN16_I Prefix##_I; \ +typedef FFX_MIN16_I2 Prefix##_I2; \ +typedef FFX_MIN16_I3 Prefix##_I3; \ +typedef FFX_MIN16_I4 Prefix##_I4; \ +typedef FFX_MIN16_U Prefix##_U; \ +typedef FFX_MIN16_U2 Prefix##_U2; \ +typedef FFX_MIN16_U3 Prefix##_U3; \ +typedef FFX_MIN16_U4 Prefix##_U4; + +#define TYPEDEF_16BIT_TYPES(Prefix) \ +typedef FFX_16BIT_F Prefix##_F; \ +typedef FFX_16BIT_F2 Prefix##_F2; \ +typedef FFX_16BIT_F3 Prefix##_F3; \ +typedef FFX_16BIT_F4 Prefix##_F4; \ +typedef FFX_16BIT_I Prefix##_I; \ +typedef FFX_16BIT_I2 Prefix##_I2; \ +typedef FFX_16BIT_I3 Prefix##_I3; \ +typedef FFX_16BIT_I4 Prefix##_I4; \ +typedef FFX_16BIT_U Prefix##_U; \ +typedef FFX_16BIT_U2 Prefix##_U2; \ +typedef FFX_16BIT_U3 Prefix##_U3; \ +typedef FFX_16BIT_U4 Prefix##_U4; + +#define TYPEDEF_FULL_PRECISION_TYPES(Prefix) \ +typedef FfxFloat32 Prefix##_F; \ +typedef FfxFloat32x2 Prefix##_F2; \ +typedef FfxFloat32x3 Prefix##_F3; \ +typedef FfxFloat32x4 Prefix##_F4; \ +typedef FfxInt32 Prefix##_I; \ +typedef FfxInt32x2 Prefix##_I2; \ +typedef FfxInt32x3 Prefix##_I3; \ +typedef FfxInt32x4 Prefix##_I4; \ +typedef FfxUInt32 Prefix##_U; \ +typedef FfxUInt32x2 Prefix##_U2; \ +typedef FfxUInt32x3 Prefix##_U3; \ +typedef FfxUInt32x4 Prefix##_U4; +#endif // #if defined(FFX_HLSL) + +#if defined(SHADER_API_PSSL) + +#define unorm +#define globallycoherent + +#if FFX_HALF + +#define FFX_MIN16_F half +#define FFX_MIN16_F2 half2 +#define FFX_MIN16_F3 half3 +#define FFX_MIN16_F4 half4 + +#define FFX_MIN16_I short +#define FFX_MIN16_I2 short2 +#define FFX_MIN16_I3 short3 +#define FFX_MIN16_I4 short4 + +#define FFX_MIN16_U ushort +#define FFX_MIN16_U2 ushort2 +#define FFX_MIN16_U3 ushort3 +#define FFX_MIN16_U4 ushort4 + +#define FFX_16BIT_F half +#define FFX_16BIT_F2 half2 +#define FFX_16BIT_F3 half3 +#define FFX_16BIT_F4 half4 + +#define FFX_16BIT_I short +#define FFX_16BIT_I2 short2 +#define FFX_16BIT_I3 short3 +#define FFX_16BIT_I4 short4 + +#define FFX_16BIT_U ushort +#define FFX_16BIT_U2 ushort2 +#define FFX_16BIT_U3 ushort3 +#define FFX_16BIT_U4 ushort4 + +#else // FFX_HALF + +#define FFX_MIN16_F float +#define FFX_MIN16_F2 float2 +#define FFX_MIN16_F3 float3 +#define FFX_MIN16_F4 float4 + +#define FFX_MIN16_I int +#define FFX_MIN16_I2 int2 +#define FFX_MIN16_I3 int3 +#define FFX_MIN16_I4 int4 + +#define FFX_MIN16_U uint +#define FFX_MIN16_U2 uint2 +#define FFX_MIN16_U3 uint3 +#define FFX_MIN16_U4 uint4 + +#define FFX_16BIT_F float +#define FFX_16BIT_F2 float2 +#define FFX_16BIT_F3 float3 +#define FFX_16BIT_F4 float4 + +#define FFX_16BIT_I int +#define FFX_16BIT_I2 int2 +#define FFX_16BIT_I3 int3 +#define FFX_16BIT_I4 int4 + +#define FFX_16BIT_U uint +#define FFX_16BIT_U2 uint2 +#define FFX_16BIT_U3 uint3 +#define FFX_16BIT_U4 uint4 + +#endif // FFX_HALF + +#endif // #if defined(SHADER_API_PSSL) + +#if defined(FFX_GLSL) + +#if FFX_HALF + +#define FFX_MIN16_F float16_t +#define FFX_MIN16_F2 f16vec2 +#define FFX_MIN16_F3 f16vec3 +#define FFX_MIN16_F4 f16vec4 + +#define FFX_MIN16_I int16_t +#define FFX_MIN16_I2 i16vec2 +#define FFX_MIN16_I3 i16vec3 +#define FFX_MIN16_I4 i16vec4 + +#define FFX_MIN16_U uint16_t +#define FFX_MIN16_U2 u16vec2 +#define FFX_MIN16_U3 u16vec3 +#define FFX_MIN16_U4 u16vec4 + +#define FFX_16BIT_F float16_t +#define FFX_16BIT_F2 f16vec2 +#define FFX_16BIT_F3 f16vec3 +#define FFX_16BIT_F4 f16vec4 + +#define FFX_16BIT_I int16_t +#define FFX_16BIT_I2 i16vec2 +#define FFX_16BIT_I3 i16vec3 +#define FFX_16BIT_I4 i16vec4 + +#define FFX_16BIT_U uint16_t +#define FFX_16BIT_U2 u16vec2 +#define FFX_16BIT_U3 u16vec3 +#define FFX_16BIT_U4 u16vec4 + +#else // FFX_HALF + +#define FFX_MIN16_F float +#define FFX_MIN16_F2 vec2 +#define FFX_MIN16_F3 vec3 +#define FFX_MIN16_F4 vec4 + +#define FFX_MIN16_I int +#define FFX_MIN16_I2 ivec2 +#define FFX_MIN16_I3 ivec3 +#define FFX_MIN16_I4 ivec4 + +#define FFX_MIN16_U uint +#define FFX_MIN16_U2 uvec2 +#define FFX_MIN16_U3 uvec3 +#define FFX_MIN16_U4 uvec4 + +#define FFX_16BIT_F float +#define FFX_16BIT_F2 vec2 +#define FFX_16BIT_F3 vec3 +#define FFX_16BIT_F4 vec4 + +#define FFX_16BIT_I int +#define FFX_16BIT_I2 ivec2 +#define FFX_16BIT_I3 ivec3 +#define FFX_16BIT_I4 ivec4 + +#define FFX_16BIT_U uint +#define FFX_16BIT_U2 uvec2 +#define FFX_16BIT_U3 uvec3 +#define FFX_16BIT_U4 uvec4 + +#endif // FFX_HALF + +#endif // #if defined(FFX_GLSL) + +#endif // #if defined(FFX_GPU) +#endif // #ifndef FFX_COMMON_TYPES_H diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_common_types.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_common_types.h.meta new file mode 100644 index 0000000..d0f05eb --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_common_types.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: ec0f8c94ee9930b438b99b82735d181b +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core.h b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core.h new file mode 100644 index 0000000..d1ed144 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core.h @@ -0,0 +1,80 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +/// @defgroup FfxGPU GPU +/// The FidelityFX SDK GPU References +/// +/// @ingroup ffxSDK + +/// @defgroup FfxHLSL HLSL References +/// FidelityFX SDK HLSL GPU References +/// +/// @ingroup FfxGPU + +/// @defgroup FfxGLSL GLSL References +/// FidelityFX SDK GLSL GPU References +/// +/// @ingroup FfxGPU + +/// @defgroup FfxGPUEffects FidelityFX GPU References +/// FidelityFX Effect GPU Reference Documentation +/// +/// @ingroup FfxGPU + +/// @defgroup GPUCore GPU Core +/// GPU defines and functions +/// +/// @ingroup FfxGPU + +#if !defined(FFX_CORE_H) +#define FFX_CORE_H + +#ifdef __hlsl_dx_compiler +#pragma dxc diagnostic push +#pragma dxc diagnostic ignored "-Wambig-lit-shift" +#endif //__hlsl_dx_compiler + +#include "ffx_common_types.h" + +#if defined(FFX_CPU) + #include "ffx_core_cpu.h" +#endif // #if defined(FFX_CPU) + +#if defined(FFX_GLSL) && defined(FFX_GPU) + #include "ffx_core_glsl.h" +#endif // #if defined(FFX_GLSL) && defined(FFX_GPU) + +#if defined(FFX_HLSL) && defined(FFX_GPU) + #include "ffx_core_hlsl.h" +#endif // #if defined(FFX_HLSL) && defined(FFX_GPU) + +#if defined(FFX_GPU) + #include "ffx_core_gpu_common.h" + #include "ffx_core_gpu_common_half.h" + #include "ffx_core_portability.h" +#endif // #if defined(FFX_GPU) + +#ifdef __hlsl_dx_compiler +#pragma dxc diagnostic pop +#endif //__hlsl_dx_compiler + +#endif // #if !defined(FFX_CORE_H) diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core.h.meta new file mode 100644 index 0000000..18282d1 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 223ad96bb47790e4d8658dd82ba950f3 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_gpu_common.h b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_gpu_common.h new file mode 100644 index 0000000..9f88c94 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_gpu_common.h @@ -0,0 +1,2736 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +/// A define for a true value in a boolean expression. +/// +/// @ingroup GPUCore +#define FFX_TRUE (true) + +/// A define for a false value in a boolean expression. +/// +/// @ingroup GPUCore +#define FFX_FALSE (false) + +/// A define value for positive infinity. +/// +/// @ingroup GPUCore +#define FFX_POSITIVE_INFINITY_FLOAT ffxAsFloat(0x7f800000u) + +/// A define value for negative infinity. +/// +/// @ingroup GPUCore +#define FFX_NEGATIVE_INFINITY_FLOAT ffxAsFloat(0xff800000u) + +/// A define value for PI. +/// +/// @ingroup GPUCore +#define FFX_PI (3.14159) + +FFX_STATIC const FfxFloat32 FFX_FP16_MIN = 6.10e-05f; +FFX_STATIC const FfxFloat32 FFX_FP16_MAX = 65504.0f; +FFX_STATIC const FfxFloat32 FFX_TONEMAP_EPSILON = 1.0f / FFX_FP16_MAX; + +#define FFX_HAS_FLAG(v, f) ((v & f) == f) + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxFloat32 ffxMin(FfxFloat32 x, FfxFloat32 y) +{ + return min(x, y); +} + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxMin(FfxFloat32x2 x, FfxFloat32x2 y) +{ + return min(x, y); +} + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxMin(FfxFloat32x3 x, FfxFloat32x3 y) +{ + return min(x, y); +} + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxMin(FfxFloat32x4 x, FfxFloat32x4 y) +{ + return min(x, y); +} + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxInt32 ffxMin(FfxInt32 x, FfxInt32 y) +{ + return min(x, y); +} + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxInt32x2 ffxMin(FfxInt32x2 x, FfxInt32x2 y) +{ + return min(x, y); +} + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxInt32x3 ffxMin(FfxInt32x3 x, FfxInt32x3 y) +{ + return min(x, y); +} + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxInt32x4 ffxMin(FfxInt32x4 x, FfxInt32x4 y) +{ + return min(x, y); +} + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxUInt32 ffxMin(FfxUInt32 x, FfxUInt32 y) +{ + return min(x, y); +} + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxUInt32x2 ffxMin(FfxUInt32x2 x, FfxUInt32x2 y) +{ + return min(x, y); +} + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxUInt32x3 ffxMin(FfxUInt32x3 x, FfxUInt32x3 y) +{ + return min(x, y); +} + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxUInt32x4 ffxMin(FfxUInt32x4 x, FfxUInt32x4 y) +{ + return min(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxFloat32 ffxMax(FfxFloat32 x, FfxFloat32 y) +{ + return max(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxMax(FfxFloat32x2 x, FfxFloat32x2 y) +{ + return max(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxMax(FfxFloat32x3 x, FfxFloat32x3 y) +{ + return max(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxMax(FfxFloat32x4 x, FfxFloat32x4 y) +{ + return max(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxInt32 ffxMax(FfxInt32 x, FfxInt32 y) +{ + return max(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxInt32x2 ffxMax(FfxInt32x2 x, FfxInt32x2 y) +{ + return max(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxInt32x3 ffxMax(FfxInt32x3 x, FfxInt32x3 y) +{ + return max(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxInt32x4 ffxMax(FfxInt32x4 x, FfxInt32x4 y) +{ + return max(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxUInt32 ffxMax(FfxUInt32 x, FfxUInt32 y) +{ + return max(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxUInt32x2 ffxMax(FfxUInt32x2 x, FfxUInt32x2 y) +{ + return max(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxUInt32x3 ffxMax(FfxUInt32x3 x, FfxUInt32x3 y) +{ + return max(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxUInt32x4 ffxMax(FfxUInt32x4 x, FfxUInt32x4 y) +{ + return max(x, y); +} + +/// Compute the value of the first parameter raised to the power of the second. +/// +/// @param [in] x The value to raise to the power y. +/// @param [in] y The power to which to raise x. +/// +/// @returns +/// The value of the first parameter raised to the power of the second. +/// +/// @ingroup GPUCore +FfxFloat32 ffxPow(FfxFloat32 x, FfxFloat32 y) +{ + return pow(x, y); +} + +/// Compute the value of the first parameter raised to the power of the second. +/// +/// @param [in] x The value to raise to the power y. +/// @param [in] y The power to which to raise x. +/// +/// @returns +/// The value of the first parameter raised to the power of the second. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxPow(FfxFloat32x2 x, FfxFloat32x2 y) +{ + return pow(x, y); +} + +/// Compute the value of the first parameter raised to the power of the second. +/// +/// @param [in] x The value to raise to the power y. +/// @param [in] y The power to which to raise x. +/// +/// @returns +/// The value of the first parameter raised to the power of the second. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxPow(FfxFloat32x3 x, FfxFloat32x3 y) +{ + return pow(x, y); +} + +/// Compute the value of the first parameter raised to the power of the second. +/// +/// @param [in] x The value to raise to the power y. +/// @param [in] y The power to which to raise x. +/// +/// @returns +/// The value of the first parameter raised to the power of the second. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxPow(FfxFloat32x4 x, FfxFloat32x4 y) +{ + return pow(x, y); +} + +/// Compute the square root of a value. +/// +/// @param [in] x The first value to compute the min of. +/// +/// @returns +/// The the square root of x. +/// +/// @ingroup GPUCore +FfxFloat32 ffxSqrt(FfxFloat32 x) +{ + return sqrt(x); +} + +/// Compute the square root of a value. +/// +/// @param [in] x The first value to compute the min of. +/// +/// @returns +/// The the square root of x. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxSqrt(FfxFloat32x2 x) +{ + return sqrt(x); +} + +/// Compute the square root of a value. +/// +/// @param [in] x The first value to compute the min of. +/// +/// @returns +/// The the square root of x. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxSqrt(FfxFloat32x3 x) +{ + return sqrt(x); +} + +/// Compute the square root of a value. +/// +/// @param [in] x The first value to compute the min of. +/// +/// @returns +/// The the square root of x. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxSqrt(FfxFloat32x4 x) +{ + return sqrt(x); +} + +/// Copy the sign bit from 's' to positive 'd'. +/// +/// @param [in] d The value to copy the sign bit into. +/// @param [in] s The value to copy the sign bit from. +/// +/// @returns +/// The value of d with the sign bit from s. +/// +/// @ingroup GPUCore +FfxFloat32 ffxCopySignBit(FfxFloat32 d, FfxFloat32 s) +{ + return ffxAsFloat(ffxAsUInt32(d) | (ffxAsUInt32(s) & FfxUInt32(0x80000000u))); +} + +/// Copy the sign bit from 's' to positive 'd'. +/// +/// @param [in] d The value to copy the sign bit into. +/// @param [in] s The value to copy the sign bit from. +/// +/// @returns +/// The value of d with the sign bit from s. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxCopySignBit(FfxFloat32x2 d, FfxFloat32x2 s) +{ + return ffxAsFloat(ffxAsUInt32(d) | (ffxAsUInt32(s) & ffxBroadcast2(0x80000000u))); +} + +/// Copy the sign bit from 's' to positive 'd'. +/// +/// @param [in] d The value to copy the sign bit into. +/// @param [in] s The value to copy the sign bit from. +/// +/// @returns +/// The value of d with the sign bit from s. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxCopySignBit(FfxFloat32x3 d, FfxFloat32x3 s) +{ + return ffxAsFloat(ffxAsUInt32(d) | (ffxAsUInt32(s) & ffxBroadcast3(0x80000000u))); +} + +/// Copy the sign bit from 's' to positive 'd'. +/// +/// @param [in] d The value to copy the sign bit into. +/// @param [in] s The value to copy the sign bit from. +/// +/// @returns +/// The value of d with the sign bit from s. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxCopySignBit(FfxFloat32x4 d, FfxFloat32x4 s) +{ + return ffxAsFloat(ffxAsUInt32(d) | (ffxAsUInt32(s) & ffxBroadcast4(0x80000000u))); +} + +/// A single operation to return the following: +/// m = NaN := 0 +/// m >= 0 := 0 +/// m < 0 := 1 +/// +/// Uses the following useful floating point logic, +/// saturate(+a*(-INF)==-INF) := 0 +/// saturate( 0*(-INF)== NaN) := 0 +/// saturate(-a*(-INF)==+INF) := 1 +/// +/// This function is useful when creating masks for branch-free logic. +/// +/// @param [in] m The value to test against 0. +/// +/// @returns +/// 1.0 when the value is negative, or 0.0 when the value is 0 or position. +/// +/// @ingroup GPUCore +FfxFloat32 ffxIsSigned(FfxFloat32 m) +{ + return ffxSaturate(m * FfxFloat32(FFX_NEGATIVE_INFINITY_FLOAT)); +} + +/// A single operation to return the following: +/// m = NaN := 0 +/// m >= 0 := 0 +/// m < 0 := 1 +/// +/// Uses the following useful floating point logic, +/// saturate(+a*(-INF)==-INF) := 0 +/// saturate( 0*(-INF)== NaN) := 0 +/// saturate(-a*(-INF)==+INF) := 1 +/// +/// This function is useful when creating masks for branch-free logic. +/// +/// @param [in] m The value to test against 0. +/// +/// @returns +/// 1.0 when the value is negative, or 0.0 when the value is 0 or position. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxIsSigned(FfxFloat32x2 m) +{ + return ffxSaturate(m * ffxBroadcast2(FFX_NEGATIVE_INFINITY_FLOAT)); +} + +/// A single operation to return the following: +/// m = NaN := 0 +/// m >= 0 := 0 +/// m < 0 := 1 +/// +/// Uses the following useful floating point logic, +/// saturate(+a*(-INF)==-INF) := 0 +/// saturate( 0*(-INF)== NaN) := 0 +/// saturate(-a*(-INF)==+INF) := 1 +/// +/// This function is useful when creating masks for branch-free logic. +/// +/// @param [in] m The value to test against 0. +/// +/// @returns +/// 1.0 when the value is negative, or 0.0 when the value is 0 or position. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxIsSigned(FfxFloat32x3 m) +{ + return ffxSaturate(m * ffxBroadcast3(FFX_NEGATIVE_INFINITY_FLOAT)); +} + +/// A single operation to return the following: +/// m = NaN := 0 +/// m >= 0 := 0 +/// m < 0 := 1 +/// +/// Uses the following useful floating point logic, +/// saturate(+a*(-INF)==-INF) := 0 +/// saturate( 0*(-INF)== NaN) := 0 +/// saturate(-a*(-INF)==+INF) := 1 +/// +/// This function is useful when creating masks for branch-free logic. +/// +/// @param [in] m The value to test against for have the sign set. +/// +/// @returns +/// 1.0 when the value is negative, or 0.0 when the value is 0 or positive. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxIsSigned(FfxFloat32x4 m) +{ + return ffxSaturate(m * ffxBroadcast4(FFX_NEGATIVE_INFINITY_FLOAT)); +} + +/// A single operation to return the following: +/// m = NaN := 1 +/// m > 0 := 0 +/// m <= 0 := 1 +/// +/// This function is useful when creating masks for branch-free logic. +/// +/// @param [in] m The value to test against zero. +/// +/// @returns +/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. +/// +/// @ingroup GPUCore +FfxFloat32 ffxIsGreaterThanZero(FfxFloat32 m) +{ + return ffxSaturate(m * FfxFloat32(FFX_POSITIVE_INFINITY_FLOAT)); +} + +/// A single operation to return the following: +/// m = NaN := 1 +/// m > 0 := 0 +/// m <= 0 := 1 +/// +/// This function is useful when creating masks for branch-free logic. +/// +/// @param [in] m The value to test against zero. +/// +/// @returns +/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxIsGreaterThanZero(FfxFloat32x2 m) +{ + return ffxSaturate(m * ffxBroadcast2(FFX_POSITIVE_INFINITY_FLOAT)); +} + +/// A single operation to return the following: +/// m = NaN := 1 +/// m > 0 := 0 +/// m <= 0 := 1 +/// +/// This function is useful when creating masks for branch-free logic. +/// +/// @param [in] m The value to test against zero. +/// +/// @returns +/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxIsGreaterThanZero(FfxFloat32x3 m) +{ + return ffxSaturate(m * ffxBroadcast3(FFX_POSITIVE_INFINITY_FLOAT)); +} + +/// A single operation to return the following: +/// m = NaN := 1 +/// m > 0 := 0 +/// m <= 0 := 1 +/// +/// This function is useful when creating masks for branch-free logic. +/// +/// @param [in] m The value to test against zero. +/// +/// @returns +/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxIsGreaterThanZero(FfxFloat32x4 m) +{ + return ffxSaturate(m * ffxBroadcast4(FFX_POSITIVE_INFINITY_FLOAT)); +} + +/// Convert a 32bit floating point value to sortable integer. +/// +/// - If sign bit=0, flip the sign bit (positives). +/// - If sign bit=1, flip all bits (negatives). +/// +/// The function has the side effects that: +/// - Larger integers are more positive values. +/// - Float zero is mapped to center of integers (so clear to integer zero is a nice default for atomic max usage). +/// +/// @param [in] value The floating point value to make sortable. +/// +/// @returns +/// The sortable integer value. +/// +/// @ingroup GPUCore +FfxUInt32 ffxFloatToSortableInteger(FfxUInt32 value) +{ + return value ^ ((ffxAShrSU1(value, FfxUInt32(31))) | FfxUInt32(0x80000000)); +} + +/// Convert a sortable integer to a 32bit floating point value. +/// +/// The function has the side effects that: +/// - If sign bit=1, flip the sign bit (positives). +/// - If sign bit=0, flip all bits (negatives). +/// +/// @param [in] value The floating point value to make sortable. +/// +/// @returns +/// The sortable integer value. +/// +/// @ingroup GPUCore +FfxUInt32 ffxSortableIntegerToFloat(FfxUInt32 value) +{ + return value ^ ((~ffxAShrSU1(value, FfxUInt32(31))) | FfxUInt32(0x80000000)); +} + +/// Calculate a low-quality approximation for the square root of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] value The value to calculate an approximate to the square root for. +/// +/// @returns +/// An approximation of the square root, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat32 ffxApproximateSqrt(FfxFloat32 value) +{ + return ffxAsFloat((ffxAsUInt32(value) >> FfxUInt32(1)) + FfxUInt32(0x1fbc4639)); +} + +/// Calculate a low-quality approximation for the reciprocal of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] value The value to calculate an approximate to the reciprocal for. +/// +/// @returns +/// An approximation of the reciprocal, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat32 ffxApproximateReciprocal(FfxFloat32 value) +{ + return ffxAsFloat(FfxUInt32(0x7ef07ebb) - ffxAsUInt32(value)); +} + +/// Calculate a medium-quality approximation for the reciprocal of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] value The value to calculate an approximate to the reciprocal for. +/// +/// @returns +/// An approximation of the reciprocal, estimated to medium quality. +/// +/// @ingroup GPUCore +FfxFloat32 ffxApproximateReciprocalMedium(FfxFloat32 value) +{ + FfxFloat32 b = ffxAsFloat(FfxUInt32(0x7ef19fff) - ffxAsUInt32(value)); + return b * (-b * value + FfxFloat32(2.0)); +} + +/// Calculate a low-quality approximation for the reciprocal of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] value The value to calculate an approximate to the reciprocal square root for. +/// +/// @returns +/// An approximation of the reciprocal square root, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat32 ffxApproximateReciprocalSquareRoot(FfxFloat32 value) +{ + return ffxAsFloat(FfxUInt32(0x5f347d74) - (ffxAsUInt32(value) >> FfxUInt32(1))); +} + +/// Calculate a low-quality approximation for the square root of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] value The value to calculate an approximate to the square root for. +/// +/// @returns +/// An approximation of the square root, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxApproximateSqrt(FfxFloat32x2 value) +{ + return ffxAsFloat((ffxAsUInt32(value) >> ffxBroadcast2(1u)) + ffxBroadcast2(0x1fbc4639u)); +} + +/// Calculate a low-quality approximation for the reciprocal of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] value The value to calculate an approximate to the reciprocal for. +/// +/// @returns +/// An approximation of the reciprocal, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxApproximateReciprocal(FfxFloat32x2 value) +{ + return ffxAsFloat(ffxBroadcast2(0x7ef07ebbu) - ffxAsUInt32(value)); +} + +/// Calculate a medium-quality approximation for the reciprocal of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] value The value to calculate an approximate to the reciprocal for. +/// +/// @returns +/// An approximation of the reciprocal, estimated to medium quality. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxApproximateReciprocalMedium(FfxFloat32x2 value) +{ + FfxFloat32x2 b = ffxAsFloat(ffxBroadcast2(0x7ef19fffu) - ffxAsUInt32(value)); + return b * (-b * value + ffxBroadcast2(2.0f)); +} + +/// Calculate a low-quality approximation for the square root of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] value The value to calculate an approximate to the square root for. +/// +/// @returns +/// An approximation of the square root, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxApproximateReciprocalSquareRoot(FfxFloat32x2 value) +{ + return ffxAsFloat(ffxBroadcast2(0x5f347d74u) - (ffxAsUInt32(value) >> ffxBroadcast2(1u))); +} + +/// Calculate a low-quality approximation for the square root of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] value The value to calculate an approximate to the square root for. +/// +/// @returns +/// An approximation of the square root, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxApproximateSqrt(FfxFloat32x3 value) +{ + return ffxAsFloat((ffxAsUInt32(value) >> ffxBroadcast3(1u)) + ffxBroadcast3(0x1fbc4639u)); +} + +/// Calculate a low-quality approximation for the reciprocal of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] value The value to calculate an approximate to the reciprocal for. +/// +/// @returns +/// An approximation of the reciprocal, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxApproximateReciprocal(FfxFloat32x3 value) +{ + return ffxAsFloat(ffxBroadcast3(0x7ef07ebbu) - ffxAsUInt32(value)); +} + +/// Calculate a medium-quality approximation for the reciprocal of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] value The value to calculate an approximate to the reciprocal for. +/// +/// @returns +/// An approximation of the reciprocal, estimated to medium quality. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxApproximateReciprocalMedium(FfxFloat32x3 value) +{ + FfxFloat32x3 b = ffxAsFloat(ffxBroadcast3(0x7ef19fffu) - ffxAsUInt32(value)); + return b * (-b * value + ffxBroadcast3(2.0f)); +} + +/// Calculate a low-quality approximation for the square root of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] value The value to calculate an approximate to the square root for. +/// +/// @returns +/// An approximation of the square root, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxApproximateReciprocalSquareRoot(FfxFloat32x3 value) +{ + return ffxAsFloat(ffxBroadcast3(0x5f347d74u) - (ffxAsUInt32(value) >> ffxBroadcast3(1u))); +} + +/// Calculate a low-quality approximation for the square root of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] value The value to calculate an approximate to the square root for. +/// +/// @returns +/// An approximation of the square root, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxApproximateSqrt(FfxFloat32x4 value) +{ + return ffxAsFloat((ffxAsUInt32(value) >> ffxBroadcast4(1u)) + ffxBroadcast4(0x1fbc4639u)); +} + +/// Calculate a low-quality approximation for the reciprocal of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] value The value to calculate an approximate to the reciprocal for. +/// +/// @returns +/// An approximation of the reciprocal, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxApproximateReciprocal(FfxFloat32x4 value) +{ + return ffxAsFloat(ffxBroadcast4(0x7ef07ebbu) - ffxAsUInt32(value)); +} + +/// Calculate a medium-quality approximation for the reciprocal of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] value The value to calculate an approximate to the reciprocal for. +/// +/// @returns +/// An approximation of the reciprocal, estimated to medium quality. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxApproximateReciprocalMedium(FfxFloat32x4 value) +{ + FfxFloat32x4 b = ffxAsFloat(ffxBroadcast4(0x7ef19fffu) - ffxAsUInt32(value)); + return b * (-b * value + ffxBroadcast4(2.0f)); +} + +/// Calculate a low-quality approximation for the square root of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] value The value to calculate an approximate to the square root for. +/// +/// @returns +/// An approximation of the square root, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxApproximateReciprocalSquareRoot(FfxFloat32x4 value) +{ + return ffxAsFloat(ffxBroadcast4(0x5f347d74u) - (ffxAsUInt32(value) >> ffxBroadcast4(1u))); +} + +/// Calculate dot product of 'a' and 'b'. +/// +/// @param [in] a First vector input. +/// @param [in] b Second vector input. +/// +/// @returns +/// The value of a dot b. +/// +/// @ingroup GPUCore +FfxFloat32 ffxDot2(FfxFloat32x2 a, FfxFloat32x2 b) +{ + return dot(a, b); +} + +/// Calculate dot product of 'a' and 'b'. +/// +/// @param [in] a First vector input. +/// @param [in] b Second vector input. +/// +/// @returns +/// The value of a dot b. +/// +/// @ingroup GPUCore +FfxFloat32 ffxDot3(FfxFloat32x3 a, FfxFloat32x3 b) +{ + return dot(a, b); +} + +/// Calculate dot product of 'a' and 'b'. +/// +/// @param [in] a First vector input. +/// @param [in] b Second vector input. +/// +/// @returns +/// The value of a dot b. +/// +/// @ingroup GPUCore +FfxFloat32 ffxDot4(FfxFloat32x4 a, FfxFloat32x4 b) +{ + return dot(a, b); +} + + +/// Compute an approximate conversion from PQ to Gamma2 space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between PQ and Gamma2. +/// +/// @returns +/// The value a converted into Gamma2. +/// +/// @ingroup GPUCore +FfxFloat32 ffxApproximatePQToGamma2Medium(FfxFloat32 a) +{ + return a * a * a * a; +} + +/// Compute an approximate conversion from PQ to linear space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between PQ and linear. +/// +/// @returns +/// The value a converted into linear. +/// +/// @ingroup GPUCore +FfxFloat32 ffxApproximatePQToLinear(FfxFloat32 a) +{ + return a * a * a * a * a * a * a * a; +} + +/// Compute an approximate conversion from gamma2 to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between gamma2 and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32 ffxApproximateGamma2ToPQ(FfxFloat32 a) +{ + return ffxAsFloat((ffxAsUInt32(a) >> FfxUInt32(2)) + FfxUInt32(0x2F9A4E46)); +} + +/// Compute a more accurate approximate conversion from gamma2 to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between gamma2 and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32 ffxApproximateGamma2ToPQMedium(FfxFloat32 a) +{ + FfxFloat32 b = ffxAsFloat((ffxAsUInt32(a) >> FfxUInt32(2)) + FfxUInt32(0x2F9A4E46)); + FfxFloat32 b4 = b * b * b * b; + return b - b * (b4 - a) / (FfxFloat32(4.0) * b4); +} + +/// Compute a high accuracy approximate conversion from gamma2 to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between gamma2 and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32 ffxApproximateGamma2ToPQHigh(FfxFloat32 a) +{ + return ffxSqrt(ffxSqrt(a)); +} + +/// Compute an approximate conversion from linear to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between linear and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32 ffxApproximateLinearToPQ(FfxFloat32 a) +{ + return ffxAsFloat((ffxAsUInt32(a) >> FfxUInt32(3)) + FfxUInt32(0x378D8723)); +} + +/// Compute a more accurate approximate conversion from linear to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between linear and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32 ffxApproximateLinearToPQMedium(FfxFloat32 a) +{ + FfxFloat32 b = ffxAsFloat((ffxAsUInt32(a) >> FfxUInt32(3)) + FfxUInt32(0x378D8723)); + FfxFloat32 b8 = b * b * b * b * b * b * b * b; + return b - b * (b8 - a) / (FfxFloat32(8.0) * b8); +} + +/// Compute a very accurate approximate conversion from linear to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between linear and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32 ffxApproximateLinearToPQHigh(FfxFloat32 a) +{ + return ffxSqrt(ffxSqrt(ffxSqrt(a))); +} + +/// Compute an approximate conversion from PQ to Gamma2 space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between PQ and Gamma2. +/// +/// @returns +/// The value a converted into Gamma2. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxApproximatePQToGamma2Medium(FfxFloat32x2 a) +{ + return a * a * a * a; +} + +/// Compute an approximate conversion from PQ to linear space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between PQ and linear. +/// +/// @returns +/// The value a converted into linear. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxApproximatePQToLinear(FfxFloat32x2 a) +{ + return a * a * a * a * a * a * a * a; +} + +/// Compute an approximate conversion from gamma2 to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between gamma2 and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxApproximateGamma2ToPQ(FfxFloat32x2 a) +{ + return ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast2(2u)) + ffxBroadcast2(0x2F9A4E46u)); +} + +/// Compute a more accurate approximate conversion from gamma2 to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between gamma2 and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxApproximateGamma2ToPQMedium(FfxFloat32x2 a) +{ + FfxFloat32x2 b = ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast2(2u)) + ffxBroadcast2(0x2F9A4E46u)); + FfxFloat32x2 b4 = b * b * b * b; + return b - b * (b4 - a) / (FfxFloat32(4.0) * b4); +} + +/// Compute a high accuracy approximate conversion from gamma2 to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between gamma2 and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxApproximateGamma2ToPQHigh(FfxFloat32x2 a) +{ + return ffxSqrt(ffxSqrt(a)); +} + +/// Compute an approximate conversion from linear to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between linear and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxApproximateLinearToPQ(FfxFloat32x2 a) +{ + return ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast2(3u)) + ffxBroadcast2(0x378D8723u)); +} + +/// Compute a more accurate approximate conversion from linear to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between linear and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxApproximateLinearToPQMedium(FfxFloat32x2 a) +{ + FfxFloat32x2 b = ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast2(3u)) + ffxBroadcast2(0x378D8723u)); + FfxFloat32x2 b8 = b * b * b * b * b * b * b * b; + return b - b * (b8 - a) / (FfxFloat32(8.0) * b8); +} + +/// Compute a very accurate approximate conversion from linear to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between linear and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxApproximateLinearToPQHigh(FfxFloat32x2 a) +{ + return ffxSqrt(ffxSqrt(ffxSqrt(a))); +} + +/// Compute an approximate conversion from PQ to Gamma2 space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between PQ and Gamma2. +/// +/// @returns +/// The value a converted into Gamma2. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxApproximatePQToGamma2Medium(FfxFloat32x3 a) +{ + return a * a * a * a; +} + +/// Compute an approximate conversion from PQ to linear space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between PQ and linear. +/// +/// @returns +/// The value a converted into linear. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxApproximatePQToLinear(FfxFloat32x3 a) +{ + return a * a * a * a * a * a * a * a; +} + +/// Compute an approximate conversion from gamma2 to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between gamma2 and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxApproximateGamma2ToPQ(FfxFloat32x3 a) +{ + return ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast3(2u)) + ffxBroadcast3(0x2F9A4E46u)); +} + +/// Compute a more accurate approximate conversion from gamma2 to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between gamma2 and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxApproximateGamma2ToPQMedium(FfxFloat32x3 a) +{ + FfxFloat32x3 b = ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast3(2u)) + ffxBroadcast3(0x2F9A4E46u)); + FfxFloat32x3 b4 = b * b * b * b; + return b - b * (b4 - a) / (FfxFloat32(4.0) * b4); +} + +/// Compute a high accuracy approximate conversion from gamma2 to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between gamma2 and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxApproximateGamma2ToPQHigh(FfxFloat32x3 a) +{ + return ffxSqrt(ffxSqrt(a)); +} + +/// Compute an approximate conversion from linear to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between linear and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxApproximateLinearToPQ(FfxFloat32x3 a) +{ + return ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast3(3u)) + ffxBroadcast3(0x378D8723u)); +} + +/// Compute a more accurate approximate conversion from linear to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between linear and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxApproximateLinearToPQMedium(FfxFloat32x3 a) +{ + FfxFloat32x3 b = ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast3(3u)) + ffxBroadcast3(0x378D8723u)); + FfxFloat32x3 b8 = b * b * b * b * b * b * b * b; + return b - b * (b8 - a) / (FfxFloat32(8.0) * b8); +} + +/// Compute a very accurate approximate conversion from linear to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between linear and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxApproximateLinearToPQHigh(FfxFloat32x3 a) +{ + return ffxSqrt(ffxSqrt(ffxSqrt(a))); +} + +/// Compute an approximate conversion from PQ to Gamma2 space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between PQ and Gamma2. +/// +/// @returns +/// The value a converted into Gamma2. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxApproximatePQToGamma2Medium(FfxFloat32x4 a) +{ + return a * a * a * a; +} + +/// Compute an approximate conversion from PQ to linear space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between PQ and linear. +/// +/// @returns +/// The value a converted into linear. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxApproximatePQToLinear(FfxFloat32x4 a) +{ + return a * a * a * a * a * a * a * a; +} + +/// Compute an approximate conversion from gamma2 to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between gamma2 and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxApproximateGamma2ToPQ(FfxFloat32x4 a) +{ + return ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast4(2u)) + ffxBroadcast4(0x2F9A4E46u)); +} + +/// Compute a more accurate approximate conversion from gamma2 to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between gamma2 and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxApproximateGamma2ToPQMedium(FfxFloat32x4 a) +{ + FfxFloat32x4 b = ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast4(2u)) + ffxBroadcast4(0x2F9A4E46u)); + FfxFloat32x4 b4 = b * b * b * b * b * b * b * b; + return b - b * (b4 - a) / (FfxFloat32(4.0) * b4); +} + +/// Compute a high accuracy approximate conversion from gamma2 to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between gamma2 and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxApproximateGamma2ToPQHigh(FfxFloat32x4 a) +{ + return ffxSqrt(ffxSqrt(a)); +} + +/// Compute an approximate conversion from linear to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between linear and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxApproximateLinearToPQ(FfxFloat32x4 a) +{ + return ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast4(3u)) + ffxBroadcast4(0x378D8723u)); +} + +/// Compute a more accurate approximate conversion from linear to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between linear and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxApproximateLinearToPQMedium(FfxFloat32x4 a) +{ + FfxFloat32x4 b = ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast4(3u)) + ffxBroadcast4(0x378D8723u)); + FfxFloat32x4 b8 = b * b * b * b * b * b * b * b; + return b - b * (b8 - a) / (FfxFloat32(8.0) * b8); +} + +/// Compute a very accurate approximate conversion from linear to PQ space. +/// +/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do +/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear +/// (8th power and fast 8th root). The maximum error is approximately 0.2%. +/// +/// @param a The value to convert between linear and PQ. +/// +/// @returns +/// The value a converted into PQ. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxApproximateLinearToPQHigh(FfxFloat32x4 a) +{ + return ffxSqrt(ffxSqrt(ffxSqrt(a))); +} + +// An approximation of sine. +// +// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range +// is {-1/4 to 1/4} representing {-1 to 1}. +// +// @param [in] value The value to calculate approximate sine for. +// +// @returns +// The approximate sine of value. +FfxFloat32 ffxParabolicSin(FfxFloat32 value) +{ + return value * abs(value) - value; +} + +// An approximation of sine. +// +// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range +// is {-1/4 to 1/4} representing {-1 to 1}. +// +// @param [in] value The value to calculate approximate sine for. +// +// @returns +// The approximate sine of value. +FfxFloat32x2 ffxParabolicSin(FfxFloat32x2 x) +{ + return x * abs(x) - x; +} + +// An approximation of cosine. +// +// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range +// is {-1/4 to 1/4} representing {-1 to 1}. +// +// @param [in] value The value to calculate approximate cosine for. +// +// @returns +// The approximate cosine of value. +FfxFloat32 ffxParabolicCos(FfxFloat32 x) +{ + x = ffxFract(x * FfxFloat32(0.5) + FfxFloat32(0.75)); + x = x * FfxFloat32(2.0) - FfxFloat32(1.0); + return ffxParabolicSin(x); +} + +// An approximation of cosine. +// +// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range +// is {-1/4 to 1/4} representing {-1 to 1}. +// +// @param [in] value The value to calculate approximate cosine for. +// +// @returns +// The approximate cosine of value. +FfxFloat32x2 ffxParabolicCos(FfxFloat32x2 x) +{ + x = ffxFract(x * ffxBroadcast2(0.5f) + ffxBroadcast2(0.75f)); + x = x * ffxBroadcast2(2.0f) - ffxBroadcast2(1.0f); + return ffxParabolicSin(x); +} + +// An approximation of both sine and cosine. +// +// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range +// is {-1/4 to 1/4} representing {-1 to 1}. +// +// @param [in] value The value to calculate approximate cosine for. +// +// @returns +// A FfxFloat32x2 containing approximations of both sine and cosine of value. +FfxFloat32x2 ffxParabolicSinCos(FfxFloat32 x) +{ + FfxFloat32 y = ffxFract(x * FfxFloat32(0.5) + FfxFloat32(0.75)); + y = y * FfxFloat32(2.0) - FfxFloat32(1.0); + return ffxParabolicSin(FfxFloat32x2(x, y)); +} + +/// Conditional free logic AND operation using values. +/// +/// @param [in] x The first value to be fed into the AND operator. +/// @param [in] y The second value to be fed into the AND operator. +/// +/// @returns +/// Result of the AND operation. +/// +/// @ingroup GPUCore +FfxUInt32 ffxZeroOneAnd(FfxUInt32 x, FfxUInt32 y) +{ + return min(x, y); +} + +/// Conditional free logic AND operation using two values. +/// +/// @param [in] x The first value to be fed into the AND operator. +/// @param [in] y The second value to be fed into the AND operator. +/// +/// @returns +/// Result of the AND operation. +/// +/// @ingroup GPUCore +FfxUInt32x2 ffxZeroOneAnd(FfxUInt32x2 x, FfxUInt32x2 y) +{ + return min(x, y); +} + +/// Conditional free logic AND operation using two values. +/// +/// @param [in] x The first value to be fed into the AND operator. +/// @param [in] y The second value to be fed into the AND operator. +/// +/// @returns +/// Result of the AND operation. +/// +/// @ingroup GPUCore +FfxUInt32x3 ffxZeroOneAnd(FfxUInt32x3 x, FfxUInt32x3 y) +{ + return min(x, y); +} + +/// Conditional free logic AND operation using two values. +/// +/// @param [in] x The first value to be fed into the AND operator. +/// @param [in] y The second value to be fed into the AND operator. +/// +/// @returns +/// Result of the AND operation. +/// +/// @ingroup GPUCore +FfxUInt32x4 ffxZeroOneAnd(FfxUInt32x4 x, FfxUInt32x4 y) +{ + return min(x, y); +} + +/// Conditional free logic NOT operation using two values. +/// +/// @param [in] x The first value to be fed into the NOT operator. +/// +/// @returns +/// Result of the NOT operation. +/// +/// @ingroup GPUCore +FfxUInt32 ffxZeroOneAnd(FfxUInt32 x) +{ + return x ^ FfxUInt32(1); +} + +/// Conditional free logic NOT operation using two values. +/// +/// @param [in] x The first value to be fed into the NOT operator. +/// +/// @returns +/// Result of the NOT operation. +/// +/// @ingroup GPUCore +FfxUInt32x2 ffxZeroOneAnd(FfxUInt32x2 x) +{ + return x ^ ffxBroadcast2(1u); +} + +/// Conditional free logic NOT operation using two values. +/// +/// @param [in] x The first value to be fed into the NOT operator. +/// +/// @returns +/// Result of the NOT operation. +/// +/// @ingroup GPUCore +FfxUInt32x3 ffxZeroOneAnd(FfxUInt32x3 x) +{ + return x ^ ffxBroadcast3(1u); +} + +/// Conditional free logic NOT operation using two values. +/// +/// @param [in] x The first value to be fed into the NOT operator. +/// +/// @returns +/// Result of the NOT operation. +/// +/// @ingroup GPUCore +FfxUInt32x4 ffxZeroOneAnd(FfxUInt32x4 x) +{ + return x ^ ffxBroadcast4(1u); +} + +/// Conditional free logic OR operation using two values. +/// +/// @param [in] x The first value to be fed into the OR operator. +/// @param [in] y The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the OR operation. +/// +/// @ingroup GPUCore +FfxUInt32 ffxZeroOneOr(FfxUInt32 x, FfxUInt32 y) +{ + return max(x, y); +} + +/// Conditional free logic OR operation using two values. +/// +/// @param [in] x The first value to be fed into the OR operator. +/// @param [in] y The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the OR operation. +/// +/// @ingroup GPUCore +FfxUInt32x2 ffxZeroOneOr(FfxUInt32x2 x, FfxUInt32x2 y) +{ + return max(x, y); +} + +/// Conditional free logic OR operation using two values. +/// +/// @param [in] x The first value to be fed into the OR operator. +/// @param [in] y The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the OR operation. +/// +/// @ingroup GPUCore +FfxUInt32x3 ffxZeroOneOr(FfxUInt32x3 x, FfxUInt32x3 y) +{ + return max(x, y); +} + +/// Conditional free logic OR operation using two values. +/// +/// @param [in] x The first value to be fed into the OR operator. +/// @param [in] y The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the OR operation. +/// +/// @ingroup GPUCore +FfxUInt32x4 ffxZeroOneOr(FfxUInt32x4 x, FfxUInt32x4 y) +{ + return max(x, y); +} + +/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. +/// +/// @param [in] x The first value to be fed into the AND OR operator. +/// +/// @returns +/// Result of the AND OR operation. +/// +/// @ingroup GPUCore +FfxUInt32 ffxZeroOneAndToU1(FfxFloat32 x) +{ + return FfxUInt32(FfxFloat32(1.0) - x); +} + +/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. +/// +/// @param [in] x The first value to be fed into the AND OR operator. +/// +/// @returns +/// Result of the AND OR operation. +/// +/// @ingroup GPUCore +FfxUInt32x2 ffxZeroOneAndToU2(FfxFloat32x2 x) +{ + return FfxUInt32x2(ffxBroadcast2(1.0) - x); +} + +/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. +/// +/// @param [in] x The first value to be fed into the AND OR operator. +/// +/// @returns +/// Result of the AND OR operation. +/// +/// @ingroup GPUCore +FfxUInt32x3 ffxZeroOneAndToU3(FfxFloat32x3 x) +{ + return FfxUInt32x3(ffxBroadcast3(1.0) - x); +} + +/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. +/// +/// @param [in] x The first value to be fed into the AND OR operator. +/// +/// @returns +/// Result of the AND OR operation. +/// +/// @ingroup GPUCore +FfxUInt32x4 ffxZeroOneAndToU4(FfxFloat32x4 x) +{ + return FfxUInt32x4(ffxBroadcast4(1.0) - x); +} + +/// Conditional free logic AND operation using two values followed by a NOT operation +/// using the resulting value and a third value. +/// +/// @param [in] x The first value to be fed into the AND operator. +/// @param [in] y The second value to be fed into the AND operator. +/// @param [in] z The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the AND OR operation. +/// +/// @ingroup GPUCore +FfxFloat32 ffxZeroOneAndOr(FfxFloat32 x, FfxFloat32 y, FfxFloat32 z) +{ + return ffxSaturate(x * y + z); +} + +/// Conditional free logic AND operation using two values followed by a NOT operation +/// using the resulting value and a third value. +/// +/// @param [in] x The first value to be fed into the AND operator. +/// @param [in] y The second value to be fed into the AND operator. +/// @param [in] z The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the AND OR operation. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxZeroOneAndOr(FfxFloat32x2 x, FfxFloat32x2 y, FfxFloat32x2 z) +{ + return ffxSaturate(x * y + z); +} + +/// Conditional free logic AND operation using two values followed by a NOT operation +/// using the resulting value and a third value. +/// +/// @param [in] x The first value to be fed into the AND operator. +/// @param [in] y The second value to be fed into the AND operator. +/// @param [in] z The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the AND OR operation. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxZeroOneAndOr(FfxFloat32x3 x, FfxFloat32x3 y, FfxFloat32x3 z) +{ + return ffxSaturate(x * y + z); +} + +/// Conditional free logic AND operation using two values followed by a NOT operation +/// using the resulting value and a third value. +/// +/// @param [in] x The first value to be fed into the AND operator. +/// @param [in] y The second value to be fed into the AND operator. +/// @param [in] z The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the AND OR operation. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxZeroOneAndOr(FfxFloat32x4 x, FfxFloat32x4 y, FfxFloat32x4 z) +{ + return ffxSaturate(x * y + z); +} + +/// Given a value, returns 1.0 if greater than zero and 0.0 if not. +/// +/// @param [in] x The value to be compared. +/// +/// @returns +/// Result of the greater than zero comparison. +/// +/// @ingroup GPUCore +FfxFloat32 ffxZeroOneIsGreaterThanZero(FfxFloat32 x) +{ + return ffxSaturate(x * FfxFloat32(FFX_POSITIVE_INFINITY_FLOAT)); +} + +/// Given a value, returns 1.0 if greater than zero and 0.0 if not. +/// +/// @param [in] x The value to be compared. +/// +/// @returns +/// Result of the greater than zero comparison. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxZeroOneIsGreaterThanZero(FfxFloat32x2 x) +{ + return ffxSaturate(x * ffxBroadcast2(FFX_POSITIVE_INFINITY_FLOAT)); +} + +/// Given a value, returns 1.0 if greater than zero and 0.0 if not. +/// +/// @param [in] x The value to be compared. +/// +/// @returns +/// Result of the greater than zero comparison. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxZeroOneIsGreaterThanZero(FfxFloat32x3 x) +{ + return ffxSaturate(x * ffxBroadcast3(FFX_POSITIVE_INFINITY_FLOAT)); +} + +/// Given a value, returns 1.0 if greater than zero and 0.0 if not. +/// +/// @param [in] x The value to be compared. +/// +/// @returns +/// Result of the greater than zero comparison. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxZeroOneIsGreaterThanZero(FfxFloat32x4 x) +{ + return ffxSaturate(x * ffxBroadcast4(FFX_POSITIVE_INFINITY_FLOAT)); +} + +/// Conditional free logic signed NOT operation using two FfxFloat32 values. +/// +/// @param [in] x The first value to be fed into the AND OR operator. +/// +/// @returns +/// Result of the AND OR operation. +/// +/// @ingroup GPUCore +FfxFloat32 ffxZeroOneAnd(FfxFloat32 x) +{ + return FfxFloat32(1.0) - x; +} + +/// Conditional free logic signed NOT operation using two FfxFloat32 values. +/// +/// @param [in] x The first value to be fed into the AND OR operator. +/// +/// @returns +/// Result of the AND OR operation. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxZeroOneAnd(FfxFloat32x2 x) +{ + return ffxBroadcast2(1.0) - x; +} + +/// Conditional free logic signed NOT operation using two FfxFloat32 values. +/// +/// @param [in] x The first value to be fed into the AND OR operator. +/// +/// @returns +/// Result of the AND OR operation. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxZeroOneAnd(FfxFloat32x3 x) +{ + return ffxBroadcast3(1.0) - x; +} + +/// Conditional free logic signed NOT operation using two FfxFloat32 values. +/// +/// @param [in] x The first value to be fed into the AND OR operator. +/// +/// @returns +/// Result of the AND OR operation. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxZeroOneAnd(FfxFloat32x4 x) +{ + return ffxBroadcast4(1.0) - x; +} + +/// Conditional free logic OR operation using two FfxFloat32 values. +/// +/// @param [in] x The first value to be fed into the OR operator. +/// @param [in] y The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the OR operation. +/// +/// @ingroup GPUCore +FfxFloat32 ffxZeroOneOr(FfxFloat32 x, FfxFloat32 y) +{ + return max(x, y); +} + +/// Conditional free logic OR operation using two FfxFloat32 values. +/// +/// @param [in] x The first value to be fed into the OR operator. +/// @param [in] y The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the OR operation. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxZeroOneOr(FfxFloat32x2 x, FfxFloat32x2 y) +{ + return max(x, y); +} + +/// Conditional free logic OR operation using two FfxFloat32 values. +/// +/// @param [in] x The first value to be fed into the OR operator. +/// @param [in] y The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the OR operation. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxZeroOneOr(FfxFloat32x3 x, FfxFloat32x3 y) +{ + return max(x, y); +} + +/// Conditional free logic OR operation using two FfxFloat32 values. +/// +/// @param [in] x The first value to be fed into the OR operator. +/// @param [in] y The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the OR operation. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxZeroOneOr(FfxFloat32x4 x, FfxFloat32x4 y) +{ + return max(x, y); +} + +/// Choose between two FfxFloat32 values if the first paramter is greater than zero. +/// +/// @param [in] x The value to compare against zero. +/// @param [in] y The value to return if the comparision is greater than zero. +/// @param [in] z The value to return if the comparision is less than or equal to zero. +/// +/// @returns +/// The selected value. +/// +/// @ingroup GPUCore +FfxFloat32 ffxZeroOneSelect(FfxFloat32 x, FfxFloat32 y, FfxFloat32 z) +{ + FfxFloat32 r = (-x) * z + z; + return x * y + r; +} + +/// Choose between two FfxFloat32 values if the first paramter is greater than zero. +/// +/// @param [in] x The value to compare against zero. +/// @param [in] y The value to return if the comparision is greater than zero. +/// @param [in] z The value to return if the comparision is less than or equal to zero. +/// +/// @returns +/// The selected value. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxZeroOneSelect(FfxFloat32x2 x, FfxFloat32x2 y, FfxFloat32x2 z) +{ + FfxFloat32x2 r = (-x) * z + z; + return x * y + r; +} + +/// Choose between two FfxFloat32 values if the first paramter is greater than zero. +/// +/// @param [in] x The value to compare against zero. +/// @param [in] y The value to return if the comparision is greater than zero. +/// @param [in] z The value to return if the comparision is less than or equal to zero. +/// +/// @returns +/// The selected value. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxZeroOneSelect(FfxFloat32x3 x, FfxFloat32x3 y, FfxFloat32x3 z) +{ + FfxFloat32x3 r = (-x) * z + z; + return x * y + r; +} + +/// Choose between two FfxFloat32 values if the first paramter is greater than zero. +/// +/// @param [in] x The value to compare against zero. +/// @param [in] y The value to return if the comparision is greater than zero. +/// @param [in] z The value to return if the comparision is less than or equal to zero. +/// +/// @returns +/// The selected value. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxZeroOneSelect(FfxFloat32x4 x, FfxFloat32x4 y, FfxFloat32x4 z) +{ + FfxFloat32x4 r = (-x) * z + z; + return x * y + r; +} + +/// Given a value, returns 1.0 if less than zero and 0.0 if not. +/// +/// @param [in] x The value to be compared. +/// +/// @returns +/// Result of the sign value. +/// +/// @ingroup GPUCore +FfxFloat32 ffxZeroOneIsSigned(FfxFloat32 x) +{ + return ffxSaturate(x * FfxFloat32(FFX_NEGATIVE_INFINITY_FLOAT)); +} + +/// Given a value, returns 1.0 if less than zero and 0.0 if not. +/// +/// @param [in] x The value to be compared. +/// +/// @returns +/// Result of the sign value. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxZeroOneIsSigned(FfxFloat32x2 x) +{ + return ffxSaturate(x * ffxBroadcast2(FFX_NEGATIVE_INFINITY_FLOAT)); +} + +/// Given a value, returns 1.0 if less than zero and 0.0 if not. +/// +/// @param [in] x The value to be compared. +/// +/// @returns +/// Result of the sign value. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxZeroOneIsSigned(FfxFloat32x3 x) +{ + return ffxSaturate(x * ffxBroadcast3(FFX_NEGATIVE_INFINITY_FLOAT)); +} + +/// Given a value, returns 1.0 if less than zero and 0.0 if not. +/// +/// @param [in] x The value to be compared. +/// +/// @returns +/// Result of the sign value. +/// +/// @ingroup GPUCore +FfxFloat32x4 ffxZeroOneIsSigned(FfxFloat32x4 x) +{ + return ffxSaturate(x * ffxBroadcast4(FFX_NEGATIVE_INFINITY_FLOAT)); +} + +/// Compute a Rec.709 color space. +/// +/// Rec.709 is used for some HDTVs. +/// +/// Both Rec.709 and sRGB have a linear segment which as spec'ed would intersect the curved segment 2 times. +/// (a.) For 8-bit sRGB, steps {0 to 10.3} are in the linear region (4% of the encoding range). +/// (b.) For 8-bit 709, steps {0 to 20.7} are in the linear region (8% of the encoding range). +/// +/// @param [in] color The color to convert to Rec. 709. +/// +/// @returns +/// The color in linear space. +/// +/// @ingroup GPUCore +FfxFloat32 ffxRec709FromLinear(FfxFloat32 color) +{ + FfxFloat32x3 j = FfxFloat32x3(0.018 * 4.5, 4.5, 0.45); + FfxFloat32x2 k = FfxFloat32x2(1.099, -0.099); + return clamp(j.x, color * j.y, pow(color, j.z) * k.x + k.y); +} + +/// Compute a Rec.709 color space. +/// +/// Rec.709 is used for some HDTVs. +/// +/// Both Rec.709 and sRGB have a linear segment which as spec'ed would intersect the curved segment 2 times. +/// (a.) For 8-bit sRGB, steps {0 to 10.3} are in the linear region (4% of the encoding range). +/// (b.) For 8-bit 709, steps {0 to 20.7} are in the linear region (8% of the encoding range). +/// +/// @param [in] color The color to convert to Rec. 709. +/// +/// @returns +/// The color in linear space. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxRec709FromLinear(FfxFloat32x2 color) +{ + FfxFloat32x3 j = FfxFloat32x3(0.018 * 4.5, 4.5, 0.45); + FfxFloat32x2 k = FfxFloat32x2(1.099, -0.099); + return clamp(j.xx, color * j.yy, pow(color, j.zz) * k.xx + k.yy); +} + +/// Compute a Rec.709 color space. +/// +/// Rec.709 is used for some HDTVs. +/// +/// Both Rec.709 and sRGB have a linear segment which as spec'ed would intersect the curved segment 2 times. +/// (a.) For 8-bit sRGB, steps {0 to 10.3} are in the linear region (4% of the encoding range). +/// (b.) For 8-bit 709, steps {0 to 20.7} are in the linear region (8% of the encoding range). +/// +/// @param [in] color The color to convert to Rec. 709. +/// +/// @returns +/// The color in linear space. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxRec709FromLinear(FfxFloat32x3 color) +{ + FfxFloat32x3 j = FfxFloat32x3(0.018 * 4.5, 4.5, 0.45); + FfxFloat32x2 k = FfxFloat32x2(1.099, -0.099); + return clamp(j.xxx, color * j.yyy, pow(color, j.zzz) * k.xxx + k.yyy); +} + +/// Compute a linear value from a REC.709 value. +/// +/// @param [in] color The value to convert to linear from REC.709. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat32 ffxLinearFromRec709(FfxFloat32 color) +{ + FfxFloat32x3 j = FfxFloat32x3(0.081 / 4.5, 1.0 / 4.5, 1.0 / 0.45); + FfxFloat32x2 k = FfxFloat32x2(1.0 / 1.099, 0.099 / 1.099); + return ffxZeroOneSelect(ffxZeroOneIsSigned(color - j.x), color * j.y, pow(color * k.x + k.y, j.z)); +} + +/// Compute a linear value from a REC.709 value. +/// +/// @param [in] color The value to convert to linear from REC.709. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxLinearFromRec709(FfxFloat32x2 color) +{ + FfxFloat32x3 j = FfxFloat32x3(0.081 / 4.5, 1.0 / 4.5, 1.0 / 0.45); + FfxFloat32x2 k = FfxFloat32x2(1.0 / 1.099, 0.099 / 1.099); + return ffxZeroOneSelect(ffxZeroOneIsSigned(color - j.xx), color * j.yy, pow(color * k.xx + k.yy, j.zz)); +} + +/// Compute a linear value from a REC.709 value. +/// +/// @param [in] color The value to convert to linear from REC.709. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxLinearFromRec709(FfxFloat32x3 color) +{ + FfxFloat32x3 j = FfxFloat32x3(0.081 / 4.5, 1.0 / 4.5, 1.0 / 0.45); + FfxFloat32x2 k = FfxFloat32x2(1.0 / 1.099, 0.099 / 1.099); + return ffxZeroOneSelect(ffxZeroOneIsSigned(color - j.xxx), color * j.yyy, pow(color * k.xxx + k.yyy, j.zzz)); +} + +/// Compute a gamma value from a linear value. +/// +/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. +/// +/// Note: 'rcpX' is '1/x', where the 'x' is what would be used in ffxLinearFromGamma. +/// +/// @param [in] value The value to convert to gamma space from linear. +/// @param [in] power The reciprocal of power value used for the gamma curve. +/// +/// @returns +/// A value in gamma space. +/// +/// @ingroup GPUCore +FfxFloat32 ffxGammaFromLinear(FfxFloat32 value, FfxFloat32 power) +{ + return pow(value, FfxFloat32(power)); +} + +/// Compute a gamma value from a linear value. +/// +/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. +/// +/// Note: 'rcpX' is '1/x', where the 'x' is what would be used in ffxLinearFromGamma. +/// +/// @param [in] value The value to convert to gamma space from linear. +/// @param [in] power The reciprocal of power value used for the gamma curve. +/// +/// @returns +/// A value in gamma space. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxGammaFromLinear(FfxFloat32x2 value, FfxFloat32 power) +{ + return pow(value, ffxBroadcast2(power)); +} + +/// Compute a gamma value from a linear value. +/// +/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. +/// +/// Note: 'rcpX' is '1/x', where the 'x' is what would be used in ffxLinearFromGamma. +/// +/// @param [in] value The value to convert to gamma space from linear. +/// @param [in] power The reciprocal of power value used for the gamma curve. +/// +/// @returns +/// A value in gamma space. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxGammaFromLinear(FfxFloat32x3 value, FfxFloat32 power) +{ + return pow(value, ffxBroadcast3(power)); +} + +/// Compute a linear value from a value in a gamma space. +/// +/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. +/// +/// @param [in] color The value to convert to linear in gamma space. +/// @param [in] power The power value used for the gamma curve. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat32 ffxLinearFromGamma(FfxFloat32 color, FfxFloat32 power) +{ + return pow(color, FfxFloat32(power)); +} + +/// Compute a linear value from a value in a gamma space. +/// +/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. +/// +/// @param [in] color The value to convert to linear in gamma space. +/// @param [in] power The power value used for the gamma curve. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxLinearFromGamma(FfxFloat32x2 color, FfxFloat32 power) +{ + return pow(color, ffxBroadcast2(power)); +} + +/// Compute a linear value from a value in a gamma space. +/// +/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. +/// +/// @param [in] color The value to convert to linear in gamma space. +/// @param [in] power The power value used for the gamma curve. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxLinearFromGamma(FfxFloat32x3 color, FfxFloat32 power) +{ + return pow(color, ffxBroadcast3(power)); +} + +/// Compute a PQ value from a linear value. +/// +/// @param [in] value The value to convert to PQ from linear. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat32 ffxPQFromLinear(FfxFloat32 value) +{ + FfxFloat32 p = pow(value, FfxFloat32(0.159302)); + return pow((FfxFloat32(0.835938) + FfxFloat32(18.8516) * p) / (FfxFloat32(1.0) + FfxFloat32(18.6875) * p), FfxFloat32(78.8438)); +} + +/// Compute a PQ value from a linear value. +/// +/// @param [in] value The value to convert to PQ from linear. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxPQFromLinear(FfxFloat32x2 value) +{ + FfxFloat32x2 p = pow(value, ffxBroadcast2(0.159302)); + return pow((ffxBroadcast2(0.835938) + ffxBroadcast2(18.8516) * p) / (ffxBroadcast2(1.0) + ffxBroadcast2(18.6875) * p), ffxBroadcast2(78.8438)); +} + +/// Compute a PQ value from a linear value. +/// +/// @param [in] value The value to convert to PQ from linear. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxPQFromLinear(FfxFloat32x3 value) +{ + FfxFloat32x3 p = pow(value, ffxBroadcast3(0.159302)); + return pow((ffxBroadcast3(0.835938) + ffxBroadcast3(18.8516) * p) / (ffxBroadcast3(1.0) + ffxBroadcast3(18.6875) * p), ffxBroadcast3(78.8438)); +} + +/// Compute a linear value from a value in a PQ space. +/// +/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. +/// +/// @param [in] value The value to convert to linear in PQ space. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat32 ffxLinearFromPQ(FfxFloat32 value) +{ + FfxFloat32 p = pow(value, FfxFloat32(0.0126833)); + return pow(ffxSaturate(p - FfxFloat32(0.835938)) / (FfxFloat32(18.8516) - FfxFloat32(18.6875) * p), FfxFloat32(6.27739)); +} + +/// Compute a linear value from a value in a PQ space. +/// +/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. +/// +/// @param [in] value The value to convert to linear in PQ space. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxLinearFromPQ(FfxFloat32x2 value) +{ + FfxFloat32x2 p = pow(value, ffxBroadcast2(0.0126833)); + return pow(ffxSaturate(p - ffxBroadcast2(0.835938)) / (ffxBroadcast2(18.8516) - ffxBroadcast2(18.6875) * p), ffxBroadcast2(6.27739)); +} + +/// Compute a linear value from a value in a PQ space. +/// +/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. +/// +/// @param [in] value The value to convert to linear in PQ space. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxLinearFromPQ(FfxFloat32x3 value) +{ + FfxFloat32x3 p = pow(value, ffxBroadcast3(0.0126833)); + return pow(ffxSaturate(p - ffxBroadcast3(0.835938)) / (ffxBroadcast3(18.8516) - ffxBroadcast3(18.6875) * p), ffxBroadcast3(6.27739)); +} + +/// Compute an SRGB value from a linear value. +/// +/// @param [in] value The value to convert to SRGB from linear. +/// +/// @returns +/// A value in SRGB space. +/// +/// @ingroup GPUCore +FfxFloat32 ffxSrgbFromLinear(FfxFloat32 value) +{ + FfxFloat32x3 j = FfxFloat32x3(0.0031308 * 12.92, 12.92, 1.0 / 2.4); + FfxFloat32x2 k = FfxFloat32x2(1.055, -0.055); + return clamp(j.x, value * j.y, pow(value, j.z) * k.x + k.y); +} + +/// Compute an SRGB value from a linear value. +/// +/// @param [in] value The value to convert to SRGB from linear. +/// +/// @returns +/// A value in SRGB space. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxSrgbFromLinear(FfxFloat32x2 value) +{ + FfxFloat32x3 j = FfxFloat32x3(0.0031308 * 12.92, 12.92, 1.0 / 2.4); + FfxFloat32x2 k = FfxFloat32x2(1.055, -0.055); + return clamp(j.xx, value * j.yy, pow(value, j.zz) * k.xx + k.yy); +} + +/// Compute an SRGB value from a linear value. +/// +/// @param [in] value The value to convert to SRGB from linear. +/// +/// @returns +/// A value in SRGB space. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxSrgbFromLinear(FfxFloat32x3 value) +{ + FfxFloat32x3 j = FfxFloat32x3(0.0031308 * 12.92, 12.92, 1.0 / 2.4); + FfxFloat32x2 k = FfxFloat32x2(1.055, -0.055); + return clamp(j.xxx, value * j.yyy, pow(value, j.zzz) * k.xxx + k.yyy); +} + +/// Compute a linear value from a value in a SRGB space. +/// +/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. +/// +/// @param [in] value The value to convert to linear in SRGB space. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat32 ffxLinearFromSrgb(FfxFloat32 value) +{ + FfxFloat32x3 j = FfxFloat32x3(0.04045 / 12.92, 1.0 / 12.92, 2.4); + FfxFloat32x2 k = FfxFloat32x2(1.0 / 1.055, 0.055 / 1.055); + return ffxZeroOneSelect(ffxZeroOneIsSigned(value - j.x), value * j.y, pow(value * k.x + k.y, j.z)); +} + +/// Compute a linear value from a value in a SRGB space. +/// +/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. +/// +/// @param [in] value The value to convert to linear in SRGB space. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat32x2 ffxLinearFromSrgb(FfxFloat32x2 value) +{ + FfxFloat32x3 j = FfxFloat32x3(0.04045 / 12.92, 1.0 / 12.92, 2.4); + FfxFloat32x2 k = FfxFloat32x2(1.0 / 1.055, 0.055 / 1.055); + return ffxZeroOneSelect(ffxZeroOneIsSigned(value - j.xx), value * j.yy, pow(value * k.xx + k.yy, j.zz)); +} + +/// Compute a linear value from a value in a SRGB space. +/// +/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. +/// +/// @param [in] value The value to convert to linear in SRGB space. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat32x3 ffxLinearFromSrgb(FfxFloat32x3 value) +{ + FfxFloat32x3 j = FfxFloat32x3(0.04045 / 12.92, 1.0 / 12.92, 2.4); + FfxFloat32x2 k = FfxFloat32x2(1.0 / 1.055, 0.055 / 1.055); + return ffxZeroOneSelect(ffxZeroOneIsSigned(value - j.xxx), value * j.yyy, pow(value * k.xxx + k.yyy, j.zzz)); +} + +/// A remapping of 64x1 to 8x8 imposing rotated 2x2 pixel quads in quad linear. +/// +/// Remap illustration: +/// +/// 543210 +/// ~~~~~~ +/// ..xxx. +/// yy...y +/// +/// @param [in] a The input 1D coordinates to remap. +/// +/// @returns +/// The remapped 2D coordinates. +/// +/// @ingroup GPUCore +FfxUInt32x2 ffxRemapForQuad(FfxUInt32 a) +{ + return FfxUInt32x2(ffxBitfieldExtract(a, 1u, 3u), ffxBitfieldInsertMask(ffxBitfieldExtract(a, 3u, 3u), a, 1u)); +} + +/// A helper function performing a remap 64x1 to 8x8 remapping which is necessary for 2D wave reductions. +/// +/// The 64-wide lane indices to 8x8 remapping is performed as follows: +/// +/// 00 01 08 09 10 11 18 19 +/// 02 03 0a 0b 12 13 1a 1b +/// 04 05 0c 0d 14 15 1c 1d +/// 06 07 0e 0f 16 17 1e 1f +/// 20 21 28 29 30 31 38 39 +/// 22 23 2a 2b 32 33 3a 3b +/// 24 25 2c 2d 34 35 3c 3d +/// 26 27 2e 2f 36 37 3e 3f +/// +/// @param [in] a The input 1D coordinate to remap. +/// +/// @returns +/// The remapped 2D coordinates. +/// +/// @ingroup GPUCore +FfxUInt32x2 ffxRemapForWaveReduction(FfxUInt32 a) +{ + return FfxUInt32x2(ffxBitfieldInsertMask(ffxBitfieldExtract(a, 2u, 3u), a, 1u), ffxBitfieldInsertMask(ffxBitfieldExtract(a, 3u, 3u), ffxBitfieldExtract(a, 1u, 2u), 2u)); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_gpu_common.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_gpu_common.h.meta new file mode 100644 index 0000000..070d7a5 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_gpu_common.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 4950d6c78609df549a0ed96137bf3bf1 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_gpu_common_half.h b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_gpu_common_half.h new file mode 100644 index 0000000..1cb780b --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_gpu_common_half.h @@ -0,0 +1,2981 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#if FFX_HALF +#if FFX_HLSL_SM >= 62 +/// A define value for 16bit positive infinity. +/// +/// @ingroup GPUCore +#define FFX_POSITIVE_INFINITY_HALF FFX_TO_FLOAT16((uint16_t)0x7c00u) + +/// A define value for 16bit negative infinity. +/// +/// @ingroup GPUCore +#define FFX_NEGATIVE_INFINITY_HALF FFX_TO_FLOAT16((uint16_t)0xfc00u) +#else +/// A define value for 16bit positive infinity. +/// +/// @ingroup GPUCore +#define FFX_POSITIVE_INFINITY_HALF FFX_TO_FLOAT16(0x7c00u) + +/// A define value for 16bit negative infinity. +/// +/// @ingroup GPUCore +#define FFX_NEGATIVE_INFINITY_HALF FFX_TO_FLOAT16(0xfc00u) +#endif // #if FFX_HLSL_SM>=62 + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxFloat16 ffxMin(FfxFloat16 x, FfxFloat16 y) +{ + return min(x, y); +} + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxMin(FfxFloat16x2 x, FfxFloat16x2 y) +{ + return min(x, y); +} + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxMin(FfxFloat16x3 x, FfxFloat16x3 y) +{ + return min(x, y); +} + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxFloat16x4 ffxMin(FfxFloat16x4 x, FfxFloat16x4 y) +{ + return min(x, y); +} + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxInt16 ffxMin(FfxInt16 x, FfxInt16 y) +{ + return min(x, y); +} + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxInt16x2 ffxMin(FfxInt16x2 x, FfxInt16x2 y) +{ + return min(x, y); +} + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxInt16x3 ffxMin(FfxInt16x3 x, FfxInt16x3 y) +{ + return min(x, y); +} + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxInt16x4 ffxMin(FfxInt16x4 x, FfxInt16x4 y) +{ + return min(x, y); +} + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxUInt16 ffxMin(FfxUInt16 x, FfxUInt16 y) +{ + return min(x, y); +} + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxUInt16x2 ffxMin(FfxUInt16x2 x, FfxUInt16x2 y) +{ + return min(x, y); +} + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxUInt16x3 ffxMin(FfxUInt16x3 x, FfxUInt16x3 y) +{ + return min(x, y); +} + +/// Compute the min of two values. +/// +/// @param [in] x The first value to compute the min of. +/// @param [in] y The second value to compute the min of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxUInt16x4 ffxMin(FfxUInt16x4 x, FfxUInt16x4 y) +{ + return min(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxFloat16 ffxMax(FfxFloat16 x, FfxFloat16 y) +{ + return max(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxMax(FfxFloat16x2 x, FfxFloat16x2 y) +{ + return max(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxMax(FfxFloat16x3 x, FfxFloat16x3 y) +{ + return max(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxFloat16x4 ffxMax(FfxFloat16x4 x, FfxFloat16x4 y) +{ + return max(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxInt16 ffxMax(FfxInt16 x, FfxInt16 y) +{ + return max(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxInt16x2 ffxMax(FfxInt16x2 x, FfxInt16x2 y) +{ + return max(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxInt16x3 ffxMax(FfxInt16x3 x, FfxInt16x3 y) +{ + return max(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxInt16x4 ffxMax(FfxInt16x4 x, FfxInt16x4 y) +{ + return max(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxUInt16 ffxMax(FfxUInt16 x, FfxUInt16 y) +{ + return max(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxUInt16x2 ffxMax(FfxUInt16x2 x, FfxUInt16x2 y) +{ + return max(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxUInt16x3 ffxMax(FfxUInt16x3 x, FfxUInt16x3 y) +{ + return max(x, y); +} + +/// Compute the max of two values. +/// +/// @param [in] x The first value to compute the max of. +/// @param [in] y The second value to compute the max of. +/// +/// @returns +/// The the lowest of two values. +/// +/// @ingroup GPUCore +FfxUInt16x4 ffxMax(FfxUInt16x4 x, FfxUInt16x4 y) +{ + return max(x, y); +} + +/// Compute the value of the first parameter raised to the power of the second. +/// +/// @param [in] x The value to raise to the power y. +/// @param [in] y The power to which to raise x. +/// +/// @returns +/// The value of the first parameter raised to the power of the second. +/// +/// @ingroup GPUCore +FfxFloat16 ffxPow(FfxFloat16 x, FfxFloat16 y) +{ + return pow(x, y); +} + +/// Compute the value of the first parameter raised to the power of the second. +/// +/// @param [in] x The value to raise to the power y. +/// @param [in] y The power to which to raise x. +/// +/// @returns +/// The value of the first parameter raised to the power of the second. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxPow(FfxFloat16x2 x, FfxFloat16x2 y) +{ + return pow(x, y); +} + +/// Compute the value of the first parameter raised to the power of the second. +/// +/// @param [in] x The value to raise to the power y. +/// @param [in] y The power to which to raise x. +/// +/// @returns +/// The value of the first parameter raised to the power of the second. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxPow(FfxFloat16x3 x, FfxFloat16x3 y) +{ + return pow(x, y); +} + +/// Compute the value of the first parameter raised to the power of the second. +/// +/// @param [in] x The value to raise to the power y. +/// @param [in] y The power to which to raise x. +/// +/// @returns +/// The value of the first parameter raised to the power of the second. +/// +/// @ingroup GPUCore +FfxFloat16x4 ffxPow(FfxFloat16x4 x, FfxFloat16x4 y) +{ + return pow(x, y); +} + +/// Compute the square root of a value. +/// +/// @param [in] x The first value to compute the min of. +/// +/// @returns +/// The the square root of x. +/// +/// @ingroup GPUCore +FfxFloat16 ffxSqrt(FfxFloat16 x) +{ + return sqrt(x); +} + +/// Compute the square root of a value. +/// +/// @param [in] x The first value to compute the min of. +/// +/// @returns +/// The the square root of x. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxSqrt(FfxFloat16x2 x) +{ + return sqrt(x); +} + +/// Compute the square root of a value. +/// +/// @param [in] x The first value to compute the min of. +/// +/// @returns +/// The the square root of x. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxSqrt(FfxFloat16x3 x) +{ + return sqrt(x); +} + +/// Compute the square root of a value. +/// +/// @param [in] x The first value to compute the min of. +/// +/// @returns +/// The the square root of x. +/// +/// @ingroup GPUCore +FfxFloat16x4 ffxSqrt(FfxFloat16x4 x) +{ + return sqrt(x); +} + +/// Copy the sign bit from 's' to positive 'd'. +/// +/// @param [in] d The value to copy the sign bit into. +/// @param [in] s The value to copy the sign bit from. +/// +/// @returns +/// The value of d with the sign bit from s. +/// +/// @ingroup GPUCore +FfxFloat16 ffxCopySignBitHalf(FfxFloat16 d, FfxFloat16 s) +{ + return FFX_TO_FLOAT16(FFX_TO_UINT16(d) | (FFX_TO_UINT16(s) & FFX_BROADCAST_UINT16(0x8000u))); +} + +/// Copy the sign bit from 's' to positive 'd'. +/// +/// @param [in] d The value to copy the sign bit into. +/// @param [in] s The value to copy the sign bit from. +/// +/// @returns +/// The value of d with the sign bit from s. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxCopySignBitHalf(FfxFloat16x2 d, FfxFloat16x2 s) +{ + return FFX_TO_FLOAT16X2(FFX_TO_UINT16X2(d) | (FFX_TO_UINT16X2(s) & FFX_BROADCAST_UINT16X2(0x8000u))); +} + +/// Copy the sign bit from 's' to positive 'd'. +/// +/// @param [in] d The value to copy the sign bit into. +/// @param [in] s The value to copy the sign bit from. +/// +/// @returns +/// The value of d with the sign bit from s. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxCopySignBitHalf(FfxFloat16x3 d, FfxFloat16x3 s) +{ + return FFX_TO_FLOAT16X3(FFX_TO_UINT16X3(d) | (FFX_TO_UINT16X3(s) & FFX_BROADCAST_UINT16X3(0x8000u))); +} + +/// Copy the sign bit from 's' to positive 'd'. +/// +/// @param [in] d The value to copy the sign bit into. +/// @param [in] s The value to copy the sign bit from. +/// +/// @returns +/// The value of d with the sign bit from s. +/// +/// @ingroup GPUCore +FfxFloat16x4 ffxCopySignBitHalf(FfxFloat16x4 d, FfxFloat16x4 s) +{ + return FFX_TO_FLOAT16X4(FFX_TO_UINT16X4(d) | (FFX_TO_UINT16X4(s) & FFX_BROADCAST_UINT16X4(0x8000u))); +} + +/// A single operation to return the following: +/// m = NaN := 0 +/// m >= 0 := 0 +/// m < 0 := 1 +/// +/// Uses the following useful floating point logic, +/// saturate(+a*(-INF)==-INF) := 0 +/// saturate( 0*(-INF)== NaN) := 0 +/// saturate(-a*(-INF)==+INF) := 1 +/// +/// This function is useful when creating masks for branch-free logic. +/// +/// @param [in] m The value to test against 0. +/// +/// @returns +/// 1.0 when the value is negative, or 0.0 when the value is 0 or position. +/// +/// @ingroup GPUCore +FfxFloat16 ffxIsSignedHalf(FfxFloat16 m) +{ + return ffxSaturate(m * FFX_BROADCAST_FLOAT16(FFX_NEGATIVE_INFINITY_HALF)); +} + +/// A single operation to return the following: +/// m = NaN := 0 +/// m >= 0 := 0 +/// m < 0 := 1 +/// +/// Uses the following useful floating point logic, +/// saturate(+a*(-INF)==-INF) := 0 +/// saturate( 0*(-INF)== NaN) := 0 +/// saturate(-a*(-INF)==+INF) := 1 +/// +/// This function is useful when creating masks for branch-free logic. +/// +/// @param [in] m The value to test against 0. +/// +/// @returns +/// 1.0 when the value is negative, or 0.0 when the value is 0 or position. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxIsSignedHalf(FfxFloat16x2 m) +{ + return ffxSaturate(m * FFX_BROADCAST_FLOAT16X2(FFX_NEGATIVE_INFINITY_HALF)); +} + +/// A single operation to return the following: +/// m = NaN := 0 +/// m >= 0 := 0 +/// m < 0 := 1 +/// +/// Uses the following useful floating point logic, +/// saturate(+a*(-INF)==-INF) := 0 +/// saturate( 0*(-INF)== NaN) := 0 +/// saturate(-a*(-INF)==+INF) := 1 +/// +/// This function is useful when creating masks for branch-free logic. +/// +/// @param [in] m The value to test against 0. +/// +/// @returns +/// 1.0 when the value is negative, or 0.0 when the value is 0 or position. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxIsSignedHalf(FfxFloat16x3 m) +{ + return ffxSaturate(m * FFX_BROADCAST_FLOAT16X3(FFX_NEGATIVE_INFINITY_HALF)); +} + +/// A single operation to return the following: +/// m = NaN := 0 +/// m >= 0 := 0 +/// m < 0 := 1 +/// +/// Uses the following useful floating point logic, +/// saturate(+a*(-INF)==-INF) := 0 +/// saturate( 0*(-INF)== NaN) := 0 +/// saturate(-a*(-INF)==+INF) := 1 +/// +/// This function is useful when creating masks for branch-free logic. +/// +/// @param [in] m The value to test against 0. +/// +/// @returns +/// 1.0 when the value is negative, or 0.0 when the value is 0 or position. +/// +/// @ingroup GPUCore +FfxFloat16x4 ffxIsSignedHalf(FfxFloat16x4 m) +{ + return ffxSaturate(m * FFX_BROADCAST_FLOAT16X4(FFX_NEGATIVE_INFINITY_HALF)); +} + +/// A single operation to return the following: +/// m = NaN := 1 +/// m > 0 := 0 +/// m <= 0 := 1 +/// +/// This function is useful when creating masks for branch-free logic. +/// +/// @param [in] m The value to test against zero. +/// +/// @returns +/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. +/// +/// @ingroup GPUCore +FfxFloat16 ffxIsGreaterThanZeroHalf(FfxFloat16 m) +{ + return ffxSaturate(m * FFX_BROADCAST_FLOAT16(FFX_POSITIVE_INFINITY_HALF)); +} + +/// A single operation to return the following: +/// m = NaN := 1 +/// m > 0 := 0 +/// m <= 0 := 1 +/// +/// This function is useful when creating masks for branch-free logic. +/// +/// @param [in] m The value to test against zero. +/// +/// @returns +/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxIsGreaterThanZeroHalf(FfxFloat16x2 m) +{ + return ffxSaturate(m * FFX_BROADCAST_FLOAT16X2(FFX_POSITIVE_INFINITY_HALF)); +} + +/// A single operation to return the following: +/// m = NaN := 1 +/// m > 0 := 0 +/// m <= 0 := 1 +/// +/// This function is useful when creating masks for branch-free logic. +/// +/// @param [in] m The value to test against zero. +/// +/// @returns +/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxIsGreaterThanZeroHalf(FfxFloat16x3 m) +{ + return ffxSaturate(m * FFX_BROADCAST_FLOAT16X3(FFX_POSITIVE_INFINITY_HALF)); +} + +/// A single operation to return the following: +/// m = NaN := 1 +/// m > 0 := 0 +/// m <= 0 := 1 +/// +/// This function is useful when creating masks for branch-free logic. +/// +/// @param [in] m The value to test against zero. +/// +/// @returns +/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. +/// +/// @ingroup GPUCore +FfxFloat16x4 ffxIsGreaterThanZeroHalf(FfxFloat16x4 m) +{ + return ffxSaturate(m * FFX_BROADCAST_FLOAT16X4(FFX_POSITIVE_INFINITY_HALF)); +} + +/// Convert a 16bit floating point value to sortable integer. +/// +/// - If sign bit=0, flip the sign bit (positives). +/// - If sign bit=1, flip all bits (negatives). +/// +/// The function has the side effects that: +/// - Larger integers are more positive values. +/// - Float zero is mapped to center of integers (so clear to integer zero is a nice default for atomic max usage). +/// +/// @param [in] x The floating point value to make sortable. +/// +/// @returns +/// The sortable integer value. +/// +/// @ingroup GPUCore +FfxUInt16 ffxFloatToSortableIntegerHalf(FfxUInt16 x) +{ + return x ^ ((ffxBitShiftRightHalf(x, FFX_BROADCAST_UINT16(15))) | FFX_BROADCAST_UINT16(0x8000)); +} + +/// Convert a sortable integer to a 16bit floating point value. +/// +/// The function has the side effects that: +/// - If sign bit=1, flip the sign bit (positives). +/// - If sign bit=0, flip all bits (negatives). +/// +/// @param [in] x The sortable integer value to make floating point. +/// +/// @returns +/// The floating point value. +/// +/// @ingroup GPUCore +FfxUInt16 ffxSortableIntegerToFloatHalf(FfxUInt16 x) +{ + return x ^ ((~ffxBitShiftRightHalf(x, FFX_BROADCAST_UINT16(15))) | FFX_BROADCAST_UINT16(0x8000)); +} + +/// Convert a pair of 16bit floating point values to a pair of sortable integers. +/// +/// - If sign bit=0, flip the sign bit (positives). +/// - If sign bit=1, flip all bits (negatives). +/// +/// The function has the side effects that: +/// - Larger integers are more positive values. +/// - Float zero is mapped to center of integers (so clear to integer zero is a nice default for atomic max usage). +/// +/// @param [in] x The floating point values to make sortable. +/// +/// @returns +/// The sortable integer values. +/// +/// @ingroup GPUCore +FfxUInt16x2 ffxFloatToSortableIntegerHalf(FfxUInt16x2 x) +{ + return x ^ ((ffxBitShiftRightHalf(x, FFX_BROADCAST_UINT16X2(15))) | FFX_BROADCAST_UINT16X2(0x8000)); +} + +/// Convert a pair of sortable integers to a pair of 16bit floating point values. +/// +/// The function has the side effects that: +/// - If sign bit=1, flip the sign bit (positives). +/// - If sign bit=0, flip all bits (negatives). +/// +/// @param [in] x The sortable integer values to make floating point. +/// +/// @returns +/// The floating point values. +/// +/// @ingroup GPUCore +FfxUInt16x2 ffxSortableIntegerToFloatHalf(FfxUInt16x2 x) +{ + return x ^ ((~ffxBitShiftRightHalf(x, FFX_BROADCAST_UINT16X2(15))) | FFX_BROADCAST_UINT16X2(0x8000)); +} + +/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. +/// +/// The resulting integer will contain bytes in the following order, from most to least significant: +/// [Zero] Y0 [Zero] X0 +/// +/// @param [in] i The integer pair to pack. +/// +/// @returns +/// The packed integer value. +/// +/// @ingroup GPUCore +FfxUInt32 ffxPackBytesZeroY0ZeroX0(FfxUInt32x2 i) +{ + return ((i.x) & 0xffu) | ((i.y << 16) & 0xff0000u); +} + +/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. +/// +/// The resulting integer will contain bytes in the following order, from most to least significant: +/// [Zero] Y1 [Zero] X1 +/// +/// @param [in] i The integer pair to pack. +/// +/// @returns +/// The packed integer value. +/// +/// @ingroup GPUCore +FfxUInt32 ffxPackBytesZeroY1ZeroX1(FfxUInt32x2 i) +{ + return ((i.x >> 8) & 0xffu) | ((i.y << 8) & 0xff0000u); +} + +/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. +/// +/// The resulting integer will contain bytes in the following order, from most to least significant: +/// [Zero] Y2 [Zero] X2 +/// +/// @param [in] i The integer pair to pack. +/// +/// @returns +/// The packed integer value. +/// +/// @ingroup GPUCore +FfxUInt32 ffxPackBytesZeroY2ZeroX2(FfxUInt32x2 i) +{ + return ((i.x >> 16) & 0xffu) | ((i.y) & 0xff0000u); +} + +/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. +/// +/// The resulting integer will contain bytes in the following order, from most to least significant: +/// [Zero] Y3 [Zero] X3 +/// +/// @param [in] i The integer pair to pack. +/// +/// @returns +/// The packed integer value. +/// +/// @ingroup GPUCore +FfxUInt32 ffxPackBytesZeroY3ZeroX3(FfxUInt32x2 i) +{ + return ((i.x >> 24) & 0xffu) | ((i.y >> 8) & 0xff0000u); +} + +/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. +/// +/// The resulting integer will contain bytes in the following order, from most to least significant: +/// Y3 Y2 Y1 X0 +/// +/// @param [in] i The integer pair to pack. +/// +/// @returns +/// The packed integer value. +/// +/// @ingroup GPUCore +FfxUInt32 ffxPackBytesY3Y2Y1X0(FfxUInt32x2 i) +{ + return ((i.x) & 0x000000ffu) | (i.y & 0xffffff00u); +} + +/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. +/// +/// The resulting integer will contain bytes in the following order, from most to least significant: +/// Y3 Y2 Y1 X2 +/// +/// @param [in] i The integer pair to pack. +/// +/// @returns +/// The packed integer value. +/// +/// @ingroup GPUCore +FfxUInt32 ffxPackBytesY3Y2Y1X2(FfxUInt32x2 i) +{ + return ((i.x >> 16) & 0x000000ffu) | (i.y & 0xffffff00u); +} + +/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. +/// +/// The resulting integer will contain bytes in the following order, from most to least significant: +/// Y3 Y2 X0 Y0 +/// +/// @param [in] i The integer pair to pack. +/// +/// @returns +/// The packed integer value. +/// +/// @ingroup GPUCore +FfxUInt32 ffxPackBytesY3Y2X0Y0(FfxUInt32x2 i) +{ + return ((i.x << 8) & 0x0000ff00u) | (i.y & 0xffff00ffu); +} + +/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. +/// +/// The resulting integer will contain bytes in the following order, from most to least significant: +/// Y3 Y2 X2 Y0 +/// +/// @param [in] i The integer pair to pack. +/// +/// @returns +/// The packed integer value. +/// +/// @ingroup GPUCore +FfxUInt32 ffxPackBytesY3Y2X2Y0(FfxUInt32x2 i) +{ + return ((i.x >> 8) & 0x0000ff00u) | (i.y & 0xffff00ffu); +} + +/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. +/// +/// The resulting integer will contain bytes in the following order, from most to least significant: +/// Y3 X0 Y1 Y0 +/// +/// @param [in] i The integer pair to pack. +/// +/// @returns +/// The packed integer value. +/// +/// @ingroup GPUCore +FfxUInt32 ffxPackBytesY3X0Y1Y0(FfxUInt32x2 i) +{ + return ((i.x << 16) & 0x00ff0000u) | (i.y & 0xff00ffffu); +} + +/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. +/// +/// The resulting integer will contain bytes in the following order, from most to least significant: +/// Y3 X2 Y1 Y0 +/// +/// @param [in] i The integer pair to pack. +/// +/// @returns +/// The packed integer value. +/// +/// @ingroup GPUCore +FfxUInt32 ffxPackBytesY3X2Y1Y0(FfxUInt32x2 i) +{ + return ((i.x) & 0x00ff0000u) | (i.y & 0xff00ffffu); +} + +/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. +/// +/// The resulting integer will contain bytes in the following order, from most to least significant: +/// X0 Y2 Y1 Y0 +/// +/// @param [in] i The integer pair to pack. +/// +/// @returns +/// The packed integer value. +/// +/// @ingroup GPUCore +FfxUInt32 ffxPackBytesX0Y2Y1Y0(FfxUInt32x2 i) +{ + return ((i.x << 24) & 0xff000000u) | (i.y & 0x00ffffffu); +} + +/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. +/// +/// The resulting integer will contain bytes in the following order, from most to least significant: +/// X2 Y2 Y1 Y0 +/// +/// @param [in] i The integer pair to pack. +/// +/// @returns +/// The packed integer value. +/// +/// @ingroup GPUCore +FfxUInt32 ffxPackBytesX2Y2Y1Y0(FfxUInt32x2 i) +{ + return ((i.x << 8) & 0xff000000u) | (i.y & 0x00ffffffu); +} + +/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. +/// +/// The resulting integer will contain bytes in the following order, from most to least significant: +/// Y2 X2 Y0 X0 +/// +/// @param [in] i The integer pair to pack. +/// +/// @returns +/// The packed integer value. +/// +/// @ingroup GPUCore +FfxUInt32 ffxPackBytesY2X2Y0X0(FfxUInt32x2 i) +{ + return ((i.x) & 0x00ff00ffu) | ((i.y << 8) & 0xff00ff00u); +} + +/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. +/// +/// The resulting integer will contain bytes in the following order, from most to least significant: +/// Y2 Y0 X2 X0 +/// +/// @param [in] i The integer pair to pack. +/// +/// @returns +/// The packed integer value. +/// +/// @ingroup GPUCore +FfxUInt32 ffxPackBytesY2Y0X2X0(FfxUInt32x2 i) +{ + return (((i.x) & 0xffu) | ((i.x >> 8) & 0xff00u) | ((i.y << 16) & 0xff0000u) | ((i.y << 8) & 0xff000000u)); +} + +/// Takes two Float16x2 values x and y, normalizes them and builds a single Uint16x2 value in the format {{x0,y0},{x1,y1}}. +/// +/// @param [in] x The first float16x2 value to pack. +/// @param [in] y The second float16x2 value to pack. +/// +/// @returns +/// The packed FfxUInt32x2 value. +/// +/// @ingroup GPUCore +FfxUInt16x2 ffxPackX0Y0X1Y1UnsignedToUint16x2(FfxFloat16x2 x, FfxFloat16x2 y) +{ + x *= FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0); + y *= FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0); + return FFX_UINT32_TO_UINT16X2(ffxPackBytesY2X2Y0X0(FfxUInt32x2(FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(x)), FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(y))))); +} + +/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[0:7], +/// d.y[0:7] into r.y[0:7], i.x[8:15] into r.x[8:15], r.y[8:15] and i.y[0:15] into r.x[16:31], r.y[16:31] using 3 ops. +/// +/// r=ffxPermuteUByte0Float16x2ToUint2(d,i) +/// Where 'k0' is an SGPR with {1.0/32768.0} packed into the lower 16-bits +/// Where 'k1' is an SGPR with 0x???? +/// Where 'k2' is an SGPR with 0x???? +/// V_PK_FMA_F16 i,i,k0.x,0 +/// V_PERM_B32 r.x,i,i,k1 +/// V_PERM_B32 r.y,i,i,k2 +/// +/// @param [in] d The FfxUInt32x2 value to be packed. +/// @param [in] i The FfxFloat16x2 value to be packed. +/// +/// @returns +/// The packed FfxUInt32x2 value. +/// +/// @ingroup GPUCore +FfxUInt32x2 ffxPermuteUByte0Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) +{ + FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0))); + return FfxUInt32x2(ffxPackBytesY3Y2Y1X0(FfxUInt32x2(d.x, b)), ffxPackBytesY3Y2Y1X2(FfxUInt32x2(d.y, b))); +} + +/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[8:15], +/// d.y[0:7] into r.y[8:15], i.x[0:7] into r.x[0:7], r.y[0:7] and i.y[0:15] into r.x[16:31], r.y[16:31] using 3 ops. +/// +/// r=ffxPermuteUByte1Float16x2ToUint2(d,i) +/// Where 'k0' is an SGPR with {1.0/32768.0} packed into the lower 16-bits +/// Where 'k1' is an SGPR with 0x???? +/// Where 'k2' is an SGPR with 0x???? +/// V_PK_FMA_F16 i,i,k0.x,0 +/// V_PERM_B32 r.x,i,i,k1 +/// V_PERM_B32 r.y,i,i,k2 +/// +/// @param [in] d The FfxUInt32x2 value to be packed. +/// @param [in] i The FfxFloat16x2 value to be packed. +/// +/// @returns +/// The packed FfxUInt32x2 value. +/// +/// @ingroup GPUCore +FfxUInt32x2 ffxPermuteUByte1Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) +{ + FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0))); + return FfxUInt32x2(ffxPackBytesY3Y2X0Y0(FfxUInt32x2(d.x, b)), ffxPackBytesY3Y2X2Y0(FfxUInt32x2(d.y, b))); +} + +/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[16:23], +/// d.y[0:7] into r.y[16:23], i.x[0:15] into r.x[0:15], r.y[0:15] and i.y[8:15] into r.x[24:31], r.y[24:31] using 3 ops. +/// +/// r=ffxPermuteUByte2Float16x2ToUint2(d,i) +/// Where 'k0' is an SGPR with {1.0/32768.0} packed into the lower 16-bits +/// Where 'k1' is an SGPR with 0x???? +/// Where 'k2' is an SGPR with 0x???? +/// V_PK_FMA_F16 i,i,k0.x,0 +/// V_PERM_B32 r.x,i,i,k1 +/// V_PERM_B32 r.y,i,i,k2 +/// +/// @param [in] d The FfxUInt32x2 value to be packed. +/// @param [in] i The FfxFloat16x2 value to be packed. +/// +/// @returns +/// The packed FfxUInt32x2 value. +/// +/// @ingroup GPUCore +FfxUInt32x2 ffxPermuteUByte2Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) +{ + FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0))); + return FfxUInt32x2(ffxPackBytesY3X0Y1Y0(FfxUInt32x2(d.x, b)), ffxPackBytesY3X2Y1Y0(FfxUInt32x2(d.y, b))); +} + +/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[24:31], +/// d.y[0:7] into r.y[24:31], i.x[0:15] into r.x[0:15], r.y[0:15] and i.y[0:7] into r.x[16:23], r.y[16:23] using 3 ops. +/// +/// r=ffxPermuteUByte3Float16x2ToUint2(d,i) +/// Where 'k0' is an SGPR with {1.0/32768.0} packed into the lower 16-bits +/// Where 'k1' is an SGPR with 0x???? +/// Where 'k2' is an SGPR with 0x???? +/// V_PK_FMA_F16 i,i,k0.x,0 +/// V_PERM_B32 r.x,i,i,k1 +/// V_PERM_B32 r.y,i,i,k2 +/// +/// @param [in] d The FfxUInt32x2 value to be packed. +/// @param [in] i The FfxFloat16x2 value to be packed. +/// +/// @returns +/// The packed FfxUInt32x2 value. +/// +/// @ingroup GPUCore +FfxUInt32x2 ffxPermuteUByte3Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) +{ + FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0))); + return FfxUInt32x2(ffxPackBytesX0Y2Y1Y0(FfxUInt32x2(d.x, b)), ffxPackBytesX2Y2Y1Y0(FfxUInt32x2(d.y, b))); +} + +/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[0:7] into r.x[0:7] and i.y[0:7] into r.y[0:7] using 2 ops. +/// +/// @param [in] i The FfxUInt32x2 value to be unpacked. +/// +/// @returns +/// The unpacked FfxFloat16x2. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxPermuteUByte0Uint2ToFloat16x2(FfxUInt32x2 i) +{ + return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY0ZeroX0(i))) * FFX_BROADCAST_FLOAT16X2(32768.0); +} + +/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[8:15] into r.x[0:7] and i.y[8:15] into r.y[0:7] using 2 ops. +/// +/// @param [in] i The FfxUInt32x2 value to be unpacked. +/// +/// @returns +/// The unpacked FfxFloat16x2. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxPermuteUByte1Uint2ToFloat16x2(FfxUInt32x2 i) +{ + return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY1ZeroX1(i))) * FFX_BROADCAST_FLOAT16X2(32768.0); +} + +/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[16:23] into r.x[0:7] and i.y[16:23] into r.y[0:7] using 2 ops. +/// +/// @param [in] i The FfxUInt32x2 value to be unpacked. +/// +/// @returns +/// The unpacked FfxFloat16x2. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxPermuteUByte2Uint2ToFloat16x2(FfxUInt32x2 i) +{ + return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY2ZeroX2(i))) * FFX_BROADCAST_FLOAT16X2(32768.0); +} + +/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[24:31] into r.x[0:7] and i.y[24:31] into r.y[0:7] using 2 ops. +/// +/// @param [in] i The FfxUInt32x2 value to be unpacked. +/// +/// @returns +/// The unpacked FfxFloat16x2. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxPermuteUByte3Uint2ToFloat16x2(FfxUInt32x2 i) +{ + return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY3ZeroX3(i))) * FFX_BROADCAST_FLOAT16X2(32768.0); +} + +/// Takes two Float16x2 values x and y, normalizes them and builds a single Uint16x2 value in the format {{x0,y0},{x1,y1}}. +/// +/// @param [in] x The first float16x2 value to pack. +/// @param [in] y The second float16x2 value to pack. +/// +/// @returns +/// The packed FfxUInt32x2 value. +/// +/// @ingroup GPUCore +FfxUInt16x2 ffxPackX0Y0X1Y1SignedToUint16x2(FfxFloat16x2 x, FfxFloat16x2 y) +{ + x = x * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0); + y = y * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0); + return FFX_UINT32_TO_UINT16X2(ffxPackBytesY2X2Y0X0(FfxUInt32x2(FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(x)), FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(y))))); +} + +/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[0:7], +/// d.y[0:7] into r.y[0:7], i.x[8:15] into r.x[8:15], r.y[8:15] and i.y[0:15] into r.x[16:31], r.y[16:31] using 3 ops. +/// +/// Handles signed byte values. +/// +/// @param [in] d The FfxUInt32x2 value to be packed. +/// @param [in] i The FfxFloat16x2 value to be packed. +/// +/// @returns +/// The packed FfxUInt32x2 value. +/// +/// @ingroup GPUCore +FfxUInt32x2 ffxPermuteSByte0Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) +{ + FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))); + return FfxUInt32x2(ffxPackBytesY3Y2Y1X0(FfxUInt32x2(d.x, b)), ffxPackBytesY3Y2Y1X2(FfxUInt32x2(d.y, b))); +} + +/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[8:15], +/// d.y[0:7] into r.y[8:15], i.x[0:7] into r.x[0:7], r.y[0:7] and i.y[0:15] into r.x[16:31], r.y[16:31] using 3 ops. +/// +/// Handles signed byte values. +/// +/// @param [in] d The FfxUInt32x2 value to be packed. +/// @param [in] i The FfxFloat16x2 value to be packed. +/// +/// @returns +/// The packed FfxUInt32x2 value. +/// +/// @ingroup GPUCore +FfxUInt32x2 ffxPermuteSByte1Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) +{ + FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))); + return FfxUInt32x2(ffxPackBytesY3Y2X0Y0(FfxUInt32x2(d.x, b)), ffxPackBytesY3Y2X2Y0(FfxUInt32x2(d.y, b))); +} + +/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[16:23], +/// d.y[0:7] into r.y[16:23], i.x[0:15] into r.x[0:15], r.y[0:15] and i.y[8:15] into r.x[24:31], r.y[24:31] using 3 ops. +/// +/// Handles signed byte values. +/// +/// @param [in] d The FfxUInt32x2 value to be packed. +/// @param [in] i The FfxFloat16x2 value to be packed. +/// +/// @returns +/// The packed FfxUInt32x2 value. +/// +/// @ingroup GPUCore +FfxUInt32x2 ffxPermuteSByte2Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) +{ + FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))); + return FfxUInt32x2(ffxPackBytesY3X0Y1Y0(FfxUInt32x2(d.x, b)), ffxPackBytesY3X2Y1Y0(FfxUInt32x2(d.y, b))); +} + +/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[24:31], +/// d.y[0:7] into r.y[24:31], i.x[0:15] into r.x[0:15], r.y[0:15] and i.y[0:7] into r.x[16:23], r.y[16:23] using 3 ops. +/// +/// Handles signed byte values. +/// +/// @param [in] d The FfxUInt32x2 value to be packed. +/// @param [in] i The FfxFloat16x2 value to be packed. +/// +/// @returns +/// The packed FfxUInt32x2 value. +/// +/// @ingroup GPUCore +FfxUInt32x2 ffxPermuteSByte3Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) +{ + FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))); + return FfxUInt32x2(ffxPackBytesX0Y2Y1Y0(FfxUInt32x2(d.x, b)), ffxPackBytesX2Y2Y1Y0(FfxUInt32x2(d.y, b))); +} + +/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[0:7], +/// d.y[0:7] into r.y[0:7], i.x[8:15] into r.x[8:15], r.y[8:15] and i.y[0:15] into r.x[16:31], r.y[16:31] using 3 ops. +/// +/// Zero-based flips the MSB bit of the byte (making 128 "exact zero" actually zero). +/// This is useful if there is a desire for cleared values to decode as zero. +/// +/// Handles signed byte values. +/// +/// @param [in] d The FfxUInt32x2 value to be packed. +/// @param [in] i The FfxFloat16x2 value to be packed. +/// +/// @returns +/// The packed FfxUInt32x2 value. +/// +/// @ingroup GPUCore +FfxUInt32x2 ffxPermuteZeroBasedSByte0Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) +{ + FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))) ^ 0x00800080u; + return FfxUInt32x2(ffxPackBytesY3Y2Y1X0(FfxUInt32x2(d.x, b)), ffxPackBytesY3Y2Y1X2(FfxUInt32x2(d.y, b))); +} + +/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[8:15], +/// d.y[0:7] into r.y[8:15], i.x[0:7] into r.x[0:7], r.y[0:7] and i.y[0:15] into r.x[16:31], r.y[16:31] using 3 ops. +/// +/// Zero-based flips the MSB bit of the byte (making 128 "exact zero" actually zero). +/// This is useful if there is a desire for cleared values to decode as zero. +/// +/// Handles signed byte values. +/// +/// @param [in] d The FfxUInt32x2 value to be packed. +/// @param [in] i The FfxFloat16x2 value to be packed. +/// +/// @returns +/// The packed FfxUInt32x2 value. +/// +/// @ingroup GPUCore +FfxUInt32x2 ffxPermuteZeroBasedSByte1Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) +{ + FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))) ^ 0x00800080u; + return FfxUInt32x2(ffxPackBytesY3Y2X0Y0(FfxUInt32x2(d.x, b)), ffxPackBytesY3Y2X2Y0(FfxUInt32x2(d.y, b))); +} + +/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[16:23], +/// d.y[0:7] into r.y[16:23], i.x[0:15] into r.x[0:15], r.y[0:15] and i.y[8:15] into r.x[24:31], r.y[24:31] using 3 ops. +/// +/// Zero-based flips the MSB bit of the byte (making 128 "exact zero" actually zero). +/// This is useful if there is a desire for cleared values to decode as zero. +/// +/// Handles signed byte values. +/// +/// @param [in] d The FfxUInt32x2 value to be packed. +/// @param [in] i The FfxFloat16x2 value to be packed. +/// +/// @returns +/// The packed FfxUInt32x2 value. +/// +/// @ingroup GPUCore +FfxUInt32x2 ffxPermuteZeroBasedSByte2Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) +{ + FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))) ^ 0x00800080u; + return FfxUInt32x2(ffxPackBytesY3X0Y1Y0(FfxUInt32x2(d.x, b)), ffxPackBytesY3X2Y1Y0(FfxUInt32x2(d.y, b))); +} + +/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[24:31], +/// d.y[0:7] into r.y[24:31], i.x[0:15] into r.x[0:15], r.y[0:15] and i.y[0:7] into r.x[16:23], r.y[16:23] using 3 ops. +/// +/// Zero-based flips the MSB bit of the byte (making 128 "exact zero" actually zero). +/// This is useful if there is a desire for cleared values to decode as zero. +/// +/// Handles signed byte values. +/// +/// @param [in] d The FfxUInt32x2 value to be packed. +/// @param [in] i The FfxFloat16x2 value to be packed. +/// +/// @returns +/// The packed FfxUInt32x2 value. +/// +/// @ingroup GPUCore +FfxUInt32x2 ffxPermuteZeroBasedSByte3Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) +{ + FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))) ^ 0x00800080u; + return FfxUInt32x2(ffxPackBytesX0Y2Y1Y0(FfxUInt32x2(d.x, b)), ffxPackBytesX2Y2Y1Y0(FfxUInt32x2(d.y, b))); +} + +/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[0:7] into r.x[0:7] and i.y[0:7] into r.y[0:7] using 2 ops. +/// +/// Handles signed byte values. +/// +/// @param [in] i The FfxUInt32x2 value to be unpacked. +/// +/// @returns +/// The unpacked FfxFloat16x2. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxPermuteSByte0Uint2ToFloat16x2(FfxUInt32x2 i) +{ + return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY0ZeroX0(i))) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); +} + +/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[8:15] into r.x[0:7] and i.y[8:15] into r.y[0:7] using 2 ops. +/// +/// Handles signed byte values. +/// +/// @param [in] i The FfxUInt32x2 value to be unpacked. +/// +/// @returns +/// The unpacked FfxFloat16x2. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxPermuteSByte1Uint2ToFloat16x2(FfxUInt32x2 i) +{ + return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY1ZeroX1(i))) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); +} + +/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[16:23] into r.x[0:7] and i.y[16:23] into r.y[0:7] using 2 ops. +/// +/// Handles signed byte values. +/// +/// @param [in] i The FfxUInt32x2 value to be unpacked. +/// +/// @returns +/// The unpacked FfxFloat16x2. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxPermuteSByte2Uint2ToFloat16x2(FfxUInt32x2 i) +{ + return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY2ZeroX2(i))) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); +} + +/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[24:31] into r.x[0:7] and i.y[24:31] into r.y[0:7] using 2 ops. +/// +/// Handles signed byte values. +/// +/// @param [in] i The FfxUInt32x2 value to be unpacked. +/// +/// @returns +/// The unpacked FfxFloat16x2. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxPermuteSByte3Uint2ToFloat16x2(FfxUInt32x2 i) +{ + return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY3ZeroX3(i))) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); +} + +/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[0:7] into r.x[0:7] and i.y[0:7] into r.y[0:7] using 2 ops. +/// +/// Handles signed byte values. +/// +/// @param [in] i The FfxUInt32x2 value to be unpacked. +/// +/// @returns +/// The unpacked FfxFloat16x2. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxPermuteZeroBasedSByte0Uint2ToFloat16x2(FfxUInt32x2 i) +{ + return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY0ZeroX0(i) ^ 0x00800080u)) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); +} + +/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[8:15] into r.x[0:7] and i.y[8:15] into r.y[0:7] using 2 ops. +/// +/// Handles signed byte values. +/// +/// @param [in] i The FfxUInt32x2 value to be unpacked. +/// +/// @returns +/// The unpacked FfxFloat16x2. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxPermuteZeroBasedSByte1Uint2ToFloat16x2(FfxUInt32x2 i) +{ + return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY1ZeroX1(i) ^ 0x00800080u)) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); +} + +/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[16:23] into r.x[0:7] and i.y[16:23] into r.y[0:7] using 2 ops. +/// +/// Handles signed byte values. +/// +/// @param [in] i The FfxUInt32x2 value to be unpacked. +/// +/// @returns +/// The unpacked FfxFloat16x2. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxPermuteZeroBasedSByte2Uint2ToFloat16x2(FfxUInt32x2 i) +{ + return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY2ZeroX2(i) ^ 0x00800080u)) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); +} + +/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[24:31] into r.x[0:7] and i.y[24:31] into r.y[0:7] using 2 ops. +/// +/// Handles signed byte values. +/// +/// @param [in] i The FfxUInt32x2 value to be unpacked. +/// +/// @returns +/// The unpacked FfxFloat16x2. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxPermuteZeroBasedSByte3Uint2ToFloat16x2(FfxUInt32x2 i) +{ + return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY3ZeroX3(i) ^ 0x00800080u)) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); +} + +/// Calculate a half-precision low-quality approximation for the square root of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] a The value to calculate an approximate to the square root for. +/// +/// @returns +/// An approximation of the square root, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat16 ffxApproximateSqrtHalf(FfxFloat16 a) +{ + return FFX_TO_FLOAT16((FFX_TO_UINT16(a) >> FFX_BROADCAST_UINT16(1)) + FFX_BROADCAST_UINT16(0x1de2)); +} + +/// Calculate a half-precision low-quality approximation for the square root of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] a The value to calculate an approximate to the square root for. +/// +/// @returns +/// An approximation of the square root, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxApproximateSqrtHalf(FfxFloat16x2 a) +{ + return FFX_TO_FLOAT16X2((FFX_TO_UINT16X2(a) >> FFX_BROADCAST_UINT16X2(1)) + FFX_BROADCAST_UINT16X2(0x1de2)); +} + +/// Calculate a half-precision low-quality approximation for the square root of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] a The value to calculate an approximate to the square root for. +/// +/// @returns +/// An approximation of the square root, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxApproximateSqrtHalf(FfxFloat16x3 a) +{ + return FFX_TO_FLOAT16X3((FFX_TO_UINT16X3(a) >> FFX_BROADCAST_UINT16X3(1)) + FFX_BROADCAST_UINT16X3(0x1de2)); +} + +/// Calculate a half-precision low-quality approximation for the reciprocal of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] a The value to calculate an approximate to the reciprocal for. +/// +/// @returns +/// An approximation of the reciprocal, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat16 ffxApproximateReciprocalHalf(FfxFloat16 a) +{ + return FFX_TO_FLOAT16(FFX_BROADCAST_UINT16(0x7784) - FFX_TO_UINT16(a)); +} + +/// Calculate a half-precision low-quality approximation for the reciprocal of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] a The value to calculate an approximate to the reciprocal for. +/// +/// @returns +/// An approximation of the reciprocal, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxApproximateReciprocalHalf(FfxFloat16x2 a) +{ + return FFX_TO_FLOAT16X2(FFX_BROADCAST_UINT16X2(0x7784) - FFX_TO_UINT16X2(a)); +} + +/// Calculate a half-precision low-quality approximation for the reciprocal of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] a The value to calculate an approximate to the reciprocal for. +/// +/// @returns +/// An approximation of the reciprocal, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxApproximateReciprocalHalf(FfxFloat16x3 a) +{ + return FFX_TO_FLOAT16X3(FFX_BROADCAST_UINT16X3(0x7784) - FFX_TO_UINT16X3(a)); +} + +/// Calculate a half-precision low-quality approximation for the reciprocal of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] a The value to calculate an approximate to the reciprocal for. +/// +/// @returns +/// An approximation of the reciprocal, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat16x4 ffxApproximateReciprocalHalf(FfxFloat16x4 a) +{ + return FFX_TO_FLOAT16X4(FFX_BROADCAST_UINT16X4(0x7784) - FFX_TO_UINT16X4(a)); +} + +/// Calculate a half-precision medium-quality approximation for the reciprocal of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] a The value to calculate an approximate to the reciprocal for. +/// +/// @returns +/// An approximation of the reciprocal, estimated to medium quality. +/// +/// @ingroup GPUCore +FfxFloat16 ffxApproximateReciprocalMediumHalf(FfxFloat16 a) +{ + FfxFloat16 b = FFX_TO_FLOAT16(FFX_BROADCAST_UINT16(0x778d) - FFX_TO_UINT16(a)); + return b * (-b * a + FFX_BROADCAST_FLOAT16(2.0)); +} + +/// Calculate a half-precision medium-quality approximation for the reciprocal of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] a The value to calculate an approximate to the reciprocal for. +/// +/// @returns +/// An approximation of the reciprocal, estimated to medium quality. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxApproximateReciprocalMediumHalf(FfxFloat16x2 a) +{ + FfxFloat16x2 b = FFX_TO_FLOAT16X2(FFX_BROADCAST_UINT16X2(0x778d) - FFX_TO_UINT16X2(a)); + return b * (-b * a + FFX_BROADCAST_FLOAT16X2(2.0)); +} + +/// Calculate a half-precision medium-quality approximation for the reciprocal of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] a The value to calculate an approximate to the reciprocal for. +/// +/// @returns +/// An approximation of the reciprocal, estimated to medium quality. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxApproximateReciprocalMediumHalf(FfxFloat16x3 a) +{ + FfxFloat16x3 b = FFX_TO_FLOAT16X3(FFX_BROADCAST_UINT16X3(0x778d) - FFX_TO_UINT16X3(a)); + return b * (-b * a + FFX_BROADCAST_FLOAT16X3(2.0)); +} + +/// Calculate a half-precision medium-quality approximation for the reciprocal of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] a The value to calculate an approximate to the reciprocal for. +/// +/// @returns +/// An approximation of the reciprocal, estimated to medium quality. +/// +/// @ingroup GPUCore +FfxFloat16x4 ffxApproximateReciprocalMediumHalf(FfxFloat16x4 a) +{ + FfxFloat16x4 b = FFX_TO_FLOAT16X4(FFX_BROADCAST_UINT16X4(0x778d) - FFX_TO_UINT16X4(a)); + return b * (-b * a + FFX_BROADCAST_FLOAT16X4(2.0)); +} + +/// Calculate a half-precision low-quality approximation for the reciprocal of the square root of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] a The value to calculate an approximate to the reciprocal of the square root for. +/// +/// @returns +/// An approximation of the reciprocal of the square root, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat16 ffxApproximateReciprocalSquareRootHalf(FfxFloat16 a) +{ + return FFX_TO_FLOAT16(FFX_BROADCAST_UINT16(0x59a3) - (FFX_TO_UINT16(a) >> FFX_BROADCAST_UINT16(1))); +} + +/// Calculate a half-precision low-quality approximation for the reciprocal of the square root of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] a The value to calculate an approximate to the reciprocal of the square root for. +/// +/// @returns +/// An approximation of the reciprocal of the square root, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxApproximateReciprocalSquareRootHalf(FfxFloat16x2 a) +{ + return FFX_TO_FLOAT16X2(FFX_BROADCAST_UINT16X2(0x59a3) - (FFX_TO_UINT16X2(a) >> FFX_BROADCAST_UINT16X2(1))); +} + +/// Calculate a half-precision low-quality approximation for the reciprocal of the square root of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] a The value to calculate an approximate to the reciprocal of the square root for. +/// +/// @returns +/// An approximation of the reciprocal of the square root, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxApproximateReciprocalSquareRootHalf(FfxFloat16x3 a) +{ + return FFX_TO_FLOAT16X3(FFX_BROADCAST_UINT16X3(0x59a3) - (FFX_TO_UINT16X3(a) >> FFX_BROADCAST_UINT16X3(1))); +} + +/// Calculate a half-precision low-quality approximation for the reciprocal of the square root of a value. +/// +/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent +/// presentation materials: +/// +/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf +/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h +/// +/// @param [in] a The value to calculate an approximate to the reciprocal of the square root for. +/// +/// @returns +/// An approximation of the reciprocal of the square root, estimated to low quality. +/// +/// @ingroup GPUCore +FfxFloat16x4 ffxApproximateReciprocalSquareRootHalf(FfxFloat16x4 a) +{ + return FFX_TO_FLOAT16X4(FFX_BROADCAST_UINT16X4(0x59a3) - (FFX_TO_UINT16X4(a) >> FFX_BROADCAST_UINT16X4(1))); +} + +/// An approximation of sine. +/// +/// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range +/// is {-1/4 to 1/4} representing {-1 to 1}. +/// +/// @param [in] x The value to calculate approximate sine for. +/// +/// @returns +/// The approximate sine of value. +FfxFloat16 ffxParabolicSinHalf(FfxFloat16 x) +{ + return x * abs(x) - x; +} + +/// An approximation of sine. +/// +/// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range +/// is {-1/4 to 1/4} representing {-1 to 1}. +/// +/// @param [in] x The value to calculate approximate sine for. +/// +/// @returns +/// The approximate sine of value. +FfxFloat16x2 ffxParabolicSinHalf(FfxFloat16x2 x) +{ + return x * abs(x) - x; +} + +/// An approximation of cosine. +/// +/// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range +/// is {-1/4 to 1/4} representing {-1 to 1}. +/// +/// @param [in] x The value to calculate approximate cosine for. +/// +/// @returns +/// The approximate cosine of value. +FfxFloat16 ffxParabolicCosHalf(FfxFloat16 x) +{ + x = ffxFract(x * FFX_BROADCAST_FLOAT16(0.5) + FFX_BROADCAST_FLOAT16(0.75)); + x = x * FFX_BROADCAST_FLOAT16(2.0) - FFX_BROADCAST_FLOAT16(1.0); + return ffxParabolicSinHalf(x); +} + +/// An approximation of cosine. +/// +/// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range +/// is {-1/4 to 1/4} representing {-1 to 1}. +/// +/// @param [in] x The value to calculate approximate cosine for. +/// +/// @returns +/// The approximate cosine of value. +FfxFloat16x2 ffxParabolicCosHalf(FfxFloat16x2 x) +{ + x = ffxFract(x * FFX_BROADCAST_FLOAT16X2(0.5) + FFX_BROADCAST_FLOAT16X2(0.75)); + x = x * FFX_BROADCAST_FLOAT16X2(2.0) - FFX_BROADCAST_FLOAT16X2(1.0); + return ffxParabolicSinHalf(x); +} + +/// An approximation of both sine and cosine. +/// +/// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range +/// is {-1/4 to 1/4} representing {-1 to 1}. +/// +/// @param [in] x The value to calculate approximate cosine for. +/// +/// @returns +/// A FfxFloat32x2 containing approximations of both sine and cosine of value. +FfxFloat16x2 ffxParabolicSinCosHalf(FfxFloat16 x) +{ + FfxFloat16 y = ffxFract(x * FFX_BROADCAST_FLOAT16(0.5) + FFX_BROADCAST_FLOAT16(0.75)); + y = y * FFX_BROADCAST_FLOAT16(2.0) - FFX_BROADCAST_FLOAT16(1.0); + return ffxParabolicSinHalf(FfxFloat16x2(x, y)); +} + +/// Conditional free logic AND operation using two half-precision values. +/// +/// @param [in] x The first value to be fed into the AND operator. +/// @param [in] y The second value to be fed into the AND operator. +/// +/// @returns +/// Result of the AND operation. +/// +/// @ingroup GPUCore +FfxUInt16 ffxZeroOneAndHalf(FfxUInt16 x, FfxUInt16 y) +{ + return min(x, y); +} + +/// Conditional free logic AND operation using two half-precision values. +/// +/// @param [in] x The first value to be fed into the AND operator. +/// @param [in] y The second value to be fed into the AND operator. +/// +/// @returns +/// Result of the AND operation. +/// +/// @ingroup GPUCore +FfxUInt16x2 ffxZeroOneAndHalf(FfxUInt16x2 x, FfxUInt16x2 y) +{ + return min(x, y); +} + +/// Conditional free logic AND operation using two half-precision values. +/// +/// @param [in] x The first value to be fed into the AND operator. +/// @param [in] y The second value to be fed into the AND operator. +/// +/// @returns +/// Result of the AND operation. +/// +/// @ingroup GPUCore +FfxUInt16x3 ffxZeroOneAndHalf(FfxUInt16x3 x, FfxUInt16x3 y) +{ + return min(x, y); +} + +/// Conditional free logic AND operation using two half-precision values. +/// +/// @param [in] x The first value to be fed into the AND operator. +/// @param [in] y The second value to be fed into the AND operator. +/// +/// @returns +/// Result of the AND operation. +/// +/// @ingroup GPUCore +FfxUInt16x4 ffxZeroOneAndHalf(FfxUInt16x4 x, FfxUInt16x4 y) +{ + return min(x, y); +} + +/// Conditional free logic NOT operation using two half-precision values. +/// +/// @param [in] x The first value to be fed into the NOT operator. +/// @param [in] y The second value to be fed into the NOT operator. +/// +/// @returns +/// Result of the NOT operation. +/// +/// @ingroup GPUCore +FfxUInt16 ffxZeroOneNotHalf(FfxUInt16 x) +{ + return x ^ FFX_BROADCAST_UINT16(1); +} + +/// Conditional free logic NOT operation using two half-precision values. +/// +/// @param [in] x The first value to be fed into the NOT operator. +/// @param [in] y The second value to be fed into the NOT operator. +/// +/// @returns +/// Result of the NOT operation. +/// +/// @ingroup GPUCore +FfxUInt16x2 ffxZeroOneNotHalf(FfxUInt16x2 x) +{ + return x ^ FFX_BROADCAST_UINT16X2(1); +} + +/// Conditional free logic NOT operation using two half-precision values. +/// +/// @param [in] x The first value to be fed into the NOT operator. +/// @param [in] y The second value to be fed into the NOT operator. +/// +/// @returns +/// Result of the NOT operation. +/// +/// @ingroup GPUCore +FfxUInt16x3 ffxZeroOneNotHalf(FfxUInt16x3 x) +{ + return x ^ FFX_BROADCAST_UINT16X3(1); +} + +/// Conditional free logic NOT operation using two half-precision values. +/// +/// @param [in] x The first value to be fed into the NOT operator. +/// @param [in] y The second value to be fed into the NOT operator. +/// +/// @returns +/// Result of the NOT operation. +/// +/// @ingroup GPUCore +FfxUInt16x4 ffxZeroOneNotHalf(FfxUInt16x4 x) +{ + return x ^ FFX_BROADCAST_UINT16X4(1); +} + +/// Conditional free logic OR operation using two half-precision values. +/// +/// @param [in] x The first value to be fed into the OR operator. +/// @param [in] y The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the OR operation. +/// +/// @ingroup GPUCore +FfxUInt16 ffxZeroOneOrHalf(FfxUInt16 x, FfxUInt16 y) +{ + return max(x, y); +} + +/// Conditional free logic OR operation using two half-precision values. +/// +/// @param [in] x The first value to be fed into the OR operator. +/// @param [in] y The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the OR operation. +/// +/// @ingroup GPUCore +FfxUInt16x2 ffxZeroOneOrHalf(FfxUInt16x2 x, FfxUInt16x2 y) +{ + return max(x, y); +} + +/// Conditional free logic OR operation using two half-precision values. +/// +/// @param [in] x The first value to be fed into the OR operator. +/// @param [in] y The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the OR operation. +/// +/// @ingroup GPUCore +FfxUInt16x3 ffxZeroOneOrHalf(FfxUInt16x3 x, FfxUInt16x3 y) +{ + return max(x, y); +} + +/// Conditional free logic OR operation using two half-precision values. +/// +/// @param [in] x The first value to be fed into the OR operator. +/// @param [in] y The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the OR operation. +/// +/// @ingroup GPUCore +FfxUInt16x4 ffxZeroOneOrHalf(FfxUInt16x4 x, FfxUInt16x4 y) +{ + return max(x, y); +} + +/// Convert a half-precision FfxFloat32 value between 0.0f and 1.0f to a half-precision Uint. +/// +/// @param [in] x The value to converted to a Uint. +/// +/// @returns +/// The converted Uint value. +/// +/// @ingroup GPUCore +FfxUInt16 ffxZeroOneFloat16ToUint16(FfxFloat16 x) +{ + return FFX_TO_UINT16(x * FFX_TO_FLOAT16(FFX_TO_UINT16(1))); +} + +/// Convert a half-precision FfxFloat32 value between 0.0f and 1.0f to a half-precision Uint. +/// +/// @param [in] x The value to converted to a Uint. +/// +/// @returns +/// The converted Uint value. +/// +/// @ingroup GPUCore +FfxUInt16x2 ffxZeroOneFloat16x2ToUint16x2(FfxFloat16x2 x) +{ + return FFX_TO_UINT16X2(x * FFX_TO_FLOAT16X2(FfxUInt16x2(1, 1))); +} + +/// Convert a half-precision FfxFloat32 value between 0.0f and 1.0f to a half-precision Uint. +/// +/// @param [in] x The value to converted to a Uint. +/// +/// @returns +/// The converted Uint value. +/// +/// @ingroup GPUCore +FfxUInt16x3 ffxZeroOneFloat16x3ToUint16x3(FfxFloat16x3 x) +{ + return FFX_TO_UINT16X3(x * FFX_TO_FLOAT16X3(FfxUInt16x3(1, 1, 1))); +} + +/// Convert a half-precision FfxFloat32 value between 0.0f and 1.0f to a half-precision Uint. +/// +/// @param [in] x The value to converted to a Uint. +/// +/// @returns +/// The converted Uint value. +/// +/// @ingroup GPUCore +FfxUInt16x4 ffxZeroOneFloat16x4ToUint16x4(FfxFloat16x4 x) +{ + return FFX_TO_UINT16X4(x * FFX_TO_FLOAT16X4(FfxUInt16x4(1, 1, 1, 1))); +} + +/// Convert a half-precision FfxUInt32 value between 0 and 1 to a half-precision FfxFloat32. +/// +/// @param [in] x The value to converted to a half-precision FfxFloat32. +/// +/// @returns +/// The converted half-precision FfxFloat32 value. +/// +/// @ingroup GPUCore +FfxFloat16 ffxZeroOneUint16ToFloat16(FfxUInt16 x) +{ + return FFX_TO_FLOAT16(x * FFX_TO_UINT16(FFX_TO_FLOAT16(1.0))); +} + +/// Convert a half-precision FfxUInt32 value between 0 and 1 to a half-precision FfxFloat32. +/// +/// @param [in] x The value to converted to a half-precision FfxFloat32. +/// +/// @returns +/// The converted half-precision FfxFloat32 value. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxZeroOneUint16x2ToFloat16x2(FfxUInt16x2 x) +{ + return FFX_TO_FLOAT16X2(x * FFX_TO_UINT16X2(FfxUInt16x2(FFX_TO_FLOAT16(1.0), FFX_TO_FLOAT16(1.0)))); +} + +/// Convert a half-precision FfxUInt32 value between 0 and 1 to a half-precision FfxFloat32. +/// +/// @param [in] x The value to converted to a half-precision FfxFloat32. +/// +/// @returns +/// The converted half-precision FfxFloat32 value. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxZeroOneUint16x3ToFloat16x3(FfxUInt16x3 x) +{ + return FFX_TO_FLOAT16X3(x * FFX_TO_UINT16X3(FfxUInt16x3(FFX_TO_FLOAT16(1.0), FFX_TO_FLOAT16(1.0), FFX_TO_FLOAT16(1.0)))); +} + +/// Convert a half-precision FfxUInt32 value between 0 and 1 to a half-precision FfxFloat32. +/// +/// @param [in] x The value to converted to a half-precision FfxFloat32. +/// +/// @returns +/// The converted half-precision FfxFloat32 value. +/// +/// @ingroup GPUCore +FfxFloat16x4 ffxZeroOneUint16x4ToFloat16x4(FfxUInt16x4 x) +{ + return FFX_TO_FLOAT16X4(x * FFX_TO_UINT16X4(FfxUInt16x4(FFX_TO_FLOAT16(1.0), FFX_TO_FLOAT16(1.0), FFX_TO_FLOAT16(1.0), FFX_TO_FLOAT16(1.0)))); +} + +/// Conditional free logic AND operation using two half-precision values. +/// +/// @param [in] x The first value to be fed into the AND operator. +/// @param [in] y The second value to be fed into the AND operator. +/// +/// @returns +/// Result of the AND operation. +/// +/// @ingroup GPUCore +FfxFloat16 ffxZeroOneAndHalf(FfxFloat16 x, FfxFloat16 y) +{ + return min(x, y); +} + +/// Conditional free logic AND operation using two half-precision values. +/// +/// @param [in] x The first value to be fed into the AND operator. +/// @param [in] y The second value to be fed into the AND operator. +/// +/// @returns +/// Result of the AND operation. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxZeroOneAndHalf(FfxFloat16x2 x, FfxFloat16x2 y) +{ + return min(x, y); +} + +/// Conditional free logic AND operation using two half-precision values. +/// +/// @param [in] x The first value to be fed into the AND operator. +/// @param [in] y The second value to be fed into the AND operator. +/// +/// @returns +/// Result of the AND operation. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxZeroOneAndHalf(FfxFloat16x3 x, FfxFloat16x3 y) +{ + return min(x, y); +} + +/// Conditional free logic AND operation using two half-precision values. +/// +/// @param [in] x The first value to be fed into the AND operator. +/// @param [in] y The second value to be fed into the AND operator. +/// +/// @returns +/// Result of the AND operation. +/// +/// @ingroup GPUCore +FfxFloat16x4 ffxZeroOneAndHalf(FfxFloat16x4 x, FfxFloat16x4 y) +{ + return min(x, y); +} + +/// Conditional free logic AND NOT operation using two half-precision values. +/// +/// @param [in] x The first value to be fed into the AND NOT operator. +/// @param [in] y The second value to be fed into the AND NOT operator. +/// +/// @returns +/// Result of the AND NOT operation. +/// +/// @ingroup GPUCore +FfxFloat16 ffxSignedZeroOneAndOrHalf(FfxFloat16 x, FfxFloat16 y) +{ + return (-x) * y + FFX_BROADCAST_FLOAT16(1.0); +} + +/// Conditional free logic AND NOT operation using two half-precision values. +/// +/// @param [in] x The first value to be fed into the AND NOT operator. +/// @param [in] y The second value to be fed into the AND NOT operator. +/// +/// @returns +/// Result of the AND NOT operation. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxSignedZeroOneAndOrHalf(FfxFloat16x2 x, FfxFloat16x2 y) +{ + return (-x) * y + FFX_BROADCAST_FLOAT16X2(1.0); +} + +/// Conditional free logic AND NOT operation using two half-precision values. +/// +/// @param [in] x The first value to be fed into the AND NOT operator. +/// @param [in] y The second value to be fed into the AND NOT operator. +/// +/// @returns +/// Result of the AND NOT operation. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxSignedZeroOneAndOrHalf(FfxFloat16x3 x, FfxFloat16x3 y) +{ + return (-x) * y + FFX_BROADCAST_FLOAT16X3(1.0); +} + +/// Conditional free logic AND NOT operation using two half-precision values. +/// +/// @param [in] x The first value to be fed into the AND NOT operator. +/// @param [in] y The second value to be fed into the AND NOT operator. +/// +/// @returns +/// Result of the AND NOT operation. +/// +/// @ingroup GPUCore +FfxFloat16x4 ffxSignedZeroOneAndOrHalf(FfxFloat16x4 x, FfxFloat16x4 y) +{ + return (-x) * y + FFX_BROADCAST_FLOAT16X4(1.0); +} + +/// Conditional free logic AND operation using two half-precision values followed by +/// a NOT operation using the resulting value and a third half-precision value. +/// +/// @param [in] x The first value to be fed into the AND operator. +/// @param [in] y The second value to be fed into the AND operator. +/// @param [in] z The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the AND OR operation. +/// +/// @ingroup GPUCore +FfxFloat16 ffxZeroOneAndOrHalf(FfxFloat16 x, FfxFloat16 y, FfxFloat16 z) +{ + return ffxSaturate(x * y + z); +} + +/// Conditional free logic AND operation using two half-precision values followed by +/// a NOT operation using the resulting value and a third half-precision value. +/// +/// @param [in] x The first value to be fed into the AND operator. +/// @param [in] y The second value to be fed into the AND operator. +/// @param [in] z The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the AND OR operation. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxZeroOneAndOrHalf(FfxFloat16x2 x, FfxFloat16x2 y, FfxFloat16x2 z) +{ + return ffxSaturate(x * y + z); +} + +/// Conditional free logic AND operation using two half-precision values followed by +/// a NOT operation using the resulting value and a third half-precision value. +/// +/// @param [in] x The first value to be fed into the AND operator. +/// @param [in] y The second value to be fed into the AND operator. +/// @param [in] z The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the AND OR operation. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxZeroOneAndOrHalf(FfxFloat16x3 x, FfxFloat16x3 y, FfxFloat16x3 z) +{ + return ffxSaturate(x * y + z); +} + +/// Conditional free logic AND operation using two half-precision values followed by +/// a NOT operation using the resulting value and a third half-precision value. +/// +/// @param [in] x The first value to be fed into the AND operator. +/// @param [in] y The second value to be fed into the AND operator. +/// @param [in] z The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the AND OR operation. +/// +/// @ingroup GPUCore +FfxFloat16x4 ffxZeroOneAndOrHalf(FfxFloat16x4 x, FfxFloat16x4 y, FfxFloat16x4 z) +{ + return ffxSaturate(x * y + z); +} + +/// Given a half-precision value, returns 1.0 if greater than zero and 0.0 if not. +/// +/// @param [in] x The value to be compared. +/// +/// @returns +/// Result of the greater than zero comparison. +/// +/// @ingroup GPUCore +FfxFloat16 ffxZeroOneIsGreaterThanZeroHalf(FfxFloat16 x) +{ + return ffxSaturate(x * FFX_BROADCAST_FLOAT16(FFX_POSITIVE_INFINITY_HALF)); +} + +/// Given a half-precision value, returns 1.0 if greater than zero and 0.0 if not. +/// +/// @param [in] x The value to be compared. +/// +/// @returns +/// Result of the greater than zero comparison. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxZeroOneIsGreaterThanZeroHalf(FfxFloat16x2 x) +{ + return ffxSaturate(x * FFX_BROADCAST_FLOAT16X2(FFX_POSITIVE_INFINITY_HALF)); +} + +/// Given a half-precision value, returns 1.0 if greater than zero and 0.0 if not. +/// +/// @param [in] x The value to be compared. +/// +/// @returns +/// Result of the greater than zero comparison. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxZeroOneIsGreaterThanZeroHalf(FfxFloat16x3 x) +{ + return ffxSaturate(x * FFX_BROADCAST_FLOAT16X3(FFX_POSITIVE_INFINITY_HALF)); +} + +/// Given a half-precision value, returns 1.0 if greater than zero and 0.0 if not. +/// +/// @param [in] x The value to be compared. +/// +/// @returns +/// Result of the greater than zero comparison. +/// +/// @ingroup GPUCore +FfxFloat16x4 ffxZeroOneIsGreaterThanZeroHalf(FfxFloat16x4 x) +{ + return ffxSaturate(x * FFX_BROADCAST_FLOAT16X4(FFX_POSITIVE_INFINITY_HALF)); +} + +/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. +/// +/// @param [in] x The first value to be fed into the AND OR operator. +/// +/// @returns +/// Result of the AND OR operation. +/// +/// @ingroup GPUCore +FfxFloat16 ffxZeroOneNotHalf(FfxFloat16 x) +{ + return FFX_BROADCAST_FLOAT16(1.0) - x; +} + +/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. +/// +/// @param [in] x The first value to be fed into the AND OR operator. +/// +/// @returns +/// Result of the AND OR operation. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxZeroOneNotHalf(FfxFloat16x2 x) +{ + return FFX_BROADCAST_FLOAT16X2(1.0) - x; +} + +/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. +/// +/// @param [in] x The first value to be fed into the AND OR operator. +/// +/// @returns +/// Result of the AND OR operation. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxZeroOneNotHalf(FfxFloat16x3 x) +{ + return FFX_BROADCAST_FLOAT16X3(1.0) - x; +} + +/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. +/// +/// @param [in] x The first value to be fed into the AND OR operator. +/// +/// @returns +/// Result of the AND OR operation. +/// +/// @ingroup GPUCore +FfxFloat16x4 ffxZeroOneNotHalf(FfxFloat16x4 x) +{ + return FFX_BROADCAST_FLOAT16X4(1.0) - x; +} + +/// Conditional free logic OR operation using two half-precision FfxFloat32 values. +/// +/// @param [in] x The first value to be fed into the OR operator. +/// @param [in] y The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the OR operation. +/// +/// @ingroup GPUCore +FfxFloat16 ffxZeroOneOrHalf(FfxFloat16 x, FfxFloat16 y) +{ + return max(x, y); +} + +/// Conditional free logic OR operation using two half-precision FfxFloat32 values. +/// +/// @param [in] x The first value to be fed into the OR operator. +/// @param [in] y The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the OR operation. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxZeroOneOrHalf(FfxFloat16x2 x, FfxFloat16x2 y) +{ + return max(x, y); +} + +/// Conditional free logic OR operation using two half-precision FfxFloat32 values. +/// +/// @param [in] x The first value to be fed into the OR operator. +/// @param [in] y The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the OR operation. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxZeroOneOrHalf(FfxFloat16x3 x, FfxFloat16x3 y) +{ + return max(x, y); +} + +/// Conditional free logic OR operation using two half-precision FfxFloat32 values. +/// +/// @param [in] x The first value to be fed into the OR operator. +/// @param [in] y The second value to be fed into the OR operator. +/// +/// @returns +/// Result of the OR operation. +/// +/// @ingroup GPUCore +FfxFloat16x4 ffxZeroOneOrHalf(FfxFloat16x4 x, FfxFloat16x4 y) +{ + return max(x, y); +} + +/// Choose between two half-precision FfxFloat32 values if the first paramter is greater than zero. +/// +/// @param [in] x The value to compare against zero. +/// @param [in] y The value to return if the comparision is greater than zero. +/// @param [in] z The value to return if the comparision is less than or equal to zero. +/// +/// @returns +/// The selected value. +/// +/// @ingroup GPUCore +FfxFloat16 ffxZeroOneSelectHalf(FfxFloat16 x, FfxFloat16 y, FfxFloat16 z) +{ + FfxFloat16 r = (-x) * z + z; + return x * y + r; +} + +/// Choose between two half-precision FfxFloat32 values if the first paramter is greater than zero. +/// +/// @param [in] x The value to compare against zero. +/// @param [in] y The value to return if the comparision is greater than zero. +/// @param [in] z The value to return if the comparision is less than or equal to zero. +/// +/// @returns +/// The selected value. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxZeroOneSelectHalf(FfxFloat16x2 x, FfxFloat16x2 y, FfxFloat16x2 z) +{ + FfxFloat16x2 r = (-x) * z + z; + return x * y + r; +} + +/// Choose between two half-precision FfxFloat32 values if the first paramter is greater than zero. +/// +/// @param [in] x The value to compare against zero. +/// @param [in] y The value to return if the comparision is greater than zero. +/// @param [in] z The value to return if the comparision is less than or equal to zero. +/// +/// @returns +/// The selected value. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxZeroOneSelectHalf(FfxFloat16x3 x, FfxFloat16x3 y, FfxFloat16x3 z) +{ + FfxFloat16x3 r = (-x) * z + z; + return x * y + r; +} + +/// Choose between two half-precision FfxFloat32 values if the first paramter is greater than zero. +/// +/// @param [in] x The value to compare against zero. +/// @param [in] y The value to return if the comparision is greater than zero. +/// @param [in] z The value to return if the comparision is less than or equal to zero. +/// +/// @returns +/// The selected value. +/// +/// @ingroup GPUCore +FfxFloat16x4 ffxZeroOneSelectHalf(FfxFloat16x4 x, FfxFloat16x4 y, FfxFloat16x4 z) +{ + FfxFloat16x4 r = (-x) * z + z; + return x * y + r; +} + +/// Given a half-precision value, returns 1.0 if less than zero and 0.0 if not. +/// +/// @param [in] x The value to be compared. +/// +/// @returns +/// Result of the sign value. +/// +/// @ingroup GPUCore +FfxFloat16 ffxZeroOneIsSignedHalf(FfxFloat16 x) +{ + return ffxSaturate(x * FFX_BROADCAST_FLOAT16(FFX_NEGATIVE_INFINITY_HALF)); +} + +/// Given a half-precision value, returns 1.0 if less than zero and 0.0 if not. +/// +/// @param [in] x The value to be compared. +/// +/// @returns +/// Result of the sign value. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxZeroOneIsSignedHalf(FfxFloat16x2 x) +{ + return ffxSaturate(x * FFX_BROADCAST_FLOAT16X2(FFX_NEGATIVE_INFINITY_HALF)); +} + +/// Given a half-precision value, returns 1.0 if less than zero and 0.0 if not. +/// +/// @param [in] x The value to be compared. +/// +/// @returns +/// Result of the sign value. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxZeroOneIsSignedHalf(FfxFloat16x3 x) +{ + return ffxSaturate(x * FFX_BROADCAST_FLOAT16X3(FFX_NEGATIVE_INFINITY_HALF)); +} + +/// Given a half-precision value, returns 1.0 if less than zero and 0.0 if not. +/// +/// @param [in] x The value to be compared. +/// +/// @returns +/// Result of the sign value. +/// +/// @ingroup GPUCore +FfxFloat16x4 ffxZeroOneIsSignedHalf(FfxFloat16x4 x) +{ + return ffxSaturate(x * FFX_BROADCAST_FLOAT16X4(FFX_NEGATIVE_INFINITY_HALF)); +} + +/// Compute a Rec.709 color space. +/// +/// Rec.709 is used for some HDTVs. +/// +/// Both Rec.709 and sRGB have a linear segment which as spec'ed would intersect the curved segment 2 times. +/// (a.) For 8-bit sRGB, steps {0 to 10.3} are in the linear region (4% of the encoding range). +/// (b.) For 8-bit 709, steps {0 to 20.7} are in the linear region (8% of the encoding range). +/// +/// @param [in] c The color to convert to Rec. 709. +/// +/// @returns +/// The color in Rec.709 space. +/// +/// @ingroup GPUCore +FfxFloat16 ffxRec709FromLinearHalf(FfxFloat16 c) +{ + FfxFloat16x3 j = FfxFloat16x3(0.018 * 4.5, 4.5, 0.45); + FfxFloat16x2 k = FfxFloat16x2(1.099, -0.099); + return clamp(j.x, c * j.y, pow(c, j.z) * k.x + k.y); +} + +/// Compute a Rec.709 color space. +/// +/// Rec.709 is used for some HDTVs. +/// +/// Both Rec.709 and sRGB have a linear segment which as spec'ed would intersect the curved segment 2 times. +/// (a.) For 8-bit sRGB, steps {0 to 10.3} are in the linear region (4% of the encoding range). +/// (b.) For 8-bit 709, steps {0 to 20.7} are in the linear region (8% of the encoding range). +/// +/// @param [in] c The color to convert to Rec. 709. +/// +/// @returns +/// The color in Rec.709 space. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxRec709FromLinearHalf(FfxFloat16x2 c) +{ + FfxFloat16x3 j = FfxFloat16x3(0.018 * 4.5, 4.5, 0.45); + FfxFloat16x2 k = FfxFloat16x2(1.099, -0.099); + return clamp(j.xx, c * j.yy, pow(c, j.zz) * k.xx + k.yy); +} + +/// Compute a Rec.709 color space. +/// +/// Rec.709 is used for some HDTVs. +/// +/// Both Rec.709 and sRGB have a linear segment which as spec'ed would intersect the curved segment 2 times. +/// (a.) For 8-bit sRGB, steps {0 to 10.3} are in the linear region (4% of the encoding range). +/// (b.) For 8-bit 709, steps {0 to 20.7} are in the linear region (8% of the encoding range). +/// +/// @param [in] c The color to convert to Rec. 709. +/// +/// @returns +/// The color in Rec.709 space. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxRec709FromLinearHalf(FfxFloat16x3 c) +{ + FfxFloat16x3 j = FfxFloat16x3(0.018 * 4.5, 4.5, 0.45); + FfxFloat16x2 k = FfxFloat16x2(1.099, -0.099); + return clamp(j.xxx, c * j.yyy, pow(c, j.zzz) * k.xxx + k.yyy); +} + +/// Compute a gamma value from a linear value. +/// +/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. +/// +/// Note: 'rcpX' is '1/x', where the 'x' is what would be used in ffxLinearFromGammaHalf. +/// +/// @param [in] c The value to convert to gamma space from linear. +/// @param [in] rcpX The reciprocal of power value used for the gamma curve. +/// +/// @returns +/// A value in gamma space. +/// +/// @ingroup GPUCore +FfxFloat16 ffxGammaFromLinearHalf(FfxFloat16 c, FfxFloat16 rcpX) +{ + return pow(c, FFX_BROADCAST_FLOAT16(rcpX)); +} + +/// Compute a gamma value from a linear value. +/// +/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. +/// +/// Note: 'rcpX' is '1/x', where the 'x' is what would be used in ffxLinearFromGammaHalf. +/// +/// @param [in] c The value to convert to gamma space from linear. +/// @param [in] rcpX The reciprocal of power value used for the gamma curve. +/// +/// @returns +/// A value in gamma space. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxGammaFromLinearHalf(FfxFloat16x2 c, FfxFloat16 rcpX) +{ + return pow(c, FFX_BROADCAST_FLOAT16X2(rcpX)); +} + +/// Compute a gamma value from a linear value. +/// +/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. +/// +/// Note: 'rcpX' is '1/x', where the 'x' is what would be used in ffxLinearFromGammaHalf. +/// +/// @param [in] c The value to convert to gamma space from linear. +/// @param [in] rcpX The reciprocal of power value used for the gamma curve. +/// +/// @returns +/// A value in gamma space. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxGammaFromLinearHalf(FfxFloat16x3 c, FfxFloat16 rcpX) +{ + return pow(c, FFX_BROADCAST_FLOAT16X3(rcpX)); +} + +/// Compute an SRGB value from a linear value. +/// +/// @param [in] c The value to convert to SRGB from linear. +/// +/// @returns +/// A value in SRGB space. +/// +/// @ingroup GPUCore +FfxFloat16 ffxSrgbFromLinearHalf(FfxFloat16 c) +{ + FfxFloat16x3 j = FfxFloat16x3(0.0031308 * 12.92, 12.92, 1.0 / 2.4); + FfxFloat16x2 k = FfxFloat16x2(1.055, -0.055); + return clamp(j.x, c * j.y, pow(c, j.z) * k.x + k.y); +} + +/// Compute an SRGB value from a linear value. +/// +/// @param [in] c The value to convert to SRGB from linear. +/// +/// @returns +/// A value in SRGB space. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxSrgbFromLinearHalf(FfxFloat16x2 c) +{ + FfxFloat16x3 j = FfxFloat16x3(0.0031308 * 12.92, 12.92, 1.0 / 2.4); + FfxFloat16x2 k = FfxFloat16x2(1.055, -0.055); + return clamp(j.xx, c * j.yy, pow(c, j.zz) * k.xx + k.yy); +} + +/// Compute an SRGB value from a linear value. +/// +/// @param [in] c The value to convert to SRGB from linear. +/// +/// @returns +/// A value in SRGB space. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxSrgbFromLinearHalf(FfxFloat16x3 c) +{ + FfxFloat16x3 j = FfxFloat16x3(0.0031308 * 12.92, 12.92, 1.0 / 2.4); + FfxFloat16x2 k = FfxFloat16x2(1.055, -0.055); + return clamp(j.xxx, c * j.yyy, pow(c, j.zzz) * k.xxx + k.yyy); +} + +/// Compute the square root of a value. +/// +/// @param [in] c The value to compute the square root for. +/// +/// @returns +/// A square root of the input value. +/// +/// @ingroup GPUCore +FfxFloat16 ffxSquareRootHalf(FfxFloat16 c) +{ + return sqrt(c); +} + +/// Compute the square root of a value. +/// +/// @param [in] c The value to compute the square root for. +/// +/// @returns +/// A square root of the input value. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxSquareRootHalf(FfxFloat16x2 c) +{ + return sqrt(c); +} + +/// Compute the square root of a value. +/// +/// @param [in] c The value to compute the square root for. +/// +/// @returns +/// A square root of the input value. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxSquareRootHalf(FfxFloat16x3 c) +{ + return sqrt(c); +} + +/// Compute the cube root of a value. +/// +/// @param [in] c The value to compute the cube root for. +/// +/// @returns +/// A cube root of the input value. +/// +/// @ingroup GPUCore +FfxFloat16 ffxCubeRootHalf(FfxFloat16 c) +{ + return pow(c, FFX_BROADCAST_FLOAT16(1.0 / 3.0)); +} + +/// Compute the cube root of a value. +/// +/// @param [in] c The value to compute the cube root for. +/// +/// @returns +/// A cube root of the input value. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxCubeRootHalf(FfxFloat16x2 c) +{ + return pow(c, FFX_BROADCAST_FLOAT16X2(1.0 / 3.0)); +} + +/// Compute the cube root of a value. +/// +/// @param [in] c The value to compute the cube root for. +/// +/// @returns +/// A cube root of the input value. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxCubeRootHalf(FfxFloat16x3 c) +{ + return pow(c, FFX_BROADCAST_FLOAT16X3(1.0 / 3.0)); +} + +/// Compute a linear value from a REC.709 value. +/// +/// @param [in] c The value to convert to linear from REC.709. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat16 ffxLinearFromRec709Half(FfxFloat16 c) +{ + FfxFloat16x3 j = FfxFloat16x3(0.081 / 4.5, 1.0 / 4.5, 1.0 / 0.45); + FfxFloat16x2 k = FfxFloat16x2(1.0 / 1.099, 0.099 / 1.099); + return ffxZeroOneSelectHalf(ffxZeroOneIsSignedHalf(c - j.x), c * j.y, pow(c * k.x + k.y, j.z)); +} + +/// Compute a linear value from a REC.709 value. +/// +/// @param [in] c The value to convert to linear from REC.709. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxLinearFromRec709Half(FfxFloat16x2 c) +{ + FfxFloat16x3 j = FfxFloat16x3(0.081 / 4.5, 1.0 / 4.5, 1.0 / 0.45); + FfxFloat16x2 k = FfxFloat16x2(1.0 / 1.099, 0.099 / 1.099); + return ffxZeroOneSelectHalf(ffxZeroOneIsSignedHalf(c - j.xx), c * j.yy, pow(c * k.xx + k.yy, j.zz)); +} + +/// Compute a linear value from a REC.709 value. +/// +/// @param [in] c The value to convert to linear from REC.709. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxLinearFromRec709Half(FfxFloat16x3 c) +{ + FfxFloat16x3 j = FfxFloat16x3(0.081 / 4.5, 1.0 / 4.5, 1.0 / 0.45); + FfxFloat16x2 k = FfxFloat16x2(1.0 / 1.099, 0.099 / 1.099); + return ffxZeroOneSelectHalf(ffxZeroOneIsSignedHalf(c - j.xxx), c * j.yyy, pow(c * k.xxx + k.yyy, j.zzz)); +} + +/// Compute a linear value from a value in a gamma space. +/// +/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. +/// +/// @param [in] c The value to convert to linear in gamma space. +/// @param [in] x The power value used for the gamma curve. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat16 ffxLinearFromGammaHalf(FfxFloat16 c, FfxFloat16 x) +{ + return pow(c, FFX_BROADCAST_FLOAT16(x)); +} + +/// Compute a linear value from a value in a gamma space. +/// +/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. +/// +/// @param [in] c The value to convert to linear in gamma space. +/// @param [in] x The power value used for the gamma curve. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxLinearFromGammaHalf(FfxFloat16x2 c, FfxFloat16 x) +{ + return pow(c, FFX_BROADCAST_FLOAT16X2(x)); +} + +/// Compute a linear value from a value in a gamma space. +/// +/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. +/// +/// @param [in] c The value to convert to linear in gamma space. +/// @param [in] x The power value used for the gamma curve. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxLinearFromGammaHalf(FfxFloat16x3 c, FfxFloat16 x) +{ + return pow(c, FFX_BROADCAST_FLOAT16X3(x)); +} + +/// Compute a linear value from a value in a SRGB space. +/// +/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. +/// +/// @param [in] c The value to convert to linear in SRGB space. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat16 ffxLinearFromSrgbHalf(FfxFloat16 c) +{ + FfxFloat16x3 j = FfxFloat16x3(0.04045 / 12.92, 1.0 / 12.92, 2.4); + FfxFloat16x2 k = FfxFloat16x2(1.0 / 1.055, 0.055 / 1.055); + return ffxZeroOneSelectHalf(ffxZeroOneIsSignedHalf(c - j.x), c * j.y, pow(c * k.x + k.y, j.z)); +} + +/// Compute a linear value from a value in a SRGB space. +/// +/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. +/// +/// @param [in] c The value to convert to linear in SRGB space. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat16x2 ffxLinearFromSrgbHalf(FfxFloat16x2 c) +{ + FfxFloat16x3 j = FfxFloat16x3(0.04045 / 12.92, 1.0 / 12.92, 2.4); + FfxFloat16x2 k = FfxFloat16x2(1.0 / 1.055, 0.055 / 1.055); + return ffxZeroOneSelectHalf(ffxZeroOneIsSignedHalf(c - j.xx), c * j.yy, pow(c * k.xx + k.yy, j.zz)); +} + +/// Compute a linear value from a value in a SRGB space. +/// +/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. +/// +/// @param [in] c The value to convert to linear in SRGB space. +/// +/// @returns +/// A value in linear space. +/// +/// @ingroup GPUCore +FfxFloat16x3 ffxLinearFromSrgbHalf(FfxFloat16x3 c) +{ + FfxFloat16x3 j = FfxFloat16x3(0.04045 / 12.92, 1.0 / 12.92, 2.4); + FfxFloat16x2 k = FfxFloat16x2(1.0 / 1.055, 0.055 / 1.055); + return ffxZeroOneSelectHalf(ffxZeroOneIsSignedHalf(c - j.xxx), c * j.yyy, pow(c * k.xxx + k.yyy, j.zzz)); +} + +/// A remapping of 64x1 to 8x8 imposing rotated 2x2 pixel quads in quad linear. +/// +/// Remap illustration: +/// +/// 543210 +/// ~~~~~~ +/// ..xxx. +/// yy...y +/// +/// @param [in] a The input 1D coordinates to remap. +/// +/// @returns +/// The remapped 2D coordinates. +/// +/// @ingroup GPUCore +FfxUInt16x2 ffxRemapForQuadHalf(FfxUInt32 a) +{ + return FfxUInt16x2(ffxBitfieldExtract(a, 1u, 3u), ffxBitfieldInsertMask(ffxBitfieldExtract(a, 3u, 3u), a, 1u)); +} + +/// A helper function performing a remap 64x1 to 8x8 remapping which is necessary for 2D wave reductions. +/// +/// The 64-wide lane indices to 8x8 remapping is performed as follows: +/// +/// 00 01 08 09 10 11 18 19 +/// 02 03 0a 0b 12 13 1a 1b +/// 04 05 0c 0d 14 15 1c 1d +/// 06 07 0e 0f 16 17 1e 1f +/// 20 21 28 29 30 31 38 39 +/// 22 23 2a 2b 32 33 3a 3b +/// 24 25 2c 2d 34 35 3c 3d +/// 26 27 2e 2f 36 37 3e 3f +/// +/// @param [in] a The input 1D coordinate to remap. +/// +/// @returns +/// The remapped 2D coordinates. +/// +/// @ingroup GPUCore +FfxUInt16x2 ffxRemapForWaveReductionHalf(FfxUInt32 a) +{ + return FfxUInt16x2(ffxBitfieldInsertMask(ffxBitfieldExtract(a, 2u, 3u), a, 1u), ffxBitfieldInsertMask(ffxBitfieldExtract(a, 3u, 3u), ffxBitfieldExtract(a, 1u, 2u), 2u)); +} + +#endif // FFX_HALF diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_gpu_common_half.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_gpu_common_half.h.meta new file mode 100644 index 0000000..cda0dd4 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_gpu_common_half.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 93c5f50cadb9ff14cbf03fa7cb1de897 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_hlsl.h b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_hlsl.h new file mode 100644 index 0000000..28827d9 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_hlsl.h @@ -0,0 +1,1898 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +/// @defgroup HLSLCore HLSL Core +/// HLSL core defines and functions +/// +/// @ingroup FfxHLSL + +#define DECLARE_SRV_REGISTER(regIndex) t##regIndex +#define DECLARE_UAV_REGISTER(regIndex) u##regIndex +#define DECLARE_CB_REGISTER(regIndex) b##regIndex +#define FFX_DECLARE_SRV(regIndex) register(DECLARE_SRV_REGISTER(regIndex)) +#define FFX_DECLARE_UAV(regIndex) register(DECLARE_UAV_REGISTER(regIndex)) +#define FFX_DECLARE_CB(regIndex) register(DECLARE_CB_REGISTER(regIndex)) + +/// A define for abstracting select functionality for pre/post HLSL 21 +/// +/// @ingroup HLSLCore +#if __HLSL_VERSION >= 2021 + +#define FFX_SELECT(cond, arg1, arg2) select(cond, arg1, arg2) + +#else // #if __HLSL_VERSION >= 2021 + +#define FFX_SELECT(cond, arg1, arg2) cond ? arg1 : arg2 + +#endif // #if __HLSL_VERSION >= 2021 + +/// A define for abstracting shared memory between shading languages. +/// +/// @ingroup HLSLCore +#define FFX_GROUPSHARED groupshared + +/// A define for abstracting compute memory barriers between shading languages. +/// +/// @ingroup HLSLCore +#define FFX_GROUP_MEMORY_BARRIER GroupMemoryBarrierWithGroupSync() + +/// A define for abstracting compute atomic additions between shading languages. +/// +/// @ingroup HLSLCore +#define FFX_ATOMIC_ADD(x, y) InterlockedAdd(x, y) + +/// A define for abstracting compute atomic additions between shading languages. +/// +/// @ingroup HLSLCore +#define FFX_ATOMIC_ADD_RETURN(x, y, r) InterlockedAdd(x, y, r) + +/// A define for abstracting compute atomic OR between shading languages. +/// +/// @ingroup HLSLCore +#define FFX_ATOMIC_OR(x, y) InterlockedOr(x, y) + +/// A define for abstracting compute atomic min between shading languages. +/// +/// @ingroup HLSLCore +#define FFX_ATOMIC_MIN(x, y) InterlockedMin(x, y) + +/// A define for abstracting compute atomic max between shading languages. +/// +/// @ingroup HLSLCore +#define FFX_ATOMIC_MAX(x, y) InterlockedMax(x, y) + +/// A define added to accept static markup on functions to aid CPU/GPU portability of code. +/// +/// @ingroup HLSLCore +#define FFX_STATIC static + +/// A define for abstracting loop unrolling between shading languages. +/// +/// @ingroup HLSLCore +#define FFX_UNROLL [unroll] + +/// A define for abstracting a 'greater than' comparison operator between two types. +/// +/// @ingroup HLSLCore +#define FFX_GREATER_THAN(x, y) x > y + +/// A define for abstracting a 'greater than or equal' comparison operator between two types. +/// +/// @ingroup HLSLCore +#define FFX_GREATER_THAN_EQUAL(x, y) x >= y + +/// A define for abstracting a 'less than' comparison operator between two types. +/// +/// @ingroup HLSLCore +#define FFX_LESS_THAN(x, y) x < y + +/// A define for abstracting a 'less than or equal' comparison operator between two types. +/// +/// @ingroup HLSLCore +#define FFX_LESS_THAN_EQUAL(x, y) x <= y + +/// A define for abstracting an 'equal' comparison operator between two types. +/// +/// @ingroup HLSLCore +#define FFX_EQUAL(x, y) x == y + +/// A define for abstracting a 'not equal' comparison operator between two types. +/// +/// @ingroup HLSLCore +#define FFX_NOT_EQUAL(x, y) x != y + +/// A define for abstracting matrix multiply operations between shading languages. +/// +/// @ingroup HLSLCore +#define FFX_MATRIX_MULTIPLY(a, b) mul(a, b) + +/// A define for abstracting vector transformations between shading languages. +/// +/// @ingroup HLSLCore +#define FFX_TRANSFORM_VECTOR(a, b) mul(a, b) + +/// A define for abstracting modulo operations between shading languages. +/// +/// @ingroup HLSLCore +#define FFX_MODULO(a, b) (fmod(a, b)) + +/// Broadcast a scalar value to a 1-dimensional floating point vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_FLOAT32(x) FfxFloat32(x) + +/// Broadcast a scalar value to a 2-dimensional floating point vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_FLOAT32X2(x) FfxFloat32(x) + +/// Broadcast a scalar value to a 3-dimensional floating point vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_FLOAT32X3(x) FfxFloat32(x) + +/// Broadcast a scalar value to a 4-dimensional floating point vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_FLOAT32X4(x) FfxFloat32(x) + +/// Broadcast a scalar value to a 1-dimensional unsigned integer vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_UINT32(x) FfxUInt32(x) + +/// Broadcast a scalar value to a 2-dimensional unsigned integer vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_UINT32X2(x) FfxUInt32(x) + +/// Broadcast a scalar value to a 4-dimensional unsigned integer vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_UINT32X3(x) FfxUInt32(x) + +/// Broadcast a scalar value to a 4-dimensional unsigned integer vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_UINT32X4(x) FfxUInt32(x) + +/// Broadcast a scalar value to a 1-dimensional signed integer vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_INT32(x) FfxInt32(x) + +/// Broadcast a scalar value to a 2-dimensional signed integer vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_INT32X2(x) FfxInt32(x) + +/// Broadcast a scalar value to a 3-dimensional signed integer vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_INT32X3(x) FfxInt32(x) + +/// Broadcast a scalar value to a 4-dimensional signed integer vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_INT32X4(x) FfxInt32(x) + +/// Broadcast a scalar value to a 1-dimensional half-precision floating point vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_MIN_FLOAT16(a) FFX_MIN16_F(a) + +/// Broadcast a scalar value to a 2-dimensional half-precision floating point vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_MIN_FLOAT16X2(a) FFX_MIN16_F(a) + +/// Broadcast a scalar value to a 3-dimensional half-precision floating point vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_MIN_FLOAT16X3(a) FFX_MIN16_F(a) + +/// Broadcast a scalar value to a 4-dimensional half-precision floating point vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_MIN_FLOAT16X4(a) FFX_MIN16_F(a) + +/// Broadcast a scalar value to a 1-dimensional half-precision unsigned integer vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_MIN_UINT16(a) FFX_MIN16_U(a) + +/// Broadcast a scalar value to a 2-dimensional half-precision unsigned integer vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_MIN_UINT16X2(a) FFX_MIN16_U(a) + +/// Broadcast a scalar value to a 3-dimensional half-precision unsigned integer vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_MIN_UINT16X3(a) FFX_MIN16_U(a) + +/// Broadcast a scalar value to a 4-dimensional half-precision unsigned integer vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_MIN_UINT16X4(a) FFX_MIN16_U(a) + +/// Broadcast a scalar value to a 1-dimensional half-precision signed integer vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_MIN_INT16(a) FFX_MIN16_I(a) + +/// Broadcast a scalar value to a 2-dimensional half-precision signed integer vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_MIN_INT16X2(a) FFX_MIN16_I(a) + +/// Broadcast a scalar value to a 3-dimensional half-precision signed integer vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_MIN_INT16X3(a) FFX_MIN16_I(a) + +/// Broadcast a scalar value to a 4-dimensional half-precision signed integer vector. +/// +/// @ingroup HLSLCore +#define FFX_BROADCAST_MIN_INT16X4(a) FFX_MIN16_I(a) + +/// Convert FfxFloat32 to half (in lower 16-bits of output). +/// +/// This function implements the same fast technique that is documented here: ftp://ftp.fox-toolkit.org/pub/fasthalffloatconversion.pdf +/// +/// The function supports denormals. +/// +/// Some conversion rules are to make computations possibly "safer" on the GPU, +/// -INF & -NaN -> -65504 +/// +INF & +NaN -> +65504 +/// +/// @param [in] f The 32bit floating point value to convert. +/// +/// @returns +/// The closest 16bit floating point value to f. +/// +/// @ingroup HLSLCore +#define ffxF32ToF16 f32tof16 + +/// Pack 2x32-bit floating point values in a single 32bit value. +/// +/// This function first converts each component of value into their nearest 16-bit floating +/// point representation, and then stores the X and Y components in the lower and upper 16 bits of the +/// 32bit unsigned integer respectively. +/// +/// @param [in] value A 2-dimensional floating point value to convert and pack. +/// +/// @returns +/// A packed 32bit value containing 2 16bit floating point values. +/// +/// @ingroup HLSLCore +FfxUInt32 ffxPackHalf2x16(FfxFloat32x2 value) +{ + return ffxF32ToF16(value.x) | (ffxF32ToF16(value.y) << 16); +} + +/// Broadcast a scalar value to a 2-dimensional floating point vector. +/// +/// @param [in] value The value to to broadcast. +/// +/// @returns +/// A 2-dimensional floating point vector with value in each component. +/// +/// @ingroup HLSLCore +FfxFloat32x2 ffxBroadcast2(FfxFloat32 value) +{ + return FfxFloat32x2(value, value); +} + +/// Broadcast a scalar value to a 3-dimensional floating point vector. +/// +/// @param [in] value The value to to broadcast. +/// +/// @returns +/// A 3-dimensional floating point vector with value in each component. +/// +/// @ingroup HLSLCore +FfxFloat32x3 ffxBroadcast3(FfxFloat32 value) +{ + return FfxFloat32x3(value, value, value); +} + +/// Broadcast a scalar value to a 4-dimensional floating point vector. +/// +/// @param [in] value The value to to broadcast. +/// +/// @returns +/// A 4-dimensional floating point vector with value in each component. +/// +/// @ingroup HLSLCore +FfxFloat32x4 ffxBroadcast4(FfxFloat32 value) +{ + return FfxFloat32x4(value, value, value, value); +} + +/// Broadcast a scalar value to a 2-dimensional signed integer vector. +/// +/// @param [in] value The value to to broadcast. +/// +/// @returns +/// A 2-dimensional signed integer vector with value in each component. +/// +/// @ingroup HLSLCore +FfxInt32x2 ffxBroadcast2(FfxInt32 value) +{ + return FfxInt32x2(value, value); +} + +/// Broadcast a scalar value to a 3-dimensional signed integer vector. +/// +/// @param [in] value The value to to broadcast. +/// +/// @returns +/// A 3-dimensional signed integer vector with value in each component. +/// +/// @ingroup HLSLCore +FfxInt32x3 ffxBroadcast3(FfxInt32 value) +{ + return FfxInt32x3(value, value, value); +} + +/// Broadcast a scalar value to a 4-dimensional signed integer vector. +/// +/// @param [in] value The value to to broadcast. +/// +/// @returns +/// A 4-dimensional signed integer vector with value in each component. +/// +/// @ingroup HLSLCore +FfxInt32x4 ffxBroadcast4(FfxInt32 value) +{ + return FfxInt32x4(value, value, value, value); +} + +/// Broadcast a scalar value to a 2-dimensional unsigned integer vector. +/// +/// @param [in] value The value to to broadcast. +/// +/// @returns +/// A 2-dimensional unsigned integer vector with value in each component. +/// +/// @ingroup HLSLCore +FfxUInt32x2 ffxBroadcast2(FfxUInt32 value) +{ + return FfxUInt32x2(value, value); +} + +/// Broadcast a scalar value to a 3-dimensional unsigned integer vector. +/// +/// @param [in] value The value to to broadcast. +/// +/// @returns +/// A 3-dimensional unsigned integer vector with value in each component. +/// +/// @ingroup HLSLCore +FfxUInt32x3 ffxBroadcast3(FfxUInt32 value) +{ + return FfxUInt32x3(value, value, value); +} + +/// Broadcast a scalar value to a 4-dimensional unsigned integer vector. +/// +/// @param [in] value The value to to broadcast. +/// +/// @returns +/// A 4-dimensional unsigned integer vector with value in each component. +/// +/// @ingroup HLSLCore +FfxUInt32x4 ffxBroadcast4(FfxUInt32 value) +{ + return FfxUInt32x4(value, value, value, value); +} + +FfxUInt32 ffxBitfieldExtract(FfxUInt32 src, FfxUInt32 off, FfxUInt32 bits) +{ + FfxUInt32 mask = (1u << bits) - 1; + return (src >> off) & mask; +} + +FfxUInt32 ffxBitfieldInsert(FfxUInt32 src, FfxUInt32 ins, FfxUInt32 mask) +{ + return (ins & mask) | (src & (~mask)); +} + +FfxUInt32 ffxBitfieldInsertMask(FfxUInt32 src, FfxUInt32 ins, FfxUInt32 bits) +{ + FfxUInt32 mask = (1u << bits) - 1; + return (ins & mask) | (src & (~mask)); +} + +/// Interprets the bit pattern of x as an unsigned integer. +/// +/// @param [in] x The input value. +/// +/// @returns +/// The input interpreted as an unsigned integer. +/// +/// @ingroup HLSLCore +FfxUInt32 ffxAsUInt32(FfxFloat32 x) +{ + return asuint(x); +} + +/// Interprets the bit pattern of x as an unsigned integer. +/// +/// @param [in] x The input value. +/// +/// @returns +/// The input interpreted as an unsigned integer. +/// +/// @ingroup HLSLCore +FfxUInt32x2 ffxAsUInt32(FfxFloat32x2 x) +{ + return asuint(x); +} + +/// Interprets the bit pattern of x as an unsigned integer. +/// +/// @param [in] x The input value. +/// +/// @returns +/// The input interpreted as an unsigned integer. +/// +/// @ingroup HLSLCore +FfxUInt32x3 ffxAsUInt32(FfxFloat32x3 x) +{ + return asuint(x); +} + +/// Interprets the bit pattern of x as an unsigned integer. +/// +/// @param [in] x The input value. +/// +/// @returns +/// The input interpreted as an unsigned integer. +/// +/// @ingroup HLSLCore +FfxUInt32x4 ffxAsUInt32(FfxFloat32x4 x) +{ + return asuint(x); +} + +/// Interprets the bit pattern of x as a floating-point number. +/// +/// @param [in] x The input value. +/// +/// @returns +/// The input interpreted as a floating-point number. +/// +/// @ingroup HLSLCore +FfxFloat32 ffxAsFloat(FfxUInt32 x) +{ + return asfloat(x); +} + +/// Interprets the bit pattern of x as a floating-point number. +/// +/// @param [in] x The input value. +/// +/// @returns +/// The input interpreted as a floating-point number. +/// +/// @ingroup HLSLCore +FfxFloat32x2 ffxAsFloat(FfxUInt32x2 x) +{ + return asfloat(x); +} + +/// Interprets the bit pattern of x as a floating-point number. +/// +/// @param [in] x The input value. +/// +/// @returns +/// The input interpreted as a floating-point number. +/// +/// @ingroup HLSLCore +FfxFloat32x3 ffxAsFloat(FfxUInt32x3 x) +{ + return asfloat(x); +} + +/// Interprets the bit pattern of x as a floating-point number. +/// +/// @param [in] x The input value. +/// +/// @returns +/// The input interpreted as a floating-point number. +/// +/// @ingroup HLSLCore +FfxFloat32x4 ffxAsFloat(FfxUInt32x4 x) +{ + return asfloat(x); +} + +/// Compute the inverse of a value. +/// +/// @param [in] x The value to calulate the inverse of. +/// +/// @returns +/// The inverse of x. +/// +/// @ingroup HLSLCore +FfxFloat32 ffxReciprocal(FfxFloat32 x) +{ + return rcp(x); +} + +/// Compute the inverse of a value. +/// +/// @param [in] x The value to calulate the inverse of. +/// +/// @returns +/// The inverse of x. +/// +/// @ingroup HLSLCore +FfxFloat32x2 ffxReciprocal(FfxFloat32x2 x) +{ + return rcp(x); +} + +/// Compute the inverse of a value. +/// +/// @param [in] x The value to calulate the inverse of. +/// +/// @returns +/// The inverse of x. +/// +/// @ingroup HLSLCore +FfxFloat32x3 ffxReciprocal(FfxFloat32x3 x) +{ + return rcp(x); +} + +/// Compute the inverse of a value. +/// +/// @param [in] x The value to calulate the inverse of. +/// +/// @returns +/// The inverse of x. +/// +/// @ingroup HLSLCore +FfxFloat32x4 ffxReciprocal(FfxFloat32x4 x) +{ + return rcp(x); +} + +/// Compute the inverse square root of a value. +/// +/// @param [in] x The value to calulate the inverse square root of. +/// +/// @returns +/// The inverse square root of x. +/// +/// @ingroup HLSLCore +FfxFloat32 ffxRsqrt(FfxFloat32 x) +{ + return rsqrt(x); +} + +/// Compute the inverse square root of a value. +/// +/// @param [in] x The value to calulate the inverse square root of. +/// +/// @returns +/// The inverse square root of x. +/// +/// @ingroup HLSLCore +FfxFloat32x2 ffxRsqrt(FfxFloat32x2 x) +{ + return rsqrt(x); +} + +/// Compute the inverse square root of a value. +/// +/// @param [in] x The value to calulate the inverse square root of. +/// +/// @returns +/// The inverse square root of x. +/// +/// @ingroup HLSLCore +FfxFloat32x3 ffxRsqrt(FfxFloat32x3 x) +{ + return rsqrt(x); +} + +/// Compute the inverse square root of a value. +/// +/// @param [in] x The value to calulate the inverse square root of. +/// +/// @returns +/// The inverse square root of x. +/// +/// @ingroup HLSLCore +FfxFloat32x4 ffxRsqrt(FfxFloat32x4 x) +{ + return rsqrt(x); +} + +/// Compute the linear interopation between two values. +/// +/// Implemented by calling the HLSL mix instrinsic function. Implements the +/// following math: +/// +/// (1 - t) * x + t * y +/// +/// @param [in] x The first value to lerp between. +/// @param [in] y The second value to lerp between. +/// @param [in] t The value to determine how much of x and how much of y. +/// +/// @returns +/// A linearly interpolated value between x and y according to t. +/// +/// @ingroup HLSLCore +FfxFloat32 ffxLerp(FfxFloat32 x, FfxFloat32 y, FfxFloat32 t) +{ + return lerp(x, y, t); +} + +/// Compute the linear interopation between two values. +/// +/// Implemented by calling the HLSL mix instrinsic function. Implements the +/// following math: +/// +/// (1 - t) * x + t * y +/// +/// @param [in] x The first value to lerp between. +/// @param [in] y The second value to lerp between. +/// @param [in] t The value to determine how much of x and how much of y. +/// +/// @returns +/// A linearly interpolated value between x and y according to t. +/// +/// @ingroup HLSLCore +FfxFloat32x2 ffxLerp(FfxFloat32x2 x, FfxFloat32x2 y, FfxFloat32 t) +{ + return lerp(x, y, t); +} + +/// Compute the linear interopation between two values. +/// +/// Implemented by calling the HLSL mix instrinsic function. Implements the +/// following math: +/// +/// (1 - t) * x + t * y +/// +/// @param [in] x The first value to lerp between. +/// @param [in] y The second value to lerp between. +/// @param [in] t The value to determine how much of x and how much of y. +/// +/// @returns +/// A linearly interpolated value between x and y according to t. +/// +/// @ingroup HLSLCore +FfxFloat32x2 ffxLerp(FfxFloat32x2 x, FfxFloat32x2 y, FfxFloat32x2 t) +{ + return lerp(x, y, t); +} + +/// Compute the linear interopation between two values. +/// +/// Implemented by calling the HLSL mix instrinsic function. Implements the +/// following math: +/// +/// (1 - t) * x + t * y +/// +/// @param [in] x The first value to lerp between. +/// @param [in] y The second value to lerp between. +/// @param [in] t The value to determine how much of x and how much of y. +/// +/// @returns +/// A linearly interpolated value between x and y according to t. +/// +/// @ingroup HLSLCore +FfxFloat32x3 ffxLerp(FfxFloat32x3 x, FfxFloat32x3 y, FfxFloat32 t) +{ + return lerp(x, y, t); +} + +/// Compute the linear interopation between two values. +/// +/// Implemented by calling the HLSL mix instrinsic function. Implements the +/// following math: +/// +/// (1 - t) * x + t * y +/// +/// @param [in] x The first value to lerp between. +/// @param [in] y The second value to lerp between. +/// @param [in] t The value to determine how much of x and how much of y. +/// +/// @returns +/// A linearly interpolated value between x and y according to t. +/// +/// @ingroup HLSLCore +FfxFloat32x3 ffxLerp(FfxFloat32x3 x, FfxFloat32x3 y, FfxFloat32x3 t) +{ + return lerp(x, y, t); +} + +/// Compute the linear interopation between two values. +/// +/// Implemented by calling the HLSL mix instrinsic function. Implements the +/// following math: +/// +/// (1 - t) * x + t * y +/// +/// @param [in] x The first value to lerp between. +/// @param [in] y The second value to lerp between. +/// @param [in] t The value to determine how much of x and how much of y. +/// +/// @returns +/// A linearly interpolated value between x and y according to t. +/// +/// @ingroup HLSLCore +FfxFloat32x4 ffxLerp(FfxFloat32x4 x, FfxFloat32x4 y, FfxFloat32 t) +{ + return lerp(x, y, t); +} + +/// Compute the linear interopation between two values. +/// +/// Implemented by calling the HLSL mix instrinsic function. Implements the +/// following math: +/// +/// (1 - t) * x + t * y +/// +/// @param [in] x The first value to lerp between. +/// @param [in] y The second value to lerp between. +/// @param [in] t The value to determine how much of x and how much of y. +/// +/// @returns +/// A linearly interpolated value between x and y according to t. +/// +/// @ingroup HLSLCore +FfxFloat32x4 ffxLerp(FfxFloat32x4 x, FfxFloat32x4 y, FfxFloat32x4 t) +{ + return lerp(x, y, t); +} + +/// Clamp a value to a [0..1] range. +/// +/// @param [in] x The value to clamp to [0..1] range. +/// +/// @returns +/// The clamped version of x. +/// +/// @ingroup HLSLCore +FfxFloat32 ffxSaturate(FfxFloat32 x) +{ + return saturate(x); +} + +/// Clamp a value to a [0..1] range. +/// +/// @param [in] x The value to clamp to [0..1] range. +/// +/// @returns +/// The clamped version of x. +/// +/// @ingroup HLSLCore +FfxFloat32x2 ffxSaturate(FfxFloat32x2 x) +{ + return saturate(x); +} + +/// Clamp a value to a [0..1] range. +/// +/// @param [in] x The value to clamp to [0..1] range. +/// +/// @returns +/// The clamped version of x. +/// +/// @ingroup HLSLCore +FfxFloat32x3 ffxSaturate(FfxFloat32x3 x) +{ + return saturate(x); +} + +/// Clamp a value to a [0..1] range. +/// +/// @param [in] x The value to clamp to [0..1] range. +/// +/// @returns +/// The clamped version of x. +/// +/// @ingroup HLSLCore +FfxFloat32x4 ffxSaturate(FfxFloat32x4 x) +{ + return saturate(x); +} + +/// Compute the factional part of a decimal value. +/// +/// This function calculates x - floor(x). Where floor is the intrinsic HLSL function. +/// +/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. It is +/// worth further noting that this function is intentionally distinct from the HLSL frac intrinsic +/// function. +/// +/// @param [in] x The value to compute the fractional part from. +/// +/// @returns +/// The fractional part of x. +/// +/// @ingroup HLSLCore +FfxFloat32 ffxFract(FfxFloat32 x) +{ + return x - floor(x); +} + +/// Compute the factional part of a decimal value. +/// +/// This function calculates x - floor(x). Where floor is the intrinsic HLSL function. +/// +/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. It is +/// worth further noting that this function is intentionally distinct from the HLSL frac intrinsic +/// function. +/// +/// @param [in] x The value to compute the fractional part from. +/// +/// @returns +/// The fractional part of x. +/// +/// @ingroup HLSLCore +FfxFloat32x2 ffxFract(FfxFloat32x2 x) +{ + return x - floor(x); +} + +/// Compute the factional part of a decimal value. +/// +/// This function calculates x - floor(x). Where floor is the intrinsic HLSL function. +/// +/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. It is +/// worth further noting that this function is intentionally distinct from the HLSL frac intrinsic +/// function. +/// +/// @param [in] x The value to compute the fractional part from. +/// +/// @returns +/// The fractional part of x. +/// +/// @ingroup HLSLCore +FfxFloat32x3 ffxFract(FfxFloat32x3 x) +{ + return x - floor(x); +} + +/// Compute the factional part of a decimal value. +/// +/// This function calculates x - floor(x). Where floor is the intrinsic HLSL function. +/// +/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. It is +/// worth further noting that this function is intentionally distinct from the HLSL frac intrinsic +/// function. +/// +/// @param [in] x The value to compute the fractional part from. +/// +/// @returns +/// The fractional part of x. +/// +/// @ingroup HLSLCore +FfxFloat32x4 ffxFract(FfxFloat32x4 x) +{ + return x - floor(x); +} + +/// Rounds to the nearest integer. In case the fractional part is 0.5, it will round to the nearest even integer. +/// +/// @param [in] x The value to be rounded. +/// +/// @returns +/// The nearest integer from x. The nearest even integer from x if equidistant from 2 integer. +/// +/// @ingroup HLSLCore +FfxFloat32 ffxRound(FfxFloat32 x) +{ + return round(x); +} + +/// Rounds to the nearest integer. In case the fractional part is 0.5, it will round to the nearest even integer. +/// +/// @param [in] x The value to be rounded. +/// +/// @returns +/// The nearest integer from x. The nearest even integer from x if equidistant from 2 integer. +/// +/// @ingroup HLSLCore +FfxFloat32x2 ffxRound(FfxFloat32x2 x) +{ + return round(x); +} + +/// Rounds to the nearest integer. In case the fractional part is 0.5, it will round to the nearest even integer. +/// +/// @param [in] x The value to be rounded. +/// +/// @returns +/// The nearest integer from x. The nearest even integer from x if equidistant from 2 integer. +/// +/// @ingroup HLSLCore +FfxFloat32x3 ffxRound(FfxFloat32x3 x) +{ + return round(x); +} + +/// Rounds to the nearest integer. In case the fractional part is 0.5, it will round to the nearest even integer. +/// +/// @param [in] x The value to be rounded. +/// +/// @returns +/// The nearest integer from x. The nearest even integer from x if equidistant from 2 integer. +/// +/// @ingroup HLSLCore +FfxFloat32x4 ffxRound(FfxFloat32x4 x) +{ + return round(x); +} + +/// Compute the maximum of three values. +/// +/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the max calculation. +/// @param [in] y The second value to include in the max calcuation. +/// @param [in] z The third value to include in the max calcuation. +/// +/// @returns +/// The maximum value of x, y, and z. +/// +/// @ingroup HLSLCore +FfxFloat32 ffxMax3(FfxFloat32 x, FfxFloat32 y, FfxFloat32 z) +{ + return max(x, max(y, z)); +} + +/// Compute the maximum of three values. +/// +/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the max calculation. +/// @param [in] y The second value to include in the max calcuation. +/// @param [in] z The third value to include in the max calcuation. +/// +/// @returns +/// The maximum value of x, y, and z. +/// +/// @ingroup HLSLCore +FfxFloat32x2 ffxMax3(FfxFloat32x2 x, FfxFloat32x2 y, FfxFloat32x2 z) +{ + return max(x, max(y, z)); +} + +/// Compute the maximum of three values. +/// +/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the max calculation. +/// @param [in] y The second value to include in the max calcuation. +/// @param [in] z The third value to include in the max calcuation. +/// +/// @returns +/// The maximum value of x, y, and z. +/// +/// @ingroup HLSLCore +FfxFloat32x3 ffxMax3(FfxFloat32x3 x, FfxFloat32x3 y, FfxFloat32x3 z) +{ + return max(x, max(y, z)); +} + +/// Compute the maximum of three values. +/// +/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the max calculation. +/// @param [in] y The second value to include in the max calcuation. +/// @param [in] z The third value to include in the max calcuation. +/// +/// @returns +/// The maximum value of x, y, and z. +/// +/// @ingroup HLSLCore +FfxFloat32x4 ffxMax3(FfxFloat32x4 x, FfxFloat32x4 y, FfxFloat32x4 z) +{ + return max(x, max(y, z)); +} + +/// Compute the maximum of three values. +/// +/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the max calculation. +/// @param [in] y The second value to include in the max calcuation. +/// @param [in] z The third value to include in the max calcuation. +/// +/// @returns +/// The maximum value of x, y, and z. +/// +/// @ingroup HLSLCore +FfxUInt32 ffxMax3(FfxUInt32 x, FfxUInt32 y, FfxUInt32 z) +{ + return max(x, max(y, z)); +} + +/// Compute the maximum of three values. +/// +/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the max calculation. +/// @param [in] y The second value to include in the max calcuation. +/// @param [in] z The third value to include in the max calcuation. +/// +/// @returns +/// The maximum value of x, y, and z. +/// +/// @ingroup HLSLCore +FfxUInt32x2 ffxMax3(FfxUInt32x2 x, FfxUInt32x2 y, FfxUInt32x2 z) +{ + return max(x, max(y, z)); +} + +/// Compute the maximum of three values. +/// +/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the max calculation. +/// @param [in] y The second value to include in the max calcuation. +/// @param [in] z The third value to include in the max calcuation. +/// +/// @returns +/// The maximum value of x, y, and z. +/// +/// @ingroup HLSLCore +FfxUInt32x3 ffxMax3(FfxUInt32x3 x, FfxUInt32x3 y, FfxUInt32x3 z) +{ + return max(x, max(y, z)); +} + +/// Compute the maximum of three values. +/// +/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the max calculation. +/// @param [in] y The second value to include in the max calcuation. +/// @param [in] z The third value to include in the max calcuation. +/// +/// @returns +/// The maximum value of x, y, and z. +/// +/// @ingroup HLSLCore +FfxUInt32x4 ffxMax3(FfxUInt32x4 x, FfxUInt32x4 y, FfxUInt32x4 z) +{ + return max(x, max(y, z)); +} + +/// Compute the median of three values. +/// +/// NOTE: This function should compile down to a single V_MED3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the median calculation. +/// @param [in] y The second value to include in the median calcuation. +/// @param [in] z The third value to include in the median calcuation. +/// +/// @returns +/// The median value of x, y, and z. +/// +/// @ingroup HLSLCore +FfxFloat32 ffxMed3(FfxFloat32 x, FfxFloat32 y, FfxFloat32 z) +{ + return max(min(x, y), min(max(x, y), z)); +} + +/// Compute the median of three values. +/// +/// NOTE: This function should compile down to a single V_MED3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the median calculation. +/// @param [in] y The second value to include in the median calcuation. +/// @param [in] z The third value to include in the median calcuation. +/// +/// @returns +/// The median value of x, y, and z. +/// +/// @ingroup HLSLCore +FfxFloat32x2 ffxMed3(FfxFloat32x2 x, FfxFloat32x2 y, FfxFloat32x2 z) +{ + return max(min(x, y), min(max(x, y), z)); +} + +/// Compute the median of three values. +/// +/// NOTE: This function should compile down to a single V_MED3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the median calculation. +/// @param [in] y The second value to include in the median calcuation. +/// @param [in] z The third value to include in the median calcuation. +/// +/// @returns +/// The median value of x, y, and z. +/// +/// @ingroup HLSLCore +FfxFloat32x3 ffxMed3(FfxFloat32x3 x, FfxFloat32x3 y, FfxFloat32x3 z) +{ + return max(min(x, y), min(max(x, y), z)); +} + +/// Compute the median of three values. +/// +/// NOTE: This function should compile down to a single V_MED3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the median calculation. +/// @param [in] y The second value to include in the median calcuation. +/// @param [in] z The third value to include in the median calcuation. +/// +/// @returns +/// The median value of x, y, and z. +/// +/// @ingroup HLSLCore +FfxFloat32x4 ffxMed3(FfxFloat32x4 x, FfxFloat32x4 y, FfxFloat32x4 z) +{ + return max(min(x, y), min(max(x, y), z)); +} + +/// Compute the median of three values. +/// +/// NOTE: This function should compile down to a single V_MED3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the median calculation. +/// @param [in] y The second value to include in the median calcuation. +/// @param [in] z The third value to include in the median calcuation. +/// +/// @returns +/// The median value of x, y, and z. +/// +/// @ingroup HLSL +FfxInt32 ffxMed3(FfxInt32 x, FfxInt32 y, FfxInt32 z) +{ + return max(min(x, y), min(max(x, y), z)); + // return min(max(min(y, z), x), max(y, z)); + // return max(max(x, y), z) == x ? max(y, z) : (max(max(x, y), z) == y ? max(x, z) : max(x, y)); +} + +/// Compute the median of three values. +/// +/// NOTE: This function should compile down to a single V_MED3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the median calculation. +/// @param [in] y The second value to include in the median calcuation. +/// @param [in] z The third value to include in the median calcuation. +/// +/// @returns +/// The median value of x, y, and z. +/// +/// @ingroup HLSL +FfxInt32x2 ffxMed3(FfxInt32x2 x, FfxInt32x2 y, FfxInt32x2 z) +{ + return max(min(x, y), min(max(x, y), z)); + // return min(max(min(y, z), x), max(y, z)); + // return max(max(x, y), z) == x ? max(y, z) : (max(max(x, y), z) == y ? max(x, z) : max(x, y)); +} + +/// Compute the median of three values. +/// +/// NOTE: This function should compile down to a single V_MED3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the median calculation. +/// @param [in] y The second value to include in the median calcuation. +/// @param [in] z The third value to include in the median calcuation. +/// +/// @returns +/// The median value of x, y, and z. +/// +/// @ingroup HLSL +FfxInt32x3 ffxMed3(FfxInt32x3 x, FfxInt32x3 y, FfxInt32x3 z) +{ + return max(min(x, y), min(max(x, y), z)); +} + +/// Compute the median of three values. +/// +/// NOTE: This function should compile down to a single V_MED3_I32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the median calculation. +/// @param [in] y The second value to include in the median calcuation. +/// @param [in] z The third value to include in the median calcuation. +/// +/// @returns +/// The median value of x, y, and z. +/// +/// @ingroup HLSL +FfxInt32x4 ffxMed3(FfxInt32x4 x, FfxInt32x4 y, FfxInt32x4 z) +{ + return max(min(x, y), min(max(x, y), z)); +} + +/// Compute the minimum of three values. +/// +/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the min calculation. +/// @param [in] y The second value to include in the min calcuation. +/// @param [in] z The third value to include in the min calcuation. +/// +/// @returns +/// The minimum value of x, y, and z. +/// +/// @ingroup HLSLCore +FfxFloat32 ffxMin3(FfxFloat32 x, FfxFloat32 y, FfxFloat32 z) +{ + return min(x, min(y, z)); +} + +/// Compute the minimum of three values. +/// +/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the min calculation. +/// @param [in] y The second value to include in the min calcuation. +/// @param [in] z The third value to include in the min calcuation. +/// +/// @returns +/// The minimum value of x, y, and z. +/// +/// @ingroup HLSLCore +FfxFloat32x2 ffxMin3(FfxFloat32x2 x, FfxFloat32x2 y, FfxFloat32x2 z) +{ + return min(x, min(y, z)); +} + +/// Compute the minimum of three values. +/// +/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the min calculation. +/// @param [in] y The second value to include in the min calcuation. +/// @param [in] z The third value to include in the min calcuation. +/// +/// @returns +/// The minimum value of x, y, and z. +/// +/// @ingroup HLSLCore +FfxFloat32x3 ffxMin3(FfxFloat32x3 x, FfxFloat32x3 y, FfxFloat32x3 z) +{ + return min(x, min(y, z)); +} + +/// Compute the minimum of three values. +/// +/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the min calculation. +/// @param [in] y The second value to include in the min calcuation. +/// @param [in] z The third value to include in the min calcuation. +/// +/// @returns +/// The minimum value of x, y, and z. +/// +/// @ingroup HLSLCore +FfxFloat32x4 ffxMin3(FfxFloat32x4 x, FfxFloat32x4 y, FfxFloat32x4 z) +{ + return min(x, min(y, z)); +} + +/// Compute the minimum of three values. +/// +/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the min calculation. +/// @param [in] y The second value to include in the min calcuation. +/// @param [in] z The third value to include in the min calcuation. +/// +/// @returns +/// The minimum value of x, y, and z. +/// +/// @ingroup HLSLCore +FfxUInt32 ffxMin3(FfxUInt32 x, FfxUInt32 y, FfxUInt32 z) +{ + return min(x, min(y, z)); +} + +/// Compute the minimum of three values. +/// +/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the min calculation. +/// @param [in] y The second value to include in the min calcuation. +/// @param [in] z The third value to include in the min calcuation. +/// +/// @returns +/// The minimum value of x, y, and z. +/// +/// @ingroup HLSLCore +FfxUInt32x2 ffxMin3(FfxUInt32x2 x, FfxUInt32x2 y, FfxUInt32x2 z) +{ + return min(x, min(y, z)); +} + +/// Compute the minimum of three values. +/// +/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the min calculation. +/// @param [in] y The second value to include in the min calculation. +/// @param [in] z The third value to include in the min calculation. +/// +/// @returns +/// The minimum value of x, y, and z. +/// +/// @ingroup HLSLCore +FfxUInt32x3 ffxMin3(FfxUInt32x3 x, FfxUInt32x3 y, FfxUInt32x3 z) +{ + return min(x, min(y, z)); +} + +/// Compute the minimum of three values. +/// +/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. +/// +/// @param [in] x The first value to include in the min calculation. +/// @param [in] y The second value to include in the min calcuation. +/// @param [in] z The third value to include in the min calcuation. +/// +/// @returns +/// The minimum value of x, y, and z. +/// +/// @ingroup HLSLCore +FfxUInt32x4 ffxMin3(FfxUInt32x4 x, FfxUInt32x4 y, FfxUInt32x4 z) +{ + return min(x, min(y, z)); +} + + +FfxUInt32 ffxAShrSU1(FfxUInt32 a, FfxUInt32 b) +{ + return FfxUInt32(FfxInt32(a) >> FfxInt32(b)); +} + +FfxUInt32 ffxPackF32(FfxFloat32x2 v){ + FfxUInt32x2 p = FfxUInt32x2(ffxF32ToF16(FfxFloat32x2(v).x), ffxF32ToF16(FfxFloat32x2(v).y)); + return p.x | (p.y << 16); +} + +FfxFloat32x2 ffxUnpackF32(FfxUInt32 a){ + return f16tof32(FfxUInt32x2(a & 0xFFFF, a >> 16)); +} + +FfxUInt32x2 ffxPackF32x2(FfxFloat32x4 v){ + return FfxUInt32x2(ffxPackF32(v.xy), ffxPackF32(v.zw)); +} + +FfxFloat32x4 ffxUnpackF32x2(FfxUInt32x2 a){ + return FfxFloat32x4(ffxUnpackF32(a.x), ffxUnpackF32(a.y)); +} + +//============================================================================================================================== +// HLSL HALF +//============================================================================================================================== +//============================================================================================================================== +// Need to use manual unpack to get optimal execution (don't use packed types in buffers directly). +// Unpack requires this pattern: https://gpuopen.com/first-steps-implementing-fp16/ +FFX_MIN16_F2 ffxUint32ToFloat16x2(FfxUInt32 x) +{ + FfxFloat32x2 t = f16tof32(FfxUInt32x2(x & 0xFFFF, x >> 16)); + return FFX_MIN16_F2(t); +} +FFX_MIN16_F4 ffxUint32x2ToFloat16x4(FfxUInt32x2 x) +{ + return FFX_MIN16_F4(ffxUint32ToFloat16x2(x.x), ffxUint32ToFloat16x2(x.y)); +} +FFX_MIN16_U2 ffxUint32ToUint16x2(FfxUInt32 x) +{ + FfxUInt32x2 t = FfxUInt32x2(x & 0xFFFF, x >> 16); + return FFX_MIN16_U2(t); +} +FFX_MIN16_U4 ffxUint32x2ToUint16x4(FfxUInt32x2 x) +{ + return FFX_MIN16_U4(ffxUint32ToUint16x2(x.x), ffxUint32ToUint16x2(x.y)); +} + +FfxUInt32x2 ffxFloat16x4ToUint32x2(FFX_MIN16_F4 v) +{ + FfxUInt32x2 result; + result.x = ffxF32ToF16(v.x) | (ffxF32ToF16(v.y) << 16); + result.y = ffxF32ToF16(v.z) | (ffxF32ToF16(v.w) << 16); + return result; +} + +/// @brief Inverts the value while avoiding division by zero. If the value is zero, zero is returned. +/// @param v Value to invert. +/// @return If v = 0 returns 0. If v != 0 returns 1/v. +FfxFloat32 ffxInvertSafe(FfxFloat32 v){ + FfxFloat32 s = FfxFloat32(sign(v)); + FfxFloat32 s2 = s*s; + return s2/(v + s2 - 1.0); +} + +/// @brief Inverts the value while avoiding division by zero. If the value is zero, zero is returned. +/// @param v Value to invert. +/// @return If v = 0 returns 0. If v != 0 returns 1/v. +FfxFloat32x2 ffxInvertSafe(FfxFloat32x2 v){ + FfxFloat32x2 s = FfxFloat32x2(sign(v)); + FfxFloat32x2 s2 = s*s; + return s2/(v + s2 - FfxFloat32x2(1.0, 1.0)); +} + +/// @brief Inverts the value while avoiding division by zero. If the value is zero, zero is returned. +/// @param v Value to invert. +/// @return If v = 0 returns 0. If v != 0 returns 1/v. +FfxFloat32x3 ffxInvertSafe(FfxFloat32x3 v){ + FfxFloat32x3 s = FfxFloat32x3(sign(v)); + FfxFloat32x3 s2 = s*s; + return s2/(v + s2 - FfxFloat32x3(1.0, 1.0, 1.0)); +} + +/// @brief Inverts the value while avoiding division by zero. If the value is zero, zero is returned. +/// @param v Value to invert. +/// @return If v = 0 returns 0. If v != 0 returns 1/v. +FfxFloat32x4 ffxInvertSafe(FfxFloat32x4 v){ + FfxFloat32x4 s = FfxFloat32x4(sign(v)); + FfxFloat32x4 s2 = s*s; + return s2/(v + s2 - FfxFloat32x4(1.0, 1.0, 1.0, 1.0)); +} + +#define FFX_UINT32_TO_FLOAT16X2(x) ffxUint32ToFloat16x2(FfxUInt32(x)) +#if FFX_HALF + +#define FFX_UINT32X2_TO_FLOAT16X4(x) ffxUint32x2ToFloat16x4(FfxUInt32x2(x)) +#define FFX_UINT32_TO_UINT16X2(x) ffxUint32ToUint16x2(FfxUInt32(x)) +#define FFX_UINT32X2_TO_UINT16X4(x) ffxUint32x2ToUint16x4(FfxUInt32x2(x)) + +FfxUInt32 ffxPackF16(FfxFloat16x2 v){ + FfxUInt32x2 p = FfxUInt32x2(ffxF32ToF16(FfxFloat32x2(v).x), ffxF32ToF16(FfxFloat32x2(v).y)); + return p.x | (p.y << 16); +} + +FfxFloat16x2 ffxUnpackF16(FfxUInt32 a){ + return FfxFloat16x2(f16tof32(FfxUInt32x2(a & 0xFFFF, a >> 16))); +} + +//------------------------------------------------------------------------------------------------------------------------------ +FfxUInt32 FFX_MIN16_F2ToUint32(FFX_MIN16_F2 x) +{ + return ffxF32ToF16(x.x) + (ffxF32ToF16(x.y) << 16); +} +FfxUInt32x2 FFX_MIN16_F4ToUint32x2(FFX_MIN16_F4 x) +{ + return FfxUInt32x2(FFX_MIN16_F2ToUint32(x.xy), FFX_MIN16_F2ToUint32(x.zw)); +} +FfxUInt32 FFX_MIN16_U2ToUint32(FFX_MIN16_U2 x) +{ + return FfxUInt32(x.x) + (FfxUInt32(x.y) << 16); +} +FfxUInt32x2 FFX_MIN16_U4ToUint32x2(FFX_MIN16_U4 x) +{ + return FfxUInt32x2(FFX_MIN16_U2ToUint32(x.xy), FFX_MIN16_U2ToUint32(x.zw)); +} +#define FFX_FLOAT16X2_TO_UINT32(x) FFX_MIN16_F2ToUint32(FFX_MIN16_F2(x)) +#define FFX_FLOAT16X4_TO_UINT32X2(x) FFX_MIN16_F4ToUint32x2(FFX_MIN16_F4(x)) +#define FFX_UINT16X2_TO_UINT32(x) FFX_MIN16_U2ToUint32(FFX_MIN16_U2(x)) +#define FFX_UINT16X4_TO_UINT32X2(x) FFX_MIN16_U4ToUint32x2(FFX_MIN16_U4(x)) + +#if (FFX_HLSL_SM >= 62) && !defined(FFX_NO_16_BIT_CAST) +#define FFX_TO_UINT16(x) asuint16(x) +#define FFX_TO_UINT16X2(x) asuint16(x) +#define FFX_TO_UINT16X3(x) asuint16(x) +#define FFX_TO_UINT16X4(x) asuint16(x) +#else +#define FFX_TO_UINT16(a) FFX_MIN16_U(ffxF32ToF16(FfxFloat32(a))) +#define FFX_TO_UINT16X2(a) FFX_MIN16_U2(FFX_TO_UINT16((a).x), FFX_TO_UINT16((a).y)) +#define FFX_TO_UINT16X3(a) FFX_MIN16_U3(FFX_TO_UINT16((a).x), FFX_TO_UINT16((a).y), FFX_TO_UINT16((a).z)) +#define FFX_TO_UINT16X4(a) FFX_MIN16_U4(FFX_TO_UINT16((a).x), FFX_TO_UINT16((a).y), FFX_TO_UINT16((a).z), FFX_TO_UINT16((a).w)) +#endif // #if (FFX_HLSL_SM>=62) && !defined(FFX_NO_16_BIT_CAST) + +#if (FFX_HLSL_SM >= 62) && !defined(FFX_NO_16_BIT_CAST) +#define FFX_TO_FLOAT16(x) asfloat16(x) +#define FFX_TO_FLOAT16X2(x) asfloat16(x) +#define FFX_TO_FLOAT16X3(x) asfloat16(x) +#define FFX_TO_FLOAT16X4(x) asfloat16(x) +#else +#define FFX_TO_FLOAT16(a) FFX_MIN16_F(f16tof32(FfxUInt32(a))) +#define FFX_TO_FLOAT16X2(a) FFX_MIN16_F2(FFX_TO_FLOAT16((a).x), FFX_TO_FLOAT16((a).y)) +#define FFX_TO_FLOAT16X3(a) FFX_MIN16_F3(FFX_TO_FLOAT16((a).x), FFX_TO_FLOAT16((a).y), FFX_TO_FLOAT16((a).z)) +#define FFX_TO_FLOAT16X4(a) FFX_MIN16_F4(FFX_TO_FLOAT16((a).x), FFX_TO_FLOAT16((a).y), FFX_TO_FLOAT16((a).z), FFX_TO_FLOAT16((a).w)) +#endif // #if (FFX_HLSL_SM>=62) && !defined(FFX_NO_16_BIT_CAST) + +//============================================================================================================================== +#define FFX_BROADCAST_FLOAT16(a) FFX_MIN16_F(a) +#define FFX_BROADCAST_FLOAT16X2(a) FFX_MIN16_F(a) +#define FFX_BROADCAST_FLOAT16X3(a) FFX_MIN16_F(a) +#define FFX_BROADCAST_FLOAT16X4(a) FFX_MIN16_F(a) + +//------------------------------------------------------------------------------------------------------------------------------ +#define FFX_BROADCAST_INT16(a) FFX_MIN16_I(a) +#define FFX_BROADCAST_INT16X2(a) FFX_MIN16_I(a) +#define FFX_BROADCAST_INT16X3(a) FFX_MIN16_I(a) +#define FFX_BROADCAST_INT16X4(a) FFX_MIN16_I(a) + +//------------------------------------------------------------------------------------------------------------------------------ +#define FFX_BROADCAST_UINT16(a) FFX_MIN16_U(a) +#define FFX_BROADCAST_UINT16X2(a) FFX_MIN16_U(a) +#define FFX_BROADCAST_UINT16X3(a) FFX_MIN16_U(a) +#define FFX_BROADCAST_UINT16X4(a) FFX_MIN16_U(a) + +//============================================================================================================================== +FFX_MIN16_U ffxAbsHalf(FFX_MIN16_U a) +{ + return FFX_MIN16_U(abs(FFX_MIN16_I(a))); +} +FFX_MIN16_U2 ffxAbsHalf(FFX_MIN16_U2 a) +{ + return FFX_MIN16_U2(abs(FFX_MIN16_I2(a))); +} +FFX_MIN16_U3 ffxAbsHalf(FFX_MIN16_U3 a) +{ + return FFX_MIN16_U3(abs(FFX_MIN16_I3(a))); +} +FFX_MIN16_U4 ffxAbsHalf(FFX_MIN16_U4 a) +{ + return FFX_MIN16_U4(abs(FFX_MIN16_I4(a))); +} +//------------------------------------------------------------------------------------------------------------------------------ +FFX_MIN16_F ffxClampHalf(FFX_MIN16_F x, FFX_MIN16_F n, FFX_MIN16_F m) +{ + return max(n, min(x, m)); +} +FFX_MIN16_F2 ffxClampHalf(FFX_MIN16_F2 x, FFX_MIN16_F2 n, FFX_MIN16_F2 m) +{ + return max(n, min(x, m)); +} +FFX_MIN16_F3 ffxClampHalf(FFX_MIN16_F3 x, FFX_MIN16_F3 n, FFX_MIN16_F3 m) +{ + return max(n, min(x, m)); +} +FFX_MIN16_F4 ffxClampHalf(FFX_MIN16_F4 x, FFX_MIN16_F4 n, FFX_MIN16_F4 m) +{ + return max(n, min(x, m)); +} +//------------------------------------------------------------------------------------------------------------------------------ +// V_FRACT_F16 (note DX frac() is different). +FFX_MIN16_F ffxFract(FFX_MIN16_F x) +{ + return x - floor(x); +} +FFX_MIN16_F2 ffxFract(FFX_MIN16_F2 x) +{ + return x - floor(x); +} +FFX_MIN16_F3 ffxFract(FFX_MIN16_F3 x) +{ + return x - floor(x); +} +FFX_MIN16_F4 ffxFract(FFX_MIN16_F4 x) +{ + return x - floor(x); +} +//------------------------------------------------------------------------------------------------------------------------------ +FFX_MIN16_F ffxLerp(FFX_MIN16_F x, FFX_MIN16_F y, FFX_MIN16_F a) +{ + return lerp(x, y, a); +} +FFX_MIN16_F2 ffxLerp(FFX_MIN16_F2 x, FFX_MIN16_F2 y, FFX_MIN16_F a) +{ + return lerp(x, y, a); +} +FFX_MIN16_F2 ffxLerp(FFX_MIN16_F2 x, FFX_MIN16_F2 y, FFX_MIN16_F2 a) +{ + return lerp(x, y, a); +} +FFX_MIN16_F3 ffxLerp(FFX_MIN16_F3 x, FFX_MIN16_F3 y, FFX_MIN16_F a) +{ + return lerp(x, y, a); +} +FFX_MIN16_F3 ffxLerp(FFX_MIN16_F3 x, FFX_MIN16_F3 y, FFX_MIN16_F3 a) +{ + return lerp(x, y, a); +} +FFX_MIN16_F4 ffxLerp(FFX_MIN16_F4 x, FFX_MIN16_F4 y, FFX_MIN16_F a) +{ + return lerp(x, y, a); +} +FFX_MIN16_F4 ffxLerp(FFX_MIN16_F4 x, FFX_MIN16_F4 y, FFX_MIN16_F4 a) +{ + return lerp(x, y, a); +} +//------------------------------------------------------------------------------------------------------------------------------ +FFX_MIN16_F ffxMax3Half(FFX_MIN16_F x, FFX_MIN16_F y, FFX_MIN16_F z) +{ + return max(x, max(y, z)); +} +FFX_MIN16_F2 ffxMax3Half(FFX_MIN16_F2 x, FFX_MIN16_F2 y, FFX_MIN16_F2 z) +{ + return max(x, max(y, z)); +} +FFX_MIN16_F3 ffxMax3Half(FFX_MIN16_F3 x, FFX_MIN16_F3 y, FFX_MIN16_F3 z) +{ + return max(x, max(y, z)); +} +FFX_MIN16_F4 ffxMax3Half(FFX_MIN16_F4 x, FFX_MIN16_F4 y, FFX_MIN16_F4 z) +{ + return max(x, max(y, z)); +} +//------------------------------------------------------------------------------------------------------------------------------ +FFX_MIN16_F ffxMin3Half(FFX_MIN16_F x, FFX_MIN16_F y, FFX_MIN16_F z) +{ + return min(x, min(y, z)); +} +FFX_MIN16_F2 ffxMin3Half(FFX_MIN16_F2 x, FFX_MIN16_F2 y, FFX_MIN16_F2 z) +{ + return min(x, min(y, z)); +} +FFX_MIN16_F3 ffxMin3Half(FFX_MIN16_F3 x, FFX_MIN16_F3 y, FFX_MIN16_F3 z) +{ + return min(x, min(y, z)); +} +FFX_MIN16_F4 ffxMin3Half(FFX_MIN16_F4 x, FFX_MIN16_F4 y, FFX_MIN16_F4 z) +{ + return min(x, min(y, z)); +} +//------------------------------------------------------------------------------------------------------------------------------ +FFX_MIN16_F ffxMed3Half(FFX_MIN16_F x, FFX_MIN16_F y, FFX_MIN16_F z) +{ + return max(min(x, y), min(max(x, y), z)); +} +FFX_MIN16_F2 ffxMed3Half(FFX_MIN16_F2 x, FFX_MIN16_F2 y, FFX_MIN16_F2 z) +{ + return max(min(x, y), min(max(x, y), z)); +} +FFX_MIN16_F3 ffxMed3Half(FFX_MIN16_F3 x, FFX_MIN16_F3 y, FFX_MIN16_F3 z) +{ + return max(min(x, y), min(max(x, y), z)); +} +FFX_MIN16_F4 ffxMed3Half(FFX_MIN16_F4 x, FFX_MIN16_F4 y, FFX_MIN16_F4 z) +{ + return max(min(x, y), min(max(x, y), z)); +} +//------------------------------------------------------------------------------------------------------------------------------ +FFX_MIN16_I ffxMed3Half(FFX_MIN16_I x, FFX_MIN16_I y, FFX_MIN16_I z) +{ + return max(min(x, y), min(max(x, y), z)); +} +FFX_MIN16_I2 ffxMed3Half(FFX_MIN16_I2 x, FFX_MIN16_I2 y, FFX_MIN16_I2 z) +{ + return max(min(x, y), min(max(x, y), z)); +} +FFX_MIN16_I3 ffxMed3Half(FFX_MIN16_I3 x, FFX_MIN16_I3 y, FFX_MIN16_I3 z) +{ + return max(min(x, y), min(max(x, y), z)); +} +FFX_MIN16_I4 ffxMed3Half(FFX_MIN16_I4 x, FFX_MIN16_I4 y, FFX_MIN16_I4 z) +{ + return max(min(x, y), min(max(x, y), z)); +} +//------------------------------------------------------------------------------------------------------------------------------ +FFX_MIN16_F ffxReciprocalHalf(FFX_MIN16_F x) +{ + return rcp(x); +} +FFX_MIN16_F2 ffxReciprocalHalf(FFX_MIN16_F2 x) +{ + return rcp(x); +} +FFX_MIN16_F3 ffxReciprocalHalf(FFX_MIN16_F3 x) +{ + return rcp(x); +} +FFX_MIN16_F4 ffxReciprocalHalf(FFX_MIN16_F4 x) +{ + return rcp(x); +} +//------------------------------------------------------------------------------------------------------------------------------ +FFX_MIN16_F ffxReciprocalSquareRootHalf(FFX_MIN16_F x) +{ + return rsqrt(x); +} +FFX_MIN16_F2 ffxReciprocalSquareRootHalf(FFX_MIN16_F2 x) +{ + return rsqrt(x); +} +FFX_MIN16_F3 ffxReciprocalSquareRootHalf(FFX_MIN16_F3 x) +{ + return rsqrt(x); +} +FFX_MIN16_F4 ffxReciprocalSquareRootHalf(FFX_MIN16_F4 x) +{ + return rsqrt(x); +} +//------------------------------------------------------------------------------------------------------------------------------ +FFX_MIN16_F ffxSaturate(FFX_MIN16_F x) +{ + return saturate(x); +} +FFX_MIN16_F2 ffxSaturate(FFX_MIN16_F2 x) +{ + return saturate(x); +} +FFX_MIN16_F3 ffxSaturate(FFX_MIN16_F3 x) +{ + return saturate(x); +} +FFX_MIN16_F4 ffxSaturate(FFX_MIN16_F4 x) +{ + return saturate(x); +} +//------------------------------------------------------------------------------------------------------------------------------ +FFX_MIN16_U ffxBitShiftRightHalf(FFX_MIN16_U a, FFX_MIN16_U b) +{ + return FFX_MIN16_U(FFX_MIN16_I(a) >> FFX_MIN16_I(b)); +} +FFX_MIN16_U2 ffxBitShiftRightHalf(FFX_MIN16_U2 a, FFX_MIN16_U2 b) +{ + return FFX_MIN16_U2(FFX_MIN16_I2(a) >> FFX_MIN16_I2(b)); +} +FFX_MIN16_U3 ffxBitShiftRightHalf(FFX_MIN16_U3 a, FFX_MIN16_U3 b) +{ + return FFX_MIN16_U3(FFX_MIN16_I3(a) >> FFX_MIN16_I3(b)); +} +FFX_MIN16_U4 ffxBitShiftRightHalf(FFX_MIN16_U4 a, FFX_MIN16_U4 b) +{ + return FFX_MIN16_U4(FFX_MIN16_I4(a) >> FFX_MIN16_I4(b)); +} +#endif // FFX_HALF + +//============================================================================================================================== +// HLSL WAVE +//============================================================================================================================== +#if defined(FFX_WAVE) +// Where 'x' must be a compile time literal. +FfxFloat32 ffxWaveXorF1(FfxFloat32 v, FfxUInt32 x) +{ + return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); +} +FfxFloat32x2 ffxWaveXorF2(FfxFloat32x2 v, FfxUInt32 x) +{ + return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); +} +FfxFloat32x3 ffxWaveXorF3(FfxFloat32x3 v, FfxUInt32 x) +{ + return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); +} +FfxFloat32x4 ffxWaveXorF4(FfxFloat32x4 v, FfxUInt32 x) +{ + return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); +} +FfxUInt32 ffxWaveXorU1(FfxUInt32 v, FfxUInt32 x) +{ + return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); +} +FfxUInt32x2 ffxWaveXorU1(FfxUInt32x2 v, FfxUInt32 x) +{ + return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); +} +FfxUInt32x3 ffxWaveXorU1(FfxUInt32x3 v, FfxUInt32 x) +{ + return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); +} +FfxUInt32x4 ffxWaveXorU1(FfxUInt32x4 v, FfxUInt32 x) +{ + return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); +} +FfxBoolean ffxWaveIsFirstLane() +{ + return WaveIsFirstLane(); +} +FfxUInt32 ffxWaveLaneIndex() +{ + return WaveGetLaneIndex(); +} +FfxBoolean ffxWaveReadAtLaneIndexB1(FfxBoolean v, FfxUInt32 x) +{ + return WaveReadLaneAt(v, x); +} +FfxUInt32 ffxWavePrefixCountBits(FfxBoolean v) +{ + return WavePrefixCountBits(v); +} +FfxUInt32 ffxWaveActiveCountBits(FfxBoolean v) +{ + return WaveActiveCountBits(v); +} +FfxUInt32 ffxWaveReadLaneFirstU1(FfxUInt32 v) +{ + return WaveReadLaneFirst(v); +} +FfxUInt32x2 ffxWaveReadLaneFirstU2(FfxUInt32x2 v) +{ + return WaveReadLaneFirst(v); +} +FfxBoolean ffxWaveReadLaneFirstB1(FfxBoolean v) +{ + return WaveReadLaneFirst(v); +} +FfxUInt32 ffxWaveOr(FfxUInt32 a) +{ + return WaveActiveBitOr(a); +} +FfxUInt32 ffxWaveMin(FfxUInt32 a) +{ + return WaveActiveMin(a); +} +FfxFloat32 ffxWaveMin(FfxFloat32 a) +{ + return WaveActiveMin(a); +} +FfxUInt32 ffxWaveMax(FfxUInt32 a) +{ + return WaveActiveMax(a); +} +FfxFloat32 ffxWaveMax(FfxFloat32 a) +{ + return WaveActiveMax(a); +} +FfxUInt32 ffxWaveSum(FfxUInt32 a) +{ + return WaveActiveSum(a); +} +FfxFloat32 ffxWaveSum(FfxFloat32 a) +{ + return WaveActiveSum(a); +} +FfxUInt32 ffxWaveLaneCount() +{ + return WaveGetLaneCount(); +} +FfxBoolean ffxWaveAllTrue(FfxBoolean v) +{ + return WaveActiveAllTrue(v); +} +FfxFloat32 ffxQuadReadX(FfxFloat32 v) +{ + return QuadReadAcrossX(v); +} +FfxFloat32x2 ffxQuadReadX(FfxFloat32x2 v) +{ + return QuadReadAcrossX(v); +} +FfxFloat32 ffxQuadReadY(FfxFloat32 v) +{ + return QuadReadAcrossY(v); +} +FfxFloat32x2 ffxQuadReadY(FfxFloat32x2 v) +{ + return QuadReadAcrossY(v); +} + +#if FFX_HALF +FfxFloat16x2 ffxWaveXorFloat16x2(FfxFloat16x2 v, FfxUInt32 x) +{ + return FFX_UINT32_TO_FLOAT16X2(WaveReadLaneAt(FFX_FLOAT16X2_TO_UINT32(v), WaveGetLaneIndex() ^ x)); +} +FfxFloat16x4 ffxWaveXorFloat16x4(FfxFloat16x4 v, FfxUInt32 x) +{ + return FFX_UINT32X2_TO_FLOAT16X4(WaveReadLaneAt(FFX_FLOAT16X4_TO_UINT32X2(v), WaveGetLaneIndex() ^ x)); +} +FfxUInt16x2 ffxWaveXorUint16x2(FfxUInt16x2 v, FfxUInt32 x) +{ + return FFX_UINT32_TO_UINT16X2(WaveReadLaneAt(FFX_UINT16X2_TO_UINT32(v), WaveGetLaneIndex() ^ x)); +} +FfxUInt16x4 ffxWaveXorUint16x4(FfxUInt16x4 v, FfxUInt32 x) +{ + return FFX_UINT32X2_TO_UINT16X4(WaveReadLaneAt(FFX_UINT16X4_TO_UINT32X2(v), WaveGetLaneIndex() ^ x)); +} +#endif // FFX_HALF +#endif // #if defined(FFX_WAVE) diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_hlsl.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_hlsl.h.meta new file mode 100644 index 0000000..7c889bd --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_hlsl.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 47a04cb434355164b8da169fbd474688 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_portability.h b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_portability.h new file mode 100644 index 0000000..12147b9 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_portability.h @@ -0,0 +1,46 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +void ffxOpAAddOneF3(FFX_PARAMETER_OUT FfxFloat32x3 d, FfxFloat32x3 a, FfxFloat32 b) +{ + d = a + ffxBroadcast3(b); +} + +void ffxOpACpyF3(FFX_PARAMETER_OUT FfxFloat32x3 d, FfxFloat32x3 a) +{ + d = a; +} + +void ffxOpAMulF3(FFX_PARAMETER_OUT FfxFloat32x3 d, FfxFloat32x3 a, FfxFloat32x3 b) +{ + d = a * b; +} + +void ffxOpAMulOneF3(FFX_PARAMETER_OUT FfxFloat32x3 d, FfxFloat32x3 a, FfxFloat32 b) +{ + d = a * b; +} + +void ffxOpARcpF3(FFX_PARAMETER_OUT FfxFloat32x3 d, FfxFloat32x3 a) +{ + d = ffxReciprocal(a); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_portability.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_portability.h.meta new file mode 100644 index 0000000..2f87125 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_core_portability.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 9dc8a5ba90431934588adf0dabbf0cdd +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_accumulate_pass.hlsl b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_accumulate_pass.hlsl new file mode 100644 index 0000000..59cfd80 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_accumulate_pass.hlsl @@ -0,0 +1,79 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#define FSR2_BIND_SRV_INPUT_EXPOSURE 0 +#define FSR2_BIND_SRV_DILATED_REACTIVE_MASKS 1 +#if FFX_FSR2_OPTION_LOW_RESOLUTION_MOTION_VECTORS +#define FSR2_BIND_SRV_DILATED_MOTION_VECTORS 2 +#else +#define FSR2_BIND_SRV_INPUT_MOTION_VECTORS 2 +#endif +#define FSR2_BIND_SRV_INTERNAL_UPSCALED 3 +#define FSR2_BIND_SRV_LOCK_STATUS 4 +#define FSR2_BIND_SRV_PREPARED_INPUT_COLOR 5 +#define FSR2_BIND_SRV_LANCZOS_LUT 6 +#define FSR2_BIND_SRV_UPSCALE_MAXIMUM_BIAS_LUT 7 +#define FSR2_BIND_SRV_SCENE_LUMINANCE_MIPS 8 +#define FSR2_BIND_SRV_AUTO_EXPOSURE 9 +#define FSR2_BIND_SRV_LUMA_HISTORY 10 + +#define FSR2_BIND_UAV_INTERNAL_UPSCALED 0 +#define FSR2_BIND_UAV_LOCK_STATUS 1 +#define FSR2_BIND_UAV_UPSCALED_OUTPUT 2 +#define FSR2_BIND_UAV_NEW_LOCKS 3 +#define FSR2_BIND_UAV_LUMA_HISTORY 4 + +#define FSR2_BIND_CB_FSR2 0 + +#include "fsr2/ffx_fsr2_callbacks_hlsl.h" +#include "fsr2/ffx_fsr2_common.h" +#include "fsr2/ffx_fsr2_sample.h" +#include "fsr2/ffx_fsr2_upsample.h" +#include "fsr2/ffx_fsr2_postprocess_lock_status.h" +#include "fsr2/ffx_fsr2_reproject.h" +#include "fsr2/ffx_fsr2_accumulate.h" + +#ifndef FFX_FSR2_THREAD_GROUP_WIDTH +#define FFX_FSR2_THREAD_GROUP_WIDTH 8 +#endif // #ifndef FFX_FSR2_THREAD_GROUP_WIDTH +#ifndef FFX_FSR2_THREAD_GROUP_HEIGHT +#define FFX_FSR2_THREAD_GROUP_HEIGHT 8 +#endif // FFX_FSR2_THREAD_GROUP_HEIGHT +#ifndef FFX_FSR2_THREAD_GROUP_DEPTH +#define FFX_FSR2_THREAD_GROUP_DEPTH 1 +#endif // #ifndef FFX_FSR2_THREAD_GROUP_DEPTH +#ifndef FFX_FSR2_NUM_THREADS +#define FFX_FSR2_NUM_THREADS [numthreads(FFX_FSR2_THREAD_GROUP_WIDTH, FFX_FSR2_THREAD_GROUP_HEIGHT, FFX_FSR2_THREAD_GROUP_DEPTH)] +#endif // #ifndef FFX_FSR2_NUM_THREADS + +FFX_PREFER_WAVE64 +FFX_FSR2_NUM_THREADS +FFX_FSR2_EMBED_ROOTSIG_CONTENT +void CS(uint2 uGroupId : SV_GroupID, uint2 uGroupThreadId : SV_GroupThreadID) +{ + const uint GroupRows = (uint(DisplaySize().y) + FFX_FSR2_THREAD_GROUP_HEIGHT - 1) / FFX_FSR2_THREAD_GROUP_HEIGHT; + uGroupId.y = GroupRows - uGroupId.y - 1; + + uint2 uDispatchThreadId = uGroupId * uint2(FFX_FSR2_THREAD_GROUP_WIDTH, FFX_FSR2_THREAD_GROUP_HEIGHT) + uGroupThreadId; + + Accumulate(uDispatchThreadId); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_accumulate_pass.hlsl.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_accumulate_pass.hlsl.meta new file mode 100644 index 0000000..a5a43a9 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_accumulate_pass.hlsl.meta @@ -0,0 +1,7 @@ +fileFormatVersion: 2 +guid: 2587b11e2d882c649ac7fd1a51f6dc8c +ShaderIncludeImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_autogen_reactive_pass.hlsl b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_autogen_reactive_pass.hlsl new file mode 100644 index 0000000..1a5995b --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_autogen_reactive_pass.hlsl @@ -0,0 +1,78 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#define FSR2_BIND_SRV_INPUT_OPAQUE_ONLY 0 +#define FSR2_BIND_SRV_INPUT_COLOR 1 + +#define FSR2_BIND_UAV_AUTOREACTIVE 0 + +#define FSR2_BIND_CB_FSR2 0 +#define FSR2_BIND_CB_REACTIVE 1 + +#include "fsr2/ffx_fsr2_callbacks_hlsl.h" +#include "fsr2/ffx_fsr2_common.h" + +#ifndef FFX_FSR2_THREAD_GROUP_WIDTH +#define FFX_FSR2_THREAD_GROUP_WIDTH 8 +#endif // #ifndef FFX_FSR2_THREAD_GROUP_WIDTH +#ifndef FFX_FSR2_THREAD_GROUP_HEIGHT +#define FFX_FSR2_THREAD_GROUP_HEIGHT 8 +#endif // FFX_FSR2_THREAD_GROUP_HEIGHT +#ifndef FFX_FSR2_THREAD_GROUP_DEPTH +#define FFX_FSR2_THREAD_GROUP_DEPTH 1 +#endif // #ifndef FFX_FSR2_THREAD_GROUP_DEPTH +#ifndef FFX_FSR2_NUM_THREADS +#define FFX_FSR2_NUM_THREADS [numthreads(FFX_FSR2_THREAD_GROUP_WIDTH, FFX_FSR2_THREAD_GROUP_HEIGHT, FFX_FSR2_THREAD_GROUP_DEPTH)] +#endif // #ifndef FFX_FSR2_NUM_THREADS + +FFX_PREFER_WAVE64 +FFX_FSR2_NUM_THREADS +FFX_FSR2_EMBED_ROOTSIG_REACTIVE_CONTENT +void CS(uint2 uGroupId : SV_GroupID, uint2 uGroupThreadId : SV_GroupThreadID) +{ + uint2 uDispatchThreadId = uGroupId * uint2(FFX_FSR2_THREAD_GROUP_WIDTH, FFX_FSR2_THREAD_GROUP_HEIGHT) + uGroupThreadId; + + float3 ColorPreAlpha = LoadOpaqueOnly( FFX_MIN16_I2(uDispatchThreadId) ).rgb; + float3 ColorPostAlpha = LoadInputColor(uDispatchThreadId).rgb; + + if (GenReactiveFlags() & FFX_FSR2_AUTOREACTIVEFLAGS_APPLY_TONEMAP) + { + ColorPreAlpha = Tonemap(ColorPreAlpha); + ColorPostAlpha = Tonemap(ColorPostAlpha); + } + + if (GenReactiveFlags() & FFX_FSR2_AUTOREACTIVEFLAGS_APPLY_INVERSETONEMAP) + { + ColorPreAlpha = InverseTonemap(ColorPreAlpha); + ColorPostAlpha = InverseTonemap(ColorPostAlpha); + } + + float out_reactive_value = 0.f; + float3 delta = abs(ColorPostAlpha - ColorPreAlpha); + + out_reactive_value = (GenReactiveFlags() & FFX_FSR2_AUTOREACTIVEFLAGS_USE_COMPONENTS_MAX) ? max(delta.x, max(delta.y, delta.z)) : length(delta); + out_reactive_value *= GenReactiveScale(); + + out_reactive_value = (GenReactiveFlags() & FFX_FSR2_AUTOREACTIVEFLAGS_APPLY_THRESHOLD) ? (out_reactive_value < GenReactiveThreshold() ? 0 : GenReactiveBinaryValue()) : out_reactive_value; + + rw_output_autoreactive[uDispatchThreadId] = out_reactive_value; +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_autogen_reactive_pass.hlsl.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_autogen_reactive_pass.hlsl.meta new file mode 100644 index 0000000..776a7af --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_autogen_reactive_pass.hlsl.meta @@ -0,0 +1,7 @@ +fileFormatVersion: 2 +guid: 15c74efa821aaf3429f92b8d5d9901d5 +ShaderIncludeImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_compute_luminance_pyramid_pass.hlsl b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_compute_luminance_pyramid_pass.hlsl new file mode 100644 index 0000000..31de05f --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_compute_luminance_pyramid_pass.hlsl @@ -0,0 +1,56 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#define FSR2_BIND_SRV_INPUT_COLOR 0 + +#define FSR2_BIND_UAV_SPD_GLOBAL_ATOMIC 0 +#define FSR2_BIND_UAV_EXPOSURE_MIP_LUMA_CHANGE 1 +#define FSR2_BIND_UAV_EXPOSURE_MIP_5 2 +#define FSR2_BIND_UAV_AUTO_EXPOSURE 3 + +#define FSR2_BIND_CB_FSR2 0 +#define FSR2_BIND_CB_SPD 1 + +#include "fsr2/ffx_fsr2_callbacks_hlsl.h" +#include "fsr2/ffx_fsr2_common.h" +#include "fsr2/ffx_fsr2_compute_luminance_pyramid.h" + +#ifndef FFX_FSR2_THREAD_GROUP_WIDTH +#define FFX_FSR2_THREAD_GROUP_WIDTH 256 +#endif // #ifndef FFX_FSR2_THREAD_GROUP_WIDTH +#ifndef FFX_FSR2_THREAD_GROUP_HEIGHT +#define FFX_FSR2_THREAD_GROUP_HEIGHT 1 +#endif // #ifndef FFX_FSR2_THREAD_GROUP_HEIGHT +#ifndef FFX_FSR2_THREAD_GROUP_DEPTH +#define FFX_FSR2_THREAD_GROUP_DEPTH 1 +#endif // #ifndef FFX_FSR2_THREAD_GROUP_DEPTH +#ifndef FFX_FSR2_NUM_THREADS +#define FFX_FSR2_NUM_THREADS [numthreads(FFX_FSR2_THREAD_GROUP_WIDTH, FFX_FSR2_THREAD_GROUP_HEIGHT, FFX_FSR2_THREAD_GROUP_DEPTH)] +#endif // #ifndef FFX_FSR2_NUM_THREADS + +FFX_PREFER_WAVE64 +FFX_FSR2_NUM_THREADS +FFX_FSR2_EMBED_CB2_ROOTSIG_CONTENT +void CS(uint3 WorkGroupId : SV_GroupID, uint LocalThreadIndex : SV_GroupIndex) +{ + ComputeAutoExposure(WorkGroupId, LocalThreadIndex); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_compute_luminance_pyramid_pass.hlsl.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_compute_luminance_pyramid_pass.hlsl.meta new file mode 100644 index 0000000..8ed0c35 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_compute_luminance_pyramid_pass.hlsl.meta @@ -0,0 +1,7 @@ +fileFormatVersion: 2 +guid: 630d959ddf3e02e4087e181b7d7cd7ed +ShaderIncludeImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_depth_clip_pass.hlsl b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_depth_clip_pass.hlsl new file mode 100644 index 0000000..30bbcc0 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_depth_clip_pass.hlsl @@ -0,0 +1,67 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#define FSR2_BIND_SRV_RECONSTRUCTED_PREV_NEAREST_DEPTH 0 +#define FSR2_BIND_SRV_DILATED_MOTION_VECTORS 1 +#define FSR2_BIND_SRV_DILATED_DEPTH 2 +#define FSR2_BIND_SRV_REACTIVE_MASK 3 +#define FSR2_BIND_SRV_TRANSPARENCY_AND_COMPOSITION_MASK 4 +#define FSR2_BIND_SRV_PREVIOUS_DILATED_MOTION_VECTORS 5 +#define FSR2_BIND_SRV_INPUT_MOTION_VECTORS 6 +#define FSR2_BIND_SRV_INPUT_COLOR 7 +#define FSR2_BIND_SRV_INPUT_DEPTH 8 +#define FSR2_BIND_SRV_INPUT_EXPOSURE 9 + +#define FSR2_BIND_UAV_DILATED_REACTIVE_MASKS 0 +#define FSR2_BIND_UAV_PREPARED_INPUT_COLOR 1 + +#define FSR2_BIND_CB_FSR2 0 + +#include "fsr2/ffx_fsr2_callbacks_hlsl.h" +#include "fsr2/ffx_fsr2_common.h" +#include "fsr2/ffx_fsr2_sample.h" +#include "fsr2/ffx_fsr2_depth_clip.h" + +#ifndef FFX_FSR2_THREAD_GROUP_WIDTH +#define FFX_FSR2_THREAD_GROUP_WIDTH 8 +#endif // #ifndef FFX_FSR2_THREAD_GROUP_WIDTH +#ifndef FFX_FSR2_THREAD_GROUP_HEIGHT +#define FFX_FSR2_THREAD_GROUP_HEIGHT 8 +#endif // #ifndef FFX_FSR2_THREAD_GROUP_HEIGHT +#ifndef FFX_FSR2_THREAD_GROUP_DEPTH +#define FFX_FSR2_THREAD_GROUP_DEPTH 1 +#endif // #ifndef FFX_FSR2_THREAD_GROUP_DEPTH +#ifndef FFX_FSR2_NUM_THREADS +#define FFX_FSR2_NUM_THREADS [numthreads(FFX_FSR2_THREAD_GROUP_WIDTH, FFX_FSR2_THREAD_GROUP_HEIGHT, FFX_FSR2_THREAD_GROUP_DEPTH)] +#endif // #ifndef FFX_FSR2_NUM_THREADS + +FFX_PREFER_WAVE64 +FFX_FSR2_NUM_THREADS +FFX_FSR2_EMBED_ROOTSIG_CONTENT +void CS( + int2 iGroupId : SV_GroupID, + int2 iDispatchThreadId : SV_DispatchThreadID, + int2 iGroupThreadId : SV_GroupThreadID, + int iGroupIndex : SV_GroupIndex) +{ + DepthClip(iDispatchThreadId); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_depth_clip_pass.hlsl.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_depth_clip_pass.hlsl.meta new file mode 100644 index 0000000..794fe20 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_depth_clip_pass.hlsl.meta @@ -0,0 +1,7 @@ +fileFormatVersion: 2 +guid: 38a518d5929d9134096f87f66f59a285 +ShaderIncludeImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_lock_pass.hlsl b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_lock_pass.hlsl new file mode 100644 index 0000000..dec17fe --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_lock_pass.hlsl @@ -0,0 +1,56 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#define FSR2_BIND_SRV_LOCK_INPUT_LUMA 0 + +#define FSR2_BIND_UAV_NEW_LOCKS 0 +#define FSR2_BIND_UAV_RECONSTRUCTED_PREV_NEAREST_DEPTH 1 + +#define FSR2_BIND_CB_FSR2 0 + +#include "fsr2/ffx_fsr2_callbacks_hlsl.h" +#include "fsr2/ffx_fsr2_common.h" +#include "fsr2/ffx_fsr2_sample.h" +#include "fsr2/ffx_fsr2_lock.h" + +#ifndef FFX_FSR2_THREAD_GROUP_WIDTH +#define FFX_FSR2_THREAD_GROUP_WIDTH 8 +#endif // #ifndef FFX_FSR2_THREAD_GROUP_WIDTH +#ifndef FFX_FSR2_THREAD_GROUP_HEIGHT +#define FFX_FSR2_THREAD_GROUP_HEIGHT 8 +#endif // #ifndef FFX_FSR2_THREAD_GROUP_HEIGHT +#ifndef FFX_FSR2_THREAD_GROUP_DEPTH +#define FFX_FSR2_THREAD_GROUP_DEPTH 1 +#endif // #ifndef FFX_FSR2_THREAD_GROUP_DEPTH +#ifndef FFX_FSR2_NUM_THREADS +#define FFX_FSR2_NUM_THREADS [numthreads(FFX_FSR2_THREAD_GROUP_WIDTH, FFX_FSR2_THREAD_GROUP_HEIGHT, FFX_FSR2_THREAD_GROUP_DEPTH)] +#endif // #ifndef FFX_FSR2_NUM_THREADS + +FFX_PREFER_WAVE64 +FFX_FSR2_NUM_THREADS +FFX_FSR2_EMBED_ROOTSIG_CONTENT +void CS(uint2 uGroupId : SV_GroupID, uint2 uGroupThreadId : SV_GroupThreadID) +{ + uint2 uDispatchThreadId = uGroupId * uint2(FFX_FSR2_THREAD_GROUP_WIDTH, FFX_FSR2_THREAD_GROUP_HEIGHT) + uGroupThreadId; + + ComputeLock(uDispatchThreadId); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_lock_pass.hlsl.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_lock_pass.hlsl.meta new file mode 100644 index 0000000..7dcfaed --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_lock_pass.hlsl.meta @@ -0,0 +1,7 @@ +fileFormatVersion: 2 +guid: c3bf93953aee91541a1be089c2ed268f +ShaderIncludeImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_rcas_pass.hlsl b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_rcas_pass.hlsl new file mode 100644 index 0000000..6fd6d29 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_rcas_pass.hlsl @@ -0,0 +1,54 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#define FSR2_BIND_SRV_INPUT_EXPOSURE 0 +#define FSR2_BIND_SRV_RCAS_INPUT 1 + +#define FSR2_BIND_UAV_UPSCALED_OUTPUT 0 + +#define FSR2_BIND_CB_FSR2 0 +#define FSR2_BIND_CB_RCAS 1 + +#include "fsr2/ffx_fsr2_callbacks_hlsl.h" +#include "fsr2/ffx_fsr2_common.h" +#include "fsr2/ffx_fsr2_rcas.h" + +#ifndef FFX_FSR2_THREAD_GROUP_WIDTH +#define FFX_FSR2_THREAD_GROUP_WIDTH 64 +#endif // #ifndef FFX_FSR2_THREAD_GROUP_WIDTH +#ifndef FFX_FSR2_THREAD_GROUP_HEIGHT +#define FFX_FSR2_THREAD_GROUP_HEIGHT 1 +#endif // #ifndef FFX_FSR2_THREAD_GROUP_HEIGHT +#ifndef FFX_FSR2_THREAD_GROUP_DEPTH +#define FFX_FSR2_THREAD_GROUP_DEPTH 1 +#endif // #ifndef FFX_FSR2_THREAD_GROUP_DEPTH +#ifndef FFX_FSR2_NUM_THREADS +#define FFX_FSR2_NUM_THREADS [numthreads(FFX_FSR2_THREAD_GROUP_WIDTH, FFX_FSR2_THREAD_GROUP_HEIGHT, FFX_FSR2_THREAD_GROUP_DEPTH)] +#endif // #ifndef FFX_FSR2_NUM_THREADS + +FFX_PREFER_WAVE64 +FFX_FSR2_NUM_THREADS +FFX_FSR2_EMBED_CB2_ROOTSIG_CONTENT +void CS(uint3 LocalThreadId : SV_GroupThreadID, uint3 WorkGroupId : SV_GroupID, uint3 Dtid : SV_DispatchThreadID) +{ + RCAS(LocalThreadId, WorkGroupId, Dtid); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_rcas_pass.hlsl.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_rcas_pass.hlsl.meta new file mode 100644 index 0000000..b1198a1 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_rcas_pass.hlsl.meta @@ -0,0 +1,7 @@ +fileFormatVersion: 2 +guid: 43f9d03e57211f8478e3388eecb8d5f8 +ShaderIncludeImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_reconstruct_previous_depth_pass.hlsl b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_reconstruct_previous_depth_pass.hlsl new file mode 100644 index 0000000..7e17318 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_reconstruct_previous_depth_pass.hlsl @@ -0,0 +1,64 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#define FSR2_BIND_SRV_INPUT_MOTION_VECTORS 0 +#define FSR2_BIND_SRV_INPUT_DEPTH 1 +#define FSR2_BIND_SRV_INPUT_COLOR 2 +#define FSR2_BIND_SRV_INPUT_EXPOSURE 3 + +#define FSR2_BIND_UAV_RECONSTRUCTED_PREV_NEAREST_DEPTH 0 +#define FSR2_BIND_UAV_DILATED_MOTION_VECTORS 1 +#define FSR2_BIND_UAV_DILATED_DEPTH 2 +#define FSR2_BIND_UAV_LOCK_INPUT_LUMA 3 + +#define FSR2_BIND_CB_FSR2 0 + +#include "fsr2/ffx_fsr2_callbacks_hlsl.h" +#include "fsr2/ffx_fsr2_common.h" +#include "fsr2/ffx_fsr2_sample.h" +#include "fsr2/ffx_fsr2_reconstruct_dilated_velocity_and_previous_depth.h" + +#ifndef FFX_FSR2_THREAD_GROUP_WIDTH +#define FFX_FSR2_THREAD_GROUP_WIDTH 8 +#endif // #ifndef FFX_FSR2_THREAD_GROUP_WIDTH +#ifndef FFX_FSR2_THREAD_GROUP_HEIGHT +#define FFX_FSR2_THREAD_GROUP_HEIGHT 8 +#endif // #ifndef FFX_FSR2_THREAD_GROUP_HEIGHT +#ifndef FFX_FSR2_THREAD_GROUP_DEPTH +#define FFX_FSR2_THREAD_GROUP_DEPTH 1 +#endif // #ifndef FFX_FSR2_THREAD_GROUP_DEPTH +#ifndef FFX_FSR2_NUM_THREADS +#define FFX_FSR2_NUM_THREADS [numthreads(FFX_FSR2_THREAD_GROUP_WIDTH, FFX_FSR2_THREAD_GROUP_HEIGHT, FFX_FSR2_THREAD_GROUP_DEPTH)] +#endif // #ifndef FFX_FSR2_NUM_THREADS + +FFX_PREFER_WAVE64 +FFX_FSR2_NUM_THREADS +FFX_FSR2_EMBED_ROOTSIG_CONTENT +void CS( + int2 iGroupId : SV_GroupID, + int2 iDispatchThreadId : SV_DispatchThreadID, + int2 iGroupThreadId : SV_GroupThreadID, + int iGroupIndex : SV_GroupIndex +) +{ + ReconstructAndDilate(iDispatchThreadId); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_reconstruct_previous_depth_pass.hlsl.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_reconstruct_previous_depth_pass.hlsl.meta new file mode 100644 index 0000000..622615b --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_reconstruct_previous_depth_pass.hlsl.meta @@ -0,0 +1,7 @@ +fileFormatVersion: 2 +guid: c7220f38e87579a4788ddaabba0a6d97 +ShaderIncludeImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_tcr_autogen_pass.hlsl b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_tcr_autogen_pass.hlsl new file mode 100644 index 0000000..765cc8b --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_tcr_autogen_pass.hlsl @@ -0,0 +1,92 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#define FSR2_BIND_SRV_INPUT_OPAQUE_ONLY 0 +#define FSR2_BIND_SRV_INPUT_COLOR 1 +#define FSR2_BIND_SRV_INPUT_MOTION_VECTORS 2 +#define FSR2_BIND_SRV_PREV_PRE_ALPHA_COLOR 3 +#define FSR2_BIND_SRV_PREV_POST_ALPHA_COLOR 4 +#define FSR2_BIND_SRV_REACTIVE_MASK 4 +#define FSR2_BIND_SRV_TRANSPARENCY_AND_COMPOSITION_MASK 5 + +#define FSR2_BIND_UAV_AUTOREACTIVE 0 +#define FSR2_BIND_UAV_AUTOCOMPOSITION 1 +#define FSR2_BIND_UAV_PREV_PRE_ALPHA_COLOR 2 +#define FSR2_BIND_UAV_PREV_POST_ALPHA_COLOR 3 + +#define FSR2_BIND_CB_FSR2 0 +#define FSR2_BIND_CB_AUTOREACTIVE 1 + +#include "fsr2/ffx_fsr2_callbacks_hlsl.h" +#include "fsr2/ffx_fsr2_common.h" +#include "fsr2/ffx_fsr2_tcr_autogen.h" + +#ifndef FFX_FSR2_THREAD_GROUP_WIDTH +#define FFX_FSR2_THREAD_GROUP_WIDTH 8 +#endif // #ifndef FFX_FSR2_THREAD_GROUP_WIDTH +#ifndef FFX_FSR2_THREAD_GROUP_HEIGHT +#define FFX_FSR2_THREAD_GROUP_HEIGHT 8 +#endif // FFX_FSR2_THREAD_GROUP_HEIGHT +#ifndef FFX_FSR2_THREAD_GROUP_DEPTH +#define FFX_FSR2_THREAD_GROUP_DEPTH 1 +#endif // #ifndef FFX_FSR2_THREAD_GROUP_DEPTH +#ifndef FFX_FSR2_NUM_THREADS +#define FFX_FSR2_NUM_THREADS [numthreads(FFX_FSR2_THREAD_GROUP_WIDTH, FFX_FSR2_THREAD_GROUP_HEIGHT, FFX_FSR2_THREAD_GROUP_DEPTH)] +#endif // #ifndef FFX_FSR2_NUM_THREADS + +FFX_PREFER_WAVE64 +FFX_FSR2_NUM_THREADS +FFX_FSR2_EMBED_ROOTSIG_REACTIVE_CONTENT +void CS(uint2 uGroupId : SV_GroupID, uint2 uGroupThreadId : SV_GroupThreadID) +{ + FFX_MIN16_I2 uDispatchThreadId = FFX_MIN16_I2(uGroupId * uint2(FFX_FSR2_THREAD_GROUP_WIDTH, FFX_FSR2_THREAD_GROUP_HEIGHT) + uGroupThreadId); + + // ToDo: take into account jitter (i.e. add delta of previous jitter and current jitter to previous UV + // fetch pre- and post-alpha color values + FFX_MIN16_F2 fUv = ( FFX_MIN16_F2(uDispatchThreadId) + FFX_MIN16_F2(0.5f, 0.5f) ) / FFX_MIN16_F2( RenderSize() ); + FFX_MIN16_F2 fPrevUV = fUv + FFX_MIN16_F2( LoadInputMotionVector(uDispatchThreadId) ); + FFX_MIN16_I2 iPrevIdx = FFX_MIN16_I2(fPrevUV * FFX_MIN16_F2(RenderSize()) - 0.5f); + + FFX_MIN16_F3 colorPreAlpha = FFX_MIN16_F3( LoadOpaqueOnly( uDispatchThreadId ) ); + FFX_MIN16_F3 colorPostAlpha = FFX_MIN16_F3( LoadInputColor( uDispatchThreadId ) ); + + FFX_MIN16_F2 outReactiveMask = 0; + + outReactiveMask.y = ComputeTransparencyAndComposition(uDispatchThreadId, iPrevIdx); + + if (outReactiveMask.y > 0.5f) + { + outReactiveMask.x = ComputeReactive(uDispatchThreadId, iPrevIdx); + outReactiveMask.x *= FFX_MIN16_F(ReactiveScale()); + outReactiveMask.x = outReactiveMask.x < ReactiveMax() ? outReactiveMask.x : FFX_MIN16_F(ReactiveMax()); + } + + outReactiveMask.y *= FFX_MIN16_F( TcScale() ); + + outReactiveMask.x = max( outReactiveMask.x, FFX_MIN16_F( LoadReactiveMask(uDispatchThreadId) ) ); + outReactiveMask.y = max( outReactiveMask.y, FFX_MIN16_F( LoadTransparencyAndCompositionMask(uDispatchThreadId) ) ); + + StoreAutoReactive(uDispatchThreadId, outReactiveMask); + + StorePrevPreAlpha(uDispatchThreadId, colorPreAlpha); + StorePrevPostAlpha(uDispatchThreadId, colorPostAlpha); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_tcr_autogen_pass.hlsl.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_tcr_autogen_pass.hlsl.meta new file mode 100644 index 0000000..b6fd348 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr2_tcr_autogen_pass.hlsl.meta @@ -0,0 +1,7 @@ +fileFormatVersion: 2 +guid: d96e2a0e75dd01d47a231e3c4d18be0c +ShaderIncludeImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_accumulate_pass.hlsl b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_accumulate_pass.hlsl new file mode 100644 index 0000000..c000624 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_accumulate_pass.hlsl @@ -0,0 +1,70 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#define FSR3UPSCALER_BIND_SRV_INPUT_EXPOSURE 0 +#define FSR3UPSCALER_BIND_SRV_DILATED_REACTIVE_MASKS 1 +#if FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS +#define FSR3UPSCALER_BIND_SRV_DILATED_MOTION_VECTORS 2 +#else +#define FSR3UPSCALER_BIND_SRV_INPUT_MOTION_VECTORS 2 +#endif +#define FSR3UPSCALER_BIND_SRV_INTERNAL_UPSCALED 3 +#define FSR3UPSCALER_BIND_SRV_LANCZOS_LUT 4 +#define FSR3UPSCALER_BIND_SRV_FARTHEST_DEPTH_MIP1 5 + +#define FSR3UPSCALER_BIND_SRV_CURRENT_LUMA 6 +#define FSR3UPSCALER_BIND_SRV_LUMA_INSTABILITY 7 +#define FSR3UPSCALER_BIND_SRV_INPUT_COLOR 8 + +#define FSR3UPSCALER_BIND_UAV_INTERNAL_UPSCALED 0 +#define FSR3UPSCALER_BIND_UAV_UPSCALED_OUTPUT 1 +#define FSR3UPSCALER_BIND_UAV_NEW_LOCKS 2 + +#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 + +#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" +#include "fsr3upscaler/ffx_fsr3upscaler_common.h" +#include "fsr3upscaler/ffx_fsr3upscaler_sample.h" +#include "fsr3upscaler/ffx_fsr3upscaler_upsample.h" +#include "fsr3upscaler/ffx_fsr3upscaler_reproject.h" +#include "fsr3upscaler/ffx_fsr3upscaler_accumulate.h" + +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH +#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 8 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT +#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 8 +#endif // FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH +#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH +#ifndef FFX_FSR3UPSCALER_NUM_THREADS +#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] +#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS + +FFX_PREFER_WAVE64 +FFX_FSR3UPSCALER_NUM_THREADS +FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT +void CS(int2 iDispatchThreadId : SV_DispatchThreadID) +{ + Accumulate(iDispatchThreadId); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_accumulate_pass.hlsl.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_accumulate_pass.hlsl.meta new file mode 100644 index 0000000..5cdfe03 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_accumulate_pass.hlsl.meta @@ -0,0 +1,7 @@ +fileFormatVersion: 2 +guid: 28932314459baa6449098e1b5862470c +ShaderIncludeImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_autogen_reactive_pass.hlsl b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_autogen_reactive_pass.hlsl new file mode 100644 index 0000000..c052a1d --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_autogen_reactive_pass.hlsl @@ -0,0 +1,78 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#define FSR3UPSCALER_BIND_SRV_INPUT_OPAQUE_ONLY 0 +#define FSR3UPSCALER_BIND_SRV_INPUT_COLOR 1 + +#define FSR3UPSCALER_BIND_UAV_AUTOREACTIVE 0 +#define FSR3UPSCALER_BIND_UAV_AUTOCOMPOSITION 1 + +#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 +#define FSR3UPSCALER_BIND_CB_REACTIVE 1 + +#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" +#include "fsr3upscaler/ffx_fsr3upscaler_common.h" + +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH +#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 8 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT +#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 8 +#endif // FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH +#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH +#ifndef FFX_FSR3UPSCALER_NUM_THREADS +#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] +#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS + +FFX_FSR3UPSCALER_NUM_THREADS +FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT +void CS(uint2 uGroupId : SV_GroupID, uint2 uGroupThreadId : SV_GroupThreadID) +{ + uint2 uDispatchThreadId = uGroupId * uint2(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT) + uGroupThreadId; + + float3 ColorPreAlpha = LoadOpaqueOnly( FFX_MIN16_I2(uDispatchThreadId) ).rgb; + float3 ColorPostAlpha = LoadInputColor(uDispatchThreadId).rgb; + + if (GenReactiveFlags() & FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_APPLY_TONEMAP) + { + ColorPreAlpha = Tonemap(ColorPreAlpha); + ColorPostAlpha = Tonemap(ColorPostAlpha); + } + + if (GenReactiveFlags() & FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_APPLY_INVERSETONEMAP) + { + ColorPreAlpha = InverseTonemap(ColorPreAlpha); + ColorPostAlpha = InverseTonemap(ColorPostAlpha); + } + + float out_reactive_value = 0.f; + float3 delta = abs(ColorPostAlpha - ColorPreAlpha); + + out_reactive_value = (GenReactiveFlags() & FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_USE_COMPONENTS_MAX) ? max(delta.x, max(delta.y, delta.z)) : length(delta); + out_reactive_value *= GenReactiveScale(); + + out_reactive_value = (GenReactiveFlags() & FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_APPLY_THRESHOLD) ? (out_reactive_value < GenReactiveThreshold() ? 0 : GenReactiveBinaryValue()) : out_reactive_value; + + rw_output_autoreactive[uDispatchThreadId] = out_reactive_value; +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_autogen_reactive_pass.hlsl.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_autogen_reactive_pass.hlsl.meta new file mode 100644 index 0000000..0eb056a --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_autogen_reactive_pass.hlsl.meta @@ -0,0 +1,7 @@ +fileFormatVersion: 2 +guid: 0731dfbd04968084582397e48795d9d8 +ShaderIncludeImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_debug_view_pass.hlsl b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_debug_view_pass.hlsl new file mode 100644 index 0000000..6ac5067 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_debug_view_pass.hlsl @@ -0,0 +1,56 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#define FSR3UPSCALER_BIND_SRV_DILATED_REACTIVE_MASKS 0 +#define FSR3UPSCALER_BIND_SRV_DILATED_MOTION_VECTORS 1 +#define FSR3UPSCALER_BIND_SRV_DILATED_DEPTH 2 +#define FSR3UPSCALER_BIND_SRV_INTERNAL_UPSCALED 3 +#define FSR3UPSCALER_BIND_SRV_INPUT_EXPOSURE 4 + +#define FSR3UPSCALER_BIND_UAV_UPSCALED_OUTPUT 0 + +#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 + +#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" +#include "fsr3upscaler/ffx_fsr3upscaler_common.h" +#include "fsr3upscaler/ffx_fsr3upscaler_debug_view.h" + +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH +#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 8 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT +#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 8 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH +#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH +#ifndef FFX_FSR3UPSCALER_NUM_THREADS +#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] +#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS + +FFX_PREFER_WAVE64 +FFX_FSR3UPSCALER_NUM_THREADS +FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT +void CS(FfxInt32x2 iPxPos : SV_DispatchThreadID) +{ + DebugView(iPxPos); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_debug_view_pass.hlsl.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_debug_view_pass.hlsl.meta new file mode 100644 index 0000000..2a75154 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_debug_view_pass.hlsl.meta @@ -0,0 +1,7 @@ +fileFormatVersion: 2 +guid: 00c8328353e3dba449b5e3855add5a35 +ShaderIncludeImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_luma_instability_pass.hlsl b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_luma_instability_pass.hlsl new file mode 100644 index 0000000..44d97b3 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_luma_instability_pass.hlsl @@ -0,0 +1,59 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#define FSR3UPSCALER_BIND_SRV_INPUT_EXPOSURE 0 +#define FSR3UPSCALER_BIND_SRV_DILATED_REACTIVE_MASKS 1 +#define FSR3UPSCALER_BIND_SRV_DILATED_MOTION_VECTORS 2 +#define FSR3UPSCALER_BIND_SRV_FRAME_INFO 3 +#define FSR3UPSCALER_BIND_SRV_LUMA_HISTORY 4 +#define FSR3UPSCALER_BIND_SRV_FARTHEST_DEPTH_MIP1 5 +#define FSR3UPSCALER_BIND_SRV_CURRENT_LUMA 6 + +#define FSR3UPSCALER_BIND_UAV_LUMA_HISTORY 0 +#define FSR3UPSCALER_BIND_UAV_LUMA_INSTABILITY 1 + +#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 + +#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" +#include "fsr3upscaler/ffx_fsr3upscaler_common.h" +#include "fsr3upscaler/ffx_fsr3upscaler_luma_instability.h" + +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH +#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 8 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT +#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 8 +#endif // FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH +#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH +#ifndef FFX_FSR3UPSCALER_NUM_THREADS +#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] +#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS + +FFX_PREFER_WAVE64 +FFX_FSR3UPSCALER_NUM_THREADS +FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT +void CS(int2 iDispatchThreadId : SV_DispatchThreadID) +{ + LumaInstability(iDispatchThreadId); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_luma_instability_pass.hlsl.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_luma_instability_pass.hlsl.meta new file mode 100644 index 0000000..7e261f7 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_luma_instability_pass.hlsl.meta @@ -0,0 +1,7 @@ +fileFormatVersion: 2 +guid: 68eed10cb1a8742438be0ba0c08d5b05 +ShaderIncludeImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_luma_pyramid_pass.hlsl b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_luma_pyramid_pass.hlsl new file mode 100644 index 0000000..37fa1b7 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_luma_pyramid_pass.hlsl @@ -0,0 +1,62 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#define FSR3UPSCALER_BIND_SRV_CURRENT_LUMA 0 +#define FSR3UPSCALER_BIND_SRV_FARTHEST_DEPTH 1 + +#define FSR3UPSCALER_BIND_UAV_SPD_GLOBAL_ATOMIC 0 +#define FSR3UPSCALER_BIND_UAV_FRAME_INFO 1 +#define FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_0 2 +#define FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_1 3 +#define FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_2 4 +#define FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_3 5 +#define FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_4 6 +#define FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_5 7 +#define FSR3UPSCALER_BIND_UAV_FARTHEST_DEPTH_MIP1 8 + + +#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 +#define FSR3UPSCALER_BIND_CB_SPD 1 + +#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" +#include "fsr3upscaler/ffx_fsr3upscaler_common.h" +#include "fsr3upscaler/ffx_fsr3upscaler_luma_pyramid.h" + +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH +#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 256 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT +#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 1 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH +#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH +#ifndef FFX_FSR3UPSCALER_NUM_THREADS +#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] +#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS + +FFX_FSR3UPSCALER_NUM_THREADS +FFX_FSR3UPSCALER_EMBED_CB2_ROOTSIG_CONTENT +void CS(uint3 WorkGroupId : SV_GroupID, uint LocalThreadIndex : SV_GroupIndex) +{ + ComputeAutoExposure(WorkGroupId, LocalThreadIndex); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_luma_pyramid_pass.hlsl.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_luma_pyramid_pass.hlsl.meta new file mode 100644 index 0000000..71ca10c --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_luma_pyramid_pass.hlsl.meta @@ -0,0 +1,7 @@ +fileFormatVersion: 2 +guid: 10b66892c68bbe1408580a330cccb381 +ShaderIncludeImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_prepare_inputs_pass.hlsl b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_prepare_inputs_pass.hlsl new file mode 100644 index 0000000..d40b7d2 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_prepare_inputs_pass.hlsl @@ -0,0 +1,58 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#define FSR3UPSCALER_BIND_SRV_INPUT_MOTION_VECTORS 0 +#define FSR3UPSCALER_BIND_SRV_INPUT_DEPTH 1 +#define FSR3UPSCALER_BIND_SRV_INPUT_COLOR 2 + +#define FSR3UPSCALER_BIND_UAV_DILATED_MOTION_VECTORS 0 +#define FSR3UPSCALER_BIND_UAV_DILATED_DEPTH 1 +#define FSR3UPSCALER_BIND_UAV_RECONSTRUCTED_PREV_NEAREST_DEPTH 2 +#define FSR3UPSCALER_BIND_UAV_FARTHEST_DEPTH 3 +#define FSR3UPSCALER_BIND_UAV_CURRENT_LUMA 4 + +#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 + +#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" +#include "fsr3upscaler/ffx_fsr3upscaler_common.h" +#include "fsr3upscaler/ffx_fsr3upscaler_prepare_inputs.h" + +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH +#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 8 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT +#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 8 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH +#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH +#ifndef FFX_FSR3UPSCALER_NUM_THREADS +#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] +#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS + +FFX_PREFER_WAVE64 +FFX_FSR3UPSCALER_NUM_THREADS +FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT +void CS(int2 iDispatchThreadId : SV_DispatchThreadID) +{ + PrepareInputs(iDispatchThreadId); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_prepare_inputs_pass.hlsl.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_prepare_inputs_pass.hlsl.meta new file mode 100644 index 0000000..a56ca2a --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_prepare_inputs_pass.hlsl.meta @@ -0,0 +1,7 @@ +fileFormatVersion: 2 +guid: f6bb10da6010f8a479c9560de12f92f2 +ShaderIncludeImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_prepare_reactivity_pass.hlsl b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_prepare_reactivity_pass.hlsl new file mode 100644 index 0000000..b09e9b8 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_prepare_reactivity_pass.hlsl @@ -0,0 +1,62 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#define FSR3UPSCALER_BIND_SRV_RECONSTRUCTED_PREV_NEAREST_DEPTH 0 +#define FSR3UPSCALER_BIND_SRV_DILATED_MOTION_VECTORS 1 +#define FSR3UPSCALER_BIND_SRV_DILATED_DEPTH 2 +#define FSR3UPSCALER_BIND_SRV_REACTIVE_MASK 3 +#define FSR3UPSCALER_BIND_SRV_TRANSPARENCY_AND_COMPOSITION_MASK 4 +#define FSR3UPSCALER_BIND_SRV_ACCUMULATION 5 +#define FSR3UPSCALER_BIND_SRV_SHADING_CHANGE 6 +#define FSR3UPSCALER_BIND_SRV_CURRENT_LUMA 7 +#define FSR3UPSCALER_BIND_SRV_INPUT_EXPOSURE 8 + +#define FSR3UPSCALER_BIND_UAV_DILATED_REACTIVE_MASKS 0 +#define FSR3UPSCALER_BIND_UAV_NEW_LOCKS 1 +#define FSR3UPSCALER_BIND_UAV_ACCUMULATION 2 + +#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 + +#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" +#include "fsr3upscaler/ffx_fsr3upscaler_common.h" +#include "fsr3upscaler/ffx_fsr3upscaler_prepare_reactivity.h" + +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH +#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 8 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT +#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 8 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH +#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH +#ifndef FFX_FSR3UPSCALER_NUM_THREADS +#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] +#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS + +FFX_PREFER_WAVE64 +FFX_FSR3UPSCALER_NUM_THREADS +FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT +void CS(int2 iDispatchThreadId : SV_DispatchThreadID) +{ + PrepareReactivity(iDispatchThreadId); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_prepare_reactivity_pass.hlsl.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_prepare_reactivity_pass.hlsl.meta new file mode 100644 index 0000000..6eb9fe3 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_prepare_reactivity_pass.hlsl.meta @@ -0,0 +1,7 @@ +fileFormatVersion: 2 +guid: 0ac3112d79ef65e4eb6f4787ffe37fdd +ShaderIncludeImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_rcas_pass.hlsl b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_rcas_pass.hlsl new file mode 100644 index 0000000..2963562 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_rcas_pass.hlsl @@ -0,0 +1,53 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#define FSR3UPSCALER_BIND_SRV_INPUT_EXPOSURE 0 +#define FSR3UPSCALER_BIND_SRV_RCAS_INPUT 1 + +#define FSR3UPSCALER_BIND_UAV_UPSCALED_OUTPUT 0 + +#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 +#define FSR3UPSCALER_BIND_CB_RCAS 1 + +#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" +#include "fsr3upscaler/ffx_fsr3upscaler_common.h" +#include "fsr3upscaler/ffx_fsr3upscaler_rcas.h" + +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH +#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 64 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT +#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 1 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH +#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH +#ifndef FFX_FSR3UPSCALER_NUM_THREADS +#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] +#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS + +FFX_FSR3UPSCALER_NUM_THREADS +FFX_FSR3UPSCALER_EMBED_CB2_ROOTSIG_CONTENT +void CS(uint3 LocalThreadId : SV_GroupThreadID, uint3 WorkGroupId : SV_GroupID, uint3 Dtid : SV_DispatchThreadID) +{ + RCAS(LocalThreadId, WorkGroupId, Dtid); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_rcas_pass.hlsl.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_rcas_pass.hlsl.meta new file mode 100644 index 0000000..10533cf --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_rcas_pass.hlsl.meta @@ -0,0 +1,7 @@ +fileFormatVersion: 2 +guid: 3e953864b9b462f468b64a78463b84aa +ShaderIncludeImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_shading_change_pass.hlsl b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_shading_change_pass.hlsl new file mode 100644 index 0000000..ab2b545 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_shading_change_pass.hlsl @@ -0,0 +1,52 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#define FSR3UPSCALER_BIND_SRV_SPD_MIPS 0 + +#define FSR3UPSCALER_BIND_UAV_SHADING_CHANGE 0 + +#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 + +#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" +#include "fsr3upscaler/ffx_fsr3upscaler_common.h" +#include "fsr3upscaler/ffx_fsr3upscaler_shading_change.h" + +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH +#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 8 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT +#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 8 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH +#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH +#ifndef FFX_FSR3UPSCALER_NUM_THREADS +#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] +#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS + +FFX_PREFER_WAVE64 +FFX_FSR3UPSCALER_NUM_THREADS +FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT +void CS(int2 iDispatchThreadId : SV_DispatchThreadID) +{ + ShadingChange(iDispatchThreadId); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_shading_change_pass.hlsl.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_shading_change_pass.hlsl.meta new file mode 100644 index 0000000..e8c5fe8 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_shading_change_pass.hlsl.meta @@ -0,0 +1,7 @@ +fileFormatVersion: 2 +guid: 503052090415bdf4c9307ea6c5aa94b6 +ShaderIncludeImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_shading_change_pyramid_pass.hlsl b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_shading_change_pyramid_pass.hlsl new file mode 100644 index 0000000..5403792 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_shading_change_pyramid_pass.hlsl @@ -0,0 +1,63 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#define FSR3UPSCALER_BIND_SRV_CURRENT_LUMA 0 +#define FSR3UPSCALER_BIND_SRV_PREVIOUS_LUMA 1 +#define FSR3UPSCALER_BIND_SRV_DILATED_MOTION_VECTORS 2 +#define FSR3UPSCALER_BIND_SRV_INPUT_EXPOSURE 3 + + +#define FSR3UPSCALER_BIND_UAV_SPD_GLOBAL_ATOMIC 0 +#define FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_0 1 +#define FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_1 2 +#define FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_2 3 +#define FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_3 4 +#define FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_4 5 +#define FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_5 6 + + +#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 +#define FSR3UPSCALER_BIND_CB_SPD 1 + +#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" +#include "fsr3upscaler/ffx_fsr3upscaler_common.h" +#include "fsr3upscaler/ffx_fsr3upscaler_shading_change_pyramid.h" + +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH +#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 256 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT +#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 1 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH +#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH +#ifndef FFX_FSR3UPSCALER_NUM_THREADS +#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] +#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS + +FFX_FSR3UPSCALER_NUM_THREADS +FFX_FSR3UPSCALER_EMBED_CB2_ROOTSIG_CONTENT +void CS(uint3 WorkGroupId : SV_GroupID, uint LocalThreadIndex : SV_GroupIndex) +{ + ComputeShadingChangePyramid(WorkGroupId, LocalThreadIndex); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_shading_change_pyramid_pass.hlsl.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_shading_change_pyramid_pass.hlsl.meta new file mode 100644 index 0000000..1694ed4 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_shading_change_pyramid_pass.hlsl.meta @@ -0,0 +1,7 @@ +fileFormatVersion: 2 +guid: 8a6fef31a72aae649803d98a0ced428d +ShaderIncludeImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_tcr_autogen_pass.hlsl b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_tcr_autogen_pass.hlsl new file mode 100644 index 0000000..8a3c882 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_tcr_autogen_pass.hlsl @@ -0,0 +1,90 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (c) 2022-2023 Advanced Micro Devices, Inc. All rights reserved. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files (the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions: +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#define FSR3UPSCALER_BIND_SRV_INPUT_OPAQUE_ONLY 0 +#define FSR3UPSCALER_BIND_SRV_INPUT_COLOR 1 +#define FSR3UPSCALER_BIND_SRV_INPUT_MOTION_VECTORS 2 +#define FSR3UPSCALER_BIND_SRV_PREV_PRE_ALPHA_COLOR 3 +#define FSR3UPSCALER_BIND_SRV_PREV_POST_ALPHA_COLOR 4 +#define FSR3UPSCALER_BIND_SRV_REACTIVE_MASK 5 +#define FSR3UPSCALER_BIND_SRV_TRANSPARENCY_AND_COMPOSITION_MASK 6 + +#define FSR3UPSCALER_BIND_UAV_AUTOREACTIVE 0 +#define FSR3UPSCALER_BIND_UAV_AUTOCOMPOSITION 1 +#define FSR3UPSCALER_BIND_UAV_PREV_PRE_ALPHA_COLOR 2 +#define FSR3UPSCALER_BIND_UAV_PREV_POST_ALPHA_COLOR 3 + +#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 +#define FSR3UPSCALER_BIND_CB_AUTOREACTIVE 1 + +#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" +#include "fsr3upscaler/ffx_fsr3upscaler_common.h" +#include "fsr3upscaler/ffx_fsr3upscaler_tcr_autogen.h" + +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH +#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 8 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT +#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 8 +#endif // FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT +#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH +#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 +#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH +#ifndef FFX_FSR3UPSCALER_NUM_THREADS +#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] +#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS + +FFX_FSR3UPSCALER_NUM_THREADS +FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT +void CS(uint2 uGroupId : SV_GroupID, uint2 uGroupThreadId : SV_GroupThreadID) +{ + FFX_MIN16_I2 uDispatchThreadId = FFX_MIN16_I2(uGroupId * uint2(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT) + uGroupThreadId); + + // ToDo: take into account jitter (i.e. add delta of previous jitter and current jitter to previous UV + // fetch pre- and post-alpha color values + FFX_MIN16_F2 fUv = ( FFX_MIN16_F2(uDispatchThreadId) + FFX_MIN16_F2(0.5f, 0.5f) ) / FFX_MIN16_F2( RenderSize() ); + FFX_MIN16_F2 fPrevUV = fUv + FFX_MIN16_F2( LoadInputMotionVector(uDispatchThreadId) ); + FFX_MIN16_I2 iPrevIdx = FFX_MIN16_I2(fPrevUV * FFX_MIN16_F2(RenderSize()) - 0.5f); + + FFX_MIN16_F3 colorPreAlpha = FFX_MIN16_F3( LoadOpaqueOnly( uDispatchThreadId ) ); + FFX_MIN16_F3 colorPostAlpha = FFX_MIN16_F3( LoadInputColor( uDispatchThreadId ) ); + + FFX_MIN16_F2 outReactiveMask = 0; + + outReactiveMask.y = ComputeTransparencyAndComposition(uDispatchThreadId, iPrevIdx); + + if (outReactiveMask.y > 0.5f) + { + outReactiveMask.x = ComputeReactive(uDispatchThreadId, iPrevIdx); + outReactiveMask.x *= FFX_MIN16_F(fReactiveScale); + outReactiveMask.x = outReactiveMask.x < fReactiveMax ? outReactiveMask.x : FFX_MIN16_F( fReactiveMax ); + } + + outReactiveMask.y *= FFX_MIN16_F(fTcScale ); + + outReactiveMask.x = max( outReactiveMask.x, FFX_MIN16_F( LoadReactiveMask(uDispatchThreadId) ) ); + outReactiveMask.y = max( outReactiveMask.y, FFX_MIN16_F( LoadTransparencyAndCompositionMask(uDispatchThreadId) ) ); + + StoreAutoReactive(uDispatchThreadId, outReactiveMask); + + StorePrevPreAlpha(uDispatchThreadId, colorPreAlpha); + StorePrevPostAlpha(uDispatchThreadId, colorPostAlpha); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_tcr_autogen_pass.hlsl.meta b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_tcr_autogen_pass.hlsl.meta new file mode 100644 index 0000000..c6923a5 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/ffx_fsr3upscaler_tcr_autogen_pass.hlsl.meta @@ -0,0 +1,7 @@ +fileFormatVersion: 2 +guid: 877b3428f6073db41b9fc80bb6c6d4d7 +ShaderIncludeImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr1.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr1.meta new file mode 100644 index 0000000..e52d0ac --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr1.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: dafd228fb8a2f4e478fe1df6d38589f9 +folderAsset: yes +DefaultImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr1/ffx_fsr1.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr1/ffx_fsr1.h new file mode 100644 index 0000000..82ebf21 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr1/ffx_fsr1.h @@ -0,0 +1,1252 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +/// @defgroup FfxGPUFsr1 FidelityFX FSR1 +/// FidelityFX Super Resolution 1 GPU documentation +/// +/// @ingroup FfxGPUEffects + +/// Setup required constant values for EASU (works on CPU or GPU). +/// +/// @param [out] con0 +/// @param [out] con1 +/// @param [out] con2 +/// @param [out] con3 +/// @param [in] inputViewportInPixelsX The rendered image resolution being upscaled in X dimension. +/// @param [in] inputViewportInPixelsY The rendered image resolution being upscaled in Y dimension. +/// @param [in] inputSizeInPixelsX The resolution of the resource containing the input image (useful for dynamic resolution) in X dimension. +/// @param [in] inputSizeInPixelsY The resolution of the resource containing the input image (useful for dynamic resolution) in Y dimension. +/// @param [in] outputSizeInPixelsX The display resolution which the input image gets upscaled to in X dimension. +/// @param [in] outputSizeInPixelsY The display resolution which the input image gets upscaled to in Y dimension. +/// +/// @ingroup FfxGPUFsr1 +FFX_STATIC void ffxFsrPopulateEasuConstants( + FFX_PARAMETER_INOUT FfxUInt32x4 con0, + FFX_PARAMETER_INOUT FfxUInt32x4 con1, + FFX_PARAMETER_INOUT FfxUInt32x4 con2, + FFX_PARAMETER_INOUT FfxUInt32x4 con3, + FFX_PARAMETER_IN FfxFloat32 inputViewportInPixelsX, + FFX_PARAMETER_IN FfxFloat32 inputViewportInPixelsY, + FFX_PARAMETER_IN FfxFloat32 inputSizeInPixelsX, + FFX_PARAMETER_IN FfxFloat32 inputSizeInPixelsY, + FFX_PARAMETER_IN FfxFloat32 outputSizeInPixelsX, + FFX_PARAMETER_IN FfxFloat32 outputSizeInPixelsY) +{ + // Output integer position to a pixel position in viewport. + con0[0] = ffxAsUInt32(inputViewportInPixelsX * ffxReciprocal(outputSizeInPixelsX)); + con0[1] = ffxAsUInt32(inputViewportInPixelsY * ffxReciprocal(outputSizeInPixelsY)); + con0[2] = ffxAsUInt32(FfxFloat32(0.5) * inputViewportInPixelsX * ffxReciprocal(outputSizeInPixelsX) - FfxFloat32(0.5)); + con0[3] = ffxAsUInt32(FfxFloat32(0.5) * inputViewportInPixelsY * ffxReciprocal(outputSizeInPixelsY) - FfxFloat32(0.5)); + + // Viewport pixel position to normalized image space. + // This is used to get upper-left of 'F' tap. + con1[0] = ffxAsUInt32(ffxReciprocal(inputSizeInPixelsX)); + con1[1] = ffxAsUInt32(ffxReciprocal(inputSizeInPixelsY)); + + // Centers of gather4, first offset from upper-left of 'F'. + // +---+---+ + // | | | + // +--(0)--+ + // | b | c | + // +---F---+---+---+ + // | e | f | g | h | + // +--(1)--+--(2)--+ + // | i | j | k | l | + // +---+---+---+---+ + // | n | o | + // +--(3)--+ + // | | | + // +---+---+ + con1[2] = ffxAsUInt32(FfxFloat32(1.0) * ffxReciprocal(inputSizeInPixelsX)); + con1[3] = ffxAsUInt32(FfxFloat32(-1.0) * ffxReciprocal(inputSizeInPixelsY)); + + // These are from (0) instead of 'F'. + con2[0] = ffxAsUInt32(FfxFloat32(-1.0) * ffxReciprocal(inputSizeInPixelsX)); + con2[1] = ffxAsUInt32(FfxFloat32(2.0) * ffxReciprocal(inputSizeInPixelsY)); + con2[2] = ffxAsUInt32(FfxFloat32(1.0) * ffxReciprocal(inputSizeInPixelsX)); + con2[3] = ffxAsUInt32(FfxFloat32(2.0) * ffxReciprocal(inputSizeInPixelsY)); + con3[0] = ffxAsUInt32(FfxFloat32(0.0) * ffxReciprocal(inputSizeInPixelsX)); + con3[1] = ffxAsUInt32(FfxFloat32(4.0) * ffxReciprocal(inputSizeInPixelsY)); + con3[2] = con3[3] = 0; +} + +/// Setup required constant values for EASU (works on CPU or GPU). +/// +/// @param [out] con0 +/// @param [out] con1 +/// @param [out] con2 +/// @param [out] con3 +/// @param [in] inputViewportInPixelsX The resolution of the input in the X dimension. +/// @param [in] inputViewportInPixelsY The resolution of the input in the Y dimension. +/// @param [in] inputSizeInPixelsX The input size in pixels in the X dimension. +/// @param [in] inputSizeInPixelsY The input size in pixels in the Y dimension. +/// @param [in] outputSizeInPixelsX The output size in pixels in the X dimension. +/// @param [in] outputSizeInPixelsY The output size in pixels in the Y dimension. +/// @param [in] inputOffsetInPixelsX The input image offset in the X dimension into the resource containing it (useful for dynamic resolution). +/// @param [in] inputOffsetInPixelsY The input image offset in the Y dimension into the resource containing it (useful for dynamic resolution). +/// +/// @ingroup FfxGPUFsr1 +FFX_STATIC void ffxFsrPopulateEasuConstantsOffset( + FFX_PARAMETER_INOUT FfxUInt32x4 con0, + FFX_PARAMETER_INOUT FfxUInt32x4 con1, + FFX_PARAMETER_INOUT FfxUInt32x4 con2, + FFX_PARAMETER_INOUT FfxUInt32x4 con3, + FFX_PARAMETER_IN FfxFloat32 inputViewportInPixelsX, + FFX_PARAMETER_IN FfxFloat32 inputViewportInPixelsY, + FFX_PARAMETER_IN FfxFloat32 inputSizeInPixelsX, + FFX_PARAMETER_IN FfxFloat32 inputSizeInPixelsY, + FFX_PARAMETER_IN FfxFloat32 outputSizeInPixelsX, + FFX_PARAMETER_IN FfxFloat32 outputSizeInPixelsY, + FFX_PARAMETER_IN FfxFloat32 inputOffsetInPixelsX, + FFX_PARAMETER_IN FfxFloat32 inputOffsetInPixelsY) +{ + ffxFsrPopulateEasuConstants( + con0, + con1, + con2, + con3, + inputViewportInPixelsX, + inputViewportInPixelsY, + inputSizeInPixelsX, + inputSizeInPixelsY, + outputSizeInPixelsX, + outputSizeInPixelsY); + + // override + con0[2] = ffxAsUInt32(FfxFloat32(0.5) * inputViewportInPixelsX * ffxReciprocal(outputSizeInPixelsX) - FfxFloat32(0.5) + inputOffsetInPixelsX); + con0[3] = ffxAsUInt32(FfxFloat32(0.5) * inputViewportInPixelsY * ffxReciprocal(outputSizeInPixelsY) - FfxFloat32(0.5) + inputOffsetInPixelsY); +} + +#if defined(FFX_GPU) && defined(FFX_FSR_EASU_FLOAT) +// Input callback prototypes, need to be implemented by calling shader +FfxFloat32x4 FsrEasuRF(FfxFloat32x2 p); +FfxFloat32x4 FsrEasuGF(FfxFloat32x2 p); +FfxFloat32x4 FsrEasuBF(FfxFloat32x2 p); + +// Filtering for a given tap for the scalar. +void fsrEasuTapFloat( + FFX_PARAMETER_INOUT FfxFloat32x3 accumulatedColor, // Accumulated color, with negative lobe. + FFX_PARAMETER_INOUT FfxFloat32 accumulatedWeight, // Accumulated weight. + FFX_PARAMETER_IN FfxFloat32x2 pixelOffset, // Pixel offset from resolve position to tap. + FFX_PARAMETER_IN FfxFloat32x2 gradientDirection, // Gradient direction. + FFX_PARAMETER_IN FfxFloat32x2 length, // Length. + FFX_PARAMETER_IN FfxFloat32 negativeLobeStrength, // Negative lobe strength. + FFX_PARAMETER_IN FfxFloat32 clippingPoint, // Clipping point. + FFX_PARAMETER_IN FfxFloat32x3 color) // Tap color. +{ + // Rotate offset by direction. + FfxFloat32x2 rotatedOffset; + rotatedOffset.x = (pixelOffset.x * (gradientDirection.x)) + (pixelOffset.y * gradientDirection.y); + rotatedOffset.y = (pixelOffset.x * (-gradientDirection.y)) + (pixelOffset.y * gradientDirection.x); + + // Anisotropy. + rotatedOffset *= length; + + // Compute distance^2. + FfxFloat32 distanceSquared = rotatedOffset.x * rotatedOffset.x + rotatedOffset.y * rotatedOffset.y; + + // Limit to the window as at corner, 2 taps can easily be outside. + distanceSquared = ffxMin(distanceSquared, clippingPoint); + + // Approximation of lancos2 without sin() or rcp(), or sqrt() to get x. + // (25/16 * (2/5 * x^2 - 1)^2 - (25/16 - 1)) * (1/4 * x^2 - 1)^2 + // |_______________________________________| |_______________| + // base window + // The general form of the 'base' is, + // (a*(b*x^2-1)^2-(a-1)) + // Where 'a=1/(2*b-b^2)' and 'b' moves around the negative lobe. + FfxFloat32 weightB = FfxFloat32(2.0 / 5.0) * distanceSquared + FfxFloat32(-1.0); + FfxFloat32 weightA = negativeLobeStrength * distanceSquared + FfxFloat32(-1.0); + weightB *= weightB; + weightA *= weightA; + weightB = FfxFloat32(25.0 / 16.0) * weightB + FfxFloat32(-(25.0 / 16.0 - 1.0)); + FfxFloat32 weight = weightB * weightA; + + // Do weighted average. + accumulatedColor += color * weight; + accumulatedWeight += weight; +} + +// Accumulate direction and length. +void fsrEasuSetFloat( + FFX_PARAMETER_INOUT FfxFloat32x2 direction, + FFX_PARAMETER_INOUT FfxFloat32 length, + FFX_PARAMETER_IN FfxFloat32x2 pp, + FFX_PARAMETER_IN FfxBoolean biS, + FFX_PARAMETER_IN FfxBoolean biT, + FFX_PARAMETER_IN FfxBoolean biU, + FFX_PARAMETER_IN FfxBoolean biV, + FFX_PARAMETER_IN FfxFloat32 lA, + FFX_PARAMETER_IN FfxFloat32 lB, + FFX_PARAMETER_IN FfxFloat32 lC, + FFX_PARAMETER_IN FfxFloat32 lD, + FFX_PARAMETER_IN FfxFloat32 lE) +{ + // Compute bilinear weight, branches factor out as predicates are compiler time immediates. + // s t + // u v + FfxFloat32 weight = FfxFloat32(0.0); + if (biS) + weight = (FfxFloat32(1.0) - pp.x) * (FfxFloat32(1.0) - pp.y); + if (biT) + weight = pp.x * (FfxFloat32(1.0) - pp.y); + if (biU) + weight = (FfxFloat32(1.0) - pp.x) * pp.y; + if (biV) + weight = pp.x * pp.y; + + // Direction is the '+' diff. + // a + // b c d + // e + // Then takes magnitude from abs average of both sides of 'c'. + // Length converts gradient reversal to 0, smoothly to non-reversal at 1, shaped, then adding horz and vert terms. + FfxFloat32 dc = lD - lC; + FfxFloat32 cb = lC - lB; + FfxFloat32 lengthX = max(abs(dc), abs(cb)); + lengthX = ffxApproximateReciprocal(lengthX); + FfxFloat32 directionX = lD - lB; + direction.x += directionX * weight; + lengthX = ffxSaturate(abs(directionX) * lengthX); + lengthX *= lengthX; + length += lengthX * weight; + + // Repeat for the y axis. + FfxFloat32 ec = lE - lC; + FfxFloat32 ca = lC - lA; + FfxFloat32 lengthY = max(abs(ec), abs(ca)); + lengthY = ffxApproximateReciprocal(lengthY); + FfxFloat32 directionY = lE - lA; + direction.y += directionY * weight; + lengthY = ffxSaturate(abs(directionY) * lengthY); + lengthY *= lengthY; + length += lengthY * weight; +} + +/// Apply edge-aware spatial upsampling using 32bit floating point precision calculations. +/// +/// @param [out] outPixel The computed color of a pixel. +/// @param [in] integerPosition Integer pixel position within the output. +/// @param [in] con0 The first constant value generated by ffxFsrPopulateEasuConstants. +/// @param [in] con1 The second constant value generated by ffxFsrPopulateEasuConstants. +/// @param [in] con2 The third constant value generated by ffxFsrPopulateEasuConstants. +/// @param [in] con3 The fourth constant value generated by ffxFsrPopulateEasuConstants. +/// +/// @ingroup FSR +void ffxFsrEasuFloat( + FFX_PARAMETER_OUT FfxFloat32x3 pix, + FFX_PARAMETER_IN FfxUInt32x2 ip, + FFX_PARAMETER_IN FfxUInt32x4 con0, + FFX_PARAMETER_IN FfxUInt32x4 con1, + FFX_PARAMETER_IN FfxUInt32x4 con2, + FFX_PARAMETER_IN FfxUInt32x4 con3) +{ + // Get position of 'f'. + FfxFloat32x2 pp = FfxFloat32x2(ip) * ffxAsFloat(con0.xy) + ffxAsFloat(con0.zw); + FfxFloat32x2 fp = floor(pp); + pp -= fp; + + // 12-tap kernel. + // b c + // e f g h + // i j k l + // n o + // Gather 4 ordering. + // a b + // r g + // For packed FP16, need either {rg} or {ab} so using the following setup for gather in all versions, + // a b <- unused (z) + // r g + // a b a b + // r g r g + // a b + // r g <- unused (z) + // Allowing dead-code removal to remove the 'z's. + FfxFloat32x2 p0 = fp * ffxAsFloat(con1.xy) + ffxAsFloat(con1.zw); + + // These are from p0 to avoid pulling two constants on pre-Navi hardware. + FfxFloat32x2 p1 = p0 + ffxAsFloat(con2.xy); + FfxFloat32x2 p2 = p0 + ffxAsFloat(con2.zw); + FfxFloat32x2 p3 = p0 + ffxAsFloat(con3.xy); + FfxFloat32x4 bczzR = FsrEasuRF(p0); + FfxFloat32x4 bczzG = FsrEasuGF(p0); + FfxFloat32x4 bczzB = FsrEasuBF(p0); + FfxFloat32x4 ijfeR = FsrEasuRF(p1); + FfxFloat32x4 ijfeG = FsrEasuGF(p1); + FfxFloat32x4 ijfeB = FsrEasuBF(p1); + FfxFloat32x4 klhgR = FsrEasuRF(p2); + FfxFloat32x4 klhgG = FsrEasuGF(p2); + FfxFloat32x4 klhgB = FsrEasuBF(p2); + FfxFloat32x4 zzonR = FsrEasuRF(p3); + FfxFloat32x4 zzonG = FsrEasuGF(p3); + FfxFloat32x4 zzonB = FsrEasuBF(p3); + + // Simplest multi-channel approximate luma possible (luma times 2, in 2 FMA/MAD). + FfxFloat32x4 bczzL = bczzB * ffxBroadcast4(0.5) + (bczzR * ffxBroadcast4(0.5) + bczzG); + FfxFloat32x4 ijfeL = ijfeB * ffxBroadcast4(0.5) + (ijfeR * ffxBroadcast4(0.5) + ijfeG); + FfxFloat32x4 klhgL = klhgB * ffxBroadcast4(0.5) + (klhgR * ffxBroadcast4(0.5) + klhgG); + FfxFloat32x4 zzonL = zzonB * ffxBroadcast4(0.5) + (zzonR * ffxBroadcast4(0.5) + zzonG); + + // Rename. + FfxFloat32 bL = bczzL.x; + FfxFloat32 cL = bczzL.y; + FfxFloat32 iL = ijfeL.x; + FfxFloat32 jL = ijfeL.y; + FfxFloat32 fL = ijfeL.z; + FfxFloat32 eL = ijfeL.w; + FfxFloat32 kL = klhgL.x; + FfxFloat32 lL = klhgL.y; + FfxFloat32 hL = klhgL.z; + FfxFloat32 gL = klhgL.w; + FfxFloat32 oL = zzonL.z; + FfxFloat32 nL = zzonL.w; + + // Accumulate for bilinear interpolation. + FfxFloat32x2 dir = ffxBroadcast2(0.0); + FfxFloat32 len = FfxFloat32(0.0); + fsrEasuSetFloat(dir, len, pp, FFX_TRUE, FFX_FALSE, FFX_FALSE, FFX_FALSE, bL, eL, fL, gL, jL); + fsrEasuSetFloat(dir, len, pp, FFX_FALSE, FFX_TRUE, FFX_FALSE, FFX_FALSE, cL, fL, gL, hL, kL); + fsrEasuSetFloat(dir, len, pp, FFX_FALSE, FFX_FALSE, FFX_TRUE, FFX_FALSE, fL, iL, jL, kL, nL); + fsrEasuSetFloat(dir, len, pp, FFX_FALSE, FFX_FALSE, FFX_FALSE, FFX_TRUE, gL, jL, kL, lL, oL); + + // Normalize with approximation, and cleanup close to zero. + FfxFloat32x2 dir2 = dir * dir; + FfxFloat32 dirR = dir2.x + dir2.y; + FfxBoolean zro = dirR < FfxFloat32(1.0 / 32768.0); + dirR = ffxApproximateReciprocalSquareRoot(dirR); + dirR = zro ? FfxFloat32(1.0) : dirR; + dir.x = zro ? FfxFloat32(1.0) : dir.x; + dir *= ffxBroadcast2(dirR); + + // Transform from {0 to 2} to {0 to 1} range, and shape with square. + len = len * FfxFloat32(0.5); + len *= len; + + // Stretch kernel {1.0 vert|horz, to sqrt(2.0) on diagonal}. + FfxFloat32 stretch = (dir.x * dir.x + dir.y * dir.y) * ffxApproximateReciprocal(max(abs(dir.x), abs(dir.y))); + + // Anisotropic length after rotation, + // x := 1.0 lerp to 'stretch' on edges + // y := 1.0 lerp to 2x on edges + FfxFloat32x2 len2 = FfxFloat32x2(FfxFloat32(1.0) + (stretch - FfxFloat32(1.0)) * len, FfxFloat32(1.0) + FfxFloat32(-0.5) * len); + + // Based on the amount of 'edge', + // the window shifts from +/-{sqrt(2.0) to slightly beyond 2.0}. + FfxFloat32 lob = FfxFloat32(0.5) + FfxFloat32((1.0 / 4.0 - 0.04) - 0.5) * len; + + // Set distance^2 clipping point to the end of the adjustable window. + FfxFloat32 clp = ffxApproximateReciprocal(lob); + + // Accumulation mixed with min/max of 4 nearest. + // b c + // e f g h + // i j k l + // n o + FfxFloat32x3 min4 = + ffxMin(ffxMin3(FfxFloat32x3(ijfeR.z, ijfeG.z, ijfeB.z), FfxFloat32x3(klhgR.w, klhgG.w, klhgB.w), FfxFloat32x3(ijfeR.y, ijfeG.y, ijfeB.y)), + FfxFloat32x3(klhgR.x, klhgG.x, klhgB.x)); + FfxFloat32x3 max4 = + max(ffxMax3(FfxFloat32x3(ijfeR.z, ijfeG.z, ijfeB.z), FfxFloat32x3(klhgR.w, klhgG.w, klhgB.w), FfxFloat32x3(ijfeR.y, ijfeG.y, ijfeB.y)), FfxFloat32x3(klhgR.x, klhgG.x, klhgB.x)); + + // Accumulation. + FfxFloat32x3 aC = ffxBroadcast3(0.0); + FfxFloat32 aW = FfxFloat32(0.0); + fsrEasuTapFloat(aC, aW, FfxFloat32x2(0.0, -1.0) - pp, dir, len2, lob, clp, FfxFloat32x3(bczzR.x, bczzG.x, bczzB.x)); // b + fsrEasuTapFloat(aC, aW, FfxFloat32x2(1.0, -1.0) - pp, dir, len2, lob, clp, FfxFloat32x3(bczzR.y, bczzG.y, bczzB.y)); // c + fsrEasuTapFloat(aC, aW, FfxFloat32x2(-1.0, 1.0) - pp, dir, len2, lob, clp, FfxFloat32x3(ijfeR.x, ijfeG.x, ijfeB.x)); // i + fsrEasuTapFloat(aC, aW, FfxFloat32x2(0.0, 1.0) - pp, dir, len2, lob, clp, FfxFloat32x3(ijfeR.y, ijfeG.y, ijfeB.y)); // j + fsrEasuTapFloat(aC, aW, FfxFloat32x2(0.0, 0.0) - pp, dir, len2, lob, clp, FfxFloat32x3(ijfeR.z, ijfeG.z, ijfeB.z)); // f + fsrEasuTapFloat(aC, aW, FfxFloat32x2(-1.0, 0.0) - pp, dir, len2, lob, clp, FfxFloat32x3(ijfeR.w, ijfeG.w, ijfeB.w)); // e + fsrEasuTapFloat(aC, aW, FfxFloat32x2(1.0, 1.0) - pp, dir, len2, lob, clp, FfxFloat32x3(klhgR.x, klhgG.x, klhgB.x)); // k + fsrEasuTapFloat(aC, aW, FfxFloat32x2(2.0, 1.0) - pp, dir, len2, lob, clp, FfxFloat32x3(klhgR.y, klhgG.y, klhgB.y)); // l + fsrEasuTapFloat(aC, aW, FfxFloat32x2(2.0, 0.0) - pp, dir, len2, lob, clp, FfxFloat32x3(klhgR.z, klhgG.z, klhgB.z)); // h + fsrEasuTapFloat(aC, aW, FfxFloat32x2(1.0, 0.0) - pp, dir, len2, lob, clp, FfxFloat32x3(klhgR.w, klhgG.w, klhgB.w)); // g + fsrEasuTapFloat(aC, aW, FfxFloat32x2(1.0, 2.0) - pp, dir, len2, lob, clp, FfxFloat32x3(zzonR.z, zzonG.z, zzonB.z)); // o + fsrEasuTapFloat(aC, aW, FfxFloat32x2(0.0, 2.0) - pp, dir, len2, lob, clp, FfxFloat32x3(zzonR.w, zzonG.w, zzonB.w)); // n + + // Normalize and dering. + pix = ffxMin(max4, max(min4, aC * ffxBroadcast3(ffxReciprocal(aW)))); +} +#endif // #if defined(FFX_GPU) && defined(FFX_FSR_EASU_FLOAT) + +#if defined(FFX_GPU) && FFX_HALF == 1 && defined(FFX_FSR_EASU_HALF) +// Input callback prototypes, need to be implemented by calling shader +FfxFloat16x4 FsrEasuRH(FfxFloat32x2 p); +FfxFloat16x4 FsrEasuGH(FfxFloat32x2 p); +FfxFloat16x4 FsrEasuBH(FfxFloat32x2 p); + +// This runs 2 taps in parallel. +void FsrEasuTapH( + FFX_PARAMETER_INOUT FfxFloat16x2 aCR, + FFX_PARAMETER_INOUT FfxFloat16x2 aCG, + FFX_PARAMETER_INOUT FfxFloat16x2 aCB, + FFX_PARAMETER_INOUT FfxFloat16x2 aW, + FFX_PARAMETER_IN FfxFloat16x2 offX, + FFX_PARAMETER_IN FfxFloat16x2 offY, + FFX_PARAMETER_IN FfxFloat16x2 dir, + FFX_PARAMETER_IN FfxFloat16x2 len, + FFX_PARAMETER_IN FfxFloat16 lob, + FFX_PARAMETER_IN FfxFloat16 clp, + FFX_PARAMETER_IN FfxFloat16x2 cR, + FFX_PARAMETER_IN FfxFloat16x2 cG, + FFX_PARAMETER_IN FfxFloat16x2 cB) +{ + FfxFloat16x2 vX, vY; + vX = offX * dir.xx + offY * dir.yy; + vY = offX * (-dir.yy) + offY * dir.xx; + vX *= len.x; + vY *= len.y; + FfxFloat16x2 d2 = vX * vX + vY * vY; + d2 = min(d2, FFX_BROADCAST_FLOAT16X2(clp)); + FfxFloat16x2 wB = FFX_BROADCAST_FLOAT16X2(2.0 / 5.0) * d2 + FFX_BROADCAST_FLOAT16X2(-1.0); + FfxFloat16x2 wA = FFX_BROADCAST_FLOAT16X2(lob) * d2 + FFX_BROADCAST_FLOAT16X2(-1.0); + wB *= wB; + wA *= wA; + wB = FFX_BROADCAST_FLOAT16X2(25.0 / 16.0) * wB + FFX_BROADCAST_FLOAT16X2(-(25.0 / 16.0 - 1.0)); + FfxFloat16x2 w = wB * wA; + aCR += cR * w; + aCG += cG * w; + aCB += cB * w; + aW += w; +} + +// This runs 2 taps in parallel. +void FsrEasuSetH( + FFX_PARAMETER_INOUT FfxFloat16x2 dirPX, + FFX_PARAMETER_INOUT FfxFloat16x2 dirPY, + FFX_PARAMETER_INOUT FfxFloat16x2 lenP, + FFX_PARAMETER_IN FfxFloat16x2 pp, + FFX_PARAMETER_IN FfxBoolean biST, + FFX_PARAMETER_IN FfxBoolean biUV, + FFX_PARAMETER_IN FfxFloat16x2 lA, + FFX_PARAMETER_IN FfxFloat16x2 lB, + FFX_PARAMETER_IN FfxFloat16x2 lC, + FFX_PARAMETER_IN FfxFloat16x2 lD, + FFX_PARAMETER_IN FfxFloat16x2 lE) +{ + FfxFloat16x2 w = FFX_BROADCAST_FLOAT16X2(0.0); + + if (biST) + w = (FfxFloat16x2(1.0, 0.0) + FfxFloat16x2(-pp.x, pp.x)) * FFX_BROADCAST_FLOAT16X2(FFX_BROADCAST_FLOAT16(1.0) - pp.y); + + if (biUV) + w = (FfxFloat16x2(1.0, 0.0) + FfxFloat16x2(-pp.x, pp.x)) * FFX_BROADCAST_FLOAT16X2(pp.y); + + // ABS is not free in the packed FP16 path. + FfxFloat16x2 dc = lD - lC; + FfxFloat16x2 cb = lC - lB; + FfxFloat16x2 lenX = max(abs(dc), abs(cb)); + lenX = ffxReciprocalHalf(lenX); + + FfxFloat16x2 dirX = lD - lB; + dirPX += dirX * w; + lenX = ffxSaturate(abs(dirX) * lenX); + lenX *= lenX; + lenP += lenX * w; + FfxFloat16x2 ec = lE - lC; + FfxFloat16x2 ca = lC - lA; + FfxFloat16x2 lenY = max(abs(ec), abs(ca)); + lenY = ffxReciprocalHalf(lenY); + FfxFloat16x2 dirY = lE - lA; + dirPY += dirY * w; + lenY = ffxSaturate(abs(dirY) * lenY); + lenY *= lenY; + lenP += lenY * w; +} + +void FsrEasuH( + FFX_PARAMETER_OUT FfxFloat16x3 pix, + FFX_PARAMETER_IN FfxUInt32x2 ip, + FFX_PARAMETER_IN FfxUInt32x4 con0, + FFX_PARAMETER_IN FfxUInt32x4 con1, + FFX_PARAMETER_IN FfxUInt32x4 con2, + FFX_PARAMETER_IN FfxUInt32x4 con3) +{ + FfxFloat32x2 pp = FfxFloat32x2(ip) * ffxAsFloat(con0.xy) + ffxAsFloat(con0.zw); + FfxFloat32x2 fp = floor(pp); + pp -= fp; + FfxFloat16x2 ppp = FfxFloat16x2(pp); + + FfxFloat32x2 p0 = fp * ffxAsFloat(con1.xy) + ffxAsFloat(con1.zw); + FfxFloat32x2 p1 = p0 + ffxAsFloat(con2.xy); + FfxFloat32x2 p2 = p0 + ffxAsFloat(con2.zw); + FfxFloat32x2 p3 = p0 + ffxAsFloat(con3.xy); + FfxFloat16x4 bczzR = FsrEasuRH(p0); + FfxFloat16x4 bczzG = FsrEasuGH(p0); + FfxFloat16x4 bczzB = FsrEasuBH(p0); + FfxFloat16x4 ijfeR = FsrEasuRH(p1); + FfxFloat16x4 ijfeG = FsrEasuGH(p1); + FfxFloat16x4 ijfeB = FsrEasuBH(p1); + FfxFloat16x4 klhgR = FsrEasuRH(p2); + FfxFloat16x4 klhgG = FsrEasuGH(p2); + FfxFloat16x4 klhgB = FsrEasuBH(p2); + FfxFloat16x4 zzonR = FsrEasuRH(p3); + FfxFloat16x4 zzonG = FsrEasuGH(p3); + FfxFloat16x4 zzonB = FsrEasuBH(p3); + + FfxFloat16x4 bczzL = bczzB * FFX_BROADCAST_FLOAT16X4(0.5) + (bczzR * FFX_BROADCAST_FLOAT16X4(0.5) + bczzG); + FfxFloat16x4 ijfeL = ijfeB * FFX_BROADCAST_FLOAT16X4(0.5) + (ijfeR * FFX_BROADCAST_FLOAT16X4(0.5) + ijfeG); + FfxFloat16x4 klhgL = klhgB * FFX_BROADCAST_FLOAT16X4(0.5) + (klhgR * FFX_BROADCAST_FLOAT16X4(0.5) + klhgG); + FfxFloat16x4 zzonL = zzonB * FFX_BROADCAST_FLOAT16X4(0.5) + (zzonR * FFX_BROADCAST_FLOAT16X4(0.5) + zzonG); + FfxFloat16 bL = bczzL.x; + FfxFloat16 cL = bczzL.y; + FfxFloat16 iL = ijfeL.x; + FfxFloat16 jL = ijfeL.y; + FfxFloat16 fL = ijfeL.z; + FfxFloat16 eL = ijfeL.w; + FfxFloat16 kL = klhgL.x; + FfxFloat16 lL = klhgL.y; + FfxFloat16 hL = klhgL.z; + FfxFloat16 gL = klhgL.w; + FfxFloat16 oL = zzonL.z; + FfxFloat16 nL = zzonL.w; + + // This part is different, accumulating 2 taps in parallel. + FfxFloat16x2 dirPX = FFX_BROADCAST_FLOAT16X2(0.0); + FfxFloat16x2 dirPY = FFX_BROADCAST_FLOAT16X2(0.0); + FfxFloat16x2 lenP = FFX_BROADCAST_FLOAT16X2(0.0); + FsrEasuSetH(dirPX, + dirPY, + lenP, + ppp, + FfxBoolean(true), + FfxBoolean(false), + FfxFloat16x2(bL, cL), + FfxFloat16x2(eL, fL), + FfxFloat16x2(fL, gL), + FfxFloat16x2(gL, hL), + FfxFloat16x2(jL, kL)); + FsrEasuSetH(dirPX, + dirPY, + lenP, + ppp, + FfxBoolean(false), + FfxBoolean(true), + FfxFloat16x2(fL, gL), + FfxFloat16x2(iL, jL), + FfxFloat16x2(jL, kL), + FfxFloat16x2(kL, lL), + FfxFloat16x2(nL, oL)); + FfxFloat16x2 dir = FfxFloat16x2(dirPX.r + dirPX.g, dirPY.r + dirPY.g); + FfxFloat16 len = lenP.r + lenP.g; + + FfxFloat16x2 dir2 = dir * dir; + FfxFloat16 dirR = dir2.x + dir2.y; + FfxUInt32 zro = FfxUInt32(dirR < FFX_BROADCAST_FLOAT16(1.0 / 32768.0)); + dirR = ffxApproximateReciprocalSquareRootHalf(dirR); + dirR = (zro > 0) ? FFX_BROADCAST_FLOAT16(1.0) : dirR; + dir.x = (zro > 0) ? FFX_BROADCAST_FLOAT16(1.0) : dir.x; + dir *= FFX_BROADCAST_FLOAT16X2(dirR); + len = len * FFX_BROADCAST_FLOAT16(0.5); + len *= len; + FfxFloat16 stretch = (dir.x * dir.x + dir.y * dir.y) * ffxApproximateReciprocalHalf(max(abs(dir.x), abs(dir.y))); + FfxFloat16x2 len2 = + FfxFloat16x2(FFX_BROADCAST_FLOAT16(1.0) + (stretch - FFX_BROADCAST_FLOAT16(1.0)) * len, FFX_BROADCAST_FLOAT16(1.0) + FFX_BROADCAST_FLOAT16(-0.5) * len); + FfxFloat16 lob = FFX_BROADCAST_FLOAT16(0.5) + FFX_BROADCAST_FLOAT16((1.0 / 4.0 - 0.04) - 0.5) * len; + FfxFloat16 clp = ffxApproximateReciprocalHalf(lob); + + // FP16 is different, using packed trick to do min and max in same operation. + FfxFloat16x2 bothR = + max(max(FfxFloat16x2(-ijfeR.z, ijfeR.z), FfxFloat16x2(-klhgR.w, klhgR.w)), max(FfxFloat16x2(-ijfeR.y, ijfeR.y), FfxFloat16x2(-klhgR.x, klhgR.x))); + FfxFloat16x2 bothG = + max(max(FfxFloat16x2(-ijfeG.z, ijfeG.z), FfxFloat16x2(-klhgG.w, klhgG.w)), max(FfxFloat16x2(-ijfeG.y, ijfeG.y), FfxFloat16x2(-klhgG.x, klhgG.x))); + FfxFloat16x2 bothB = + max(max(FfxFloat16x2(-ijfeB.z, ijfeB.z), FfxFloat16x2(-klhgB.w, klhgB.w)), max(FfxFloat16x2(-ijfeB.y, ijfeB.y), FfxFloat16x2(-klhgB.x, klhgB.x))); + + // This part is different for FP16, working pairs of taps at a time. + FfxFloat16x2 pR = FFX_BROADCAST_FLOAT16X2(0.0); + FfxFloat16x2 pG = FFX_BROADCAST_FLOAT16X2(0.0); + FfxFloat16x2 pB = FFX_BROADCAST_FLOAT16X2(0.0); + FfxFloat16x2 pW = FFX_BROADCAST_FLOAT16X2(0.0); + FsrEasuTapH(pR, pG, pB, pW, FfxFloat16x2(0.0, 1.0) - ppp.xx, FfxFloat16x2(-1.0, -1.0) - ppp.yy, dir, len2, lob, clp, bczzR.xy, bczzG.xy, bczzB.xy); + FsrEasuTapH(pR, pG, pB, pW, FfxFloat16x2(-1.0, 0.0) - ppp.xx, FfxFloat16x2(1.0, 1.0) - ppp.yy, dir, len2, lob, clp, ijfeR.xy, ijfeG.xy, ijfeB.xy); + FsrEasuTapH(pR, pG, pB, pW, FfxFloat16x2(0.0, -1.0) - ppp.xx, FfxFloat16x2(0.0, 0.0) - ppp.yy, dir, len2, lob, clp, ijfeR.zw, ijfeG.zw, ijfeB.zw); + FsrEasuTapH(pR, pG, pB, pW, FfxFloat16x2(1.0, 2.0) - ppp.xx, FfxFloat16x2(1.0, 1.0) - ppp.yy, dir, len2, lob, clp, klhgR.xy, klhgG.xy, klhgB.xy); + FsrEasuTapH(pR, pG, pB, pW, FfxFloat16x2(2.0, 1.0) - ppp.xx, FfxFloat16x2(0.0, 0.0) - ppp.yy, dir, len2, lob, clp, klhgR.zw, klhgG.zw, klhgB.zw); + FsrEasuTapH(pR, pG, pB, pW, FfxFloat16x2(1.0, 0.0) - ppp.xx, FfxFloat16x2(2.0, 2.0) - ppp.yy, dir, len2, lob, clp, zzonR.zw, zzonG.zw, zzonB.zw); + FfxFloat16x3 aC = FfxFloat16x3(pR.x + pR.y, pG.x + pG.y, pB.x + pB.y); + FfxFloat16 aW = pW.x + pW.y; + + // Slightly different for FP16 version due to combined min and max. + pix = min(FfxFloat16x3(bothR.y, bothG.y, bothB.y), max(-FfxFloat16x3(bothR.x, bothG.x, bothB.x), aC * FFX_BROADCAST_FLOAT16X3(ffxReciprocalHalf(aW)))); +} +#endif // #if defined(FFX_GPU) && defined(FFX_HALF) && defined(FFX_FSR_EASU_HALF) + +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//_____________________________________________________________/\_______________________________________________________________ +//============================================================================================================================== +// +// FSR - [RCAS] ROBUST CONTRAST ADAPTIVE SHARPENING +// +//------------------------------------------------------------------------------------------------------------------------------ +// CAS uses a simplified mechanism to convert local contrast into a variable amount of sharpness. +// RCAS uses a more exact mechanism, solving for the maximum local sharpness possible before clipping. +// RCAS also has a built in process to limit sharpening of what it detects as possible noise. +// RCAS sharper does not support scaling, as it should be applied after EASU scaling. +// Pass EASU output straight into RCAS, no color conversions necessary. +//------------------------------------------------------------------------------------------------------------------------------ +// RCAS is based on the following logic. +// RCAS uses a 5 tap filter in a cross pattern (same as CAS), +// w n +// w 1 w for taps w m e +// w s +// Where 'w' is the negative lobe weight. +// output = (w*(n+e+w+s)+m)/(4*w+1) +// RCAS solves for 'w' by seeing where the signal might clip out of the {0 to 1} input range, +// 0 == (w*(n+e+w+s)+m)/(4*w+1) -> w = -m/(n+e+w+s) +// 1 == (w*(n+e+w+s)+m)/(4*w+1) -> w = (1-m)/(n+e+w+s-4*1) +// Then chooses the 'w' which results in no clipping, limits 'w', and multiplies by the 'sharp' amount. +// This solution above has issues with MSAA input as the steps along the gradient cause edge detection issues. +// So RCAS uses 4x the maximum and 4x the minimum (depending on equation)in place of the individual taps. +// As well as switching from 'm' to either the minimum or maximum (depending on side), to help in energy conservation. +// This stabilizes RCAS. +// RCAS does a simple highpass which is normalized against the local contrast then shaped, +// 0.25 +// 0.25 -1 0.25 +// 0.25 +// This is used as a noise detection filter, to reduce the effect of RCAS on grain, and focus on real edges. +// +// GLSL example for the required callbacks : +// +// FfxFloat16x4 FsrRcasLoadH(FfxInt16x2 p){return FfxFloat16x4(imageLoad(imgSrc,FfxInt32x2(p)));} +// void FsrRcasInputH(inout FfxFloat16 r,inout FfxFloat16 g,inout FfxFloat16 b) +// { +// //do any simple input color conversions here or leave empty if none needed +// } +// +// FsrRcasCon need to be called from the CPU or GPU to set up constants. +// Including a GPU example here, the 'con' value would be stored out to a constant buffer. +// +// FfxUInt32x4 con; +// FsrRcasCon(con, +// 0.0); // The scale is {0.0 := maximum sharpness, to N>0, where N is the number of stops (halving) of the reduction of sharpness}. +// --------------- +// RCAS sharpening supports a CAS-like pass-through alpha via, +// #define FSR_RCAS_PASSTHROUGH_ALPHA 1 +// RCAS also supports a define to enable a more expensive path to avoid some sharpening of noise. +// Would suggest it is better to apply film grain after RCAS sharpening (and after scaling) instead of using this define, +// #define FSR_RCAS_DENOISE 1 +//============================================================================================================================== +// This is set at the limit of providing unnatural results for sharpening. +#define FSR_RCAS_LIMIT (0.25-(1.0/16.0)) +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//_____________________________________________________________/\_______________________________________________________________ +//============================================================================================================================== +// CONSTANT SETUP +//============================================================================================================================== +// Call to setup required constant values (works on CPU or GPU). + FFX_STATIC void FsrRcasCon(FfxUInt32x4 con, + // The scale is {0.0 := maximum, to N>0, where N is the number of stops (halving) of the reduction of sharpness}. + FfxFloat32 sharpness) + { + // Transform from stops to linear value. + sharpness = exp2(-sharpness); + FfxFloat32x2 hSharp = {sharpness, sharpness}; + con[0] = ffxAsUInt32(sharpness); + con[1] = ffxPackHalf2x16(hSharp); + con[2] = 0; + con[3] = 0; + } + //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//_____________________________________________________________/\_______________________________________________________________ +//============================================================================================================================== +// NON-PACKED 32-BIT VERSION +//============================================================================================================================== +#if defined(FFX_GPU)&&defined(FSR_RCAS_F) + // Input callback prototypes that need to be implemented by calling shader + FfxFloat32x4 FsrRcasLoadF(FfxInt32x2 p); + void FsrRcasInputF(inout FfxFloat32 r,inout FfxFloat32 g,inout FfxFloat32 b); +//------------------------------------------------------------------------------------------------------------------------------ + void FsrRcasF(out FfxFloat32 pixR, // Output values, non-vector so port between RcasFilter() and RcasFilterH() is easy. + out FfxFloat32 pixG, + out FfxFloat32 pixB, +#ifdef FSR_RCAS_PASSTHROUGH_ALPHA + out FfxFloat32 pixA, +#endif + FfxUInt32x2 ip, // Integer pixel position in output. + FfxUInt32x4 con) + { // Constant generated by RcasSetup(). + // Algorithm uses minimal 3x3 pixel neighborhood. + // b + // d e f + // h + FfxInt32x2 sp = FfxInt32x2(ip); + FfxFloat32x3 b = FsrRcasLoadF(sp + FfxInt32x2(0, -1)).rgb; + FfxFloat32x3 d = FsrRcasLoadF(sp + FfxInt32x2(-1, 0)).rgb; +#ifdef FSR_RCAS_PASSTHROUGH_ALPHA + FfxFloat32x4 ee = FsrRcasLoadF(sp); + FfxFloat32x3 e = ee.rgb; + pixA = ee.a; +#else + FfxFloat32x3 e = FsrRcasLoadF(sp).rgb; +#endif + FfxFloat32x3 f = FsrRcasLoadF(sp + FfxInt32x2(1, 0)).rgb; + FfxFloat32x3 h = FsrRcasLoadF(sp + FfxInt32x2(0, 1)).rgb; + // Rename (32-bit) or regroup (16-bit). + FfxFloat32 bR = b.r; + FfxFloat32 bG = b.g; + FfxFloat32 bB = b.b; + FfxFloat32 dR = d.r; + FfxFloat32 dG = d.g; + FfxFloat32 dB = d.b; + FfxFloat32 eR = e.r; + FfxFloat32 eG = e.g; + FfxFloat32 eB = e.b; + FfxFloat32 fR = f.r; + FfxFloat32 fG = f.g; + FfxFloat32 fB = f.b; + FfxFloat32 hR = h.r; + FfxFloat32 hG = h.g; + FfxFloat32 hB = h.b; + // Run optional input transform. + FsrRcasInputF(bR, bG, bB); + FsrRcasInputF(dR, dG, dB); + FsrRcasInputF(eR, eG, eB); + FsrRcasInputF(fR, fG, fB); + FsrRcasInputF(hR, hG, hB); + // Luma times 2. + FfxFloat32 bL = bB * FfxFloat32(0.5) + (bR * FfxFloat32(0.5) + bG); + FfxFloat32 dL = dB * FfxFloat32(0.5) + (dR * FfxFloat32(0.5) + dG); + FfxFloat32 eL = eB * FfxFloat32(0.5) + (eR * FfxFloat32(0.5) + eG); + FfxFloat32 fL = fB * FfxFloat32(0.5) + (fR * FfxFloat32(0.5) + fG); + FfxFloat32 hL = hB * FfxFloat32(0.5) + (hR * FfxFloat32(0.5) + hG); + // Noise detection. + FfxFloat32 nz = FfxFloat32(0.25) * bL + FfxFloat32(0.25) * dL + FfxFloat32(0.25) * fL + FfxFloat32(0.25) * hL - eL; + nz = ffxSaturate(abs(nz) * ffxApproximateReciprocalMedium(ffxMax3(ffxMax3(bL, dL, eL), fL, hL) - ffxMin3(ffxMin3(bL, dL, eL), fL, hL))); + nz = FfxFloat32(-0.5) * nz + FfxFloat32(1.0); + // Min and max of ring. + FfxFloat32 mn4R = ffxMin(ffxMin3(bR, dR, fR), hR); + FfxFloat32 mn4G = ffxMin(ffxMin3(bG, dG, fG), hG); + FfxFloat32 mn4B = ffxMin(ffxMin3(bB, dB, fB), hB); + FfxFloat32 mx4R = max(ffxMax3(bR, dR, fR), hR); + FfxFloat32 mx4G = max(ffxMax3(bG, dG, fG), hG); + FfxFloat32 mx4B = max(ffxMax3(bB, dB, fB), hB); + // Immediate constants for peak range. + FfxFloat32x2 peakC = FfxFloat32x2(1.0, -1.0 * 4.0); + // Limiters, these need to be high precision RCPs. + FfxFloat32 hitMinR = mn4R * ffxReciprocal(FfxFloat32(4.0) * mx4R); + FfxFloat32 hitMinG = mn4G * ffxReciprocal(FfxFloat32(4.0) * mx4G); + FfxFloat32 hitMinB = mn4B * ffxReciprocal(FfxFloat32(4.0) * mx4B); + FfxFloat32 hitMaxR = (peakC.x - mx4R) * ffxReciprocal(FfxFloat32(4.0) * mn4R + peakC.y); + FfxFloat32 hitMaxG = (peakC.x - mx4G) * ffxReciprocal(FfxFloat32(4.0) * mn4G + peakC.y); + FfxFloat32 hitMaxB = (peakC.x - mx4B) * ffxReciprocal(FfxFloat32(4.0) * mn4B + peakC.y); + FfxFloat32 lobeR = max(-hitMinR, hitMaxR); + FfxFloat32 lobeG = max(-hitMinG, hitMaxG); + FfxFloat32 lobeB = max(-hitMinB, hitMaxB); + FfxFloat32 lobe = max(FfxFloat32(-FSR_RCAS_LIMIT), ffxMin(ffxMax3(lobeR, lobeG, lobeB), FfxFloat32(0.0))) * ffxAsFloat + (con.x); + // Apply noise removal. +#ifdef FSR_RCAS_DENOISE + lobe *= nz; +#endif + // Resolve, which needs the medium precision rcp approximation to avoid visible tonality changes. + FfxFloat32 rcpL = ffxApproximateReciprocalMedium(FfxFloat32(4.0) * lobe + FfxFloat32(1.0)); + pixR = (lobe * bR + lobe * dR + lobe * hR + lobe * fR + eR) * rcpL; + pixG = (lobe * bG + lobe * dG + lobe * hG + lobe * fG + eG) * rcpL; + pixB = (lobe * bB + lobe * dB + lobe * hB + lobe * fB + eB) * rcpL; + return; + } +#endif +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//_____________________________________________________________/\_______________________________________________________________ +//============================================================================================================================== +// NON-PACKED 16-BIT VERSION +//============================================================================================================================== +#if defined(FFX_GPU) && FFX_HALF == 1 && defined(FSR_RCAS_H) + // Input callback prototypes that need to be implemented by calling shader + FfxFloat16x4 FsrRcasLoadH(FfxInt16x2 p); + void FsrRcasInputH(inout FfxFloat16 r,inout FfxFloat16 g,inout FfxFloat16 b); +//------------------------------------------------------------------------------------------------------------------------------ + void FsrRcasH( + out FfxFloat16 pixR, // Output values, non-vector so port between RcasFilter() and RcasFilterH() is easy. + out FfxFloat16 pixG, + out FfxFloat16 pixB, + #ifdef FSR_RCAS_PASSTHROUGH_ALPHA + out FfxFloat16 pixA, + #endif + FfxUInt32x2 ip, // Integer pixel position in output. + FfxUInt32x4 con){ // Constant generated by RcasSetup(). + // Sharpening algorithm uses minimal 3x3 pixel neighborhood. + // b + // d e f + // h + FfxInt16x2 sp=FfxInt16x2(ip); + FfxFloat16x3 b=FsrRcasLoadH(sp+FfxInt16x2( 0,-1)).rgb; + FfxFloat16x3 d=FsrRcasLoadH(sp+FfxInt16x2(-1, 0)).rgb; + #ifdef FSR_RCAS_PASSTHROUGH_ALPHA + FfxFloat16x4 ee=FsrRcasLoadH(sp); + FfxFloat16x3 e=ee.rgb;pixA=ee.a; + #else + FfxFloat16x3 e=FsrRcasLoadH(sp).rgb; + #endif + FfxFloat16x3 f=FsrRcasLoadH(sp+FfxInt16x2( 1, 0)).rgb; + FfxFloat16x3 h=FsrRcasLoadH(sp+FfxInt16x2( 0, 1)).rgb; + // Rename (32-bit) or regroup (16-bit). + FfxFloat16 bR=b.r; + FfxFloat16 bG=b.g; + FfxFloat16 bB=b.b; + FfxFloat16 dR=d.r; + FfxFloat16 dG=d.g; + FfxFloat16 dB=d.b; + FfxFloat16 eR=e.r; + FfxFloat16 eG=e.g; + FfxFloat16 eB=e.b; + FfxFloat16 fR=f.r; + FfxFloat16 fG=f.g; + FfxFloat16 fB=f.b; + FfxFloat16 hR=h.r; + FfxFloat16 hG=h.g; + FfxFloat16 hB=h.b; + // Run optional input transform. + FsrRcasInputH(bR,bG,bB); + FsrRcasInputH(dR,dG,dB); + FsrRcasInputH(eR,eG,eB); + FsrRcasInputH(fR,fG,fB); + FsrRcasInputH(hR,hG,hB); + // Luma times 2. + FfxFloat16 bL=bB*FFX_BROADCAST_FLOAT16(0.5)+(bR*FFX_BROADCAST_FLOAT16(0.5)+bG); + FfxFloat16 dL=dB*FFX_BROADCAST_FLOAT16(0.5)+(dR*FFX_BROADCAST_FLOAT16(0.5)+dG); + FfxFloat16 eL=eB*FFX_BROADCAST_FLOAT16(0.5)+(eR*FFX_BROADCAST_FLOAT16(0.5)+eG); + FfxFloat16 fL=fB*FFX_BROADCAST_FLOAT16(0.5)+(fR*FFX_BROADCAST_FLOAT16(0.5)+fG); + FfxFloat16 hL=hB*FFX_BROADCAST_FLOAT16(0.5)+(hR*FFX_BROADCAST_FLOAT16(0.5)+hG); + // Noise detection. + FfxFloat16 nz=FFX_BROADCAST_FLOAT16(0.25)*bL+FFX_BROADCAST_FLOAT16(0.25)*dL+FFX_BROADCAST_FLOAT16(0.25)*fL+FFX_BROADCAST_FLOAT16(0.25)*hL-eL; + nz=ffxSaturate(abs(nz)*ffxApproximateReciprocalMediumHalf(ffxMax3Half(ffxMax3Half(bL,dL,eL),fL,hL)-ffxMin3Half(ffxMin3Half(bL,dL,eL),fL,hL))); + nz=FFX_BROADCAST_FLOAT16(-0.5)*nz+FFX_BROADCAST_FLOAT16(1.0); + // Min and max of ring. + FfxFloat16 mn4R=min(ffxMin3Half(bR,dR,fR),hR); + FfxFloat16 mn4G=min(ffxMin3Half(bG,dG,fG),hG); + FfxFloat16 mn4B=min(ffxMin3Half(bB,dB,fB),hB); + FfxFloat16 mx4R=max(ffxMax3Half(bR,dR,fR),hR); + FfxFloat16 mx4G=max(ffxMax3Half(bG,dG,fG),hG); + FfxFloat16 mx4B=max(ffxMax3Half(bB,dB,fB),hB); + // Immediate constants for peak range. + FfxFloat16x2 peakC=FfxFloat16x2(1.0,-1.0*4.0); + // Limiters, these need to be high precision RCPs. + FfxFloat16 hitMinR=mn4R*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16(4.0)*mx4R); + FfxFloat16 hitMinG=mn4G*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16(4.0)*mx4G); + FfxFloat16 hitMinB=mn4B*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16(4.0)*mx4B); + FfxFloat16 hitMaxR=(peakC.x-mx4R)*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16(4.0)*mn4R+peakC.y); + FfxFloat16 hitMaxG=(peakC.x-mx4G)*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16(4.0)*mn4G+peakC.y); + FfxFloat16 hitMaxB=(peakC.x-mx4B)*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16(4.0)*mn4B+peakC.y); + FfxFloat16 lobeR=max(-hitMinR,hitMaxR); + FfxFloat16 lobeG=max(-hitMinG,hitMaxG); + FfxFloat16 lobeB=max(-hitMinB,hitMaxB); + FfxFloat16 lobe=max(FFX_BROADCAST_FLOAT16(-FSR_RCAS_LIMIT),min(ffxMax3Half(lobeR,lobeG,lobeB),FFX_BROADCAST_FLOAT16(0.0)))*FFX_UINT32_TO_FLOAT16X2(con.y).x; + // Apply noise removal. + #ifdef FSR_RCAS_DENOISE + lobe*=nz; + #endif + // Resolve, which needs the medium precision rcp approximation to avoid visible tonality changes. + FfxFloat16 rcpL=ffxApproximateReciprocalMediumHalf(FFX_BROADCAST_FLOAT16(4.0)*lobe+FFX_BROADCAST_FLOAT16(1.0)); + pixR=(lobe*bR+lobe*dR+lobe*hR+lobe*fR+eR)*rcpL; + pixG=(lobe*bG+lobe*dG+lobe*hG+lobe*fG+eG)*rcpL; + pixB=(lobe*bB+lobe*dB+lobe*hB+lobe*fB+eB)*rcpL; +} +#endif +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//_____________________________________________________________/\_______________________________________________________________ +//============================================================================================================================== +// PACKED 16-BIT VERSION +//============================================================================================================================== +#if defined(FFX_GPU)&& FFX_HALF == 1 && defined(FSR_RCAS_HX2) + // Input callback prototypes that need to be implemented by the calling shader + FfxFloat16x4 FsrRcasLoadHx2(FfxInt16x2 p); + void FsrRcasInputHx2(inout FfxFloat16x2 r,inout FfxFloat16x2 g,inout FfxFloat16x2 b); +//------------------------------------------------------------------------------------------------------------------------------ + // Can be used to convert from packed Structures of Arrays to Arrays of Structures for store. + void FsrRcasDepackHx2(out FfxFloat16x4 pix0,out FfxFloat16x4 pix1,FfxFloat16x2 pixR,FfxFloat16x2 pixG,FfxFloat16x2 pixB){ + #ifdef FFX_HLSL + // Invoke a slower path for DX only, since it won't allow uninitialized values. + pix0.a=pix1.a=0.0; + #endif + pix0.rgb=FfxFloat16x3(pixR.x,pixG.x,pixB.x); + pix1.rgb=FfxFloat16x3(pixR.y,pixG.y,pixB.y);} +//------------------------------------------------------------------------------------------------------------------------------ + void FsrRcasHx2( + // Output values are for 2 8x8 tiles in a 16x8 region. + // pix.x = left 8x8 tile + // pix.y = right 8x8 tile + // This enables later processing to easily be packed as well. + out FfxFloat16x2 pixR, + out FfxFloat16x2 pixG, + out FfxFloat16x2 pixB, + #ifdef FSR_RCAS_PASSTHROUGH_ALPHA + out FfxFloat16x2 pixA, + #endif + FfxUInt32x2 ip, // Integer pixel position in output. + FfxUInt32x4 con){ // Constant generated by RcasSetup(). + // No scaling algorithm uses minimal 3x3 pixel neighborhood. + FfxInt16x2 sp0=FfxInt16x2(ip); + FfxFloat16x3 b0=FsrRcasLoadHx2(sp0+FfxInt16x2( 0,-1)).rgb; + FfxFloat16x3 d0=FsrRcasLoadHx2(sp0+FfxInt16x2(-1, 0)).rgb; + #ifdef FSR_RCAS_PASSTHROUGH_ALPHA + FfxFloat16x4 ee0=FsrRcasLoadHx2(sp0); + FfxFloat16x3 e0=ee0.rgb;pixA.r=ee0.a; + #else + FfxFloat16x3 e0=FsrRcasLoadHx2(sp0).rgb; + #endif + FfxFloat16x3 f0=FsrRcasLoadHx2(sp0+FfxInt16x2( 1, 0)).rgb; + FfxFloat16x3 h0=FsrRcasLoadHx2(sp0+FfxInt16x2( 0, 1)).rgb; + FfxInt16x2 sp1=sp0+FfxInt16x2(8,0); + FfxFloat16x3 b1=FsrRcasLoadHx2(sp1+FfxInt16x2( 0,-1)).rgb; + FfxFloat16x3 d1=FsrRcasLoadHx2(sp1+FfxInt16x2(-1, 0)).rgb; + #ifdef FSR_RCAS_PASSTHROUGH_ALPHA + FfxFloat16x4 ee1=FsrRcasLoadHx2(sp1); + FfxFloat16x3 e1=ee1.rgb;pixA.g=ee1.a; + #else + FfxFloat16x3 e1=FsrRcasLoadHx2(sp1).rgb; + #endif + FfxFloat16x3 f1=FsrRcasLoadHx2(sp1+FfxInt16x2( 1, 0)).rgb; + FfxFloat16x3 h1=FsrRcasLoadHx2(sp1+FfxInt16x2( 0, 1)).rgb; + // Arrays of Structures to Structures of Arrays conversion. + FfxFloat16x2 bR=FfxFloat16x2(b0.r,b1.r); + FfxFloat16x2 bG=FfxFloat16x2(b0.g,b1.g); + FfxFloat16x2 bB=FfxFloat16x2(b0.b,b1.b); + FfxFloat16x2 dR=FfxFloat16x2(d0.r,d1.r); + FfxFloat16x2 dG=FfxFloat16x2(d0.g,d1.g); + FfxFloat16x2 dB=FfxFloat16x2(d0.b,d1.b); + FfxFloat16x2 eR=FfxFloat16x2(e0.r,e1.r); + FfxFloat16x2 eG=FfxFloat16x2(e0.g,e1.g); + FfxFloat16x2 eB=FfxFloat16x2(e0.b,e1.b); + FfxFloat16x2 fR=FfxFloat16x2(f0.r,f1.r); + FfxFloat16x2 fG=FfxFloat16x2(f0.g,f1.g); + FfxFloat16x2 fB=FfxFloat16x2(f0.b,f1.b); + FfxFloat16x2 hR=FfxFloat16x2(h0.r,h1.r); + FfxFloat16x2 hG=FfxFloat16x2(h0.g,h1.g); + FfxFloat16x2 hB=FfxFloat16x2(h0.b,h1.b); + // Run optional input transform. + FsrRcasInputHx2(bR,bG,bB); + FsrRcasInputHx2(dR,dG,dB); + FsrRcasInputHx2(eR,eG,eB); + FsrRcasInputHx2(fR,fG,fB); + FsrRcasInputHx2(hR,hG,hB); + // Luma times 2. + FfxFloat16x2 bL=bB*FFX_BROADCAST_FLOAT16X2(0.5)+(bR*FFX_BROADCAST_FLOAT16X2(0.5)+bG); + FfxFloat16x2 dL=dB*FFX_BROADCAST_FLOAT16X2(0.5)+(dR*FFX_BROADCAST_FLOAT16X2(0.5)+dG); + FfxFloat16x2 eL=eB*FFX_BROADCAST_FLOAT16X2(0.5)+(eR*FFX_BROADCAST_FLOAT16X2(0.5)+eG); + FfxFloat16x2 fL=fB*FFX_BROADCAST_FLOAT16X2(0.5)+(fR*FFX_BROADCAST_FLOAT16X2(0.5)+fG); + FfxFloat16x2 hL=hB*FFX_BROADCAST_FLOAT16X2(0.5)+(hR*FFX_BROADCAST_FLOAT16X2(0.5)+hG); + // Noise detection. + FfxFloat16x2 nz=FFX_BROADCAST_FLOAT16X2(0.25)*bL+FFX_BROADCAST_FLOAT16X2(0.25)*dL+FFX_BROADCAST_FLOAT16X2(0.25)*fL+FFX_BROADCAST_FLOAT16X2(0.25)*hL-eL; + nz=ffxSaturate(abs(nz)*ffxApproximateReciprocalMediumHalf(ffxMax3Half(ffxMax3Half(bL,dL,eL),fL,hL)-ffxMin3Half(ffxMin3Half(bL,dL,eL),fL,hL))); + nz=FFX_BROADCAST_FLOAT16X2(-0.5)*nz+FFX_BROADCAST_FLOAT16X2(1.0); + // Min and max of ring. + FfxFloat16x2 mn4R=min(ffxMin3Half(bR,dR,fR),hR); + FfxFloat16x2 mn4G=min(ffxMin3Half(bG,dG,fG),hG); + FfxFloat16x2 mn4B=min(ffxMin3Half(bB,dB,fB),hB); + FfxFloat16x2 mx4R=max(ffxMax3Half(bR,dR,fR),hR); + FfxFloat16x2 mx4G=max(ffxMax3Half(bG,dG,fG),hG); + FfxFloat16x2 mx4B=max(ffxMax3Half(bB,dB,fB),hB); + // Immediate constants for peak range. + FfxFloat16x2 peakC=FfxFloat16x2(1.0,-1.0*4.0); + // Limiters, these need to be high precision RCPs. + FfxFloat16x2 hitMinR=mn4R*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16X2(4.0)*mx4R); + FfxFloat16x2 hitMinG=mn4G*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16X2(4.0)*mx4G); + FfxFloat16x2 hitMinB=mn4B*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16X2(4.0)*mx4B); + FfxFloat16x2 hitMaxR=(peakC.x-mx4R)*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16X2(4.0)*mn4R+peakC.y); + FfxFloat16x2 hitMaxG=(peakC.x-mx4G)*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16X2(4.0)*mn4G+peakC.y); + FfxFloat16x2 hitMaxB=(peakC.x-mx4B)*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16X2(4.0)*mn4B+peakC.y); + FfxFloat16x2 lobeR=max(-hitMinR,hitMaxR); + FfxFloat16x2 lobeG=max(-hitMinG,hitMaxG); + FfxFloat16x2 lobeB=max(-hitMinB,hitMaxB); + FfxFloat16x2 lobe=max(FFX_BROADCAST_FLOAT16X2(-FSR_RCAS_LIMIT),min(ffxMax3Half(lobeR,lobeG,lobeB),FFX_BROADCAST_FLOAT16X2(0.0)))*FFX_BROADCAST_FLOAT16X2(FFX_UINT32_TO_FLOAT16X2(con.y).x); + // Apply noise removal. + #ifdef FSR_RCAS_DENOISE + lobe*=nz; + #endif + // Resolve, which needs the medium precision rcp approximation to avoid visible tonality changes. + FfxFloat16x2 rcpL=ffxApproximateReciprocalMediumHalf(FFX_BROADCAST_FLOAT16X2(4.0)*lobe+FFX_BROADCAST_FLOAT16X2(1.0)); + pixR=(lobe*bR+lobe*dR+lobe*hR+lobe*fR+eR)*rcpL; + pixG=(lobe*bG+lobe*dG+lobe*hG+lobe*fG+eG)*rcpL; + pixB=(lobe*bB+lobe*dB+lobe*hB+lobe*fB+eB)*rcpL;} +#endif +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//_____________________________________________________________/\_______________________________________________________________ +//============================================================================================================================== +// +// FSR - [LFGA] LINEAR FILM GRAIN APPLICATOR +// +//------------------------------------------------------------------------------------------------------------------------------ +// Adding output-resolution film grain after scaling is a good way to mask both rendering and scaling artifacts. +// Suggest using tiled blue noise as film grain input, with peak noise frequency set for a specific look and feel. +// The 'Lfga*()' functions provide a convenient way to introduce grain. +// These functions limit grain based on distance to signal limits. +// This is done so that the grain is temporally energy preserving, and thus won't modify image tonality. +// Grain application should be done in a linear colorspace. +// The grain should be temporally changing, but have a temporal sum per pixel that adds to zero (non-biased). +//------------------------------------------------------------------------------------------------------------------------------ +// Usage, +// FsrLfga*( +// color, // In/out linear colorspace color {0 to 1} ranged. +// grain, // Per pixel grain texture value {-0.5 to 0.5} ranged, input is 3-channel to support colored grain. +// amount); // Amount of grain (0 to 1} ranged. +//------------------------------------------------------------------------------------------------------------------------------ +// Example if grain texture is monochrome: 'FsrLfgaF(color,ffxBroadcast3(grain),amount)' +//============================================================================================================================== +#if defined(FFX_GPU) + // Maximum grain is the minimum distance to the signal limit. + void FsrLfgaF(inout FfxFloat32x3 c, FfxFloat32x3 t, FfxFloat32 a) + { + c += (t * ffxBroadcast3(a)) * ffxMin(ffxBroadcast3(1.0) - c, c); + } +#endif +//============================================================================================================================== +#if defined(FFX_GPU)&& FFX_HALF == 1 + // Half precision version (slower). + void FsrLfgaH(inout FfxFloat16x3 c, FfxFloat16x3 t, FfxFloat16 a) + { + c += (t * FFX_BROADCAST_FLOAT16X3(a)) * min(FFX_BROADCAST_FLOAT16X3(1.0) - c, c); + } + //------------------------------------------------------------------------------------------------------------------------------ + // Packed half precision version (faster). + void FsrLfgaHx2(inout FfxFloat16x2 cR,inout FfxFloat16x2 cG,inout FfxFloat16x2 cB,FfxFloat16x2 tR,FfxFloat16x2 tG,FfxFloat16x2 tB,FfxFloat16 a){ + cR+=(tR*FFX_BROADCAST_FLOAT16X2(a))*min(FFX_BROADCAST_FLOAT16X2(1.0)-cR,cR);cG+=(tG*FFX_BROADCAST_FLOAT16X2(a))*min(FFX_BROADCAST_FLOAT16X2(1.0)-cG,cG);cB+=(tB*FFX_BROADCAST_FLOAT16X2(a))*min(FFX_BROADCAST_FLOAT16X2(1.0)-cB,cB);} +#endif +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//_____________________________________________________________/\_______________________________________________________________ +//============================================================================================================================== +// +// FSR - [SRTM] SIMPLE REVERSIBLE TONE-MAPPER +// +//------------------------------------------------------------------------------------------------------------------------------ +// This provides a way to take linear HDR color {0 to FP16_MAX} and convert it into a temporary {0 to 1} ranged post-tonemapped linear. +// The tonemapper preserves RGB ratio, which helps maintain HDR color bleed during filtering. +//------------------------------------------------------------------------------------------------------------------------------ +// Reversible tonemapper usage, +// FsrSrtm*(color); // {0 to FP16_MAX} converted to {0 to 1}. +// FsrSrtmInv*(color); // {0 to 1} converted into {0 to 32768, output peak safe for FP16}. +//============================================================================================================================== +#if defined(FFX_GPU) + void FsrSrtmF(inout FfxFloat32x3 c) + { + c *= ffxBroadcast3(ffxReciprocal(ffxMax3(c.r, c.g, c.b) + FfxFloat32(1.0))); + } + // The extra max solves the c=1.0 case (which is a /0). + void FsrSrtmInvF(inout FfxFloat32x3 c){c*=ffxBroadcast3(ffxReciprocal(max(FfxFloat32(1.0/32768.0),FfxFloat32(1.0)-ffxMax3(c.r,c.g,c.b))));} +#endif +//============================================================================================================================== +#if defined(FFX_GPU )&& FFX_HALF == 1 + void FsrSrtmH(inout FfxFloat16x3 c) + { + c *= FFX_BROADCAST_FLOAT16X3(ffxReciprocalHalf(ffxMax3Half(c.r, c.g, c.b) + FFX_BROADCAST_FLOAT16(1.0))); + } + void FsrSrtmInvH(inout FfxFloat16x3 c) + { + c *= FFX_BROADCAST_FLOAT16X3(ffxReciprocalHalf(max(FFX_BROADCAST_FLOAT16(1.0 / 32768.0), FFX_BROADCAST_FLOAT16(1.0) - ffxMax3Half(c.r, c.g, c.b)))); + } + //------------------------------------------------------------------------------------------------------------------------------ + void FsrSrtmHx2(inout FfxFloat16x2 cR, inout FfxFloat16x2 cG, inout FfxFloat16x2 cB) + { + FfxFloat16x2 rcp = ffxReciprocalHalf(ffxMax3Half(cR, cG, cB) + FFX_BROADCAST_FLOAT16X2(1.0)); + cR *= rcp; + cG *= rcp; + cB *= rcp; + } + void FsrSrtmInvHx2(inout FfxFloat16x2 cR,inout FfxFloat16x2 cG,inout FfxFloat16x2 cB) + { + FfxFloat16x2 rcp=ffxReciprocalHalf(max(FFX_BROADCAST_FLOAT16X2(1.0/32768.0),FFX_BROADCAST_FLOAT16X2(1.0)-ffxMax3Half(cR,cG,cB))); + cR*=rcp; + cG*=rcp; + cB*=rcp; + } +#endif +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//_____________________________________________________________/\_______________________________________________________________ +//============================================================================================================================== +// +// FSR - [TEPD] TEMPORAL ENERGY PRESERVING DITHER +// +//------------------------------------------------------------------------------------------------------------------------------ +// Temporally energy preserving dithered {0 to 1} linear to gamma 2.0 conversion. +// Gamma 2.0 is used so that the conversion back to linear is just to square the color. +// The conversion comes in 8-bit and 10-bit modes, designed for output to 8-bit UNORM or 10:10:10:2 respectively. +// Given good non-biased temporal blue noise as dither input, +// the output dither will temporally conserve energy. +// This is done by choosing the linear nearest step point instead of perceptual nearest. +// See code below for details. +//------------------------------------------------------------------------------------------------------------------------------ +// DX SPEC RULES FOR FLOAT->UNORM 8-BIT CONVERSION +// =============================================== +// - Output is 'FfxUInt32(floor(saturate(n)*255.0+0.5))'. +// - Thus rounding is to nearest. +// - NaN gets converted to zero. +// - INF is clamped to {0.0 to 1.0}. +//============================================================================================================================== +#if defined(FFX_GPU) + // Hand tuned integer position to dither value, with more values than simple checkerboard. + // Only 32-bit has enough precision for this compddation. + // Output is {0 to <1}. + FfxFloat32 FsrTepdDitF(FfxUInt32x2 p, FfxUInt32 f) + { + FfxFloat32 x = FfxFloat32(p.x + f); + FfxFloat32 y = FfxFloat32(p.y); + // The 1.61803 golden ratio. + FfxFloat32 a = FfxFloat32((1.0 + ffxSqrt(5.0f)) / 2.0); + // Number designed to provide a good visual pattern. + FfxFloat32 b = FfxFloat32(1.0 / 3.69); + x = x * a + (y * b); + return ffxFract(x); + } + //------------------------------------------------------------------------------------------------------------------------------ + // This version is 8-bit gamma 2.0. + // The 'c' input is {0 to 1}. + // Output is {0 to 1} ready for image store. + void FsrTepdC8F(inout FfxFloat32x3 c, FfxFloat32 dit) + { + FfxFloat32x3 n = ffxSqrt(c); + n = floor(n * ffxBroadcast3(255.0)) * ffxBroadcast3(1.0 / 255.0); + FfxFloat32x3 a = n * n; + FfxFloat32x3 b = n + ffxBroadcast3(1.0 / 255.0); + b = b * b; + // Ratio of 'a' to 'b' required to produce 'c'. + // ffxApproximateReciprocal() won't work here (at least for very high dynamic ranges). + // ffxApproximateReciprocalMedium() is an IADD,FMA,MUL. + FfxFloat32x3 r = (c - b) * ffxApproximateReciprocalMedium(a - b); + // Use the ratio as a cutoff to choose 'a' or 'b'. + // ffxIsGreaterThanZero() is a MUL. + c = ffxSaturate(n + ffxIsGreaterThanZero(ffxBroadcast3(dit) - r) * ffxBroadcast3(1.0 / 255.0)); + } + //------------------------------------------------------------------------------------------------------------------------------ + // This version is 10-bit gamma 2.0. + // The 'c' input is {0 to 1}. + // Output is {0 to 1} ready for image store. + void FsrTepdC10F(inout FfxFloat32x3 c, FfxFloat32 dit) + { + FfxFloat32x3 n = ffxSqrt(c); + n = floor(n * ffxBroadcast3(1023.0)) * ffxBroadcast3(1.0 / 1023.0); + FfxFloat32x3 a = n * n; + FfxFloat32x3 b = n + ffxBroadcast3(1.0 / 1023.0); + b = b * b; + FfxFloat32x3 r = (c - b) * ffxApproximateReciprocalMedium(a - b); + c = ffxSaturate(n + ffxIsGreaterThanZero(ffxBroadcast3(dit) - r) * ffxBroadcast3(1.0 / 1023.0)); + } +#endif +//============================================================================================================================== +#if defined(FFX_GPU)&& FFX_HALF == 1 + FfxFloat16 FsrTepdDitH(FfxUInt32x2 p, FfxUInt32 f) + { + FfxFloat32 x = FfxFloat32(p.x + f); + FfxFloat32 y = FfxFloat32(p.y); + FfxFloat32 a = FfxFloat32((1.0 + ffxSqrt(5.0f)) / 2.0); + FfxFloat32 b = FfxFloat32(1.0 / 3.69); + x = x * a + (y * b); + return FfxFloat16(ffxFract(x)); + } + //------------------------------------------------------------------------------------------------------------------------------ + void FsrTepdC8H(inout FfxFloat16x3 c, FfxFloat16 dit) + { + FfxFloat16x3 n = sqrt(c); + n = floor(n * FFX_BROADCAST_FLOAT16X3(255.0)) * FFX_BROADCAST_FLOAT16X3(1.0 / 255.0); + FfxFloat16x3 a = n * n; + FfxFloat16x3 b = n + FFX_BROADCAST_FLOAT16X3(1.0 / 255.0); + b = b * b; + FfxFloat16x3 r = (c - b) * ffxApproximateReciprocalMediumHalf(a - b); + c = ffxSaturate(n + ffxIsGreaterThanZeroHalf(FFX_BROADCAST_FLOAT16X3(dit) - r) * FFX_BROADCAST_FLOAT16X3(1.0 / 255.0)); + } + //------------------------------------------------------------------------------------------------------------------------------ + void FsrTepdC10H(inout FfxFloat16x3 c, FfxFloat16 dit) + { + FfxFloat16x3 n = sqrt(c); + n = floor(n * FFX_BROADCAST_FLOAT16X3(1023.0)) * FFX_BROADCAST_FLOAT16X3(1.0 / 1023.0); + FfxFloat16x3 a = n * n; + FfxFloat16x3 b = n + FFX_BROADCAST_FLOAT16X3(1.0 / 1023.0); + b = b * b; + FfxFloat16x3 r = (c - b) * ffxApproximateReciprocalMediumHalf(a - b); + c = ffxSaturate(n + ffxIsGreaterThanZeroHalf(FFX_BROADCAST_FLOAT16X3(dit) - r) * FFX_BROADCAST_FLOAT16X3(1.0 / 1023.0)); + } + //============================================================================================================================== + // This computes dither for positions 'p' and 'p+{8,0}'. + FfxFloat16x2 FsrTepdDitHx2(FfxUInt32x2 p, FfxUInt32 f) + { + FfxFloat32x2 x; + x.x = FfxFloat32(p.x + f); + x.y = x.x + FfxFloat32(8.0); + FfxFloat32 y = FfxFloat32(p.y); + FfxFloat32 a = FfxFloat32((1.0 + ffxSqrt(5.0f)) / 2.0); + FfxFloat32 b = FfxFloat32(1.0 / 3.69); + x = x * ffxBroadcast2(a) + ffxBroadcast2(y * b); + return FfxFloat16x2(ffxFract(x)); + } + //------------------------------------------------------------------------------------------------------------------------------ + void FsrTepdC8Hx2(inout FfxFloat16x2 cR, inout FfxFloat16x2 cG, inout FfxFloat16x2 cB, FfxFloat16x2 dit) + { + FfxFloat16x2 nR = sqrt(cR); + FfxFloat16x2 nG = sqrt(cG); + FfxFloat16x2 nB = sqrt(cB); + nR = floor(nR * FFX_BROADCAST_FLOAT16X2(255.0)) * FFX_BROADCAST_FLOAT16X2(1.0 / 255.0); + nG = floor(nG * FFX_BROADCAST_FLOAT16X2(255.0)) * FFX_BROADCAST_FLOAT16X2(1.0 / 255.0); + nB = floor(nB * FFX_BROADCAST_FLOAT16X2(255.0)) * FFX_BROADCAST_FLOAT16X2(1.0 / 255.0); + FfxFloat16x2 aR = nR * nR; + FfxFloat16x2 aG = nG * nG; + FfxFloat16x2 aB = nB * nB; + FfxFloat16x2 bR = nR + FFX_BROADCAST_FLOAT16X2(1.0 / 255.0); + bR = bR * bR; + FfxFloat16x2 bG = nG + FFX_BROADCAST_FLOAT16X2(1.0 / 255.0); + bG = bG * bG; + FfxFloat16x2 bB = nB + FFX_BROADCAST_FLOAT16X2(1.0 / 255.0); + bB = bB * bB; + FfxFloat16x2 rR = (cR - bR) * ffxApproximateReciprocalMediumHalf(aR - bR); + FfxFloat16x2 rG = (cG - bG) * ffxApproximateReciprocalMediumHalf(aG - bG); + FfxFloat16x2 rB = (cB - bB) * ffxApproximateReciprocalMediumHalf(aB - bB); + cR = ffxSaturate(nR + ffxIsGreaterThanZeroHalf(dit - rR) * FFX_BROADCAST_FLOAT16X2(1.0 / 255.0)); + cG = ffxSaturate(nG + ffxIsGreaterThanZeroHalf(dit - rG) * FFX_BROADCAST_FLOAT16X2(1.0 / 255.0)); + cB = ffxSaturate(nB + ffxIsGreaterThanZeroHalf(dit - rB) * FFX_BROADCAST_FLOAT16X2(1.0 / 255.0)); + } + //------------------------------------------------------------------------------------------------------------------------------ + void FsrTepdC10Hx2(inout FfxFloat16x2 cR,inout FfxFloat16x2 cG,inout FfxFloat16x2 cB,FfxFloat16x2 dit){ + FfxFloat16x2 nR=sqrt(cR); + FfxFloat16x2 nG=sqrt(cG); + FfxFloat16x2 nB=sqrt(cB); + nR=floor(nR*FFX_BROADCAST_FLOAT16X2(1023.0))*FFX_BROADCAST_FLOAT16X2(1.0/1023.0); + nG=floor(nG*FFX_BROADCAST_FLOAT16X2(1023.0))*FFX_BROADCAST_FLOAT16X2(1.0/1023.0); + nB=floor(nB*FFX_BROADCAST_FLOAT16X2(1023.0))*FFX_BROADCAST_FLOAT16X2(1.0/1023.0); + FfxFloat16x2 aR=nR*nR; + FfxFloat16x2 aG=nG*nG; + FfxFloat16x2 aB=nB*nB; + FfxFloat16x2 bR=nR+FFX_BROADCAST_FLOAT16X2(1.0/1023.0);bR=bR*bR; + FfxFloat16x2 bG=nG+FFX_BROADCAST_FLOAT16X2(1.0/1023.0);bG=bG*bG; + FfxFloat16x2 bB=nB+FFX_BROADCAST_FLOAT16X2(1.0/1023.0);bB=bB*bB; + FfxFloat16x2 rR=(cR-bR)*ffxApproximateReciprocalMediumHalf(aR-bR); + FfxFloat16x2 rG=(cG-bG)*ffxApproximateReciprocalMediumHalf(aG-bG); + FfxFloat16x2 rB=(cB-bB)*ffxApproximateReciprocalMediumHalf(aB-bB); + cR=ffxSaturate(nR+ffxIsGreaterThanZeroHalf(dit-rR)*FFX_BROADCAST_FLOAT16X2(1.0/1023.0)); + cG=ffxSaturate(nG+ffxIsGreaterThanZeroHalf(dit-rG)*FFX_BROADCAST_FLOAT16X2(1.0/1023.0)); + cB = ffxSaturate(nB + ffxIsGreaterThanZeroHalf(dit - rB) * FFX_BROADCAST_FLOAT16X2(1.0 / 1023.0)); +} +#endif diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr1/ffx_fsr1.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr1/ffx_fsr1.h.meta new file mode 100644 index 0000000..4a8bdf5 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr1/ffx_fsr1.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: e73be06d181f19f42a1e8f1cf6ebc92b +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2.meta new file mode 100644 index 0000000..dcb75fc --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: 3aeed5c87f2e3e345adb5ef1906cd480 +folderAsset: yes +DefaultImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_accumulate.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_accumulate.h new file mode 100644 index 0000000..d87c748 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_accumulate.h @@ -0,0 +1,295 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#ifndef FFX_FSR2_ACCUMULATE_H +#define FFX_FSR2_ACCUMULATE_H + +FfxFloat32 GetPxHrVelocity(FfxFloat32x2 fMotionVector) +{ + return length(fMotionVector * DisplaySize()); +} +#if FFX_HALF +FFX_MIN16_F GetPxHrVelocity(FFX_MIN16_F2 fMotionVector) +{ + return length(fMotionVector * FFX_MIN16_F2(DisplaySize())); +} +#endif + +void Accumulate(const AccumulationPassCommonParams params, FFX_PARAMETER_INOUT FfxFloat32x3 fHistoryColor, FfxFloat32x3 fAccumulation, FFX_PARAMETER_IN FfxFloat32x4 fUpsampledColorAndWeight) +{ + // Avoid invalid values when accumulation and upsampled weight is 0 + fAccumulation = ffxMax(FSR2_EPSILON.xxx, fAccumulation + fUpsampledColorAndWeight.www); + +#if FFX_FSR2_OPTION_HDR_COLOR_INPUT + //YCoCg -> RGB -> Tonemap -> YCoCg (Use RGB tonemapper to avoid color desaturation) + fUpsampledColorAndWeight.xyz = RGBToYCoCg(Tonemap(YCoCgToRGB(fUpsampledColorAndWeight.xyz))); + fHistoryColor = RGBToYCoCg(Tonemap(YCoCgToRGB(fHistoryColor))); +#endif + + const FfxFloat32x3 fAlpha = fUpsampledColorAndWeight.www / fAccumulation; + fHistoryColor = ffxLerp(fHistoryColor, fUpsampledColorAndWeight.xyz, fAlpha); + + fHistoryColor = YCoCgToRGB(fHistoryColor); + +#if FFX_FSR2_OPTION_HDR_COLOR_INPUT + fHistoryColor = InverseTonemap(fHistoryColor); +#endif +} + +void RectifyHistory( + const AccumulationPassCommonParams params, + RectificationBox clippingBox, + FFX_PARAMETER_INOUT FfxFloat32x3 fHistoryColor, + FFX_PARAMETER_INOUT FfxFloat32x3 fAccumulation, + FfxFloat32 fLockContributionThisFrame, + FfxFloat32 fTemporalReactiveFactor, + FfxFloat32 fLumaInstabilityFactor) +{ + FfxFloat32 fScaleFactorInfluence = ffxMin(20.0f, ffxPow(FfxFloat32(1.0f / length(DownscaleFactor().x * DownscaleFactor().y)), 3.0f)); + + const FfxFloat32 fVecolityFactor = ffxSaturate(params.fHrVelocity / 20.0f); + const FfxFloat32 fBoxScaleT = ffxMax(params.fDepthClipFactor, ffxMax(params.fAccumulationMask, fVecolityFactor)); + FfxFloat32 fBoxScale = ffxLerp(fScaleFactorInfluence, 1.0f, fBoxScaleT); + + FfxFloat32x3 fScaledBoxVec = clippingBox.boxVec * fBoxScale; + FfxFloat32x3 boxMin = clippingBox.boxCenter - fScaledBoxVec; + FfxFloat32x3 boxMax = clippingBox.boxCenter + fScaledBoxVec; + FfxFloat32x3 boxCenter = clippingBox.boxCenter; + FfxFloat32 boxVecSize = length(clippingBox.boxVec); + + boxMin = ffxMax(clippingBox.aabbMin, boxMin); + boxMax = ffxMin(clippingBox.aabbMax, boxMax); + + if (any(FFX_GREATER_THAN(boxMin, fHistoryColor)) || any(FFX_GREATER_THAN(fHistoryColor, boxMax))) { + + const FfxFloat32x3 fClampedHistoryColor = clamp(fHistoryColor, boxMin, boxMax); + + FfxFloat32x3 fHistoryContribution = ffxMax(fLumaInstabilityFactor, fLockContributionThisFrame).xxx; + + const FfxFloat32 fReactiveFactor = params.fDilatedReactiveFactor; + const FfxFloat32 fReactiveContribution = 1.0f - ffxPow(fReactiveFactor, 1.0f / 2.0f); + fHistoryContribution *= fReactiveContribution; + + // Scale history color using rectification info, also using accumulation mask to avoid potential invalid color protection + fHistoryColor = ffxLerp(fClampedHistoryColor, fHistoryColor, ffxSaturate(fHistoryContribution)); + + // Scale accumulation using rectification info + const FfxFloat32x3 fAccumulationMin = ffxMin(fAccumulation, FFX_BROADCAST_FLOAT32X3(0.1f)); + fAccumulation = ffxLerp(fAccumulationMin, fAccumulation, ffxSaturate(fHistoryContribution)); + } +} + +void WriteUpscaledOutput(FfxInt32x2 iPxHrPos, FfxFloat32x3 fUpscaledColor) +{ + StoreUpscaledOutput(iPxHrPos, fUpscaledColor); +} + +void FinalizeLockStatus(const AccumulationPassCommonParams params, FfxFloat32x2 fLockStatus, FfxFloat32 fUpsampledWeight) +{ + // we expect similar motion for next frame + // kill lock if that location is outside screen, avoid locks to be clamped to screen borders + FfxFloat32x2 fEstimatedUvNextFrame = params.fHrUv - params.fMotionVector; + if (IsUvInside(fEstimatedUvNextFrame) == false) { + KillLock(fLockStatus); + } + else { + // Decrease lock lifetime + const FfxFloat32 fLifetimeDecreaseLanczosMax = FfxFloat32(JitterSequenceLength()) * FfxFloat32(fAverageLanczosWeightPerFrame); + const FfxFloat32 fLifetimeDecrease = FfxFloat32(fUpsampledWeight / fLifetimeDecreaseLanczosMax); + fLockStatus[LOCK_LIFETIME_REMAINING] = ffxMax(FfxFloat32(0), fLockStatus[LOCK_LIFETIME_REMAINING] - fLifetimeDecrease); + } + + StoreLockStatus(params.iPxHrPos, fLockStatus); +} + + +FfxFloat32x3 ComputeBaseAccumulationWeight(const AccumulationPassCommonParams params, FfxFloat32 fThisFrameReactiveFactor, FfxBoolean bInMotionLastFrame, FfxFloat32 fUpsampledWeight, LockState lockState) +{ + // Always assume max accumulation was reached + FfxFloat32 fBaseAccumulation = fMaxAccumulationLanczosWeight * FfxFloat32(params.bIsExistingSample) * (1.0f - fThisFrameReactiveFactor) * (1.0f - params.fDepthClipFactor); + + fBaseAccumulation = ffxMin(fBaseAccumulation, ffxLerp(fBaseAccumulation, fUpsampledWeight * 10.0f, ffxMax(FfxFloat32(bInMotionLastFrame), ffxSaturate(params.fHrVelocity * FfxFloat32(10))))); + + fBaseAccumulation = ffxMin(fBaseAccumulation, ffxLerp(fBaseAccumulation, fUpsampledWeight, ffxSaturate(params.fHrVelocity / FfxFloat32(20)))); + + return fBaseAccumulation.xxx; +} + +FfxFloat32 ComputeLumaInstabilityFactor(const AccumulationPassCommonParams params, RectificationBox clippingBox, FfxFloat32 fThisFrameReactiveFactor, FfxFloat32 fLuminanceDiff) +{ + const FfxFloat32 fUnormThreshold = 1.0f / 255.0f; + const FfxInt32 N_MINUS_1 = 0; + const FfxInt32 N_MINUS_2 = 1; + const FfxInt32 N_MINUS_3 = 2; + const FfxInt32 N_MINUS_4 = 3; + + FfxFloat32 fCurrentFrameLuma = clippingBox.boxCenter.x; + +#if FFX_FSR2_OPTION_HDR_COLOR_INPUT + fCurrentFrameLuma = fCurrentFrameLuma / (1.0f + ffxMax(0.0f, fCurrentFrameLuma)); +#endif + + fCurrentFrameLuma = round(fCurrentFrameLuma * 255.0f) / 255.0f; + + const FfxBoolean bSampleLumaHistory = (ffxMax(ffxMax(params.fDepthClipFactor, params.fAccumulationMask), fLuminanceDiff) < 0.1f) && (params.bIsNewSample == false); + FfxFloat32x4 fCurrentFrameLumaHistory = bSampleLumaHistory ? SampleLumaHistory(params.fReprojectedHrUv) : FFX_BROADCAST_FLOAT32X4(0.0f); + + FfxFloat32 fLumaInstability = 0.0f; + FfxFloat32 fDiffs0 = (fCurrentFrameLuma - fCurrentFrameLumaHistory[N_MINUS_1]); + + FfxFloat32 fMin = abs(fDiffs0); + + if (fMin >= fUnormThreshold) { + for (int i = N_MINUS_2; i <= N_MINUS_4; i++) { + FfxFloat32 fDiffs1 = (fCurrentFrameLuma - fCurrentFrameLumaHistory[i]); + + if (sign(fDiffs0) == sign(fDiffs1)) { + + // Scale difference to protect historically similar values + const FfxFloat32 fMinBias = 1.0f; + fMin = ffxMin(fMin, abs(fDiffs1) * fMinBias); + } + } + + const FfxFloat32 fBoxSize = clippingBox.boxVec.x; + const FfxFloat32 fBoxSizeFactor = ffxPow(ffxSaturate(fBoxSize / 0.1f), 6.0f); + + fLumaInstability = FfxFloat32(fMin != abs(fDiffs0)) * fBoxSizeFactor; + fLumaInstability = FfxFloat32(fLumaInstability > fUnormThreshold); + + fLumaInstability *= 1.0f - ffxMax(params.fAccumulationMask, ffxPow(fThisFrameReactiveFactor, 1.0f / 6.0f)); + } + + //shift history + fCurrentFrameLumaHistory[N_MINUS_4] = fCurrentFrameLumaHistory[N_MINUS_3]; + fCurrentFrameLumaHistory[N_MINUS_3] = fCurrentFrameLumaHistory[N_MINUS_2]; + fCurrentFrameLumaHistory[N_MINUS_2] = fCurrentFrameLumaHistory[N_MINUS_1]; + fCurrentFrameLumaHistory[N_MINUS_1] = fCurrentFrameLuma; + + StoreLumaHistory(params.iPxHrPos, fCurrentFrameLumaHistory); + + return fLumaInstability * FfxFloat32(fCurrentFrameLumaHistory[N_MINUS_4] != 0); +} + +FfxFloat32 ComputeTemporalReactiveFactor(const AccumulationPassCommonParams params, FfxFloat32 fTemporalReactiveFactor) +{ + FfxFloat32 fNewFactor = ffxMin(0.99f, fTemporalReactiveFactor); + + fNewFactor = ffxMax(fNewFactor, ffxLerp(fNewFactor, 0.4f, ffxSaturate(params.fHrVelocity))); + + fNewFactor = ffxMax(fNewFactor * fNewFactor, ffxMax(params.fDepthClipFactor * 0.1f, params.fDilatedReactiveFactor)); + + // Force reactive factor for new samples + fNewFactor = params.bIsNewSample ? 1.0f : fNewFactor; + + if (ffxSaturate(params.fHrVelocity * 10.0f) >= 1.0f) { + fNewFactor = ffxMax(FSR2_EPSILON, fNewFactor) * -1.0f; + } + + return fNewFactor; +} + +AccumulationPassCommonParams InitParams(FfxInt32x2 iPxHrPos) +{ + AccumulationPassCommonParams params; + + params.iPxHrPos = iPxHrPos; + const FfxFloat32x2 fHrUv = (iPxHrPos + 0.5f) / DisplaySize(); + params.fHrUv = fHrUv; + + const FfxFloat32x2 fLrUvJittered = fHrUv + Jitter() / RenderSize(); + params.fLrUv_HwSampler = ClampUv(fLrUvJittered, RenderSize(), MaxRenderSize()); + + params.fMotionVector = GetMotionVector(iPxHrPos, fHrUv); + params.fHrVelocity = GetPxHrVelocity(params.fMotionVector); + + ComputeReprojectedUVs(params, params.fReprojectedHrUv, params.bIsExistingSample); + + params.fDepthClipFactor = ffxSaturate(SampleDepthClip(params.fLrUv_HwSampler)); + + const FfxFloat32x2 fDilatedReactiveMasks = SampleDilatedReactiveMasks(params.fLrUv_HwSampler); + params.fDilatedReactiveFactor = fDilatedReactiveMasks.x; + params.fAccumulationMask = fDilatedReactiveMasks.y; + params.bIsResetFrame = (0 == FrameIndex()); + + params.bIsNewSample = (params.bIsExistingSample == false || params.bIsResetFrame); + + return params; +} + +void Accumulate(FfxInt32x2 iPxHrPos) +{ + const AccumulationPassCommonParams params = InitParams(iPxHrPos); + + FfxFloat32x3 fHistoryColor = FfxFloat32x3(0, 0, 0); + FfxFloat32x2 fLockStatus; + InitializeNewLockSample(fLockStatus); + + FfxFloat32 fTemporalReactiveFactor = 0.0f; + FfxBoolean bInMotionLastFrame = FFX_FALSE; + LockState lockState = { FFX_FALSE , FFX_FALSE }; + if (params.bIsExistingSample && !params.bIsResetFrame) { + ReprojectHistoryColor(params, fHistoryColor, fTemporalReactiveFactor, bInMotionLastFrame); + lockState = ReprojectHistoryLockStatus(params, fLockStatus); + } + + FfxFloat32 fThisFrameReactiveFactor = ffxMax(params.fDilatedReactiveFactor, fTemporalReactiveFactor); + + FfxFloat32 fLuminanceDiff = 0.0f; + FfxFloat32 fLockContributionThisFrame = 0.0f; + UpdateLockStatus(params, fThisFrameReactiveFactor, lockState, fLockStatus, fLockContributionThisFrame, fLuminanceDiff); + + // Load upsampled input color + RectificationBox clippingBox; + FfxFloat32x4 fUpsampledColorAndWeight = ComputeUpsampledColorAndWeight(params, clippingBox, fThisFrameReactiveFactor); + + const FfxFloat32 fLumaInstabilityFactor = ComputeLumaInstabilityFactor(params, clippingBox, fThisFrameReactiveFactor, fLuminanceDiff); + + + FfxFloat32x3 fAccumulation = ComputeBaseAccumulationWeight(params, fThisFrameReactiveFactor, bInMotionLastFrame, fUpsampledColorAndWeight.w, lockState); + + if (params.bIsNewSample) { + fHistoryColor = YCoCgToRGB(fUpsampledColorAndWeight.xyz); + } + else { + RectifyHistory(params, clippingBox, fHistoryColor, fAccumulation, fLockContributionThisFrame, fThisFrameReactiveFactor, fLumaInstabilityFactor); + + Accumulate(params, fHistoryColor, fAccumulation, fUpsampledColorAndWeight); + } + + fHistoryColor = UnprepareRgb(fHistoryColor, Exposure()); + + FinalizeLockStatus(params, fLockStatus, fUpsampledColorAndWeight.w); + + // Get new temporal reactive factor + fTemporalReactiveFactor = ComputeTemporalReactiveFactor(params, fThisFrameReactiveFactor); + + StoreInternalColorAndWeight(iPxHrPos, FfxFloat32x4(fHistoryColor, fTemporalReactiveFactor)); + + // Output final color when RCAS is disabled +#if FFX_FSR2_OPTION_APPLY_SHARPENING == 0 + WriteUpscaledOutput(iPxHrPos, fHistoryColor); +#endif + StoreNewLocks(iPxHrPos, 0); +} + +#endif // FFX_FSR2_ACCUMULATE_H diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_accumulate.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_accumulate.h.meta new file mode 100644 index 0000000..7b61108 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_accumulate.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: ad56ff15a05a8c1479276e80fc3ea721 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_callbacks_hlsl.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_callbacks_hlsl.h new file mode 100644 index 0000000..c52cc1a --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_callbacks_hlsl.h @@ -0,0 +1,945 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#include "ffx_fsr2_resources.h" + +#if defined(FFX_GPU) +#ifdef __hlsl_dx_compiler +#pragma dxc diagnostic push +#pragma dxc diagnostic ignored "-Wambig-lit-shift" +#endif //__hlsl_dx_compiler +#include "../ffx_core.h" +#ifdef __hlsl_dx_compiler +#pragma dxc diagnostic pop +#endif //__hlsl_dx_compiler + +#ifndef FFX_PREFER_WAVE64 +#define FFX_PREFER_WAVE64 +#endif // #ifndef FFX_PREFER_WAVE64 + +#pragma warning(disable: 3205) // conversion from larger type to smaller + +#define DECLARE_SRV_REGISTER(regIndex) t##regIndex +#define DECLARE_UAV_REGISTER(regIndex) u##regIndex +#define DECLARE_CB_REGISTER(regIndex) b##regIndex +#define FFX_FSR2_DECLARE_SRV(regIndex) register(DECLARE_SRV_REGISTER(regIndex)) +#define FFX_FSR2_DECLARE_UAV(regIndex) register(DECLARE_UAV_REGISTER(regIndex)) +#define FFX_FSR2_DECLARE_CB(regIndex) register(DECLARE_CB_REGISTER(regIndex)) + +#if defined(FSR2_BIND_CB_FSR2) + cbuffer cbFSR2 : FFX_FSR2_DECLARE_CB(FSR2_BIND_CB_FSR2) + { + FfxInt32x2 iRenderSize; + FfxInt32x2 iMaxRenderSize; + FfxInt32x2 iDisplaySize; + FfxInt32x2 iInputColorResourceDimensions; + FfxInt32x2 iLumaMipDimensions; + FfxInt32 iLumaMipLevelToUse; + FfxInt32 iFrameIndex; + + FfxFloat32x4 fDeviceToViewDepth; + FfxFloat32x2 fJitter; + FfxFloat32x2 fMotionVectorScale; + FfxFloat32x2 fDownscaleFactor; + FfxFloat32x2 fMotionVectorJitterCancellation; + FfxFloat32 fPreExposure; + FfxFloat32 fPreviousFramePreExposure; + FfxFloat32 fTanHalfFOV; + FfxFloat32 fJitterSequenceLength; + FfxFloat32 fDeltaTime; + FfxFloat32 fDynamicResChangeFactor; + FfxFloat32 fViewSpaceToMetersFactor; + FfxFloat32 fPadding; + }; + +#define FFX_FSR2_CONSTANT_BUFFER_1_SIZE 32 + +/* Define getter functions in the order they are defined in the CB! */ +FfxInt32x2 RenderSize() +{ + return iRenderSize; +} + +FfxInt32x2 MaxRenderSize() +{ + return iMaxRenderSize; +} + +FfxInt32x2 DisplaySize() +{ + return iDisplaySize; +} + +FfxInt32x2 InputColorResourceDimensions() +{ + return iInputColorResourceDimensions; +} + +FfxInt32x2 LumaMipDimensions() +{ + return iLumaMipDimensions; +} + +FfxInt32 LumaMipLevelToUse() +{ + return iLumaMipLevelToUse; +} + +FfxInt32 FrameIndex() +{ + return iFrameIndex; +} + +FfxFloat32x2 Jitter() +{ + return fJitter; +} + +FfxFloat32x4 DeviceToViewSpaceTransformFactors() +{ + return fDeviceToViewDepth; +} + +FfxFloat32x2 MotionVectorScale() +{ + return fMotionVectorScale; +} + +FfxFloat32x2 DownscaleFactor() +{ + return fDownscaleFactor; +} + +FfxFloat32x2 MotionVectorJitterCancellation() +{ + return fMotionVectorJitterCancellation; +} + +FfxFloat32 PreExposure() +{ + return fPreExposure; +} + +FfxFloat32 PreviousFramePreExposure() +{ + return fPreviousFramePreExposure; +} + +FfxFloat32 TanHalfFoV() +{ + return fTanHalfFOV; +} + +FfxFloat32 JitterSequenceLength() +{ + return fJitterSequenceLength; +} + +FfxFloat32 DeltaTime() +{ + return fDeltaTime; +} + +FfxFloat32 DynamicResChangeFactor() +{ + return fDynamicResChangeFactor; +} + +FfxFloat32 ViewSpaceToMetersFactor() +{ + return fViewSpaceToMetersFactor; +} +#endif // #if defined(FSR2_BIND_CB_FSR2) + +#define FFX_FSR2_ROOTSIG_STRINGIFY(p) FFX_FSR2_ROOTSIG_STR(p) +#define FFX_FSR2_ROOTSIG_STR(p) #p +#define FFX_FSR2_ROOTSIG [RootSignature("DescriptorTable(UAV(u0, numDescriptors = " FFX_FSR2_ROOTSIG_STRINGIFY(FFX_FSR2_RESOURCE_IDENTIFIER_COUNT) ")), " \ + "DescriptorTable(SRV(t0, numDescriptors = " FFX_FSR2_ROOTSIG_STRINGIFY(FFX_FSR2_RESOURCE_IDENTIFIER_COUNT) ")), " \ + "CBV(b0), " \ + "StaticSampler(s0, filter = FILTER_MIN_MAG_MIP_POINT, " \ + "addressU = TEXTURE_ADDRESS_CLAMP, " \ + "addressV = TEXTURE_ADDRESS_CLAMP, " \ + "addressW = TEXTURE_ADDRESS_CLAMP, " \ + "comparisonFunc = COMPARISON_NEVER, " \ + "borderColor = STATIC_BORDER_COLOR_TRANSPARENT_BLACK), " \ + "StaticSampler(s1, filter = FILTER_MIN_MAG_MIP_LINEAR, " \ + "addressU = TEXTURE_ADDRESS_CLAMP, " \ + "addressV = TEXTURE_ADDRESS_CLAMP, " \ + "addressW = TEXTURE_ADDRESS_CLAMP, " \ + "comparisonFunc = COMPARISON_NEVER, " \ + "borderColor = STATIC_BORDER_COLOR_TRANSPARENT_BLACK)" )] + +#define FFX_FSR2_CONSTANT_BUFFER_2_SIZE 6 // Number of 32-bit values. This must be kept in sync with max( cbRCAS , cbSPD) size. + +#define FFX_FSR2_CB2_ROOTSIG [RootSignature("DescriptorTable(UAV(u0, numDescriptors = " FFX_FSR2_ROOTSIG_STRINGIFY(FFX_FSR2_RESOURCE_IDENTIFIER_COUNT) ")), " \ + "DescriptorTable(SRV(t0, numDescriptors = " FFX_FSR2_ROOTSIG_STRINGIFY(FFX_FSR2_RESOURCE_IDENTIFIER_COUNT) ")), " \ + "CBV(b0), " \ + "CBV(b1), " \ + "StaticSampler(s0, filter = FILTER_MIN_MAG_MIP_POINT, " \ + "addressU = TEXTURE_ADDRESS_CLAMP, " \ + "addressV = TEXTURE_ADDRESS_CLAMP, " \ + "addressW = TEXTURE_ADDRESS_CLAMP, " \ + "comparisonFunc = COMPARISON_NEVER, " \ + "borderColor = STATIC_BORDER_COLOR_TRANSPARENT_BLACK), " \ + "StaticSampler(s1, filter = FILTER_MIN_MAG_MIP_LINEAR, " \ + "addressU = TEXTURE_ADDRESS_CLAMP, " \ + "addressV = TEXTURE_ADDRESS_CLAMP, " \ + "addressW = TEXTURE_ADDRESS_CLAMP, " \ + "comparisonFunc = COMPARISON_NEVER, " \ + "borderColor = STATIC_BORDER_COLOR_TRANSPARENT_BLACK)" )] + +#define FFX_FSR2_CONSTANT_BUFFER_3_SIZE 4 // Number of 32-bit values. This must be kept in sync with cbGenerateReactive size. + +#define FFX_FSR2_REACTIVE_ROOTSIG [RootSignature("DescriptorTable(UAV(u0, numDescriptors = " FFX_FSR2_ROOTSIG_STRINGIFY(FFX_FSR2_RESOURCE_IDENTIFIER_COUNT) ")), " \ + "DescriptorTable(SRV(t0, numDescriptors = " FFX_FSR2_ROOTSIG_STRINGIFY(FFX_FSR2_RESOURCE_IDENTIFIER_COUNT) ")), " \ + "CBV(b0), " \ + "CBV(b1), " \ + "StaticSampler(s0, filter = FILTER_MIN_MAG_MIP_POINT, " \ + "addressU = TEXTURE_ADDRESS_CLAMP, " \ + "addressV = TEXTURE_ADDRESS_CLAMP, " \ + "addressW = TEXTURE_ADDRESS_CLAMP, " \ + "comparisonFunc = COMPARISON_NEVER, " \ + "borderColor = STATIC_BORDER_COLOR_TRANSPARENT_BLACK), " \ + "StaticSampler(s1, filter = FILTER_MIN_MAG_MIP_LINEAR, " \ + "addressU = TEXTURE_ADDRESS_CLAMP, " \ + "addressV = TEXTURE_ADDRESS_CLAMP, " \ + "addressW = TEXTURE_ADDRESS_CLAMP, " \ + "comparisonFunc = COMPARISON_NEVER, " \ + "borderColor = STATIC_BORDER_COLOR_TRANSPARENT_BLACK)" )] + +#if defined(FFX_FSR2_EMBED_ROOTSIG) +#define FFX_FSR2_EMBED_ROOTSIG_CONTENT FFX_FSR2_ROOTSIG +#define FFX_FSR2_EMBED_CB2_ROOTSIG_CONTENT FFX_FSR2_CB2_ROOTSIG +#define FFX_FSR2_EMBED_ROOTSIG_REACTIVE_CONTENT FFX_FSR2_REACTIVE_ROOTSIG +#else +#define FFX_FSR2_EMBED_ROOTSIG_CONTENT +#define FFX_FSR2_EMBED_CB2_ROOTSIG_CONTENT +#define FFX_FSR2_EMBED_ROOTSIG_REACTIVE_CONTENT +#endif // #if FFX_FSR2_EMBED_ROOTSIG + +#if defined(FSR2_BIND_CB_AUTOREACTIVE) +cbuffer cbGenerateReactive : FFX_FSR2_DECLARE_CB(FSR2_BIND_CB_AUTOREACTIVE) +{ + FfxFloat32 fTcThreshold; // 0.1 is a good starting value, lower will result in more TC pixels + FfxFloat32 fTcScale; + FfxFloat32 fReactiveScale; + FfxFloat32 fReactiveMax; +}; + +FfxFloat32 TcThreshold() +{ + return fTcThreshold; +} + +FfxFloat32 TcScale() +{ + return fTcScale; +} + +FfxFloat32 ReactiveScale() +{ + return fReactiveScale; +} + +FfxFloat32 ReactiveMax() +{ + return fReactiveMax; +} +#endif // #if defined(FSR2_BIND_CB_AUTOREACTIVE) + +#if defined(FSR2_BIND_CB_RCAS) +cbuffer cbRCAS : FFX_FSR2_DECLARE_CB(FSR2_BIND_CB_RCAS) +{ + FfxUInt32x4 rcasConfig; +}; + +FfxUInt32x4 RCASConfig() +{ + return rcasConfig; +} +#endif // #if defined(FSR2_BIND_CB_RCAS) + + +#if defined(FSR2_BIND_CB_REACTIVE) +cbuffer cbGenerateReactive : FFX_FSR2_DECLARE_CB(FSR2_BIND_CB_REACTIVE) +{ + FfxFloat32 gen_reactive_scale; + FfxFloat32 gen_reactive_threshold; + FfxFloat32 gen_reactive_binaryValue; + FfxUInt32 gen_reactive_flags; +}; + +FfxFloat32 GenReactiveScale() +{ + return gen_reactive_scale; +} + +FfxFloat32 GenReactiveThreshold() +{ + return gen_reactive_threshold; +} + +FfxFloat32 GenReactiveBinaryValue() +{ + return gen_reactive_binaryValue; +} + +FfxUInt32 GenReactiveFlags() +{ + return gen_reactive_flags; +} +#endif // #if defined(FSR2_BIND_CB_REACTIVE) + +#if defined(FSR2_BIND_CB_SPD) +cbuffer cbSPD : FFX_FSR2_DECLARE_CB(FSR2_BIND_CB_SPD) { + + FfxUInt32 mips; + FfxUInt32 numWorkGroups; + FfxUInt32x2 workGroupOffset; + FfxUInt32x2 renderSize; +}; + +FfxUInt32 MipCount() +{ + return mips; +} + +FfxUInt32 NumWorkGroups() +{ + return numWorkGroups; +} + +FfxUInt32x2 WorkGroupOffset() +{ + return workGroupOffset; +} + +FfxUInt32x2 SPD_RenderSize() +{ + return renderSize; +} +#endif // #if defined(FSR2_BIND_CB_SPD) + +// Declare and sample camera buffers as regular textures, unless overridden +#if !defined(UNITY_FSR_TEX2D) +#define UNITY_FSR_TEX2D(type) Texture2D +#endif +#if !defined(UNITY_FSR_RWTEX2D) +#define UNITY_FSR_RWTEX2D(type) RWTexture2D +#endif +#if !defined(UNITY_FSR_POS) +#define UNITY_FSR_POS(pxPos) (pxPos) +#endif +#if !defined(UNITY_FSR_UV) +#define UNITY_FSR_UV(uv) (uv) +#endif + +SamplerState s_PointClamp : register(s0); +SamplerState s_LinearClamp : register(s1); + + // SRVs + #if defined FSR2_BIND_SRV_INPUT_COLOR + UNITY_FSR_TEX2D(FfxFloat32x4) r_input_color_jittered : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_INPUT_COLOR); + #endif + #if defined FSR2_BIND_SRV_INPUT_OPAQUE_ONLY + UNITY_FSR_TEX2D(FfxFloat32x4) r_input_opaque_only : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_INPUT_OPAQUE_ONLY); + #endif + #if defined FSR2_BIND_SRV_INPUT_MOTION_VECTORS + UNITY_FSR_TEX2D(FfxFloat32x4) r_input_motion_vectors : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_INPUT_MOTION_VECTORS); + #endif + #if defined FSR2_BIND_SRV_INPUT_DEPTH + UNITY_FSR_TEX2D(FfxFloat32) r_input_depth : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_INPUT_DEPTH); + #endif + #if defined FSR2_BIND_SRV_INPUT_EXPOSURE + Texture2D r_input_exposure : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_INPUT_EXPOSURE); + #endif + #if defined FSR2_BIND_SRV_AUTO_EXPOSURE + Texture2D r_auto_exposure : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_AUTO_EXPOSURE); + #endif + #if defined FSR2_BIND_SRV_REACTIVE_MASK + UNITY_FSR_TEX2D(FfxFloat32) r_reactive_mask : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_REACTIVE_MASK); + #endif + #if defined FSR2_BIND_SRV_TRANSPARENCY_AND_COMPOSITION_MASK + UNITY_FSR_TEX2D(FfxFloat32) r_transparency_and_composition_mask : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_TRANSPARENCY_AND_COMPOSITION_MASK); + #endif + #if defined FSR2_BIND_SRV_RECONSTRUCTED_PREV_NEAREST_DEPTH + Texture2D r_reconstructed_previous_nearest_depth : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_RECONSTRUCTED_PREV_NEAREST_DEPTH); + #endif + #if defined FSR2_BIND_SRV_DILATED_MOTION_VECTORS + Texture2D r_dilated_motion_vectors : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_DILATED_MOTION_VECTORS); + #endif + #if defined FSR2_BIND_SRV_PREVIOUS_DILATED_MOTION_VECTORS + Texture2D r_previous_dilated_motion_vectors : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_PREVIOUS_DILATED_MOTION_VECTORS); + #endif + #if defined FSR2_BIND_SRV_DILATED_DEPTH + Texture2D r_dilatedDepth : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_DILATED_DEPTH); + #endif + #if defined FSR2_BIND_SRV_INTERNAL_UPSCALED + Texture2D r_internal_upscaled_color : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_INTERNAL_UPSCALED); + #endif + #if defined FSR2_BIND_SRV_LOCK_STATUS + Texture2D r_lock_status : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_LOCK_STATUS); + #endif + #if defined FSR2_BIND_SRV_LOCK_INPUT_LUMA + Texture2D r_lock_input_luma : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_LOCK_INPUT_LUMA); + #endif + #if defined FSR2_BIND_SRV_NEW_LOCKS + Texture2D r_new_locks : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_NEW_LOCKS); + #endif + #if defined FSR2_BIND_SRV_PREPARED_INPUT_COLOR + Texture2D r_prepared_input_color : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_PREPARED_INPUT_COLOR); + #endif + #if defined FSR2_BIND_SRV_LUMA_HISTORY + Texture2D r_luma_history : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_LUMA_HISTORY); + #endif + #if defined FSR2_BIND_SRV_RCAS_INPUT + Texture2D r_rcas_input : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_RCAS_INPUT); + #endif + #if defined FSR2_BIND_SRV_LANCZOS_LUT + Texture2D r_lanczos_lut : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_LANCZOS_LUT); + #endif + #if defined FSR2_BIND_SRV_SCENE_LUMINANCE_MIPS + Texture2D r_imgMips : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_SCENE_LUMINANCE_MIPS); + #endif + #if defined FSR2_BIND_SRV_UPSCALE_MAXIMUM_BIAS_LUT + Texture2D r_upsample_maximum_bias_lut : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_UPSCALE_MAXIMUM_BIAS_LUT); + #endif + #if defined FSR2_BIND_SRV_DILATED_REACTIVE_MASKS + Texture2D r_dilated_reactive_masks : FFX_FSR2_DECLARE_SRV(FSR2_BIND_SRV_DILATED_REACTIVE_MASKS); + #endif + + #if defined FSR2_BIND_SRV_PREV_PRE_ALPHA_COLOR + Texture2D r_input_prev_color_pre_alpha : FFX_FSR2_DECLARE_SRV(FFX_FSR2_RESOURCE_IDENTIFIER_PREV_PRE_ALPHA_COLOR); + #endif + #if defined FSR2_BIND_SRV_PREV_POST_ALPHA_COLOR + Texture2D r_input_prev_color_post_alpha : FFX_FSR2_DECLARE_SRV(FFX_FSR2_RESOURCE_IDENTIFIER_PREV_POST_ALPHA_COLOR); + #endif + + // UAV declarations + #if defined FSR2_BIND_UAV_RECONSTRUCTED_PREV_NEAREST_DEPTH + RWTexture2D rw_reconstructed_previous_nearest_depth : FFX_FSR2_DECLARE_UAV(FSR2_BIND_UAV_RECONSTRUCTED_PREV_NEAREST_DEPTH); + #endif + #if defined FSR2_BIND_UAV_DILATED_MOTION_VECTORS + RWTexture2D rw_dilated_motion_vectors : FFX_FSR2_DECLARE_UAV(FSR2_BIND_UAV_DILATED_MOTION_VECTORS); + #endif + #if defined FSR2_BIND_UAV_DILATED_DEPTH + RWTexture2D rw_dilatedDepth : FFX_FSR2_DECLARE_UAV(FSR2_BIND_UAV_DILATED_DEPTH); + #endif + #if defined FSR2_BIND_UAV_INTERNAL_UPSCALED + RWTexture2D rw_internal_upscaled_color : FFX_FSR2_DECLARE_UAV(FSR2_BIND_UAV_INTERNAL_UPSCALED); + #endif + #if defined FSR2_BIND_UAV_LOCK_STATUS + RWTexture2D rw_lock_status : FFX_FSR2_DECLARE_UAV(FSR2_BIND_UAV_LOCK_STATUS); + #endif + #if defined FSR2_BIND_UAV_LOCK_INPUT_LUMA + RWTexture2D rw_lock_input_luma : FFX_FSR2_DECLARE_UAV(FSR2_BIND_UAV_LOCK_INPUT_LUMA); + #endif + #if defined FSR2_BIND_UAV_NEW_LOCKS + RWTexture2D rw_new_locks : FFX_FSR2_DECLARE_UAV(FSR2_BIND_UAV_NEW_LOCKS); + #endif + #if defined FSR2_BIND_UAV_PREPARED_INPUT_COLOR + RWTexture2D rw_prepared_input_color : FFX_FSR2_DECLARE_UAV(FSR2_BIND_UAV_PREPARED_INPUT_COLOR); + #endif + #if defined FSR2_BIND_UAV_LUMA_HISTORY + RWTexture2D rw_luma_history : FFX_FSR2_DECLARE_UAV(FSR2_BIND_UAV_LUMA_HISTORY); + #endif + #if defined FSR2_BIND_UAV_UPSCALED_OUTPUT + UNITY_FSR_RWTEX2D(FfxFloat32x4) rw_upscaled_output : FFX_FSR2_DECLARE_UAV(FSR2_BIND_UAV_UPSCALED_OUTPUT); + #endif + #if defined FSR2_BIND_UAV_EXPOSURE_MIP_LUMA_CHANGE + globallycoherent RWTexture2D rw_img_mip_shading_change : FFX_FSR2_DECLARE_UAV(FSR2_BIND_UAV_EXPOSURE_MIP_LUMA_CHANGE); + #endif + #if defined FSR2_BIND_UAV_EXPOSURE_MIP_5 + globallycoherent RWTexture2D rw_img_mip_5 : FFX_FSR2_DECLARE_UAV(FSR2_BIND_UAV_EXPOSURE_MIP_5); + #endif + #if defined FSR2_BIND_UAV_DILATED_REACTIVE_MASKS + RWTexture2D rw_dilated_reactive_masks : FFX_FSR2_DECLARE_UAV(FSR2_BIND_UAV_DILATED_REACTIVE_MASKS); + #endif + #if defined FSR2_BIND_UAV_EXPOSURE + RWTexture2D rw_exposure : FFX_FSR2_DECLARE_UAV(FSR2_BIND_UAV_EXPOSURE); + #endif + #if defined FSR2_BIND_UAV_AUTO_EXPOSURE + RWTexture2D rw_auto_exposure : FFX_FSR2_DECLARE_UAV(FSR2_BIND_UAV_AUTO_EXPOSURE); + #endif + #if defined FSR2_BIND_UAV_SPD_GLOBAL_ATOMIC + globallycoherent RWTexture2D rw_spd_global_atomic : FFX_FSR2_DECLARE_UAV(FSR2_BIND_UAV_SPD_GLOBAL_ATOMIC); + #endif + + #if defined FSR2_BIND_UAV_AUTOREACTIVE + RWTexture2D rw_output_autoreactive : FFX_FSR2_DECLARE_UAV(FSR2_BIND_UAV_AUTOREACTIVE); + #endif + #if defined FSR2_BIND_UAV_AUTOCOMPOSITION + RWTexture2D rw_output_autocomposition : FFX_FSR2_DECLARE_UAV(FSR2_BIND_UAV_AUTOCOMPOSITION); + #endif + #if defined FSR2_BIND_UAV_PREV_PRE_ALPHA_COLOR + RWTexture2D rw_output_prev_color_pre_alpha : FFX_FSR2_DECLARE_UAV(FSR2_BIND_UAV_PREV_PRE_ALPHA_COLOR); + #endif + #if defined FSR2_BIND_UAV_PREV_POST_ALPHA_COLOR + RWTexture2D rw_output_prev_color_post_alpha : FFX_FSR2_DECLARE_UAV(FSR2_BIND_UAV_PREV_POST_ALPHA_COLOR); + #endif + +#if defined(FSR2_BIND_SRV_SCENE_LUMINANCE_MIPS) +FfxFloat32 LoadMipLuma(FfxUInt32x2 iPxPos, FfxUInt32 mipLevel) +{ + return r_imgMips.mips[mipLevel][iPxPos]; +} +#endif + +#if defined(FSR2_BIND_SRV_SCENE_LUMINANCE_MIPS) +FfxFloat32 SampleMipLuma(FfxFloat32x2 fUV, FfxUInt32 mipLevel) +{ + return r_imgMips.SampleLevel(s_LinearClamp, fUV, mipLevel); +} +#endif + +#if defined(FSR2_BIND_SRV_INPUT_DEPTH) +FfxFloat32 LoadInputDepth(FfxUInt32x2 iPxPos) +{ + return r_input_depth[UNITY_FSR_POS(iPxPos)]; +} +#endif + +#if defined(FSR2_BIND_SRV_INPUT_DEPTH) +FfxFloat32 SampleInputDepth(FfxFloat32x2 fUV) +{ + return r_input_depth.SampleLevel(s_LinearClamp, UNITY_FSR_UV(fUV), 0).x; +} +#endif + +#if defined(FSR2_BIND_SRV_REACTIVE_MASK) +FfxFloat32 LoadReactiveMask(FfxUInt32x2 iPxPos) +{ + return r_reactive_mask[UNITY_FSR_POS(iPxPos)]; +} +#endif + +#if defined(FSR2_BIND_SRV_TRANSPARENCY_AND_COMPOSITION_MASK) +FfxFloat32 LoadTransparencyAndCompositionMask(FfxUInt32x2 iPxPos) +{ + return r_transparency_and_composition_mask[UNITY_FSR_POS(iPxPos)]; +} +#endif + +#if defined(FSR2_BIND_SRV_INPUT_COLOR) +FfxFloat32x3 LoadInputColor(FfxUInt32x2 iPxPos) +{ + return r_input_color_jittered[UNITY_FSR_POS(iPxPos)].rgb; +} +#endif + +#if defined(FSR2_BIND_SRV_INPUT_COLOR) +FfxFloat32x3 SampleInputColor(FfxFloat32x2 fUV) +{ + return r_input_color_jittered.SampleLevel(s_LinearClamp, UNITY_FSR_UV(fUV), 0).rgb; +} +#endif + +#if defined(FSR2_BIND_SRV_PREPARED_INPUT_COLOR) +FfxFloat32x3 LoadPreparedInputColor(FfxUInt32x2 iPxPos) +{ + return r_prepared_input_color[iPxPos].xyz; +} +#endif + +#if defined(FSR2_BIND_SRV_INPUT_MOTION_VECTORS) +FfxFloat32x2 LoadInputMotionVector(FfxUInt32x2 iPxDilatedMotionVectorPos) +{ + FfxFloat32x2 fSrcMotionVector = r_input_motion_vectors[UNITY_FSR_POS(iPxDilatedMotionVectorPos)].xy; + + FfxFloat32x2 fUvMotionVector = fSrcMotionVector * MotionVectorScale(); + +#if FFX_FSR2_OPTION_JITTERED_MOTION_VECTORS + fUvMotionVector -= MotionVectorJitterCancellation(); +#endif + + return fUvMotionVector; +} +#endif + +#if defined(FSR2_BIND_SRV_INTERNAL_UPSCALED) +FfxFloat32x4 LoadHistory(FfxUInt32x2 iPxHistory) +{ + return r_internal_upscaled_color[iPxHistory]; +} +#endif + +#if defined(FSR2_BIND_UAV_LUMA_HISTORY) +void StoreLumaHistory(FfxUInt32x2 iPxPos, FfxFloat32x4 fLumaHistory) +{ + rw_luma_history[iPxPos] = fLumaHistory; +} +#endif + +#if defined(FSR2_BIND_SRV_LUMA_HISTORY) +FfxFloat32x4 SampleLumaHistory(FfxFloat32x2 fUV) +{ + return r_luma_history.SampleLevel(s_LinearClamp, fUV, 0); +} +#endif + +FfxFloat32x4 LoadRCAS_Input(FfxInt32x2 iPxPos) +{ +#if defined(FSR2_BIND_SRV_RCAS_INPUT) + return r_rcas_input[iPxPos]; +#else + return 0.0; +#endif +} + +#if defined(FSR2_BIND_UAV_INTERNAL_UPSCALED) +void StoreReprojectedHistory(FfxUInt32x2 iPxHistory, FfxFloat32x4 fHistory) +{ + rw_internal_upscaled_color[iPxHistory] = fHistory; +} +#endif + +#if defined(FSR2_BIND_UAV_INTERNAL_UPSCALED) +void StoreInternalColorAndWeight(FfxUInt32x2 iPxPos, FfxFloat32x4 fColorAndWeight) +{ + rw_internal_upscaled_color[iPxPos] = fColorAndWeight; +} +#endif + +#if defined(FSR2_BIND_UAV_UPSCALED_OUTPUT) +void StoreUpscaledOutput(FfxUInt32x2 iPxPos, FfxFloat32x3 fColor) +{ + rw_upscaled_output[UNITY_FSR_POS(iPxPos)] = FfxFloat32x4(fColor, 1.f); +} +#endif + +//LOCK_LIFETIME_REMAINING == 0 +//Should make LockInitialLifetime() return a const 1.0f later +#if defined(FSR2_BIND_SRV_LOCK_STATUS) +FfxFloat32x2 LoadLockStatus(FfxUInt32x2 iPxPos) +{ + return r_lock_status[iPxPos]; +} +#endif + +#if defined(FSR2_BIND_UAV_LOCK_STATUS) +void StoreLockStatus(FfxUInt32x2 iPxPos, FfxFloat32x2 fLockStatus) +{ + rw_lock_status[iPxPos] = fLockStatus; +} +#endif + +#if defined(FSR2_BIND_SRV_LOCK_INPUT_LUMA) +FfxFloat32 LoadLockInputLuma(FfxUInt32x2 iPxPos) +{ + return r_lock_input_luma[iPxPos]; +} +#endif + +#if defined(FSR2_BIND_UAV_LOCK_INPUT_LUMA) +void StoreLockInputLuma(FfxUInt32x2 iPxPos, FfxFloat32 fLuma) +{ + rw_lock_input_luma[iPxPos] = fLuma; +} +#endif + +#if defined(FSR2_BIND_SRV_NEW_LOCKS) +FfxFloat32 LoadNewLocks(FfxUInt32x2 iPxPos) +{ + return r_new_locks[iPxPos]; +} +#endif + +#if defined(FSR2_BIND_UAV_NEW_LOCKS) +FfxFloat32 LoadRwNewLocks(FfxUInt32x2 iPxPos) +{ + return rw_new_locks[iPxPos]; +} +#endif + +#if defined(FSR2_BIND_UAV_NEW_LOCKS) +void StoreNewLocks(FfxUInt32x2 iPxPos, FfxFloat32 newLock) +{ + rw_new_locks[iPxPos] = newLock; +} +#endif + +#if defined(FSR2_BIND_UAV_PREPARED_INPUT_COLOR) +void StorePreparedInputColor(FFX_PARAMETER_IN FfxUInt32x2 iPxPos, FFX_PARAMETER_IN FfxFloat32x4 fTonemapped) +{ + rw_prepared_input_color[iPxPos] = fTonemapped; +} +#endif + +#if defined(FSR2_BIND_SRV_PREPARED_INPUT_COLOR) +FfxFloat32 SampleDepthClip(FfxFloat32x2 fUV) +{ + return r_prepared_input_color.SampleLevel(s_LinearClamp, fUV, 0).w; +} +#endif + +#if defined(FSR2_BIND_SRV_LOCK_STATUS) +FfxFloat32x2 SampleLockStatus(FfxFloat32x2 fUV) +{ + FfxFloat32x2 fLockStatus = r_lock_status.SampleLevel(s_LinearClamp, fUV, 0); + return fLockStatus; +} +#endif + +#if defined(FSR2_BIND_SRV_RECONSTRUCTED_PREV_NEAREST_DEPTH) +FfxFloat32 LoadReconstructedPrevDepth(FfxUInt32x2 iPxPos) +{ + return asfloat(r_reconstructed_previous_nearest_depth[iPxPos]); +} +#endif + +#if defined(FSR2_BIND_UAV_RECONSTRUCTED_PREV_NEAREST_DEPTH) +void StoreReconstructedDepth(FfxUInt32x2 iPxSample, FfxFloat32 fDepth) +{ + FfxUInt32 uDepth = asuint(fDepth); + + #if FFX_FSR2_OPTION_INVERTED_DEPTH + InterlockedMax(rw_reconstructed_previous_nearest_depth[iPxSample], uDepth); + #else + InterlockedMin(rw_reconstructed_previous_nearest_depth[iPxSample], uDepth); // min for standard, max for inverted depth + #endif +} +#endif + +#if defined(FSR2_BIND_UAV_RECONSTRUCTED_PREV_NEAREST_DEPTH) +void SetReconstructedDepth(FfxUInt32x2 iPxSample, const FfxUInt32 uValue) +{ + rw_reconstructed_previous_nearest_depth[iPxSample] = uValue; +} +#endif + +#if defined(FSR2_BIND_UAV_DILATED_DEPTH) +void StoreDilatedDepth(FFX_PARAMETER_IN FfxUInt32x2 iPxPos, FFX_PARAMETER_IN FfxFloat32 fDepth) +{ + rw_dilatedDepth[iPxPos] = fDepth; +} +#endif + +#if defined(FSR2_BIND_UAV_DILATED_MOTION_VECTORS) +void StoreDilatedMotionVector(FFX_PARAMETER_IN FfxUInt32x2 iPxPos, FFX_PARAMETER_IN FfxFloat32x2 fMotionVector) +{ + rw_dilated_motion_vectors[iPxPos] = fMotionVector; +} +#endif + +#if defined(FSR2_BIND_SRV_DILATED_MOTION_VECTORS) +FfxFloat32x2 LoadDilatedMotionVector(FfxUInt32x2 iPxInput) +{ + return r_dilated_motion_vectors[iPxInput].xy; +} +#endif + +#if defined(FSR2_BIND_SRV_PREVIOUS_DILATED_MOTION_VECTORS) +FfxFloat32x2 LoadPreviousDilatedMotionVector(FfxUInt32x2 iPxInput) +{ + return r_previous_dilated_motion_vectors[iPxInput].xy; +} + +FfxFloat32x2 SamplePreviousDilatedMotionVector(FfxFloat32x2 uv) +{ + return r_previous_dilated_motion_vectors.SampleLevel(s_LinearClamp, uv, 0).xy; +} +#endif + +#if defined(FSR2_BIND_SRV_DILATED_DEPTH) +FfxFloat32 LoadDilatedDepth(FfxUInt32x2 iPxInput) +{ + return r_dilatedDepth[iPxInput]; +} +#endif + +#if defined(FSR2_BIND_SRV_INPUT_EXPOSURE) +FfxFloat32 Exposure() +{ + FfxFloat32 exposure = r_input_exposure[FfxUInt32x2(0, 0)].x; + + if (exposure == 0.0f) { + exposure = 1.0f; + } + + return exposure; +} +#endif + +#if defined(FSR2_BIND_SRV_AUTO_EXPOSURE) +FfxFloat32 AutoExposure() +{ + FfxFloat32 exposure = r_auto_exposure[FfxUInt32x2(0, 0)].x; + + if (exposure == 0.0f) { + exposure = 1.0f; + } + + return exposure; +} +#endif + +FfxFloat32 SampleLanczos2Weight(FfxFloat32 x) +{ +#if defined(FSR2_BIND_SRV_LANCZOS_LUT) + return r_lanczos_lut.SampleLevel(s_LinearClamp, FfxFloat32x2(x / 2, 0.5f), 0); +#else + return 0.f; +#endif +} + +#if defined(FSR2_BIND_SRV_UPSCALE_MAXIMUM_BIAS_LUT) +FfxFloat32 SampleUpsampleMaximumBias(FfxFloat32x2 uv) +{ + // Stored as a SNORM, so make sure to multiply by 2 to retrieve the actual expected range. + return FfxFloat32(2.0) * r_upsample_maximum_bias_lut.SampleLevel(s_LinearClamp, abs(uv) * 2.0, 0); +} +#endif + +#if defined(FSR2_BIND_SRV_DILATED_REACTIVE_MASKS) +FfxFloat32x2 SampleDilatedReactiveMasks(FfxFloat32x2 fUV) +{ + return r_dilated_reactive_masks.SampleLevel(s_LinearClamp, fUV, 0); +} +#endif + +#if defined(FSR2_BIND_SRV_DILATED_REACTIVE_MASKS) +FfxFloat32x2 LoadDilatedReactiveMasks(FFX_PARAMETER_IN FfxUInt32x2 iPxPos) +{ + return r_dilated_reactive_masks[iPxPos]; +} +#endif + +#if defined(FSR2_BIND_UAV_DILATED_REACTIVE_MASKS) +void StoreDilatedReactiveMasks(FFX_PARAMETER_IN FfxUInt32x2 iPxPos, FFX_PARAMETER_IN FfxFloat32x2 fDilatedReactiveMasks) +{ + rw_dilated_reactive_masks[iPxPos] = fDilatedReactiveMasks; +} +#endif + +#if defined(FSR2_BIND_SRV_INPUT_OPAQUE_ONLY) +FfxFloat32x3 LoadOpaqueOnly(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos) +{ + return r_input_opaque_only[UNITY_FSR_POS(iPxPos)].xyz; +} +#endif + +#if defined(FSR2_BIND_SRV_PREV_PRE_ALPHA_COLOR) +FfxFloat32x3 LoadPrevPreAlpha(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos) +{ + return r_input_prev_color_pre_alpha[iPxPos]; +} +#endif + +#if defined(FSR2_BIND_SRV_PREV_POST_ALPHA_COLOR) +FfxFloat32x3 LoadPrevPostAlpha(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos) +{ + return r_input_prev_color_post_alpha[iPxPos]; +} +#endif + +#if defined(FSR2_BIND_UAV_AUTOREACTIVE) +#if defined(FSR2_BIND_UAV_AUTOCOMPOSITION) +void StoreAutoReactive(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos, FFX_PARAMETER_IN FFX_MIN16_F2 fReactive) +{ + rw_output_autoreactive[iPxPos] = fReactive.x; + + rw_output_autocomposition[iPxPos] = fReactive.y; +} +#endif +#endif + +#if defined(FSR2_BIND_UAV_PREV_PRE_ALPHA_COLOR) +void StorePrevPreAlpha(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos, FFX_PARAMETER_IN FFX_MIN16_F3 color) +{ + rw_output_prev_color_pre_alpha[iPxPos] = color; + +} +#endif + +#if defined(FSR2_BIND_UAV_PREV_POST_ALPHA_COLOR) +void StorePrevPostAlpha(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos, FFX_PARAMETER_IN FFX_MIN16_F3 color) +{ + rw_output_prev_color_post_alpha[iPxPos] = color; +} +#endif + +FfxFloat32x2 SPD_LoadExposureBuffer() +{ +#if defined FSR2_BIND_UAV_AUTO_EXPOSURE + return rw_auto_exposure[FfxInt32x2(0, 0)]; +#else + return FfxFloat32x2(0.f, 0.f); +#endif // #if defined FSR2_BIND_UAV_AUTO_EXPOSURE +} + +void SPD_SetExposureBuffer(FfxFloat32x2 value) +{ +#if defined FSR2_BIND_UAV_AUTO_EXPOSURE + rw_auto_exposure[FfxInt32x2(0, 0)] = value; +#endif // #if defined FSR2_BIND_UAV_AUTO_EXPOSURE +} + +FfxFloat32x4 SPD_LoadMipmap5(FfxInt32x2 iPxPos) +{ +#if defined FSR2_BIND_UAV_EXPOSURE_MIP_5 + return FfxFloat32x4(rw_img_mip_5[iPxPos], 0, 0, 0); +#else + return FfxFloat32x4(0.f, 0.f, 0.f, 0.f); +#endif // #if defined FSR2_BIND_UAV_EXPOSURE_MIP_5 +} + +void SPD_SetMipmap(FfxInt32x2 iPxPos, FfxUInt32 slice, FfxFloat32 value) +{ + switch (slice) + { + case FFX_FSR2_SHADING_CHANGE_MIP_LEVEL: +#if defined FSR2_BIND_UAV_EXPOSURE_MIP_LUMA_CHANGE + rw_img_mip_shading_change[iPxPos] = value; +#endif // #if defined FSR2_BIND_UAV_EXPOSURE_MIP_LUMA_CHANGE + break; + case 5: +#if defined FSR2_BIND_UAV_EXPOSURE_MIP_5 + rw_img_mip_5[iPxPos] = value; +#endif // #if defined FSR2_BIND_UAV_EXPOSURE_MIP_5 + break; + default: + + // avoid flattened side effect +#if defined(FSR2_BIND_UAV_EXPOSURE_MIP_LUMA_CHANGE) + rw_img_mip_shading_change[iPxPos] = rw_img_mip_shading_change[iPxPos]; +#elif defined(FSR2_BIND_UAV_EXPOSURE_MIP_5) + rw_img_mip_5[iPxPos] = rw_img_mip_5[iPxPos]; +#endif // #if defined FSR2_BIND_UAV_EXPOSURE_MIP_5 + break; + } +} + +void SPD_IncreaseAtomicCounter(inout FfxUInt32 spdCounter) +{ +#if defined FSR2_BIND_UAV_SPD_GLOBAL_ATOMIC + InterlockedAdd(rw_spd_global_atomic[FfxInt32x2(0, 0)], 1, spdCounter); +#endif // #if defined FSR2_BIND_UAV_SPD_GLOBAL_ATOMIC +} + +void SPD_ResetAtomicCounter() +{ +#if defined FSR2_BIND_UAV_SPD_GLOBAL_ATOMIC + rw_spd_global_atomic[FfxInt32x2(0, 0)] = 0; +#endif // #if defined FSR2_BIND_UAV_SPD_GLOBAL_ATOMIC +} + +#endif // #if defined(FFX_GPU) diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_callbacks_hlsl.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_callbacks_hlsl.h.meta new file mode 100644 index 0000000..f044e59 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_callbacks_hlsl.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: ee531e22337dd714aab845c65c30ab94 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_common.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_common.h new file mode 100644 index 0000000..e46b66c --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_common.h @@ -0,0 +1,566 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#if !defined(FFX_FSR2_COMMON_H) +#define FFX_FSR2_COMMON_H + +#if defined(FFX_CPU) || defined(FFX_GPU) +//Locks +#define LOCK_LIFETIME_REMAINING 0 +#define LOCK_TEMPORAL_LUMA 1 +#endif // #if defined(FFX_CPU) || defined(FFX_GPU) + +#if defined(FFX_GPU) +FFX_STATIC const FfxFloat32 FSR2_FP16_MIN = 6.10e-05f; +FFX_STATIC const FfxFloat32 FSR2_FP16_MAX = 65504.0f; +FFX_STATIC const FfxFloat32 FSR2_EPSILON = 1e-03f; +FFX_STATIC const FfxFloat32 FSR2_TONEMAP_EPSILON = 1.0f / FSR2_FP16_MAX; +FFX_STATIC const FfxFloat32 FSR2_FLT_MAX = 3.402823466e+38f; +FFX_STATIC const FfxFloat32 FSR2_FLT_MIN = 1.175494351e-38f; + +// treat vector truncation warnings as errors +#pragma warning(error: 3206) + +// suppress warnings +#pragma warning(disable: 3205) // conversion from larger type to smaller +#pragma warning(disable: 3571) // in ffxPow(f, e), f could be negative + +// Reconstructed depth usage +FFX_STATIC const FfxFloat32 fReconstructedDepthBilinearWeightThreshold = 0.01f; + +// Accumulation +FFX_STATIC const FfxFloat32 fUpsampleLanczosWeightScale = 1.0f / 12.0f; +FFX_STATIC const FfxFloat32 fMaxAccumulationLanczosWeight = 1.0f; +FFX_STATIC const FfxFloat32 fAverageLanczosWeightPerFrame = 0.74f * fUpsampleLanczosWeightScale; // Average lanczos weight for jitter accumulated samples +FFX_STATIC const FfxFloat32 fAccumulationMaxOnMotion = 3.0f * fUpsampleLanczosWeightScale; + +// Auto exposure +FFX_STATIC const FfxFloat32 resetAutoExposureAverageSmoothing = 1e8f; + +struct AccumulationPassCommonParams +{ + FfxInt32x2 iPxHrPos; + FfxFloat32x2 fHrUv; + FfxFloat32x2 fLrUv_HwSampler; + FfxFloat32x2 fMotionVector; + FfxFloat32x2 fReprojectedHrUv; + FfxFloat32 fHrVelocity; + FfxFloat32 fDepthClipFactor; + FfxFloat32 fDilatedReactiveFactor; + FfxFloat32 fAccumulationMask; + + FfxBoolean bIsResetFrame; + FfxBoolean bIsExistingSample; + FfxBoolean bIsNewSample; +}; + +struct LockState +{ + FfxBoolean NewLock; //Set for both unique new and re-locked new + FfxBoolean WasLockedPrevFrame; //Set to identify if the pixel was already locked (relock) +}; + +void InitializeNewLockSample(FFX_PARAMETER_OUT FfxFloat32x2 fLockStatus) +{ + fLockStatus = FfxFloat32x2(0, 0); +} + +#if FFX_HALF +void InitializeNewLockSample(FFX_PARAMETER_OUT FFX_MIN16_F2 fLockStatus) +{ + fLockStatus = FFX_MIN16_F2(0, 0); +} +#endif + + +void KillLock(FFX_PARAMETER_INOUT FfxFloat32x2 fLockStatus) +{ + fLockStatus[LOCK_LIFETIME_REMAINING] = 0; +} + +#if FFX_HALF +void KillLock(FFX_PARAMETER_INOUT FFX_MIN16_F2 fLockStatus) +{ + fLockStatus[LOCK_LIFETIME_REMAINING] = FFX_MIN16_F(0); +} +#endif + +struct RectificationBox +{ + FfxFloat32x3 boxCenter; + FfxFloat32x3 boxVec; + FfxFloat32x3 aabbMin; + FfxFloat32x3 aabbMax; + FfxFloat32 fBoxCenterWeight; +}; +#if FFX_HALF +struct RectificationBoxMin16 +{ + FFX_MIN16_F3 boxCenter; + FFX_MIN16_F3 boxVec; + FFX_MIN16_F3 aabbMin; + FFX_MIN16_F3 aabbMax; + FFX_MIN16_F fBoxCenterWeight; +}; +#endif + +void RectificationBoxReset(FFX_PARAMETER_INOUT RectificationBox rectificationBox) +{ + rectificationBox.fBoxCenterWeight = FfxFloat32(0); + + rectificationBox.boxCenter = FfxFloat32x3(0, 0, 0); + rectificationBox.boxVec = FfxFloat32x3(0, 0, 0); + rectificationBox.aabbMin = FfxFloat32x3(FSR2_FLT_MAX, FSR2_FLT_MAX, FSR2_FLT_MAX); + rectificationBox.aabbMax = -FfxFloat32x3(FSR2_FLT_MAX, FSR2_FLT_MAX, FSR2_FLT_MAX); +} +#if FFX_HALF +void RectificationBoxReset(FFX_PARAMETER_INOUT RectificationBoxMin16 rectificationBox) +{ + rectificationBox.fBoxCenterWeight = FFX_MIN16_F(0); + + rectificationBox.boxCenter = FFX_MIN16_F3(0, 0, 0); + rectificationBox.boxVec = FFX_MIN16_F3(0, 0, 0); + rectificationBox.aabbMin = FFX_MIN16_F3(FSR2_FP16_MAX, FSR2_FP16_MAX, FSR2_FP16_MAX); + rectificationBox.aabbMax = -FFX_MIN16_F3(FSR2_FP16_MAX, FSR2_FP16_MAX, FSR2_FP16_MAX); +} +#endif + +void RectificationBoxAddInitialSample(FFX_PARAMETER_INOUT RectificationBox rectificationBox, const FfxFloat32x3 colorSample, const FfxFloat32 fSampleWeight) +{ + rectificationBox.aabbMin = colorSample; + rectificationBox.aabbMax = colorSample; + + FfxFloat32x3 weightedSample = colorSample * fSampleWeight; + rectificationBox.boxCenter = weightedSample; + rectificationBox.boxVec = colorSample * weightedSample; + rectificationBox.fBoxCenterWeight = fSampleWeight; +} + +void RectificationBoxAddSample(FfxBoolean bInitialSample, FFX_PARAMETER_INOUT RectificationBox rectificationBox, const FfxFloat32x3 colorSample, const FfxFloat32 fSampleWeight) +{ + if (bInitialSample) { + RectificationBoxAddInitialSample(rectificationBox, colorSample, fSampleWeight); + } else { + rectificationBox.aabbMin = ffxMin(rectificationBox.aabbMin, colorSample); + rectificationBox.aabbMax = ffxMax(rectificationBox.aabbMax, colorSample); + + FfxFloat32x3 weightedSample = colorSample * fSampleWeight; + rectificationBox.boxCenter += weightedSample; + rectificationBox.boxVec += colorSample * weightedSample; + rectificationBox.fBoxCenterWeight += fSampleWeight; + } +} +#if FFX_HALF +void RectificationBoxAddInitialSample(FFX_PARAMETER_INOUT RectificationBoxMin16 rectificationBox, const FFX_MIN16_F3 colorSample, const FFX_MIN16_F fSampleWeight) +{ + rectificationBox.aabbMin = colorSample; + rectificationBox.aabbMax = colorSample; + + FFX_MIN16_F3 weightedSample = colorSample * fSampleWeight; + rectificationBox.boxCenter = weightedSample; + rectificationBox.boxVec = colorSample * weightedSample; + rectificationBox.fBoxCenterWeight = fSampleWeight; +} + +void RectificationBoxAddSample(FfxBoolean bInitialSample, FFX_PARAMETER_INOUT RectificationBoxMin16 rectificationBox, const FFX_MIN16_F3 colorSample, const FFX_MIN16_F fSampleWeight) +{ + if (bInitialSample) { + RectificationBoxAddInitialSample(rectificationBox, colorSample, fSampleWeight); + } else { + rectificationBox.aabbMin = ffxMin(rectificationBox.aabbMin, colorSample); + rectificationBox.aabbMax = ffxMax(rectificationBox.aabbMax, colorSample); + + FFX_MIN16_F3 weightedSample = colorSample * fSampleWeight; + rectificationBox.boxCenter += weightedSample; + rectificationBox.boxVec += colorSample * weightedSample; + rectificationBox.fBoxCenterWeight += fSampleWeight; + } +} +#endif + +void RectificationBoxComputeVarianceBoxData(FFX_PARAMETER_INOUT RectificationBox rectificationBox) +{ + rectificationBox.fBoxCenterWeight = (abs(rectificationBox.fBoxCenterWeight) > FfxFloat32(FSR2_EPSILON) ? rectificationBox.fBoxCenterWeight : FfxFloat32(1.f)); + rectificationBox.boxCenter /= rectificationBox.fBoxCenterWeight; + rectificationBox.boxVec /= rectificationBox.fBoxCenterWeight; + FfxFloat32x3 stdDev = sqrt(abs(rectificationBox.boxVec - rectificationBox.boxCenter * rectificationBox.boxCenter)); + rectificationBox.boxVec = stdDev; +} +#if FFX_HALF +void RectificationBoxComputeVarianceBoxData(FFX_PARAMETER_INOUT RectificationBoxMin16 rectificationBox) +{ + rectificationBox.fBoxCenterWeight = (abs(rectificationBox.fBoxCenterWeight) > FFX_MIN16_F(FSR2_EPSILON) ? rectificationBox.fBoxCenterWeight : FFX_MIN16_F(1.f)); + rectificationBox.boxCenter /= rectificationBox.fBoxCenterWeight; + rectificationBox.boxVec /= rectificationBox.fBoxCenterWeight; + FFX_MIN16_F3 stdDev = sqrt(abs(rectificationBox.boxVec - rectificationBox.boxCenter * rectificationBox.boxCenter)); + rectificationBox.boxVec = stdDev; +} +#endif + +FfxFloat32x3 SafeRcp3(FfxFloat32x3 v) +{ + return (all(FFX_NOT_EQUAL(v, FfxFloat32x3(0, 0, 0)))) ? (FfxFloat32x3(1, 1, 1) / v) : FfxFloat32x3(0, 0, 0); +} +#if FFX_HALF +FFX_MIN16_F3 SafeRcp3(FFX_MIN16_F3 v) +{ + return (all(FFX_NOT_EQUAL(v, FFX_MIN16_F3(0, 0, 0)))) ? (FFX_MIN16_F3(1, 1, 1) / v) : FFX_MIN16_F3(0, 0, 0); +} +#endif + +FfxFloat32 MinDividedByMax(const FfxFloat32 v0, const FfxFloat32 v1) +{ + const FfxFloat32 m = ffxMax(v0, v1); + return m != 0 ? ffxMin(v0, v1) / m : 0; +} + +#if FFX_HALF +FFX_MIN16_F MinDividedByMax(const FFX_MIN16_F v0, const FFX_MIN16_F v1) +{ + const FFX_MIN16_F m = ffxMax(v0, v1); + return m != FFX_MIN16_F(0) ? ffxMin(v0, v1) / m : FFX_MIN16_F(0); +} +#endif + +FfxFloat32x3 YCoCgToRGB(FfxFloat32x3 fYCoCg) +{ + FfxFloat32x3 fRgb; + + fRgb = FfxFloat32x3( + fYCoCg.x + fYCoCg.y - fYCoCg.z, + fYCoCg.x + fYCoCg.z, + fYCoCg.x - fYCoCg.y - fYCoCg.z); + + return fRgb; +} +#if FFX_HALF +FFX_MIN16_F3 YCoCgToRGB(FFX_MIN16_F3 fYCoCg) +{ + FFX_MIN16_F3 fRgb; + + fRgb = FFX_MIN16_F3( + fYCoCg.x + fYCoCg.y - fYCoCg.z, + fYCoCg.x + fYCoCg.z, + fYCoCg.x - fYCoCg.y - fYCoCg.z); + + return fRgb; +} +#endif + +FfxFloat32x3 RGBToYCoCg(FfxFloat32x3 fRgb) +{ + FfxFloat32x3 fYCoCg; + + fYCoCg = FfxFloat32x3( + 0.25f * fRgb.r + 0.5f * fRgb.g + 0.25f * fRgb.b, + 0.5f * fRgb.r - 0.5f * fRgb.b, + -0.25f * fRgb.r + 0.5f * fRgb.g - 0.25f * fRgb.b); + + return fYCoCg; +} +#if FFX_HALF +FFX_MIN16_F3 RGBToYCoCg(FFX_MIN16_F3 fRgb) +{ + FFX_MIN16_F3 fYCoCg; + + fYCoCg = FFX_MIN16_F3( + 0.25 * fRgb.r + 0.5 * fRgb.g + 0.25 * fRgb.b, + 0.5 * fRgb.r - 0.5 * fRgb.b, + -0.25 * fRgb.r + 0.5 * fRgb.g - 0.25 * fRgb.b); + + return fYCoCg; +} +#endif + +FfxFloat32 RGBToLuma(FfxFloat32x3 fLinearRgb) +{ + return dot(fLinearRgb, FfxFloat32x3(0.2126f, 0.7152f, 0.0722f)); +} +#if FFX_HALF +FFX_MIN16_F RGBToLuma(FFX_MIN16_F3 fLinearRgb) +{ + return dot(fLinearRgb, FFX_MIN16_F3(0.2126f, 0.7152f, 0.0722f)); +} +#endif + +FfxFloat32 RGBToPerceivedLuma(FfxFloat32x3 fLinearRgb) +{ + FfxFloat32 fLuminance = RGBToLuma(fLinearRgb); + + FfxFloat32 fPercievedLuminance = 0; + if (fLuminance <= 216.0f / 24389.0f) { + fPercievedLuminance = fLuminance * (24389.0f / 27.0f); + } + else { + fPercievedLuminance = ffxPow(fLuminance, 1.0f / 3.0f) * 116.0f - 16.0f; + } + + return fPercievedLuminance * 0.01f; +} +#if FFX_HALF +FFX_MIN16_F RGBToPerceivedLuma(FFX_MIN16_F3 fLinearRgb) +{ + FFX_MIN16_F fLuminance = RGBToLuma(fLinearRgb); + + FFX_MIN16_F fPercievedLuminance = FFX_MIN16_F(0); + if (fLuminance <= FFX_MIN16_F(216.0f / 24389.0f)) { + fPercievedLuminance = fLuminance * FFX_MIN16_F(24389.0f / 27.0f); + } + else { + fPercievedLuminance = ffxPow(fLuminance, FFX_MIN16_F(1.0f / 3.0f)) * FFX_MIN16_F(116.0f) - FFX_MIN16_F(16.0f); + } + + return fPercievedLuminance * FFX_MIN16_F(0.01f); +} +#endif + +FfxFloat32x3 Tonemap(FfxFloat32x3 fRgb) +{ + return fRgb / (ffxMax(ffxMax(0.f, fRgb.r), ffxMax(fRgb.g, fRgb.b)) + 1.f).xxx; +} + +FfxFloat32x3 InverseTonemap(FfxFloat32x3 fRgb) +{ + return fRgb / ffxMax(FSR2_TONEMAP_EPSILON, 1.f - ffxMax(fRgb.r, ffxMax(fRgb.g, fRgb.b))).xxx; +} + +#if FFX_HALF +FFX_MIN16_F3 Tonemap(FFX_MIN16_F3 fRgb) +{ + return fRgb / (ffxMax(ffxMax(FFX_MIN16_F(0.f), fRgb.r), ffxMax(fRgb.g, fRgb.b)) + FFX_MIN16_F(1.f)).xxx; +} + +FFX_MIN16_F3 InverseTonemap(FFX_MIN16_F3 fRgb) +{ + return fRgb / ffxMax(FFX_MIN16_F(FSR2_TONEMAP_EPSILON), FFX_MIN16_F(1.f) - ffxMax(fRgb.r, ffxMax(fRgb.g, fRgb.b))).xxx; +} +#endif + +FfxInt32x2 ClampLoad(FfxInt32x2 iPxSample, FfxInt32x2 iPxOffset, FfxInt32x2 iTextureSize) +{ + FfxInt32x2 result = iPxSample + iPxOffset; + result.x = (iPxOffset.x < 0) ? ffxMax(result.x, 0) : result.x; + result.x = (iPxOffset.x > 0) ? ffxMin(result.x, iTextureSize.x - 1) : result.x; + result.y = (iPxOffset.y < 0) ? ffxMax(result.y, 0) : result.y; + result.y = (iPxOffset.y > 0) ? ffxMin(result.y, iTextureSize.y - 1) : result.y; + return result; + + // return ffxMed3(iPxSample + iPxOffset, FfxInt32x2(0, 0), iTextureSize - FfxInt32x2(1, 1)); +} +#if FFX_HALF +FFX_MIN16_I2 ClampLoad(FFX_MIN16_I2 iPxSample, FFX_MIN16_I2 iPxOffset, FFX_MIN16_I2 iTextureSize) +{ + FFX_MIN16_I2 result = iPxSample + iPxOffset; + result.x = (iPxOffset.x < 0) ? ffxMax(result.x, FFX_MIN16_I(0)) : result.x; + result.x = (iPxOffset.x > 0) ? ffxMin(result.x, iTextureSize.x - FFX_MIN16_I(1)) : result.x; + result.y = (iPxOffset.y < 0) ? ffxMax(result.y, FFX_MIN16_I(0)) : result.y; + result.y = (iPxOffset.y > 0) ? ffxMin(result.y, iTextureSize.y - FFX_MIN16_I(1)) : result.y; + return result; + + // return ffxMed3Half(iPxSample + iPxOffset, FFX_MIN16_I2(0, 0), iTextureSize - FFX_MIN16_I2(1, 1)); +} +#endif + +FfxFloat32x2 ClampUv(FfxFloat32x2 fUv, FfxInt32x2 iTextureSize, FfxInt32x2 iResourceSize) +{ + const FfxFloat32x2 fSampleLocation = fUv * iTextureSize; + const FfxFloat32x2 fClampedLocation = ffxMax(FfxFloat32x2(0.5f, 0.5f), ffxMin(fSampleLocation, FfxFloat32x2(iTextureSize) - FfxFloat32x2(0.5f, 0.5f))); + const FfxFloat32x2 fClampedUv = fClampedLocation / FfxFloat32x2(iResourceSize); + + return fClampedUv; +} + +FfxBoolean IsOnScreen(FfxInt32x2 pos, FfxInt32x2 size) +{ + return all(FFX_LESS_THAN(FfxUInt32x2(pos), FfxUInt32x2(size))); +} +#if FFX_HALF +FfxBoolean IsOnScreen(FFX_MIN16_I2 pos, FFX_MIN16_I2 size) +{ + return all(FFX_LESS_THAN(FFX_MIN16_U2(pos), FFX_MIN16_U2(size))); +} +#endif + +FfxFloat32 ComputeAutoExposureFromLavg(FfxFloat32 Lavg) +{ + Lavg = exp(Lavg); + + const FfxFloat32 S = 100.0f; //ISO arithmetic speed + const FfxFloat32 K = 12.5f; + FfxFloat32 ExposureISO100 = log2((Lavg * S) / K); + + const FfxFloat32 q = 0.65f; + FfxFloat32 Lmax = (78.0f / (q * S)) * ffxPow(2.0f, ExposureISO100); + + return 1 / Lmax; +} +#if FFX_HALF +FFX_MIN16_F ComputeAutoExposureFromLavg(FFX_MIN16_F Lavg) +{ + Lavg = exp(Lavg); + + const FFX_MIN16_F S = FFX_MIN16_F(100.0f); //ISO arithmetic speed + const FFX_MIN16_F K = FFX_MIN16_F(12.5f); + const FFX_MIN16_F ExposureISO100 = log2((Lavg * S) / K); + + const FFX_MIN16_F q = FFX_MIN16_F(0.65f); + const FFX_MIN16_F Lmax = (FFX_MIN16_F(78.0f) / (q * S)) * ffxPow(FFX_MIN16_F(2.0f), ExposureISO100); + + return FFX_MIN16_F(1) / Lmax; +} +#endif + +FfxInt32x2 ComputeHrPosFromLrPos(FfxInt32x2 iPxLrPos) +{ + FfxFloat32x2 fSrcJitteredPos = FfxFloat32x2(iPxLrPos) + 0.5f - Jitter(); + FfxFloat32x2 fLrPosInHr = (fSrcJitteredPos / RenderSize()) * DisplaySize(); + FfxInt32x2 iPxHrPos = FfxInt32x2(floor(fLrPosInHr)); + return iPxHrPos; +} +#if FFX_HALF +FFX_MIN16_I2 ComputeHrPosFromLrPos(FFX_MIN16_I2 iPxLrPos) +{ + FFX_MIN16_F2 fSrcJitteredPos = FFX_MIN16_F2(iPxLrPos) + FFX_MIN16_F(0.5f) - FFX_MIN16_F2(Jitter()); + FFX_MIN16_F2 fLrPosInHr = (fSrcJitteredPos / FFX_MIN16_F2(RenderSize())) * FFX_MIN16_F2(DisplaySize()); + FFX_MIN16_I2 iPxHrPos = FFX_MIN16_I2(floor(fLrPosInHr)); + return iPxHrPos; +} +#endif + +FfxFloat32x2 ComputeNdc(FfxFloat32x2 fPxPos, FfxInt32x2 iSize) +{ + return fPxPos / FfxFloat32x2(iSize) * FfxFloat32x2(2.0f, -2.0f) + FfxFloat32x2(-1.0f, 1.0f); +} + +FfxFloat32 GetViewSpaceDepth(FfxFloat32 fDeviceDepth) +{ + const FfxFloat32x4 fDeviceToViewDepth = DeviceToViewSpaceTransformFactors(); + + // fDeviceToViewDepth details found in ffx_fsr2.cpp + return (fDeviceToViewDepth[1] / (fDeviceDepth - fDeviceToViewDepth[0])); +} + +FfxFloat32 GetViewSpaceDepthInMeters(FfxFloat32 fDeviceDepth) +{ + return GetViewSpaceDepth(fDeviceDepth) * ViewSpaceToMetersFactor(); +} + +FfxFloat32x3 GetViewSpacePosition(FfxInt32x2 iViewportPos, FfxInt32x2 iViewportSize, FfxFloat32 fDeviceDepth) +{ + const FfxFloat32x4 fDeviceToViewDepth = DeviceToViewSpaceTransformFactors(); + + const FfxFloat32 Z = GetViewSpaceDepth(fDeviceDepth); + + const FfxFloat32x2 fNdcPos = ComputeNdc(iViewportPos, iViewportSize); + const FfxFloat32 X = fDeviceToViewDepth[2] * fNdcPos.x * Z; + const FfxFloat32 Y = fDeviceToViewDepth[3] * fNdcPos.y * Z; + + return FfxFloat32x3(X, Y, Z); +} + +FfxFloat32x3 GetViewSpacePositionInMeters(FfxInt32x2 iViewportPos, FfxInt32x2 iViewportSize, FfxFloat32 fDeviceDepth) +{ + return GetViewSpacePosition(iViewportPos, iViewportSize, fDeviceDepth) * ViewSpaceToMetersFactor(); +} + +FfxFloat32 GetMaxDistanceInMeters() +{ +#if FFX_FSR2_OPTION_INVERTED_DEPTH + return GetViewSpaceDepth(0.0f) * ViewSpaceToMetersFactor(); +#else + return GetViewSpaceDepth(1.0f) * ViewSpaceToMetersFactor(); +#endif +} + +FfxFloat32x3 PrepareRgb(FfxFloat32x3 fRgb, FfxFloat32 fExposure, FfxFloat32 fPreExposure) +{ + fRgb /= fPreExposure; + fRgb *= fExposure; + + fRgb = clamp(fRgb, 0.0f, FSR2_FP16_MAX); + + return fRgb; +} + +FfxFloat32x3 UnprepareRgb(FfxFloat32x3 fRgb, FfxFloat32 fExposure) +{ + fRgb /= fExposure; + fRgb *= PreExposure(); + + return fRgb; +} + + +struct BilinearSamplingData +{ + FfxInt32x2 iOffsets[4]; + FfxFloat32 fWeights[4]; + FfxInt32x2 iBasePos; +}; + +BilinearSamplingData GetBilinearSamplingData(FfxFloat32x2 fUv, FfxInt32x2 iSize) +{ + BilinearSamplingData data; + + FfxFloat32x2 fPxSample = (fUv * iSize) - FfxFloat32x2(0.5f, 0.5f); + data.iBasePos = FfxInt32x2(floor(fPxSample)); + FfxFloat32x2 fPxFrac = ffxFract(fPxSample); + + data.iOffsets[0] = FfxInt32x2(0, 0); + data.iOffsets[1] = FfxInt32x2(1, 0); + data.iOffsets[2] = FfxInt32x2(0, 1); + data.iOffsets[3] = FfxInt32x2(1, 1); + + data.fWeights[0] = (1 - fPxFrac.x) * (1 - fPxFrac.y); + data.fWeights[1] = (fPxFrac.x) * (1 - fPxFrac.y); + data.fWeights[2] = (1 - fPxFrac.x) * (fPxFrac.y); + data.fWeights[3] = (fPxFrac.x) * (fPxFrac.y); + + return data; +} + +struct PlaneData +{ + FfxFloat32x3 fNormal; + FfxFloat32 fDistanceFromOrigin; +}; + +PlaneData GetPlaneFromPoints(FfxFloat32x3 fP0, FfxFloat32x3 fP1, FfxFloat32x3 fP2) +{ + PlaneData plane; + + FfxFloat32x3 v0 = fP0 - fP1; + FfxFloat32x3 v1 = fP0 - fP2; + plane.fNormal = normalize(cross(v0, v1)); + plane.fDistanceFromOrigin = -dot(fP0, plane.fNormal); + + return plane; +} + +FfxFloat32 PointToPlaneDistance(PlaneData plane, FfxFloat32x3 fPoint) +{ + return abs(dot(plane.fNormal, fPoint) + plane.fDistanceFromOrigin); +} + +#endif // #if defined(FFX_GPU) + +#endif //!defined(FFX_FSR2_COMMON_H) diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_common.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_common.h.meta new file mode 100644 index 0000000..1357ddf --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_common.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 261b5f3ac23701a4aa6d399d43798324 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_compute_luminance_pyramid.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_compute_luminance_pyramid.h new file mode 100644 index 0000000..d8d4820 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_compute_luminance_pyramid.h @@ -0,0 +1,176 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +FFX_GROUPSHARED FfxUInt32 spdCounter; + +void SpdIncreaseAtomicCounter(FfxUInt32 slice) +{ + SPD_IncreaseAtomicCounter(spdCounter); +} + +FfxUInt32 SpdGetAtomicCounter() +{ + return spdCounter; +} + +void SpdResetAtomicCounter(FfxUInt32 slice) +{ + SPD_ResetAtomicCounter(); +} + +#ifndef SPD_PACKED_ONLY +FFX_GROUPSHARED FfxFloat32 spdIntermediateR[16][16]; +FFX_GROUPSHARED FfxFloat32 spdIntermediateG[16][16]; +FFX_GROUPSHARED FfxFloat32 spdIntermediateB[16][16]; +FFX_GROUPSHARED FfxFloat32 spdIntermediateA[16][16]; + +FfxFloat32x4 SpdLoadSourceImage(FfxFloat32x2 tex, FfxUInt32 slice) +{ + FfxFloat32x2 fUv = (tex + 0.5f + Jitter()) / RenderSize(); + fUv = ClampUv(fUv, RenderSize(), InputColorResourceDimensions()); + FfxFloat32x3 fRgb = SampleInputColor(fUv); + + fRgb /= PreExposure(); + + //compute log luma + const FfxFloat32 fLogLuma = log(ffxMax(FSR2_EPSILON, RGBToLuma(fRgb))); + + // Make sure out of screen pixels contribute no value to the end result + const FfxFloat32 result = all(FFX_LESS_THAN(tex, RenderSize())) ? fLogLuma : 0.0f; + + return FfxFloat32x4(result, 0, 0, 0); +} + +FfxFloat32x4 SpdLoad(FfxInt32x2 tex, FfxUInt32 slice) +{ + return SPD_LoadMipmap5(tex); +} + +void SpdStore(FfxInt32x2 pix, FfxFloat32x4 outValue, FfxUInt32 index, FfxUInt32 slice) +{ + if (index == LumaMipLevelToUse() || index == 5) + { + SPD_SetMipmap(pix, index, outValue.r); + } + + if (index == MipCount() - 1) { //accumulate on 1x1 level + + if (all(FFX_EQUAL(pix, FfxInt32x2(0, 0)))) + { + FfxFloat32 prev = SPD_LoadExposureBuffer().y; + FfxFloat32 result = outValue.r; + + if (prev < resetAutoExposureAverageSmoothing) // Compare Lavg, so small or negative values + { + FfxFloat32 rate = 1.0f; + result = prev + (result - prev) * (1 - exp(-DeltaTime() * rate)); + } + FfxFloat32x2 spdOutput = FfxFloat32x2(ComputeAutoExposureFromLavg(result), result); + SPD_SetExposureBuffer(spdOutput); + } + } +} + +FfxFloat32x4 SpdLoadIntermediate(FfxUInt32 x, FfxUInt32 y) +{ + return FfxFloat32x4( + spdIntermediateR[x][y], + spdIntermediateG[x][y], + spdIntermediateB[x][y], + spdIntermediateA[x][y]); +} +void SpdStoreIntermediate(FfxUInt32 x, FfxUInt32 y, FfxFloat32x4 value) +{ + spdIntermediateR[x][y] = value.x; + spdIntermediateG[x][y] = value.y; + spdIntermediateB[x][y] = value.z; + spdIntermediateA[x][y] = value.w; +} +FfxFloat32x4 SpdReduce4(FfxFloat32x4 v0, FfxFloat32x4 v1, FfxFloat32x4 v2, FfxFloat32x4 v3) +{ + return (v0 + v1 + v2 + v3) * 0.25f; +} +#endif + +// define fetch and store functions Packed +#if FFX_HALF + +FFX_GROUPSHARED FfxFloat16x2 spdIntermediateRG[16][16]; +FFX_GROUPSHARED FfxFloat16x2 spdIntermediateBA[16][16]; + +FfxFloat16x4 SpdLoadSourceImageH(FfxFloat32x2 tex, FfxUInt32 slice) +{ + return FfxFloat16x4(0, 0, 0, 0); +} + +FfxFloat16x4 SpdLoadH(FfxInt32x2 p, FfxUInt32 slice) +{ + return FfxFloat16x4(0, 0, 0, 0); +} + +void SpdStoreH(FfxInt32x2 p, FfxFloat16x4 value, FfxUInt32 mip, FfxUInt32 slice) +{ +} + +FfxFloat16x4 SpdLoadIntermediateH(FfxUInt32 x, FfxUInt32 y) +{ + return FfxFloat16x4( + spdIntermediateRG[x][y].x, + spdIntermediateRG[x][y].y, + spdIntermediateBA[x][y].x, + spdIntermediateBA[x][y].y); +} + +void SpdStoreIntermediateH(FfxUInt32 x, FfxUInt32 y, FfxFloat16x4 value) +{ + spdIntermediateRG[x][y] = value.xy; + spdIntermediateBA[x][y] = value.zw; +} + +FfxFloat16x4 SpdReduce4H(FfxFloat16x4 v0, FfxFloat16x4 v1, FfxFloat16x4 v2, FfxFloat16x4 v3) +{ + return (v0 + v1 + v2 + v3) * FfxFloat16(0.25); +} +#endif + +#include "../spd/ffx_spd.h" + +void ComputeAutoExposure(FfxUInt32x3 WorkGroupId, FfxUInt32 LocalThreadIndex) +{ +#if FFX_HALF + SpdDownsampleH( + FfxUInt32x2(WorkGroupId.xy), + FfxUInt32(LocalThreadIndex), + FfxUInt32(MipCount()), + FfxUInt32(NumWorkGroups()), + FfxUInt32(WorkGroupId.z), + FfxUInt32x2(WorkGroupOffset())); +#else + SpdDownsample( + FfxUInt32x2(WorkGroupId.xy), + FfxUInt32(LocalThreadIndex), + FfxUInt32(MipCount()), + FfxUInt32(NumWorkGroups()), + FfxUInt32(WorkGroupId.z), + FfxUInt32x2(WorkGroupOffset())); +#endif +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_compute_luminance_pyramid.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_compute_luminance_pyramid.h.meta new file mode 100644 index 0000000..1ea247c --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_compute_luminance_pyramid.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 86bf712594bf26449924ac6bb393e498 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_depth_clip.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_depth_clip.h new file mode 100644 index 0000000..873ff4b --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_depth_clip.h @@ -0,0 +1,259 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#ifndef FFX_FSR2_DEPTH_CLIP_H +#define FFX_FSR2_DEPTH_CLIP_H + +FFX_STATIC const FfxFloat32 DepthClipBaseScale = 4.0f; + +FfxFloat32 ComputeDepthClip(FfxFloat32x2 fUvSample, FfxFloat32 fCurrentDepthSample) +{ + FfxFloat32 fCurrentDepthViewSpace = GetViewSpaceDepth(fCurrentDepthSample); + BilinearSamplingData bilinearInfo = GetBilinearSamplingData(fUvSample, RenderSize()); + + FfxFloat32 fDilatedSum = 0.0f; + FfxFloat32 fDepth = 0.0f; + FfxFloat32 fWeightSum = 0.0f; + for (FfxInt32 iSampleIndex = 0; iSampleIndex < 4; iSampleIndex++) { + + const FfxInt32x2 iOffset = bilinearInfo.iOffsets[iSampleIndex]; + const FfxInt32x2 iSamplePos = bilinearInfo.iBasePos + iOffset; + + if (IsOnScreen(iSamplePos, RenderSize())) { + const FfxFloat32 fWeight = bilinearInfo.fWeights[iSampleIndex]; + if (fWeight > fReconstructedDepthBilinearWeightThreshold) { + + const FfxFloat32 fPrevDepthSample = LoadReconstructedPrevDepth(iSamplePos); + const FfxFloat32 fPrevNearestDepthViewSpace = GetViewSpaceDepth(fPrevDepthSample); + + const FfxFloat32 fDepthDiff = fCurrentDepthViewSpace - fPrevNearestDepthViewSpace; + + if (fDepthDiff > 0.0f) { + +#if FFX_FSR2_OPTION_INVERTED_DEPTH + const FfxFloat32 fPlaneDepth = ffxMin(fPrevDepthSample, fCurrentDepthSample); +#else + const FfxFloat32 fPlaneDepth = ffxMax(fPrevDepthSample, fCurrentDepthSample); +#endif + + const FfxFloat32x3 fCenter = GetViewSpacePosition(FfxInt32x2(RenderSize() * 0.5f), RenderSize(), fPlaneDepth); + const FfxFloat32x3 fCorner = GetViewSpacePosition(FfxInt32x2(0, 0), RenderSize(), fPlaneDepth); + + const FfxFloat32 fHalfViewportWidth = length(FfxFloat32x2(RenderSize())); + const FfxFloat32 fDepthThreshold = ffxMax(fCurrentDepthViewSpace, fPrevNearestDepthViewSpace); + + const FfxFloat32 Ksep = 1.37e-05f; + const FfxFloat32 Kfov = length(fCorner) / length(fCenter); + const FfxFloat32 fRequiredDepthSeparation = Ksep * Kfov * fHalfViewportWidth * fDepthThreshold; + + const FfxFloat32 fResolutionFactor = ffxSaturate(length(FfxFloat32x2(RenderSize())) / length(FfxFloat32x2(1920.0f, 1080.0f))); + const FfxFloat32 fPower = ffxLerp(1.0f, 3.0f, fResolutionFactor); + fDepth += ffxPow(ffxSaturate(FfxFloat32(fRequiredDepthSeparation / fDepthDiff)), fPower) * fWeight; + fWeightSum += fWeight; + } + } + } + } + + return (fWeightSum > 0) ? ffxSaturate(1.0f - fDepth / fWeightSum) : 0.0f; +} + +FfxFloat32 ComputeMotionDivergence(FfxInt32x2 iPxPos, FfxInt32x2 iPxInputMotionVectorSize) +{ + FfxFloat32 minconvergence = 1.0f; + + FfxFloat32x2 fMotionVectorNucleus = LoadInputMotionVector(iPxPos); + FfxFloat32 fNucleusVelocityLr = length(fMotionVectorNucleus * RenderSize()); + FfxFloat32 fMaxVelocityUv = length(fMotionVectorNucleus); + + const FfxFloat32 MotionVectorVelocityEpsilon = 1e-02f; + + if (fNucleusVelocityLr > MotionVectorVelocityEpsilon) { + for (FfxInt32 y = -1; y <= 1; ++y) { + for (FfxInt32 x = -1; x <= 1; ++x) { + + FfxInt32x2 sp = ClampLoad(iPxPos, FfxInt32x2(x, y), iPxInputMotionVectorSize); + + FfxFloat32x2 fMotionVector = LoadInputMotionVector(sp); + FfxFloat32 fVelocityUv = length(fMotionVector); + + fMaxVelocityUv = ffxMax(fVelocityUv, fMaxVelocityUv); + fVelocityUv = ffxMax(fVelocityUv, fMaxVelocityUv); + minconvergence = ffxMin(minconvergence, dot(fMotionVector / fVelocityUv, fMotionVectorNucleus / fVelocityUv)); + } + } + } + + return ffxSaturate(1.0f - minconvergence) * ffxSaturate(fMaxVelocityUv / 0.01f); +} + +FfxFloat32 ComputeDepthDivergence(FfxInt32x2 iPxPos) +{ + const FfxFloat32 fMaxDistInMeters = GetMaxDistanceInMeters(); + FfxFloat32 fDepthMax = 0.0f; + FfxFloat32 fDepthMin = fMaxDistInMeters; + + FfxInt32 iMaxDistFound = 0; + + for (FfxInt32 y = -1; y < 2; y++) { + for (FfxInt32 x = -1; x < 2; x++) { + + const FfxInt32x2 iOffset = FfxInt32x2(x, y); + const FfxInt32x2 iSamplePos = iPxPos + iOffset; + + const FfxFloat32 fOnScreenFactor = IsOnScreen(iSamplePos, RenderSize()) ? 1.0f : 0.0f; + FfxFloat32 fDepth = GetViewSpaceDepthInMeters(LoadDilatedDepth(iSamplePos)) * fOnScreenFactor; + + iMaxDistFound |= FfxInt32(fMaxDistInMeters == fDepth); + + fDepthMin = ffxMin(fDepthMin, fDepth); + fDepthMax = ffxMax(fDepthMax, fDepth); + } + } + + return (1.0f - fDepthMin / fDepthMax) * (FfxBoolean(iMaxDistFound) ? 0.0f : 1.0f); +} + +FfxFloat32 ComputeTemporalMotionDivergence(FfxInt32x2 iPxPos) +{ + const FfxFloat32x2 fUv = FfxFloat32x2(iPxPos + 0.5f) / RenderSize(); + + FfxFloat32x2 fMotionVector = LoadDilatedMotionVector(iPxPos); + FfxFloat32x2 fReprojectedUv = fUv + fMotionVector; + fReprojectedUv = ClampUv(fReprojectedUv, RenderSize(), MaxRenderSize()); + FfxFloat32x2 fPrevMotionVector = SamplePreviousDilatedMotionVector(fReprojectedUv); + + float fPxDistance = length(fMotionVector * DisplaySize()); + return fPxDistance > 1.0f ? ffxLerp(0.0f, 1.0f - ffxSaturate(length(fPrevMotionVector) / length(fMotionVector)), ffxSaturate(ffxPow(fPxDistance / 20.0f, 3.0f))) : 0; +} + +void PreProcessReactiveMasks(FfxInt32x2 iPxLrPos, FfxFloat32 fMotionDivergence) +{ + // Compensate for bilinear sampling in accumulation pass + + FfxFloat32x3 fReferenceColor = LoadInputColor(iPxLrPos).xyz; + FfxFloat32x2 fReactiveFactor = FfxFloat32x2(0.0f, fMotionDivergence); + + float fMasksSum = 0.0f; + + FfxFloat32x3 fColorSamples[9]; + FfxFloat32 fReactiveSamples[9]; + FfxFloat32 fTransparencyAndCompositionSamples[9]; + + FFX_UNROLL + for (FfxInt32 y = -1; y < 2; y++) { + FFX_UNROLL + for (FfxInt32 x = -1; x < 2; x++) { + + const FfxInt32x2 sampleCoord = ClampLoad(iPxLrPos, FfxInt32x2(x, y), FfxInt32x2(RenderSize())); + + FfxInt32 sampleIdx = (y + 1) * 3 + x + 1; + + FfxFloat32x3 fColorSample = LoadInputColor(sampleCoord).xyz; + FfxFloat32 fReactiveSample = LoadReactiveMask(sampleCoord); + FfxFloat32 fTransparencyAndCompositionSample = LoadTransparencyAndCompositionMask(sampleCoord); + + fColorSamples[sampleIdx] = fColorSample; + fReactiveSamples[sampleIdx] = fReactiveSample; + fTransparencyAndCompositionSamples[sampleIdx] = fTransparencyAndCompositionSample; + + fMasksSum += (fReactiveSample + fTransparencyAndCompositionSample); + } + } + + if (fMasksSum > 0) + { + for (FfxInt32 sampleIdx = 0; sampleIdx < 9; sampleIdx++) + { + FfxFloat32x3 fColorSample = fColorSamples[sampleIdx]; + FfxFloat32 fReactiveSample = fReactiveSamples[sampleIdx]; + FfxFloat32 fTransparencyAndCompositionSample = fTransparencyAndCompositionSamples[sampleIdx]; + + const FfxFloat32 fMaxLenSq = ffxMax(dot(fReferenceColor, fReferenceColor), dot(fColorSample, fColorSample)); + const FfxFloat32 fSimilarity = dot(fReferenceColor, fColorSample) / fMaxLenSq; + + // Increase power for non-similar samples + const FfxFloat32 fPowerBiasMax = 6.0f; + const FfxFloat32 fSimilarityPower = 1.0f + (fPowerBiasMax - fSimilarity * fPowerBiasMax); + const FfxFloat32 fWeightedReactiveSample = ffxPow(fReactiveSample, fSimilarityPower); + const FfxFloat32 fWeightedTransparencyAndCompositionSample = ffxPow(fTransparencyAndCompositionSample, fSimilarityPower); + + fReactiveFactor = ffxMax(fReactiveFactor, FfxFloat32x2(fWeightedReactiveSample, fWeightedTransparencyAndCompositionSample)); + } + } + + StoreDilatedReactiveMasks(iPxLrPos, fReactiveFactor); +} + +FfxFloat32x3 ComputePreparedInputColor(FfxInt32x2 iPxLrPos) +{ + //We assume linear data. if non-linear input (sRGB, ...), + //then we should convert to linear first and back to sRGB on output. + FfxFloat32x3 fRgb = ffxMax(FfxFloat32x3(0, 0, 0), LoadInputColor(iPxLrPos)); + + fRgb = PrepareRgb(fRgb, Exposure(), PreExposure()); + + const FfxFloat32x3 fPreparedYCoCg = RGBToYCoCg(fRgb); + + return fPreparedYCoCg; +} + +FfxFloat32 EvaluateSurface(FfxInt32x2 iPxPos, FfxFloat32x2 fMotionVector) +{ + FfxFloat32 d0 = GetViewSpaceDepth(LoadReconstructedPrevDepth(iPxPos + FfxInt32x2(0, -1))); + FfxFloat32 d1 = GetViewSpaceDepth(LoadReconstructedPrevDepth(iPxPos + FfxInt32x2(0, 0))); + FfxFloat32 d2 = GetViewSpaceDepth(LoadReconstructedPrevDepth(iPxPos + FfxInt32x2(0, 1))); + + return 1.0f - FfxFloat32(((d0 - d1) > (d1 * 0.01f)) && ((d1 - d2) > (d2 * 0.01f))); +} + +void DepthClip(FfxInt32x2 iPxPos) +{ + FfxFloat32x2 fDepthUv = (iPxPos + 0.5f) / RenderSize(); + FfxFloat32x2 fMotionVector = LoadDilatedMotionVector(iPxPos); + + // Discard tiny mvs + fMotionVector *= FfxFloat32(length(fMotionVector * DisplaySize()) > 0.01f); + + const FfxFloat32x2 fDilatedUv = fDepthUv + fMotionVector; + const FfxFloat32 fDilatedDepth = LoadDilatedDepth(iPxPos); + const FfxFloat32 fCurrentDepthViewSpace = GetViewSpaceDepth(LoadInputDepth(iPxPos)); + + // Compute prepared input color and depth clip + FfxFloat32 fDepthClip = ComputeDepthClip(fDilatedUv, fDilatedDepth) * EvaluateSurface(iPxPos, fMotionVector); + FfxFloat32x3 fPreparedYCoCg = ComputePreparedInputColor(iPxPos); + StorePreparedInputColor(iPxPos, FfxFloat32x4(fPreparedYCoCg, fDepthClip)); + + // Compute dilated reactive mask +#if FFX_FSR2_OPTION_LOW_RESOLUTION_MOTION_VECTORS + FfxInt32x2 iSamplePos = iPxPos; +#else + FfxInt32x2 iSamplePos = ComputeHrPosFromLrPos(iPxPos); +#endif + + FfxFloat32 fMotionDivergence = ComputeMotionDivergence(iSamplePos, RenderSize()); + FfxFloat32 fTemporalMotionDifference = ffxSaturate(ComputeTemporalMotionDivergence(iPxPos) - ComputeDepthDivergence(iPxPos)); + + PreProcessReactiveMasks(iPxPos, ffxMax(fTemporalMotionDifference, fMotionDivergence)); +} + +#endif //!defined( FFX_FSR2_DEPTH_CLIPH ) diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_depth_clip.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_depth_clip.h.meta new file mode 100644 index 0000000..49ed527 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_depth_clip.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 37ba9aabf1f29eb4abc9b37df9829fde +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_lock.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_lock.h new file mode 100644 index 0000000..4a1f6d5 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_lock.h @@ -0,0 +1,116 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#ifndef FFX_FSR2_LOCK_H +#define FFX_FSR2_LOCK_H + +void ClearResourcesForNextFrame(in FfxInt32x2 iPxHrPos) +{ + if (all(FFX_LESS_THAN(iPxHrPos, FfxInt32x2(RenderSize())))) + { +#if FFX_FSR2_OPTION_INVERTED_DEPTH + const FfxUInt32 farZ = 0x0; +#else + const FfxUInt32 farZ = 0x3f800000; +#endif + SetReconstructedDepth(iPxHrPos, farZ); + } +} + +FfxBoolean ComputeThinFeatureConfidence(FfxInt32x2 pos) +{ + const FfxInt32 RADIUS = 1; + + FfxFloat32 fNucleus = LoadLockInputLuma(pos); + + FfxFloat32 similar_threshold = 1.05f; + FfxFloat32 dissimilarLumaMin = FSR2_FLT_MAX; + FfxFloat32 dissimilarLumaMax = 0; + + /* + 0 1 2 + 3 4 5 + 6 7 8 + */ + + #define SETBIT(x) (1U << x) + + FfxUInt32 mask = SETBIT(4); //flag fNucleus as similar + + const FfxUInt32 uNumRejectionMasks = 4; + const FfxUInt32 uRejectionMasks[uNumRejectionMasks] = { + SETBIT(0) | SETBIT(1) | SETBIT(3) | SETBIT(4), //Upper left + SETBIT(1) | SETBIT(2) | SETBIT(4) | SETBIT(5), //Upper right + SETBIT(3) | SETBIT(4) | SETBIT(6) | SETBIT(7), //Lower left + SETBIT(4) | SETBIT(5) | SETBIT(7) | SETBIT(8), //Lower right + }; + + FfxInt32 idx = 0; + FFX_UNROLL + for (FfxInt32 y = -RADIUS; y <= RADIUS; y++) { + FFX_UNROLL + for (FfxInt32 x = -RADIUS; x <= RADIUS; x++, idx++) { + if (x == 0 && y == 0) continue; + + FfxInt32x2 samplePos = ClampLoad(pos, FfxInt32x2(x, y), FfxInt32x2(RenderSize())); + + FfxFloat32 sampleLuma = LoadLockInputLuma(samplePos); + FfxFloat32 difference = ffxMax(sampleLuma, fNucleus) / ffxMin(sampleLuma, fNucleus); + + if (difference > 0 && (difference < similar_threshold)) { + mask |= SETBIT(idx); + } else { + dissimilarLumaMin = ffxMin(dissimilarLumaMin, sampleLuma); + dissimilarLumaMax = ffxMax(dissimilarLumaMax, sampleLuma); + } + } + } + + FfxBoolean isRidge = fNucleus > dissimilarLumaMax || fNucleus < dissimilarLumaMin; + + if (FFX_FALSE == isRidge) { + + return false; + } + + FFX_UNROLL + for (FfxInt32 i = 0; i < 4; i++) { + + if ((mask & uRejectionMasks[i]) == uRejectionMasks[i]) { + return false; + } + } + + return true; +} + +void ComputeLock(FfxInt32x2 iPxLrPos) +{ + if (ComputeThinFeatureConfidence(iPxLrPos)) + { + StoreNewLocks(ComputeHrPosFromLrPos(iPxLrPos), 1.f); + } + + // ClearResourcesForNextFrame(iPxLrPos); +} + +#endif // FFX_FSR2_LOCK_H diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_lock.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_lock.h.meta new file mode 100644 index 0000000..bf702fd --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_lock.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 77b94974e206e5045afda96cf9048139 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_postprocess_lock_status.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_postprocess_lock_status.h new file mode 100644 index 0000000..90ef344 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_postprocess_lock_status.h @@ -0,0 +1,107 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#ifndef FFX_FSR2_POSTPROCESS_LOCK_STATUS_H +#define FFX_FSR2_POSTPROCESS_LOCK_STATUS_H + +FfxFloat32x4 WrapShadingChangeLuma(FfxInt32x2 iPxSample) +{ + return FfxFloat32x4(LoadMipLuma(iPxSample, LumaMipLevelToUse()), 0, 0, 0); +} + +#if FFX_HALF +FFX_MIN16_F4 WrapShadingChangeLuma(FFX_MIN16_I2 iPxSample) +{ + return FFX_MIN16_F4(LoadMipLuma(iPxSample, LumaMipLevelToUse()), 0, 0, 0); +} +#endif + +#if FFX_FSR2_OPTION_POSTPROCESSLOCKSTATUS_SAMPLERS_USE_DATA_HALF && FFX_HALF +DeclareCustomFetchBilinearSamplesMin16(FetchShadingChangeLumaSamples, WrapShadingChangeLuma) +#else +DeclareCustomFetchBicubicSamples(FetchShadingChangeLumaSamples, WrapShadingChangeLuma) +#endif +DeclareCustomTextureSample(ShadingChangeLumaSample, Lanczos2, FetchShadingChangeLumaSamples) + +FfxFloat32 GetShadingChangeLuma(FfxInt32x2 iPxHrPos, FfxFloat32x2 fUvCoord) +{ + FfxFloat32 fShadingChangeLuma = 0; + +#if 0 + fShadingChangeLuma = Exposure() * exp(ShadingChangeLumaSample(fUvCoord, LumaMipDimensions()).x); +#else + + const FfxFloat32 fDiv = FfxFloat32(2 << LumaMipLevelToUse()); + FfxInt32x2 iMipRenderSize = FfxInt32x2(RenderSize() / fDiv); + + fUvCoord = ClampUv(fUvCoord, iMipRenderSize, LumaMipDimensions()); + fShadingChangeLuma = Exposure() * exp(FfxFloat32(SampleMipLuma(fUvCoord, LumaMipLevelToUse()))); +#endif + + fShadingChangeLuma = ffxPow(fShadingChangeLuma, 1.0f / 6.0f); + + return fShadingChangeLuma; +} + +void UpdateLockStatus(AccumulationPassCommonParams params, + FFX_PARAMETER_INOUT FfxFloat32 fReactiveFactor, LockState state, + FFX_PARAMETER_INOUT FfxFloat32x2 fLockStatus, + FFX_PARAMETER_OUT FfxFloat32 fLockContributionThisFrame, + FFX_PARAMETER_OUT FfxFloat32 fLuminanceDiff) { + + const FfxFloat32 fShadingChangeLuma = GetShadingChangeLuma(params.iPxHrPos, params.fHrUv); + + //init temporal shading change factor, init to -1 or so in reproject to know if "true new"? + fLockStatus[LOCK_TEMPORAL_LUMA] = (fLockStatus[LOCK_TEMPORAL_LUMA] == FfxFloat32(0.0f)) ? fShadingChangeLuma : fLockStatus[LOCK_TEMPORAL_LUMA]; + + FfxFloat32 fPreviousShadingChangeLuma = fLockStatus[LOCK_TEMPORAL_LUMA]; + + fLuminanceDiff = 1.0f - MinDividedByMax(fPreviousShadingChangeLuma, fShadingChangeLuma); + + if (state.NewLock) { + fLockStatus[LOCK_TEMPORAL_LUMA] = fShadingChangeLuma; + + fLockStatus[LOCK_LIFETIME_REMAINING] = (fLockStatus[LOCK_LIFETIME_REMAINING] != 0.0f) ? 2.0f : 1.0f; + } + else if(fLockStatus[LOCK_LIFETIME_REMAINING] <= 1.0f) { + fLockStatus[LOCK_TEMPORAL_LUMA] = ffxLerp(fLockStatus[LOCK_TEMPORAL_LUMA], FfxFloat32(fShadingChangeLuma), 0.5f); + } + else { + if (fLuminanceDiff > 0.1f) { + KillLock(fLockStatus); + } + } + + fReactiveFactor = ffxMax(fReactiveFactor, ffxSaturate((fLuminanceDiff - 0.1f) * 10.0f)); + fLockStatus[LOCK_LIFETIME_REMAINING] *= (1.0f - fReactiveFactor); + + fLockStatus[LOCK_LIFETIME_REMAINING] *= ffxSaturate(1.0f - params.fAccumulationMask); + fLockStatus[LOCK_LIFETIME_REMAINING] *= FfxFloat32(params.fDepthClipFactor < 0.1f); + + // Compute this frame lock contribution + const FfxFloat32 fLifetimeContribution = ffxSaturate(fLockStatus[LOCK_LIFETIME_REMAINING] - 1.0f); + const FfxFloat32 fShadingChangeContribution = ffxSaturate(MinDividedByMax(fLockStatus[LOCK_TEMPORAL_LUMA], fShadingChangeLuma)); + + fLockContributionThisFrame = ffxSaturate(ffxSaturate(fLifetimeContribution * 4.0f) * fShadingChangeContribution); +} + +#endif //!defined( FFX_FSR2_POSTPROCESS_LOCK_STATUS_H ) diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_postprocess_lock_status.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_postprocess_lock_status.h.meta new file mode 100644 index 0000000..f92a8d3 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_postprocess_lock_status.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 04dd68d20ea907f4381636f90aae5bc9 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_rcas.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_rcas.h new file mode 100644 index 0000000..fd5fd26 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_rcas.h @@ -0,0 +1,67 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#define GROUP_SIZE 8 +#define FSR_RCAS_DENOISE 1 + +#include "../ffx_core.h" + +void WriteUpscaledOutput(FFX_MIN16_U2 iPxHrPos, FfxFloat32x3 fUpscaledColor) +{ + StoreUpscaledOutput(FFX_MIN16_I2(iPxHrPos), fUpscaledColor); +} + +#define FSR_RCAS_F 1 +FfxFloat32x4 FsrRcasLoadF(FfxInt32x2 p) +{ + FfxFloat32x4 fColor = LoadRCAS_Input(p); + + fColor.rgb = PrepareRgb(fColor.rgb, Exposure(), PreExposure()); + + return fColor; +} +void FsrRcasInputF(inout FfxFloat32 r, inout FfxFloat32 g, inout FfxFloat32 b) {} + +#include "../fsr1/ffx_fsr1.h" + +void CurrFilter(FFX_MIN16_U2 pos) +{ + FfxFloat32x3 c; + FsrRcasF(c.r, c.g, c.b, pos, RCASConfig()); + + c = UnprepareRgb(c, Exposure()); + + WriteUpscaledOutput(pos, c); +} + +void RCAS(FfxUInt32x3 LocalThreadId, FfxUInt32x3 WorkGroupId, FfxUInt32x3 Dtid) +{ + // Do remapping of local xy in workgroup for a more PS-like swizzle pattern. + FfxUInt32x2 gxy = ffxRemapForQuad(LocalThreadId.x) + FfxUInt32x2(WorkGroupId.x << 4u, WorkGroupId.y << 4u); + CurrFilter(FFX_MIN16_U2(gxy)); + gxy.x += 8u; + CurrFilter(FFX_MIN16_U2(gxy)); + gxy.y += 8u; + CurrFilter(FFX_MIN16_U2(gxy)); + gxy.x -= 8u; + CurrFilter(FFX_MIN16_U2(gxy)); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_rcas.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_rcas.h.meta new file mode 100644 index 0000000..c121733 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_rcas.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 1b6b9dc81790da347b42bf6a3a132479 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_reconstruct_dilated_velocity_and_previous_depth.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_reconstruct_dilated_velocity_and_previous_depth.h new file mode 100644 index 0000000..1a4305d --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_reconstruct_dilated_velocity_and_previous_depth.h @@ -0,0 +1,146 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#ifndef FFX_FSR2_RECONSTRUCT_DILATED_VELOCITY_AND_PREVIOUS_DEPTH_H +#define FFX_FSR2_RECONSTRUCT_DILATED_VELOCITY_AND_PREVIOUS_DEPTH_H + +void ReconstructPrevDepth(FfxInt32x2 iPxPos, FfxFloat32 fDepth, FfxFloat32x2 fMotionVector, FfxInt32x2 iPxDepthSize) +{ + fMotionVector *= FfxFloat32(length(fMotionVector * DisplaySize()) > 0.1f); + + FfxFloat32x2 fUv = (iPxPos + FfxFloat32(0.5)) / iPxDepthSize; + FfxFloat32x2 fReprojectedUv = fUv + fMotionVector; + + BilinearSamplingData bilinearInfo = GetBilinearSamplingData(fReprojectedUv, RenderSize()); + + // Project current depth into previous frame locations. + // Push to all pixels having some contribution if reprojection is using bilinear logic. + for (FfxInt32 iSampleIndex = 0; iSampleIndex < 4; iSampleIndex++) { + + const FfxInt32x2 iOffset = bilinearInfo.iOffsets[iSampleIndex]; + FfxFloat32 fWeight = bilinearInfo.fWeights[iSampleIndex]; + + if (fWeight > fReconstructedDepthBilinearWeightThreshold) { + + FfxInt32x2 iStorePos = bilinearInfo.iBasePos + iOffset; + if (IsOnScreen(iStorePos, iPxDepthSize)) { + StoreReconstructedDepth(iStorePos, fDepth); + } + } + } +} + +void FindNearestDepth(FFX_PARAMETER_IN FfxInt32x2 iPxPos, FFX_PARAMETER_IN FfxInt32x2 iPxSize, FFX_PARAMETER_OUT FfxFloat32 fNearestDepth, FFX_PARAMETER_OUT FfxInt32x2 fNearestDepthCoord) +{ + const FfxInt32 iSampleCount = 9; + const FfxInt32x2 iSampleOffsets[iSampleCount] = { + FfxInt32x2(+0, +0), + FfxInt32x2(+1, +0), + FfxInt32x2(+0, +1), + FfxInt32x2(+0, -1), + FfxInt32x2(-1, +0), + FfxInt32x2(-1, +1), + FfxInt32x2(+1, +1), + FfxInt32x2(-1, -1), + FfxInt32x2(+1, -1), + }; + + // pull out the depth loads to allow SC to batch them + FfxFloat32 depth[9]; + FfxInt32 iSampleIndex = 0; + FFX_UNROLL + for (iSampleIndex = 0; iSampleIndex < iSampleCount; ++iSampleIndex) { + + FfxInt32x2 iPos = iPxPos + iSampleOffsets[iSampleIndex]; + depth[iSampleIndex] = LoadInputDepth(iPos); + } + + // find closest depth + fNearestDepthCoord = iPxPos; + fNearestDepth = depth[0]; + FFX_UNROLL + for (iSampleIndex = 1; iSampleIndex < iSampleCount; ++iSampleIndex) { + + FfxInt32x2 iPos = iPxPos + iSampleOffsets[iSampleIndex]; + if (IsOnScreen(iPos, iPxSize)) { + + FfxFloat32 fNdDepth = depth[iSampleIndex]; +#if FFX_FSR2_OPTION_INVERTED_DEPTH + if (fNdDepth > fNearestDepth) { +#else + if (fNdDepth < fNearestDepth) { +#endif + fNearestDepthCoord = iPos; + fNearestDepth = fNdDepth; + } + } + } +} + +FfxFloat32 ComputeLockInputLuma(FfxInt32x2 iPxLrPos) +{ + //We assume linear data. if non-linear input (sRGB, ...), + //then we should convert to linear first and back to sRGB on output. + FfxFloat32x3 fRgb = ffxMax(FfxFloat32x3(0, 0, 0), LoadInputColor(iPxLrPos)); + + // Use internal auto exposure for locking logic + fRgb /= PreExposure(); + fRgb *= Exposure(); + +#if FFX_FSR2_OPTION_HDR_COLOR_INPUT + fRgb = Tonemap(fRgb); +#endif + + //compute luma used to lock pixels, if used elsewhere the ffxPow must be moved! + const FfxFloat32 fLockInputLuma = ffxPow(RGBToPerceivedLuma(fRgb), FfxFloat32(1.0 / 6.0)); + + return fLockInputLuma; +} + +void ReconstructAndDilate(FfxInt32x2 iPxLrPos) +{ + FfxFloat32 fDilatedDepth; + FfxInt32x2 iNearestDepthCoord; + + FindNearestDepth(iPxLrPos, RenderSize(), fDilatedDepth, iNearestDepthCoord); + +#if FFX_FSR2_OPTION_LOW_RESOLUTION_MOTION_VECTORS + FfxInt32x2 iSamplePos = iPxLrPos; + FfxInt32x2 iMotionVectorPos = iNearestDepthCoord; +#else + FfxInt32x2 iSamplePos = ComputeHrPosFromLrPos(iPxLrPos); + FfxInt32x2 iMotionVectorPos = ComputeHrPosFromLrPos(iNearestDepthCoord); +#endif + + FfxFloat32x2 fDilatedMotionVector = LoadInputMotionVector(iMotionVectorPos); + + StoreDilatedDepth(iPxLrPos, fDilatedDepth); + StoreDilatedMotionVector(iPxLrPos, fDilatedMotionVector); + + ReconstructPrevDepth(iPxLrPos, fDilatedDepth, fDilatedMotionVector, RenderSize()); + + FfxFloat32 fLockInputLuma = ComputeLockInputLuma(iPxLrPos); + StoreLockInputLuma(iPxLrPos, fLockInputLuma); +} + + +#endif //!defined( FFX_FSR2_RECONSTRUCT_DILATED_VELOCITY_AND_PREVIOUS_DEPTH_H ) diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_reconstruct_dilated_velocity_and_previous_depth.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_reconstruct_dilated_velocity_and_previous_depth.h.meta new file mode 100644 index 0000000..3d37e71 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_reconstruct_dilated_velocity_and_previous_depth.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: c7d2ce16ffba7e5459f6e88351b86355 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_reproject.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_reproject.h new file mode 100644 index 0000000..386b297 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_reproject.h @@ -0,0 +1,137 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#ifndef FFX_FSR2_REPROJECT_H +#define FFX_FSR2_REPROJECT_H + +#ifndef FFX_FSR2_OPTION_REPROJECT_USE_LANCZOS_TYPE +#define FFX_FSR2_OPTION_REPROJECT_USE_LANCZOS_TYPE 0 // Reference +#endif + +FfxFloat32x4 WrapHistory(FfxInt32x2 iPxSample) +{ + return LoadHistory(iPxSample); +} + +#if FFX_HALF +FFX_MIN16_F4 WrapHistory(FFX_MIN16_I2 iPxSample) +{ + return FFX_MIN16_F4(LoadHistory(iPxSample)); +} +#endif + + +#if FFX_FSR2_OPTION_REPROJECT_SAMPLERS_USE_DATA_HALF && FFX_HALF +DeclareCustomFetchBicubicSamplesMin16(FetchHistorySamples, WrapHistory) +DeclareCustomTextureSampleMin16(HistorySample, FFX_FSR2_GET_LANCZOS_SAMPLER1D(FFX_FSR2_OPTION_REPROJECT_USE_LANCZOS_TYPE), FetchHistorySamples) +#else +DeclareCustomFetchBicubicSamples(FetchHistorySamples, WrapHistory) +DeclareCustomTextureSample(HistorySample, FFX_FSR2_GET_LANCZOS_SAMPLER1D(FFX_FSR2_OPTION_REPROJECT_USE_LANCZOS_TYPE), FetchHistorySamples) +#endif + +FfxFloat32x4 WrapLockStatus(FfxInt32x2 iPxSample) +{ + FfxFloat32x4 fSample = FfxFloat32x4(LoadLockStatus(iPxSample), 0.0f, 0.0f); + return fSample; +} + +#if FFX_HALF +FFX_MIN16_F4 WrapLockStatus(FFX_MIN16_I2 iPxSample) +{ + FFX_MIN16_F4 fSample = FFX_MIN16_F4(LoadLockStatus(iPxSample), 0.0, 0.0); + + return fSample; +} +#endif + +#if 1 +#if FFX_FSR2_OPTION_REPROJECT_SAMPLERS_USE_DATA_HALF && FFX_HALF +DeclareCustomFetchBilinearSamplesMin16(FetchLockStatusSamples, WrapLockStatus) +DeclareCustomTextureSampleMin16(LockStatusSample, Bilinear, FetchLockStatusSamples) +#else +DeclareCustomFetchBilinearSamples(FetchLockStatusSamples, WrapLockStatus) +DeclareCustomTextureSample(LockStatusSample, Bilinear, FetchLockStatusSamples) +#endif +#else +#if FFX_FSR2_OPTION_REPROJECT_SAMPLERS_USE_DATA_HALF && FFX_HALF +DeclareCustomFetchBicubicSamplesMin16(FetchLockStatusSamples, WrapLockStatus) +DeclareCustomTextureSampleMin16(LockStatusSample, FFX_FSR2_GET_LANCZOS_SAMPLER1D(FFX_FSR2_OPTION_REPROJECT_USE_LANCZOS_TYPE), FetchLockStatusSamples) +#else +DeclareCustomFetchBicubicSamples(FetchLockStatusSamples, WrapLockStatus) +DeclareCustomTextureSample(LockStatusSample, FFX_FSR2_GET_LANCZOS_SAMPLER1D(FFX_FSR2_OPTION_REPROJECT_USE_LANCZOS_TYPE), FetchLockStatusSamples) +#endif +#endif + +FfxFloat32x2 GetMotionVector(FfxInt32x2 iPxHrPos, FfxFloat32x2 fHrUv) +{ +#if FFX_FSR2_OPTION_LOW_RESOLUTION_MOTION_VECTORS + FfxFloat32x2 fDilatedMotionVector = LoadDilatedMotionVector(FFX_MIN16_I2(fHrUv * RenderSize())); +#else + FfxFloat32x2 fDilatedMotionVector = LoadInputMotionVector(iPxHrPos); +#endif + + return fDilatedMotionVector; +} + +FfxBoolean IsUvInside(FfxFloat32x2 fUv) +{ + return (fUv.x >= 0.0f && fUv.x <= 1.0f) && (fUv.y >= 0.0f && fUv.y <= 1.0f); +} + +void ComputeReprojectedUVs(const AccumulationPassCommonParams params, FFX_PARAMETER_OUT FfxFloat32x2 fReprojectedHrUv, FFX_PARAMETER_OUT FfxBoolean bIsExistingSample) +{ + fReprojectedHrUv = params.fHrUv + params.fMotionVector; + + bIsExistingSample = IsUvInside(fReprojectedHrUv); +} + +void ReprojectHistoryColor(const AccumulationPassCommonParams params, FFX_PARAMETER_OUT FfxFloat32x3 fHistoryColor, FFX_PARAMETER_OUT FfxFloat32 fTemporalReactiveFactor, FFX_PARAMETER_OUT FfxBoolean bInMotionLastFrame) +{ + FfxFloat32x4 fHistory = HistorySample(params.fReprojectedHrUv, DisplaySize()); + + fHistoryColor = PrepareRgb(fHistory.rgb, Exposure(), PreviousFramePreExposure()); + + fHistoryColor = RGBToYCoCg(fHistoryColor); + + //Compute temporal reactivity info + fTemporalReactiveFactor = ffxSaturate(abs(fHistory.w)); + bInMotionLastFrame = (fHistory.w < 0.0f); +} + +LockState ReprojectHistoryLockStatus(const AccumulationPassCommonParams params, FFX_PARAMETER_OUT FfxFloat32x2 fReprojectedLockStatus) +{ + LockState state = { FFX_FALSE, FFX_FALSE }; + const FfxFloat32 fNewLockIntensity = LoadRwNewLocks(params.iPxHrPos); + state.NewLock = fNewLockIntensity > (127.0f / 255.0f); + + FfxFloat32 fInPlaceLockLifetime = state.NewLock ? fNewLockIntensity : 0; + + fReprojectedLockStatus = SampleLockStatus(params.fReprojectedHrUv); + + if (fReprojectedLockStatus[LOCK_LIFETIME_REMAINING] != FfxFloat32(0.0f)) { + state.WasLockedPrevFrame = true; + } + + return state; +} + +#endif //!defined( FFX_FSR2_REPROJECT_H ) diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_reproject.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_reproject.h.meta new file mode 100644 index 0000000..d99b60e --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_reproject.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: c888fea71de852c468ab21c525ce2320 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_resources.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_resources.h new file mode 100644 index 0000000..a597c5a --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_resources.h @@ -0,0 +1,106 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#ifndef FFX_FSR2_RESOURCES_H +#define FFX_FSR2_RESOURCES_H + +#if defined(FFX_CPU) || defined(FFX_GPU) +#define FFX_FSR2_RESOURCE_IDENTIFIER_NULL 0 +#define FFX_FSR2_RESOURCE_IDENTIFIER_INPUT_OPAQUE_ONLY 1 +#define FFX_FSR2_RESOURCE_IDENTIFIER_INPUT_COLOR 2 +#define FFX_FSR2_RESOURCE_IDENTIFIER_INPUT_MOTION_VECTORS 3 +#define FFX_FSR2_RESOURCE_IDENTIFIER_INPUT_DEPTH 4 +#define FFX_FSR2_RESOURCE_IDENTIFIER_INPUT_EXPOSURE 5 +#define FFX_FSR2_RESOURCE_IDENTIFIER_INPUT_REACTIVE_MASK 6 +#define FFX_FSR2_RESOURCE_IDENTIFIER_INPUT_TRANSPARENCY_AND_COMPOSITION_MASK 7 +#define FFX_FSR2_RESOURCE_IDENTIFIER_RECONSTRUCTED_PREVIOUS_NEAREST_DEPTH 8 +#define FFX_FSR2_RESOURCE_IDENTIFIER_DILATED_MOTION_VECTORS 9 +#define FFX_FSR2_RESOURCE_IDENTIFIER_DILATED_DEPTH 10 +#define FFX_FSR2_RESOURCE_IDENTIFIER_INTERNAL_UPSCALED_COLOR 11 +#define FFX_FSR2_RESOURCE_IDENTIFIER_LOCK_STATUS 12 +#define FFX_FSR2_RESOURCE_IDENTIFIER_NEW_LOCKS 13 +#define FFX_FSR2_RESOURCE_IDENTIFIER_PREPARED_INPUT_COLOR 14 +#define FFX_FSR2_RESOURCE_IDENTIFIER_LUMA_HISTORY 15 +#define FFX_FSR2_RESOURCE_IDENTIFIER_DEBUG_OUTPUT 16 +#define FFX_FSR2_RESOURCE_IDENTIFIER_LANCZOS_LUT 17 +#define FFX_FSR2_RESOURCE_IDENTIFIER_SPD_ATOMIC_COUNT 18 +#define FFX_FSR2_RESOURCE_IDENTIFIER_UPSCALED_OUTPUT 19 +#define FFX_FSR2_RESOURCE_IDENTIFIER_RCAS_INPUT 20 +#define FFX_FSR2_RESOURCE_IDENTIFIER_LOCK_STATUS_1 21 +#define FFX_FSR2_RESOURCE_IDENTIFIER_LOCK_STATUS_2 22 +#define FFX_FSR2_RESOURCE_IDENTIFIER_INTERNAL_UPSCALED_COLOR_1 23 +#define FFX_FSR2_RESOURCE_IDENTIFIER_INTERNAL_UPSCALED_COLOR_2 24 +#define FFX_FSR2_RESOURCE_IDENTIFIER_INTERNAL_DEFAULT_REACTIVITY 25 +#define FFX_FSR2_RESOURCE_IDENTIFIER_INTERNAL_DEFAULT_TRANSPARENCY_AND_COMPOSITION 26 +#define FFX_FSR2_RESOURCE_IDENTITIER_UPSAMPLE_MAXIMUM_BIAS_LUT 27 +#define FFX_FSR2_RESOURCE_IDENTIFIER_DILATED_REACTIVE_MASKS 28 +#define FFX_FSR2_RESOURCE_IDENTIFIER_SCENE_LUMINANCE 29 // same as FFX_FSR2_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_0 +#define FFX_FSR2_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_0 29 +#define FFX_FSR2_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_1 30 +#define FFX_FSR2_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_2 31 +#define FFX_FSR2_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_3 32 +#define FFX_FSR2_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_4 33 +#define FFX_FSR2_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_5 34 +#define FFX_FSR2_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_6 35 +#define FFX_FSR2_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_7 36 +#define FFX_FSR2_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_8 37 +#define FFX_FSR2_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_9 38 +#define FFX_FSR2_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_10 39 +#define FFX_FSR2_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_11 40 +#define FFX_FSR2_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_12 41 +#define FFX_FSR2_RESOURCE_IDENTIFIER_INTERNAL_DEFAULT_EXPOSURE 42 +#define FFX_FSR2_RESOURCE_IDENTIFIER_AUTO_EXPOSURE 43 +#define FFX_FSR2_RESOURCE_IDENTIFIER_AUTOREACTIVE 44 +#define FFX_FSR2_RESOURCE_IDENTIFIER_AUTOCOMPOSITION 45 + +#define FFX_FSR2_RESOURCE_IDENTIFIER_PREV_PRE_ALPHA_COLOR 46 +#define FFX_FSR2_RESOURCE_IDENTIFIER_PREV_POST_ALPHA_COLOR 47 +#define FFX_FSR2_RESOURCE_IDENTIFIER_PREV_PRE_ALPHA_COLOR_1 48 +#define FFX_FSR2_RESOURCE_IDENTIFIER_PREV_POST_ALPHA_COLOR_1 49 +#define FFX_FSR2_RESOURCE_IDENTIFIER_PREV_PRE_ALPHA_COLOR_2 50 +#define FFX_FSR2_RESOURCE_IDENTIFIER_PREV_POST_ALPHA_COLOR_2 51 +#define FFX_FSR2_RESOURCE_IDENTIFIER_PREVIOUS_DILATED_MOTION_VECTORS 52 +#define FFX_FSR2_RESOURCE_IDENTIFIER_INTERNAL_DILATED_MOTION_VECTORS_1 53 +#define FFX_FSR2_RESOURCE_IDENTIFIER_INTERNAL_DILATED_MOTION_VECTORS_2 54 +#define FFX_FSR2_RESOURCE_IDENTIFIER_LUMA_HISTORY_1 55 +#define FFX_FSR2_RESOURCE_IDENTIFIER_LUMA_HISTORY_2 56 +#define FFX_FSR2_RESOURCE_IDENTIFIER_LOCK_INPUT_LUMA 57 + +// Shading change detection mip level setting, value must be in the range [FFX_FSR2_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_0, FFX_FSR2_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_12] +#define FFX_FSR2_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_SHADING_CHANGE FFX_FSR2_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_4 +#define FFX_FSR2_SHADING_CHANGE_MIP_LEVEL (FFX_FSR2_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_SHADING_CHANGE - FFX_FSR2_RESOURCE_IDENTIFIER_SCENE_LUMINANCE) + +#define FFX_FSR2_RESOURCE_IDENTIFIER_COUNT 58 + +#define FFX_FSR2_CONSTANTBUFFER_IDENTIFIER_FSR2 0 +#define FFX_FSR2_CONSTANTBUFFER_IDENTIFIER_SPD 1 +#define FFX_FSR2_CONSTANTBUFFER_IDENTIFIER_RCAS 2 +#define FFX_FSR2_CONSTANTBUFFER_IDENTIFIER_GENREACTIVE 3 + +#define FFX_FSR2_AUTOREACTIVEFLAGS_APPLY_TONEMAP 1 +#define FFX_FSR2_AUTOREACTIVEFLAGS_APPLY_INVERSETONEMAP 2 +#define FFX_FSR2_AUTOREACTIVEFLAGS_APPLY_THRESHOLD 4 +#define FFX_FSR2_AUTOREACTIVEFLAGS_USE_COMPONENTS_MAX 8 + +#endif // #if defined(FFX_CPU) || defined(FFX_GPU) + +#endif //!defined( FFX_FSR2_RESOURCES_H ) diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_resources.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_resources.h.meta new file mode 100644 index 0000000..4269f87 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_resources.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: af595ef438ac2b4469f50e6f92485a6c +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_sample.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_sample.h new file mode 100644 index 0000000..b75f090 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_sample.h @@ -0,0 +1,606 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#ifndef FFX_FSR2_SAMPLE_H +#define FFX_FSR2_SAMPLE_H + +// suppress warnings +#ifdef FFX_HLSL +#pragma warning(disable: 4008) // potentially divide by zero +#endif //FFX_HLSL + +struct FetchedBilinearSamples { + + FfxFloat32x4 fColor00; + FfxFloat32x4 fColor10; + + FfxFloat32x4 fColor01; + FfxFloat32x4 fColor11; +}; + +struct FetchedBicubicSamples { + + FfxFloat32x4 fColor00; + FfxFloat32x4 fColor10; + FfxFloat32x4 fColor20; + FfxFloat32x4 fColor30; + + FfxFloat32x4 fColor01; + FfxFloat32x4 fColor11; + FfxFloat32x4 fColor21; + FfxFloat32x4 fColor31; + + FfxFloat32x4 fColor02; + FfxFloat32x4 fColor12; + FfxFloat32x4 fColor22; + FfxFloat32x4 fColor32; + + FfxFloat32x4 fColor03; + FfxFloat32x4 fColor13; + FfxFloat32x4 fColor23; + FfxFloat32x4 fColor33; +}; + +#if FFX_HALF +struct FetchedBilinearSamplesMin16 { + + FFX_MIN16_F4 fColor00; + FFX_MIN16_F4 fColor10; + + FFX_MIN16_F4 fColor01; + FFX_MIN16_F4 fColor11; +}; + +struct FetchedBicubicSamplesMin16 { + + FFX_MIN16_F4 fColor00; + FFX_MIN16_F4 fColor10; + FFX_MIN16_F4 fColor20; + FFX_MIN16_F4 fColor30; + + FFX_MIN16_F4 fColor01; + FFX_MIN16_F4 fColor11; + FFX_MIN16_F4 fColor21; + FFX_MIN16_F4 fColor31; + + FFX_MIN16_F4 fColor02; + FFX_MIN16_F4 fColor12; + FFX_MIN16_F4 fColor22; + FFX_MIN16_F4 fColor32; + + FFX_MIN16_F4 fColor03; + FFX_MIN16_F4 fColor13; + FFX_MIN16_F4 fColor23; + FFX_MIN16_F4 fColor33; +}; +#else //FFX_HALF +#define FetchedBicubicSamplesMin16 FetchedBicubicSamples +#define FetchedBilinearSamplesMin16 FetchedBilinearSamples +#endif //FFX_HALF + +FfxFloat32x4 Linear(FfxFloat32x4 A, FfxFloat32x4 B, FfxFloat32 t) +{ + return A + (B - A) * t; +} + +FfxFloat32x4 Bilinear(FetchedBilinearSamples BilinearSamples, FfxFloat32x2 fPxFrac) +{ + FfxFloat32x4 fColorX0 = Linear(BilinearSamples.fColor00, BilinearSamples.fColor10, fPxFrac.x); + FfxFloat32x4 fColorX1 = Linear(BilinearSamples.fColor01, BilinearSamples.fColor11, fPxFrac.x); + FfxFloat32x4 fColorXY = Linear(fColorX0, fColorX1, fPxFrac.y); + return fColorXY; +} + +#if FFX_HALF +FFX_MIN16_F4 Linear(FFX_MIN16_F4 A, FFX_MIN16_F4 B, FFX_MIN16_F t) +{ + return A + (B - A) * t; +} + +FFX_MIN16_F4 Bilinear(FetchedBilinearSamplesMin16 BilinearSamples, FFX_MIN16_F2 fPxFrac) +{ + FFX_MIN16_F4 fColorX0 = Linear(BilinearSamples.fColor00, BilinearSamples.fColor10, fPxFrac.x); + FFX_MIN16_F4 fColorX1 = Linear(BilinearSamples.fColor01, BilinearSamples.fColor11, fPxFrac.x); + FFX_MIN16_F4 fColorXY = Linear(fColorX0, fColorX1, fPxFrac.y); + return fColorXY; +} +#endif + +FfxFloat32 Lanczos2NoClamp(FfxFloat32 x) +{ + const FfxFloat32 PI = 3.141592653589793f; // TODO: share SDK constants + return abs(x) < FSR2_EPSILON ? 1.f : (sin(PI * x) / (PI * x)) * (sin(0.5f * PI * x) / (0.5f * PI * x)); +} + +FfxFloat32 Lanczos2(FfxFloat32 x) +{ + x = ffxMin(abs(x), 2.0f); + return Lanczos2NoClamp(x); +} + +#if FFX_HALF + +#if 0 +FFX_MIN16_F Lanczos2NoClamp(FFX_MIN16_F x) +{ + const FFX_MIN16_F PI = FFX_MIN16_F(3.141592653589793f); // TODO: share SDK constants + return abs(x) < FFX_MIN16_F(FSR2_EPSILON) ? FFX_MIN16_F(1.f) : (sin(PI * x) / (PI * x)) * (sin(FFX_MIN16_F(0.5f) * PI * x) / (FFX_MIN16_F(0.5f) * PI * x)); +} +#endif + +FFX_MIN16_F Lanczos2(FFX_MIN16_F x) +{ + x = ffxMin(abs(x), FFX_MIN16_F(2.0f)); + return FFX_MIN16_F(Lanczos2NoClamp(x)); +} +#endif //FFX_HALF + +// FSR1 lanczos approximation. Input is x*x and must be <= 4. +FfxFloat32 Lanczos2ApproxSqNoClamp(FfxFloat32 x2) +{ + FfxFloat32 a = (2.0f / 5.0f) * x2 - 1; + FfxFloat32 b = (1.0f / 4.0f) * x2 - 1; + return ((25.0f / 16.0f) * a * a - (25.0f / 16.0f - 1)) * (b * b); +} + +#if FFX_HALF +FFX_MIN16_F Lanczos2ApproxSqNoClamp(FFX_MIN16_F x2) +{ + FFX_MIN16_F a = FFX_MIN16_F(2.0f / 5.0f) * x2 - FFX_MIN16_F(1); + FFX_MIN16_F b = FFX_MIN16_F(1.0f / 4.0f) * x2 - FFX_MIN16_F(1); + return (FFX_MIN16_F(25.0f / 16.0f) * a * a - FFX_MIN16_F(25.0f / 16.0f - 1)) * (b * b); +} +#endif //FFX_HALF + +FfxFloat32 Lanczos2ApproxSq(FfxFloat32 x2) +{ + x2 = ffxMin(x2, 4.0f); + return Lanczos2ApproxSqNoClamp(x2); +} + +#if FFX_HALF +FFX_MIN16_F Lanczos2ApproxSq(FFX_MIN16_F x2) +{ + x2 = ffxMin(x2, FFX_MIN16_F(4.0f)); + return Lanczos2ApproxSqNoClamp(x2); +} +#endif //FFX_HALF + +FfxFloat32 Lanczos2ApproxNoClamp(FfxFloat32 x) +{ + return Lanczos2ApproxSqNoClamp(x * x); +} + +#if FFX_HALF +FFX_MIN16_F Lanczos2ApproxNoClamp(FFX_MIN16_F x) +{ + return Lanczos2ApproxSqNoClamp(x * x); +} +#endif //FFX_HALF + +FfxFloat32 Lanczos2Approx(FfxFloat32 x) +{ + return Lanczos2ApproxSq(x * x); +} + +#if FFX_HALF +FFX_MIN16_F Lanczos2Approx(FFX_MIN16_F x) +{ + return Lanczos2ApproxSq(x * x); +} +#endif //FFX_HALF + +FfxFloat32 Lanczos2_UseLUT(FfxFloat32 x) +{ + return SampleLanczos2Weight(abs(x)); +} + +#if FFX_HALF +FFX_MIN16_F Lanczos2_UseLUT(FFX_MIN16_F x) +{ + return FFX_MIN16_F(SampleLanczos2Weight(abs(x))); +} +#endif //FFX_HALF + +FfxFloat32x4 Lanczos2_UseLUT(FfxFloat32x4 fColor0, FfxFloat32x4 fColor1, FfxFloat32x4 fColor2, FfxFloat32x4 fColor3, FfxFloat32 t) +{ + FfxFloat32 fWeight0 = Lanczos2_UseLUT(-1.f - t); + FfxFloat32 fWeight1 = Lanczos2_UseLUT(-0.f - t); + FfxFloat32 fWeight2 = Lanczos2_UseLUT(+1.f - t); + FfxFloat32 fWeight3 = Lanczos2_UseLUT(+2.f - t); + return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); +} +#if FFX_HALF +FFX_MIN16_F4 Lanczos2_UseLUT(FFX_MIN16_F4 fColor0, FFX_MIN16_F4 fColor1, FFX_MIN16_F4 fColor2, FFX_MIN16_F4 fColor3, FFX_MIN16_F t) +{ + FFX_MIN16_F fWeight0 = Lanczos2_UseLUT(FFX_MIN16_F(-1.f) - t); + FFX_MIN16_F fWeight1 = Lanczos2_UseLUT(FFX_MIN16_F(-0.f) - t); + FFX_MIN16_F fWeight2 = Lanczos2_UseLUT(FFX_MIN16_F(+1.f) - t); + FFX_MIN16_F fWeight3 = Lanczos2_UseLUT(FFX_MIN16_F(+2.f) - t); + return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); +} +#endif + +FfxFloat32x4 Lanczos2(FfxFloat32x4 fColor0, FfxFloat32x4 fColor1, FfxFloat32x4 fColor2, FfxFloat32x4 fColor3, FfxFloat32 t) +{ + FfxFloat32 fWeight0 = Lanczos2(-1.f - t); + FfxFloat32 fWeight1 = Lanczos2(-0.f - t); + FfxFloat32 fWeight2 = Lanczos2(+1.f - t); + FfxFloat32 fWeight3 = Lanczos2(+2.f - t); + return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); +} + +FfxFloat32x4 Lanczos2(FetchedBicubicSamples Samples, FfxFloat32x2 fPxFrac) +{ + FfxFloat32x4 fColorX0 = Lanczos2(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); + FfxFloat32x4 fColorX1 = Lanczos2(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); + FfxFloat32x4 fColorX2 = Lanczos2(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); + FfxFloat32x4 fColorX3 = Lanczos2(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); + FfxFloat32x4 fColorXY = Lanczos2(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); + + // Deringing + + // TODO: only use 4 by checking jitter + const FfxInt32 iDeringingSampleCount = 4; + const FfxFloat32x4 fDeringingSamples[4] = { + Samples.fColor11, + Samples.fColor21, + Samples.fColor12, + Samples.fColor22, + }; + + FfxFloat32x4 fDeringingMin = fDeringingSamples[0]; + FfxFloat32x4 fDeringingMax = fDeringingSamples[0]; + + FFX_UNROLL + for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) { + + fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); + fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); + } + + fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); + + return fColorXY; +} + +#if FFX_HALF +FFX_MIN16_F4 Lanczos2(FFX_MIN16_F4 fColor0, FFX_MIN16_F4 fColor1, FFX_MIN16_F4 fColor2, FFX_MIN16_F4 fColor3, FFX_MIN16_F t) +{ + FFX_MIN16_F fWeight0 = Lanczos2(FFX_MIN16_F(-1.f) - t); + FFX_MIN16_F fWeight1 = Lanczos2(FFX_MIN16_F(-0.f) - t); + FFX_MIN16_F fWeight2 = Lanczos2(FFX_MIN16_F(+1.f) - t); + FFX_MIN16_F fWeight3 = Lanczos2(FFX_MIN16_F(+2.f) - t); + return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); +} + +FFX_MIN16_F4 Lanczos2(FetchedBicubicSamplesMin16 Samples, FFX_MIN16_F2 fPxFrac) +{ + FFX_MIN16_F4 fColorX0 = Lanczos2(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); + FFX_MIN16_F4 fColorX1 = Lanczos2(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); + FFX_MIN16_F4 fColorX2 = Lanczos2(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); + FFX_MIN16_F4 fColorX3 = Lanczos2(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); + FFX_MIN16_F4 fColorXY = Lanczos2(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); + + // Deringing + + // TODO: only use 4 by checking jitter + const FfxInt32 iDeringingSampleCount = 4; + const FFX_MIN16_F4 fDeringingSamples[4] = { + Samples.fColor11, + Samples.fColor21, + Samples.fColor12, + Samples.fColor22, + }; + + FFX_MIN16_F4 fDeringingMin = fDeringingSamples[0]; + FFX_MIN16_F4 fDeringingMax = fDeringingSamples[0]; + + FFX_UNROLL + for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) + { + fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); + fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); + } + + fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); + + return fColorXY; +} +#endif //FFX_HALF + + +FfxFloat32x4 Lanczos2LUT(FetchedBicubicSamples Samples, FfxFloat32x2 fPxFrac) +{ + FfxFloat32x4 fColorX0 = Lanczos2_UseLUT(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); + FfxFloat32x4 fColorX1 = Lanczos2_UseLUT(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); + FfxFloat32x4 fColorX2 = Lanczos2_UseLUT(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); + FfxFloat32x4 fColorX3 = Lanczos2_UseLUT(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); + FfxFloat32x4 fColorXY = Lanczos2_UseLUT(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); + + // Deringing + + // TODO: only use 4 by checking jitter + const FfxInt32 iDeringingSampleCount = 4; + const FfxFloat32x4 fDeringingSamples[4] = { + Samples.fColor11, + Samples.fColor21, + Samples.fColor12, + Samples.fColor22, + }; + + FfxFloat32x4 fDeringingMin = fDeringingSamples[0]; + FfxFloat32x4 fDeringingMax = fDeringingSamples[0]; + + FFX_UNROLL + for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) { + + fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); + fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); + } + + fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); + + return fColorXY; +} + +#if FFX_HALF +FFX_MIN16_F4 Lanczos2LUT(FetchedBicubicSamplesMin16 Samples, FFX_MIN16_F2 fPxFrac) +{ + FFX_MIN16_F4 fColorX0 = Lanczos2_UseLUT(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); + FFX_MIN16_F4 fColorX1 = Lanczos2_UseLUT(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); + FFX_MIN16_F4 fColorX2 = Lanczos2_UseLUT(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); + FFX_MIN16_F4 fColorX3 = Lanczos2_UseLUT(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); + FFX_MIN16_F4 fColorXY = Lanczos2_UseLUT(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); + + // Deringing + + // TODO: only use 4 by checking jitter + const FfxInt32 iDeringingSampleCount = 4; + const FFX_MIN16_F4 fDeringingSamples[4] = { + Samples.fColor11, + Samples.fColor21, + Samples.fColor12, + Samples.fColor22, + }; + + FFX_MIN16_F4 fDeringingMin = fDeringingSamples[0]; + FFX_MIN16_F4 fDeringingMax = fDeringingSamples[0]; + + FFX_UNROLL + for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) + { + fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); + fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); + } + + fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); + + return fColorXY; +} +#endif //FFX_HALF + + + +FfxFloat32x4 Lanczos2Approx(FfxFloat32x4 fColor0, FfxFloat32x4 fColor1, FfxFloat32x4 fColor2, FfxFloat32x4 fColor3, FfxFloat32 t) +{ + FfxFloat32 fWeight0 = Lanczos2ApproxNoClamp(-1.f - t); + FfxFloat32 fWeight1 = Lanczos2ApproxNoClamp(-0.f - t); + FfxFloat32 fWeight2 = Lanczos2ApproxNoClamp(+1.f - t); + FfxFloat32 fWeight3 = Lanczos2ApproxNoClamp(+2.f - t); + return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); +} + +#if FFX_HALF +FFX_MIN16_F4 Lanczos2Approx(FFX_MIN16_F4 fColor0, FFX_MIN16_F4 fColor1, FFX_MIN16_F4 fColor2, FFX_MIN16_F4 fColor3, FFX_MIN16_F t) +{ + FFX_MIN16_F fWeight0 = Lanczos2ApproxNoClamp(FFX_MIN16_F(-1.f) - t); + FFX_MIN16_F fWeight1 = Lanczos2ApproxNoClamp(FFX_MIN16_F(-0.f) - t); + FFX_MIN16_F fWeight2 = Lanczos2ApproxNoClamp(FFX_MIN16_F(+1.f) - t); + FFX_MIN16_F fWeight3 = Lanczos2ApproxNoClamp(FFX_MIN16_F(+2.f) - t); + return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); +} +#endif //FFX_HALF + +FfxFloat32x4 Lanczos2Approx(FetchedBicubicSamples Samples, FfxFloat32x2 fPxFrac) +{ + FfxFloat32x4 fColorX0 = Lanczos2Approx(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); + FfxFloat32x4 fColorX1 = Lanczos2Approx(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); + FfxFloat32x4 fColorX2 = Lanczos2Approx(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); + FfxFloat32x4 fColorX3 = Lanczos2Approx(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); + FfxFloat32x4 fColorXY = Lanczos2Approx(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); + + // Deringing + + // TODO: only use 4 by checking jitter + const FfxInt32 iDeringingSampleCount = 4; + const FfxFloat32x4 fDeringingSamples[4] = { + Samples.fColor11, + Samples.fColor21, + Samples.fColor12, + Samples.fColor22, + }; + + FfxFloat32x4 fDeringingMin = fDeringingSamples[0]; + FfxFloat32x4 fDeringingMax = fDeringingSamples[0]; + + FFX_UNROLL + for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) + { + fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); + fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); + } + + fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); + + return fColorXY; +} + +#if FFX_HALF +FFX_MIN16_F4 Lanczos2Approx(FetchedBicubicSamplesMin16 Samples, FFX_MIN16_F2 fPxFrac) +{ + FFX_MIN16_F4 fColorX0 = Lanczos2Approx(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); + FFX_MIN16_F4 fColorX1 = Lanczos2Approx(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); + FFX_MIN16_F4 fColorX2 = Lanczos2Approx(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); + FFX_MIN16_F4 fColorX3 = Lanczos2Approx(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); + FFX_MIN16_F4 fColorXY = Lanczos2Approx(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); + + // Deringing + + // TODO: only use 4 by checking jitter + const FfxInt32 iDeringingSampleCount = 4; + const FFX_MIN16_F4 fDeringingSamples[4] = { + Samples.fColor11, + Samples.fColor21, + Samples.fColor12, + Samples.fColor22, + }; + + FFX_MIN16_F4 fDeringingMin = fDeringingSamples[0]; + FFX_MIN16_F4 fDeringingMax = fDeringingSamples[0]; + + FFX_UNROLL + for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) + { + fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); + fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); + } + + fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); + + return fColorXY; +} +#endif + +// Clamp by offset direction. Assuming iPxSample is already in range and iPxOffset is compile time constant. +FfxInt32x2 ClampCoord(FfxInt32x2 iPxSample, FfxInt32x2 iPxOffset, FfxInt32x2 iTextureSize) +{ + FfxInt32x2 result = iPxSample + iPxOffset; + result.x = (iPxOffset.x < 0) ? ffxMax(result.x, 0) : result.x; + result.x = (iPxOffset.x > 0) ? ffxMin(result.x, iTextureSize.x - 1) : result.x; + result.y = (iPxOffset.y < 0) ? ffxMax(result.y, 0) : result.y; + result.y = (iPxOffset.y > 0) ? ffxMin(result.y, iTextureSize.y - 1) : result.y; + return result; +} +#if FFX_HALF +FFX_MIN16_I2 ClampCoord(FFX_MIN16_I2 iPxSample, FFX_MIN16_I2 iPxOffset, FFX_MIN16_I2 iTextureSize) +{ + FFX_MIN16_I2 result = iPxSample + iPxOffset; + result.x = (iPxOffset.x < FFX_MIN16_I(0)) ? ffxMax(result.x, FFX_MIN16_I(0)) : result.x; + result.x = (iPxOffset.x > FFX_MIN16_I(0)) ? ffxMin(result.x, iTextureSize.x - FFX_MIN16_I(1)) : result.x; + result.y = (iPxOffset.y < FFX_MIN16_I(0)) ? ffxMax(result.y, FFX_MIN16_I(0)) : result.y; + result.y = (iPxOffset.y > FFX_MIN16_I(0)) ? ffxMin(result.y, iTextureSize.y - FFX_MIN16_I(1)) : result.y; + return result; +} +#endif //FFX_HALF + + +#define DeclareCustomFetchBicubicSamplesWithType(SampleType, TextureType, AddrType, Name, LoadTexture) \ + SampleType Name(AddrType iPxSample, AddrType iTextureSize) \ + { \ + SampleType Samples; \ + \ + Samples.fColor00 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(-1, -1), iTextureSize))); \ + Samples.fColor10 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, -1), iTextureSize))); \ + Samples.fColor20 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, -1), iTextureSize))); \ + Samples.fColor30 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+2, -1), iTextureSize))); \ + \ + Samples.fColor01 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(-1, +0), iTextureSize))); \ + Samples.fColor11 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, +0), iTextureSize))); \ + Samples.fColor21 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, +0), iTextureSize))); \ + Samples.fColor31 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+2, +0), iTextureSize))); \ + \ + Samples.fColor02 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(-1, +1), iTextureSize))); \ + Samples.fColor12 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, +1), iTextureSize))); \ + Samples.fColor22 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, +1), iTextureSize))); \ + Samples.fColor32 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+2, +1), iTextureSize))); \ + \ + Samples.fColor03 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(-1, +2), iTextureSize))); \ + Samples.fColor13 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, +2), iTextureSize))); \ + Samples.fColor23 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, +2), iTextureSize))); \ + Samples.fColor33 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+2, +2), iTextureSize))); \ + \ + return Samples; \ + } + +#define DeclareCustomFetchBicubicSamples(Name, LoadTexture) \ + DeclareCustomFetchBicubicSamplesWithType(FetchedBicubicSamples, FfxFloat32x4, FfxInt32x2, Name, LoadTexture) + +#define DeclareCustomFetchBicubicSamplesMin16(Name, LoadTexture) \ + DeclareCustomFetchBicubicSamplesWithType(FetchedBicubicSamplesMin16, FFX_MIN16_F4, FfxInt32x2, Name, LoadTexture) + +#define DeclareCustomFetchBilinearSamplesWithType(SampleType, TextureType,AddrType, Name, LoadTexture) \ + SampleType Name(AddrType iPxSample, AddrType iTextureSize) \ + { \ + SampleType Samples; \ + Samples.fColor00 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, +0), iTextureSize))); \ + Samples.fColor10 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, +0), iTextureSize))); \ + Samples.fColor01 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, +1), iTextureSize))); \ + Samples.fColor11 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, +1), iTextureSize))); \ + return Samples; \ + } + +#define DeclareCustomFetchBilinearSamples(Name, LoadTexture) \ + DeclareCustomFetchBilinearSamplesWithType(FetchedBilinearSamples, FfxFloat32x4, FfxInt32x2, Name, LoadTexture) + +#define DeclareCustomFetchBilinearSamplesMin16(Name, LoadTexture) \ + DeclareCustomFetchBilinearSamplesWithType(FetchedBilinearSamplesMin16, FFX_MIN16_F4, FfxInt32x2, Name, LoadTexture) + +// BE CAREFUL: there is some precision issues and (3253, 125) leading to (3252.9989778, 125.001102) +// is common, so iPxSample can "jitter" +#define DeclareCustomTextureSample(Name, InterpolateSamples, FetchSamples) \ + FfxFloat32x4 Name(FfxFloat32x2 fUvSample, FfxInt32x2 iTextureSize) \ + { \ + FfxFloat32x2 fPxSample = (fUvSample * FfxFloat32x2(iTextureSize)) - FfxFloat32x2(0.5f, 0.5f); \ + /* Clamp base coords */ \ + fPxSample.x = ffxMax(0.0f, ffxMin(FfxFloat32(iTextureSize.x), fPxSample.x)); \ + fPxSample.y = ffxMax(0.0f, ffxMin(FfxFloat32(iTextureSize.y), fPxSample.y)); \ + /* */ \ + FfxInt32x2 iPxSample = FfxInt32x2(floor(fPxSample)); \ + FfxFloat32x2 fPxFrac = ffxFract(fPxSample); \ + FfxFloat32x4 fColorXY = FfxFloat32x4(InterpolateSamples(FetchSamples(iPxSample, iTextureSize), fPxFrac)); \ + return fColorXY; \ + } + +#define DeclareCustomTextureSampleMin16(Name, InterpolateSamples, FetchSamples) \ + FFX_MIN16_F4 Name(FfxFloat32x2 fUvSample, FfxInt32x2 iTextureSize) \ + { \ + FfxFloat32x2 fPxSample = (fUvSample * FfxFloat32x2(iTextureSize)) - FfxFloat32x2(0.5f, 0.5f); \ + /* Clamp base coords */ \ + fPxSample.x = ffxMax(0.0f, ffxMin(FfxFloat32(iTextureSize.x), fPxSample.x)); \ + fPxSample.y = ffxMax(0.0f, ffxMin(FfxFloat32(iTextureSize.y), fPxSample.y)); \ + /* */ \ + FfxInt32x2 iPxSample = FfxInt32x2(floor(fPxSample)); \ + FFX_MIN16_F2 fPxFrac = FFX_MIN16_F2(ffxFract(fPxSample)); \ + FFX_MIN16_F4 fColorXY = FFX_MIN16_F4(InterpolateSamples(FetchSamples(iPxSample, iTextureSize), fPxFrac)); \ + return fColorXY; \ + } + +#define FFX_FSR2_CONCAT_ID(x, y) x ## y +#define FFX_FSR2_CONCAT(x, y) FFX_FSR2_CONCAT_ID(x, y) +#define FFX_FSR2_SAMPLER_1D_0 Lanczos2 +#define FFX_FSR2_SAMPLER_1D_1 Lanczos2LUT +#define FFX_FSR2_SAMPLER_1D_2 Lanczos2Approx + +#define FFX_FSR2_GET_LANCZOS_SAMPLER1D(x) FFX_FSR2_CONCAT(FFX_FSR2_SAMPLER_1D_, x) + +#endif //!defined( FFX_FSR2_SAMPLE_H ) diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_sample.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_sample.h.meta new file mode 100644 index 0000000..5e24564 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_sample.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 4468441a6f8d1a54c95a33bca10569a2 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_tcr_autogen.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_tcr_autogen.h new file mode 100644 index 0000000..10970ef --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_tcr_autogen.h @@ -0,0 +1,251 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#define USE_YCOCG 1 + +#define fAutogenEpsilon 0.01f + +// EXPERIMENTAL + +FFX_MIN16_F ComputeAutoTC_01(FFX_MIN16_I2 uDispatchThreadId, FFX_MIN16_I2 iPrevIdx) +{ + FfxFloat32x3 colorPreAlpha = LoadOpaqueOnly(uDispatchThreadId); + FfxFloat32x3 colorPostAlpha = LoadInputColor(uDispatchThreadId); + FfxFloat32x3 colorPrevPreAlpha = LoadPrevPreAlpha(iPrevIdx); + FfxFloat32x3 colorPrevPostAlpha = LoadPrevPostAlpha(iPrevIdx); + +#if USE_YCOCG + colorPreAlpha = RGBToYCoCg(colorPreAlpha); + colorPostAlpha = RGBToYCoCg(colorPostAlpha); + colorPrevPreAlpha = RGBToYCoCg(colorPrevPreAlpha); + colorPrevPostAlpha = RGBToYCoCg(colorPrevPostAlpha); +#endif + + FfxFloat32x3 colorDeltaCurr = colorPostAlpha - colorPreAlpha; + FfxFloat32x3 colorDeltaPrev = colorPrevPostAlpha - colorPrevPreAlpha; + bool hasAlpha = any(FFX_GREATER_THAN(abs(colorDeltaCurr), FfxFloat32x3(fAutogenEpsilon, fAutogenEpsilon, fAutogenEpsilon))); + bool hadAlpha = any(FFX_GREATER_THAN(abs(colorDeltaPrev), FfxFloat32x3(fAutogenEpsilon, fAutogenEpsilon, fAutogenEpsilon))); + + FfxFloat32x3 X = colorPreAlpha; + FfxFloat32x3 Y = colorPostAlpha; + FfxFloat32x3 Z = colorPrevPreAlpha; + FfxFloat32x3 W = colorPrevPostAlpha; + + FFX_MIN16_F retVal = FFX_MIN16_F(ffxSaturate(dot(abs(abs(Y - X) - abs(W - Z)), FfxFloat32x3(1, 1, 1)))); + + // cleanup very small values + retVal = (retVal < TcThreshold()) ? FFX_MIN16_F(0.0f) : FFX_MIN16_F(1.f); + + return retVal; +} + +// works ok: thin edges +FFX_MIN16_F ComputeAutoTC_02(FFX_MIN16_I2 uDispatchThreadId, FFX_MIN16_I2 iPrevIdx) +{ + FfxFloat32x3 colorPreAlpha = LoadOpaqueOnly(uDispatchThreadId); + FfxFloat32x3 colorPostAlpha = LoadInputColor(uDispatchThreadId); + FfxFloat32x3 colorPrevPreAlpha = LoadPrevPreAlpha(iPrevIdx); + FfxFloat32x3 colorPrevPostAlpha = LoadPrevPostAlpha(iPrevIdx); + +#if USE_YCOCG + colorPreAlpha = RGBToYCoCg(colorPreAlpha); + colorPostAlpha = RGBToYCoCg(colorPostAlpha); + colorPrevPreAlpha = RGBToYCoCg(colorPrevPreAlpha); + colorPrevPostAlpha = RGBToYCoCg(colorPrevPostAlpha); +#endif + + FfxFloat32x3 colorDelta = colorPostAlpha - colorPreAlpha; + FfxFloat32x3 colorPrevDelta = colorPrevPostAlpha - colorPrevPreAlpha; + bool hasAlpha = any(FFX_GREATER_THAN(abs(colorDelta), FfxFloat32x3(fAutogenEpsilon, fAutogenEpsilon, fAutogenEpsilon))); + bool hadAlpha = any(FFX_GREATER_THAN(abs(colorPrevDelta), FfxFloat32x3(fAutogenEpsilon, fAutogenEpsilon, fAutogenEpsilon))); + + FfxFloat32x3 delta = colorPostAlpha - colorPreAlpha; //prev+1*d = post => d = color, alpha = + FfxFloat32x3 deltaPrev = colorPrevPostAlpha - colorPrevPreAlpha; + + FfxFloat32x3 X = colorPrevPreAlpha; + FfxFloat32x3 N = colorPreAlpha - colorPrevPreAlpha; + FfxFloat32x3 YAminusXA = colorPrevPostAlpha - colorPrevPreAlpha; + FfxFloat32x3 NminusNA = colorPostAlpha - colorPrevPostAlpha; + + FfxFloat32x3 A = (hasAlpha || hadAlpha) ? NminusNA / max(FfxFloat32x3(fAutogenEpsilon, fAutogenEpsilon, fAutogenEpsilon), N) : FfxFloat32x3(0, 0, 0); + + FFX_MIN16_F retVal = FFX_MIN16_F( max(max(A.x, A.y), A.z) ); + + // only pixels that have significantly changed in color shuold be considered + retVal = ffxSaturate(retVal * FFX_MIN16_F(length(colorPostAlpha - colorPrevPostAlpha)) ); + + return retVal; +} + +// This function computes the TransparencyAndComposition mask: +// This mask indicates pixels that should discard locks and apply color clamping. +// +// Typically this is the case for translucent pixels (that don't write depth values) or pixels where the correctness of +// the MVs can not be guaranteed (e.g. procedutal movement or vegetation that does not have MVs to reduce the cost during rasterization) +// Also, large changes in color due to changed lighting should be marked to remove locks on pixels with "old" lighting. +// +// This function takes a opaque only and a final texture and uses internal copies of those textures from the last frame. +// The function tries to determine where the color changes between opaque only and final image to determine the pixels that use transparency. +// Also it uses the previous frames and detects where the use of transparency changed to mark those pixels. +// Additionally it marks pixels where the color changed significantly in the opaque only image, e.g. due to lighting or texture animation. +// +// In the final step it stores the current textures in internal textures for the next frame + +FFX_MIN16_F ComputeTransparencyAndComposition(FFX_MIN16_I2 uDispatchThreadId, FFX_MIN16_I2 iPrevIdx) +{ + FFX_MIN16_F retVal = ComputeAutoTC_02(uDispatchThreadId, iPrevIdx); + + // [branch] + if (retVal > FFX_MIN16_F(0.01f)) + { + retVal = ComputeAutoTC_01(uDispatchThreadId, iPrevIdx); + } + return retVal; +} + +float computeSolidEdge(FFX_MIN16_I2 curPos, FFX_MIN16_I2 prevPos) +{ + float lum[9]; + int i = 0; + for (int y = -1; y < 2; ++y) + { + for (int x = -1; x < 2; ++x) + { + FfxFloat32x3 curCol = LoadOpaqueOnly(curPos + FFX_MIN16_I2(x, y)).rgb; + FfxFloat32x3 prevCol = LoadPrevPreAlpha(prevPos + FFX_MIN16_I2(x, y)).rgb; + lum[i++] = length(curCol - prevCol); + } + } + + //float gradX = abs(lum[3] - lum[4]) + abs(lum[5] - lum[4]); + //float gradY = abs(lum[1] - lum[4]) + abs(lum[7] - lum[4]); + + //return sqrt(gradX * gradX + gradY * gradY); + + float gradX = abs(lum[3] - lum[4]) * abs(lum[5] - lum[4]); + float gradY = abs(lum[1] - lum[4]) * abs(lum[7] - lum[4]); + + return sqrt(sqrt(gradX * gradY)); +} + +float computeAlphaEdge(FFX_MIN16_I2 curPos, FFX_MIN16_I2 prevPos) +{ + float lum[9]; + int i = 0; + for (int y = -1; y < 2; ++y) + { + for (int x = -1; x < 2; ++x) + { + FfxFloat32x3 curCol = abs(LoadInputColor(curPos + FFX_MIN16_I2(x, y)).rgb - LoadOpaqueOnly(curPos + FFX_MIN16_I2(x, y)).rgb); + FfxFloat32x3 prevCol = abs(LoadPrevPostAlpha(prevPos + FFX_MIN16_I2(x, y)).rgb - LoadPrevPreAlpha(prevPos + FFX_MIN16_I2(x, y)).rgb); + lum[i++] = length(curCol - prevCol); + } + } + + //float gradX = abs(lum[3] - lum[4]) + abs(lum[5] - lum[4]); + //float gradY = abs(lum[1] - lum[4]) + abs(lum[7] - lum[4]); + + //return sqrt(gradX * gradX + gradY * gradY); + + float gradX = abs(lum[3] - lum[4]) * abs(lum[5] - lum[4]); + float gradY = abs(lum[1] - lum[4]) * abs(lum[7] - lum[4]); + + return sqrt(sqrt(gradX * gradY)); +} + +FFX_MIN16_F ComputeAabbOverlap(FFX_MIN16_I2 uDispatchThreadId, FFX_MIN16_I2 iPrevIdx) +{ + FFX_MIN16_F retVal = FFX_MIN16_F(0.f); + + FfxFloat32x2 fMotionVector = LoadInputMotionVector(uDispatchThreadId); + FfxFloat32x3 colorPreAlpha = LoadOpaqueOnly(uDispatchThreadId); + FfxFloat32x3 colorPostAlpha = LoadInputColor(uDispatchThreadId); + FfxFloat32x3 colorPrevPreAlpha = LoadPrevPreAlpha(iPrevIdx); + FfxFloat32x3 colorPrevPostAlpha = LoadPrevPostAlpha(iPrevIdx); + +#if USE_YCOCG + colorPreAlpha = RGBToYCoCg(colorPreAlpha); + colorPostAlpha = RGBToYCoCg(colorPostAlpha); + colorPrevPreAlpha = RGBToYCoCg(colorPrevPreAlpha); + colorPrevPostAlpha = RGBToYCoCg(colorPrevPostAlpha); +#endif + FfxFloat32x3 minPrev = FFX_MIN16_F3(+1000.f, +1000.f, +1000.f); + FfxFloat32x3 maxPrev = FFX_MIN16_F3(-1000.f, -1000.f, -1000.f); + for (int y = -1; y < 2; ++y) + { + for (int x = -1; x < 2; ++x) + { + FfxFloat32x3 W = LoadPrevPostAlpha(iPrevIdx + FFX_MIN16_I2(x, y)); + +#if USE_YCOCG + W = RGBToYCoCg(W); +#endif + minPrev = min(minPrev, W); + maxPrev = max(maxPrev, W); + } + } + // instead of computing the overlap: simply count how many samples are outside + // set reactive based on that + FFX_MIN16_F count = FFX_MIN16_F(0.f); + for (int y = -1; y < 2; ++y) + { + for (int x = -1; x < 2; ++x) + { + FfxFloat32x3 Y = LoadInputColor(uDispatchThreadId + FFX_MIN16_I2(x, y)); + +#if USE_YCOCG + Y = RGBToYCoCg(Y); +#endif + count += ((Y.x < minPrev.x) || (Y.x > maxPrev.x)) ? FFX_MIN16_F(1.f) : FFX_MIN16_F(0.f); + count += ((Y.y < minPrev.y) || (Y.y > maxPrev.y)) ? FFX_MIN16_F(1.f) : FFX_MIN16_F(0.f); + count += ((Y.z < minPrev.z) || (Y.z > maxPrev.z)) ? FFX_MIN16_F(1.f) : FFX_MIN16_F(0.f); + } + } + retVal = count / FFX_MIN16_F(27.f); + + return retVal; +} + + +// This function computes the Reactive mask: +// We want pixels marked where the alpha portion of the frame changes a lot between neighbours +// Those pixels are expected to change quickly between frames, too. (e.g. small particles, reflections on curved surfaces...) +// As a result history would not be trustworthy. +// On the other hand we don't want pixels marked where pre-alpha has a large differnce, since those would profit from accumulation +// For mirrors we may assume the pre-alpha is pretty uniform color. +// +// This works well generally, but also marks edge pixels +FFX_MIN16_F ComputeReactive(FFX_MIN16_I2 uDispatchThreadId, FFX_MIN16_I2 iPrevIdx) +{ + // we only get here if alpha has a significant contribution and has changed since last frame. + FFX_MIN16_F retVal = FFX_MIN16_F(0.f); + + // mark pixels with huge variance in alpha as reactive + FFX_MIN16_F alphaEdge = FFX_MIN16_F(computeAlphaEdge(uDispatchThreadId, iPrevIdx)); + FFX_MIN16_F opaqueEdge = FFX_MIN16_F(computeSolidEdge(uDispatchThreadId, iPrevIdx)); + retVal = ffxSaturate(alphaEdge - opaqueEdge); + + // the above also marks edge pixels due to jitter, so we need to cancel those out + + + return retVal; +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_tcr_autogen.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_tcr_autogen.h.meta new file mode 100644 index 0000000..2195de0 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_tcr_autogen.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 8b62ba5e9686a6545b8ed6a8786322d2 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_upsample.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_upsample.h new file mode 100644 index 0000000..9287185 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_upsample.h @@ -0,0 +1,191 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#ifndef FFX_FSR2_UPSAMPLE_H +#define FFX_FSR2_UPSAMPLE_H + +FFX_STATIC const FfxUInt32 iLanczos2SampleCount = 16; + +void Deringing(RectificationBox clippingBox, FFX_PARAMETER_INOUT FfxFloat32x3 fColor) +{ + fColor = clamp(fColor, clippingBox.aabbMin, clippingBox.aabbMax); +} +#if FFX_HALF +void Deringing(RectificationBoxMin16 clippingBox, FFX_PARAMETER_INOUT FFX_MIN16_F3 fColor) +{ + fColor = clamp(fColor, clippingBox.aabbMin, clippingBox.aabbMax); +} +#endif + +#ifndef FFX_FSR2_OPTION_UPSAMPLE_USE_LANCZOS_TYPE +#define FFX_FSR2_OPTION_UPSAMPLE_USE_LANCZOS_TYPE 2 // Approximate +#endif + +FfxFloat32 GetUpsampleLanczosWeight(FfxFloat32x2 fSrcSampleOffset, FfxFloat32 fKernelWeight) +{ + FfxFloat32x2 fSrcSampleOffsetBiased = fSrcSampleOffset * fKernelWeight.xx; +#if FFX_FSR2_OPTION_UPSAMPLE_USE_LANCZOS_TYPE == 0 // LANCZOS_TYPE_REFERENCE + FfxFloat32 fSampleWeight = Lanczos2(length(fSrcSampleOffsetBiased)); +#elif FFX_FSR2_OPTION_UPSAMPLE_USE_LANCZOS_TYPE == 1 // LANCZOS_TYPE_LUT + FfxFloat32 fSampleWeight = Lanczos2_UseLUT(length(fSrcSampleOffsetBiased)); +#elif FFX_FSR2_OPTION_UPSAMPLE_USE_LANCZOS_TYPE == 2 // LANCZOS_TYPE_APPROXIMATE + FfxFloat32 fSampleWeight = Lanczos2ApproxSq(dot(fSrcSampleOffsetBiased, fSrcSampleOffsetBiased)); +#else +#error "Invalid Lanczos type" +#endif + return fSampleWeight; +} + +#if FFX_HALF +FFX_MIN16_F GetUpsampleLanczosWeight(FFX_MIN16_F2 fSrcSampleOffset, FFX_MIN16_F fKernelWeight) +{ + FFX_MIN16_F2 fSrcSampleOffsetBiased = fSrcSampleOffset * fKernelWeight.xx; +#if FFX_FSR2_OPTION_UPSAMPLE_USE_LANCZOS_TYPE == 0 // LANCZOS_TYPE_REFERENCE + FFX_MIN16_F fSampleWeight = Lanczos2(length(fSrcSampleOffsetBiased)); +#elif FFX_FSR2_OPTION_UPSAMPLE_USE_LANCZOS_TYPE == 1 // LANCZOS_TYPE_LUT + FFX_MIN16_F fSampleWeight = Lanczos2_UseLUT(length(fSrcSampleOffsetBiased)); +#elif FFX_FSR2_OPTION_UPSAMPLE_USE_LANCZOS_TYPE == 2 // LANCZOS_TYPE_APPROXIMATE + FFX_MIN16_F fSampleWeight = Lanczos2ApproxSq(dot(fSrcSampleOffsetBiased, fSrcSampleOffsetBiased)); + + // To Test: Save reciproqual sqrt compute + // FfxFloat32 fSampleWeight = Lanczos2Sq_UseLUT(dot(fSrcSampleOffsetBiased, fSrcSampleOffsetBiased)); +#else +#error "Invalid Lanczos type" +#endif + return fSampleWeight; +} +#endif + +FfxFloat32 ComputeMaxKernelWeight() { + const FfxFloat32 fKernelSizeBias = 1.0f; + + FfxFloat32 fKernelWeight = FfxFloat32(1) + (FfxFloat32(1.0f) / FfxFloat32x2(DownscaleFactor()) - FfxFloat32(1)).x * FfxFloat32(fKernelSizeBias); + + return ffxMin(FfxFloat32(1.99f), fKernelWeight); +} + +FfxFloat32x4 ComputeUpsampledColorAndWeight(const AccumulationPassCommonParams params, + FFX_PARAMETER_INOUT RectificationBox clippingBox, FfxFloat32 fReactiveFactor) +{ + #if FFX_FSR2_OPTION_UPSAMPLE_SAMPLERS_USE_DATA_HALF && FFX_HALF + #include "ffx_fsr2_force16_begin.h" + #endif + // We compute a sliced lanczos filter with 2 lobes (other slices are accumulated temporaly) + FfxFloat32x2 fDstOutputPos = FfxFloat32x2(params.iPxHrPos) + FFX_BROADCAST_FLOAT32X2(0.5f); // Destination resolution output pixel center position + FfxFloat32x2 fSrcOutputPos = fDstOutputPos * DownscaleFactor(); // Source resolution output pixel center position + FfxInt32x2 iSrcInputPos = FfxInt32x2(floor(fSrcOutputPos)); // TODO: what about weird upscale factors... + + #if FFX_FSR2_OPTION_UPSAMPLE_SAMPLERS_USE_DATA_HALF && FFX_HALF + #include "ffx_fsr2_force16_end.h" + #endif + + FfxFloat32x3 fSamples[iLanczos2SampleCount]; + + FfxFloat32x2 fSrcUnjitteredPos = (FfxFloat32x2(iSrcInputPos) + FfxFloat32x2(0.5f, 0.5f)) - Jitter(); // This is the un-jittered position of the sample at offset 0,0 + + FfxInt32x2 offsetTL; + offsetTL.x = (fSrcUnjitteredPos.x > fSrcOutputPos.x) ? FfxInt32(-2) : FfxInt32(-1); + offsetTL.y = (fSrcUnjitteredPos.y > fSrcOutputPos.y) ? FfxInt32(-2) : FfxInt32(-1); + + //Load samples + // If fSrcUnjitteredPos.y > fSrcOutputPos.y, indicates offsetTL.y = -2, sample offset Y will be [-2, 1], clipbox will be rows [1, 3]. + // Flip row# for sampling offset in this case, so first 0~2 rows in the sampled array can always be used for computing the clipbox. + // This reduces branch or cmove on sampled colors, but moving this overhead to sample position / weight calculation time which apply to less values. + const FfxBoolean bFlipRow = fSrcUnjitteredPos.y > fSrcOutputPos.y; + const FfxBoolean bFlipCol = fSrcUnjitteredPos.x > fSrcOutputPos.x; + + FfxFloat32x2 fOffsetTL = FfxFloat32x2(offsetTL); + + FFX_UNROLL + for (FfxInt32 row = 0; row < 3; row++) { + + FFX_UNROLL + for (FfxInt32 col = 0; col < 3; col++) { + FfxInt32 iSampleIndex = col + (row << 2); + + FfxInt32x2 sampleColRow = FfxInt32x2(bFlipCol ? (3 - col) : col, bFlipRow ? (3 - row) : row); + FfxInt32x2 iSrcSamplePos = FfxInt32x2(iSrcInputPos) + offsetTL + sampleColRow; + + const FfxInt32x2 sampleCoord = ClampLoad(iSrcSamplePos, FfxInt32x2(0, 0), FfxInt32x2(RenderSize())); + + fSamples[iSampleIndex] = LoadPreparedInputColor(FfxInt32x2(sampleCoord)); + } + } + + FfxFloat32x4 fColorAndWeight = FfxFloat32x4(0.0f, 0.0f, 0.0f, 0.0f); + + FfxFloat32x2 fBaseSampleOffset = FfxFloat32x2(fSrcUnjitteredPos - fSrcOutputPos); + + // Identify how much of each upsampled color to be used for this frame + const FfxFloat32 fKernelReactiveFactor = ffxMax(fReactiveFactor, FfxFloat32(params.bIsNewSample)); + const FfxFloat32 fKernelBiasMax = ComputeMaxKernelWeight() * (1.0f - fKernelReactiveFactor); + + const FfxFloat32 fKernelBiasMin = ffxMax(1.0f, ((1.0f + fKernelBiasMax) * 0.3f)); + const FfxFloat32 fKernelBiasFactor = ffxMax(0.0f, ffxMax(0.25f * params.fDepthClipFactor, fKernelReactiveFactor)); + const FfxFloat32 fKernelBias = ffxLerp(fKernelBiasMax, fKernelBiasMin, fKernelBiasFactor); + + const FfxFloat32 fRectificationCurveBias = ffxLerp(-2.0f, -3.0f, ffxSaturate(params.fHrVelocity / 50.0f)); + + FFX_UNROLL + for (FfxInt32 row = 0; row < 3; row++) { + FFX_UNROLL + for (FfxInt32 col = 0; col < 3; col++) { + FfxInt32 iSampleIndex = col + (row << 2); + + const FfxInt32x2 sampleColRow = FfxInt32x2(bFlipCol ? (3 - col) : col, bFlipRow ? (3 - row) : row); + const FfxFloat32x2 fOffset = fOffsetTL + FfxFloat32x2(sampleColRow); + FfxFloat32x2 fSrcSampleOffset = fBaseSampleOffset + fOffset; + + FfxInt32x2 iSrcSamplePos = FfxInt32x2(iSrcInputPos) + FfxInt32x2(offsetTL) + sampleColRow; + + const FfxFloat32 fOnScreenFactor = FfxFloat32(IsOnScreen(FfxInt32x2(iSrcSamplePos), FfxInt32x2(RenderSize()))); + FfxFloat32 fSampleWeight = fOnScreenFactor * FfxFloat32(GetUpsampleLanczosWeight(fSrcSampleOffset, fKernelBias)); + + fColorAndWeight += FfxFloat32x4(fSamples[iSampleIndex] * fSampleWeight, fSampleWeight); + + // Update rectification box + { + const FfxFloat32 fSrcSampleOffsetSq = dot(fSrcSampleOffset, fSrcSampleOffset); + const FfxFloat32 fBoxSampleWeight = exp(fRectificationCurveBias * fSrcSampleOffsetSq); + + const FfxBoolean bInitialSample = (row == 0) && (col == 0); + RectificationBoxAddSample(bInitialSample, clippingBox, fSamples[iSampleIndex], fBoxSampleWeight); + } + } + } + + RectificationBoxComputeVarianceBoxData(clippingBox); + + fColorAndWeight.w *= FfxFloat32(fColorAndWeight.w > FSR2_EPSILON); + + if (fColorAndWeight.w > FSR2_EPSILON) { + // Normalize for deringing (we need to compare colors) + fColorAndWeight.xyz = fColorAndWeight.xyz / fColorAndWeight.w; + fColorAndWeight.w *= fUpsampleLanczosWeightScale; + + Deringing(clippingBox, fColorAndWeight.xyz); + } + + return fColorAndWeight; +} + +#endif //!defined( FFX_FSR2_UPSAMPLE_H ) diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_upsample.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_upsample.h.meta new file mode 100644 index 0000000..5c0e011 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr2/ffx_fsr2_upsample.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 02358daf003aa3d408e0922790429182 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler.meta new file mode 100644 index 0000000..b538189 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: 2c8d127ec4826a54cae3fe1e7e2c17b8 +folderAsset: yes +DefaultImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_accumulate.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_accumulate.h new file mode 100644 index 0000000..766cba3 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_accumulate.h @@ -0,0 +1,172 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +void Accumulate(const AccumulationPassCommonParams params, FFX_PARAMETER_INOUT AccumulationPassData data) +{ + // Avoid invalid values when accumulation and upsampled weight is 0 + data.fHistoryWeight *= FfxFloat32(data.fHistoryWeight > FSR3UPSCALER_FP16_MIN); + data.fHistoryWeight = ffxMax(FSR3UPSCALER_EPSILON, data.fHistoryWeight + data.fUpsampledWeight); + +#if FFX_FSR3UPSCALER_OPTION_HDR_COLOR_INPUT + //YCoCg -> RGB -> Tonemap -> YCoCg (Use RGB tonemapper to avoid color desaturation) + data.fUpsampledColor = RGBToYCoCg(Tonemap(YCoCgToRGB(data.fUpsampledColor))); + data.fHistoryColor = RGBToYCoCg(Tonemap(YCoCgToRGB(data.fHistoryColor))); +#endif + + const FfxFloat32 fAlpha = ffxSaturate(data.fUpsampledWeight / data.fHistoryWeight); + data.fHistoryColor = ffxLerp(data.fHistoryColor, data.fUpsampledColor, fAlpha); + data.fHistoryColor = YCoCgToRGB(data.fHistoryColor); + +#if FFX_FSR3UPSCALER_OPTION_HDR_COLOR_INPUT + data.fHistoryColor = InverseTonemap(data.fHistoryColor); +#endif +} + +void RectifyHistory( + const AccumulationPassCommonParams params, + FFX_PARAMETER_INOUT AccumulationPassData data +) +{ + const FfxFloat32 fVecolityFactor = ffxSaturate(params.f4KVelocity / 20.0f); + const FfxFloat32 fDistanceFactor = ffxSaturate(0.75f - params.fFarthestDepthInMeters / 20.0f); + const FfxFloat32 fAccumulationFactor = 1.0f - params.fAccumulation; + const FfxFloat32 fReactiveFactor = ffxPow(params.fReactiveMask, 1.0f / 2.0f); + const FfxFloat32 fShadingChangeFactor = params.fShadingChange; + const FfxFloat32 fBoxScaleT = ffxMax(fVecolityFactor, ffxMax(fDistanceFactor, ffxMax(fAccumulationFactor, ffxMax(fReactiveFactor, fShadingChangeFactor)))); + + const FfxFloat32 fBoxScale = ffxLerp(3.0f, 1.0f, fBoxScaleT); + + const FfxFloat32x3 fScaledBoxVec = data.clippingBox.boxVec * fBoxScale; + const FfxFloat32x3 fBoxMin = data.clippingBox.boxCenter - fScaledBoxVec; + const FfxFloat32x3 fBoxMax = data.clippingBox.boxCenter + fScaledBoxVec; + + if (any(FFX_GREATER_THAN(fBoxMin, data.fHistoryColor)) || any(FFX_GREATER_THAN(data.fHistoryColor, fBoxMax))) { + + const FfxFloat32x3 fClampedHistoryColor = clamp(data.fHistoryColor, fBoxMin, fBoxMax); + + const FfxFloat32 fHistoryContribution = ffxMax(params.fLumaInstabilityFactor, data.fLockContributionThisFrame) * params.fAccumulation * (1 - params.fDisocclusion); + + // Scale history color using rectification info, also using accumulation mask to avoid potential invalid color protection + data.fHistoryColor = ffxLerp(fClampedHistoryColor, data.fHistoryColor, ffxSaturate(fHistoryContribution)); + } +} + +void UpdateLockStatus(AccumulationPassCommonParams params, FFX_PARAMETER_INOUT AccumulationPassData data) +{ + data.fLock *= FfxFloat32(params.bIsNewSample == false); + + const FfxFloat32 fLifetimeDecreaseFactor = ffxMax(ffxSaturate(params.fShadingChange), ffxMax(params.fReactiveMask, params.fDisocclusion)); + data.fLock = ffxMax(0.0f, data.fLock - fLifetimeDecreaseFactor * fLockMax); + + // Compute this frame lock contribution + data.fLockContributionThisFrame = ffxSaturate(ffxSaturate(data.fLock - fLockThreshold) * (fLockMax - fLockThreshold)); + + const FfxFloat32 fNewLockIntensity = LoadRwNewLocks(params.iPxHrPos) * (1.0f - ffxMax(params.fShadingChange * 0, params.fReactiveMask)); + data.fLock = ffxMax(0.0f, ffxMin(data.fLock + fNewLockIntensity, fLockMax)); + + // Preparing for next frame + const FfxFloat32 fLifetimeDecrease = (0.1f / JitterSequenceLength()) * (1.0f - fLifetimeDecreaseFactor); + data.fLock = ffxMax(0.0f, data.fLock - fLifetimeDecrease); + + // we expect similar motion for next frame + // kill lock if that location is outside screen, avoid locks to be clamped to screen borders + const FfxFloat32x2 fEstimatedUvNextFrame = params.fHrUv - params.fMotionVector; + data.fLock *= FfxFloat32(IsUvInside(fEstimatedUvNextFrame) == true); +} + +void ComputeBaseAccumulationWeight(const AccumulationPassCommonParams params, FFX_PARAMETER_INOUT AccumulationPassData data) +{ + FfxFloat32 fBaseAccumulation = params.fAccumulation; + + fBaseAccumulation = ffxMin(fBaseAccumulation, ffxLerp(fBaseAccumulation, 0.15f, ffxSaturate(ffxMax(0.0f, params.f4KVelocity / 0.5f)))); + + data.fHistoryWeight = fBaseAccumulation; +} + +void InitPassData(FfxInt32x2 iPxHrPos, FFX_PARAMETER_INOUT AccumulationPassCommonParams params, FFX_PARAMETER_INOUT AccumulationPassData data) +{ + // Init constant params + params.iPxHrPos = iPxHrPos; + const FfxFloat32x2 fHrUv = (iPxHrPos + 0.5f) / UpscaleSize(); + params.fHrUv = fHrUv; + params.fLrUvJittered = fHrUv + Jitter() / RenderSize(); + params.fLrUv_HwSampler = ClampUv(params.fLrUvJittered, RenderSize(), MaxRenderSize()); + + params.fMotionVector = GetMotionVector(iPxHrPos, fHrUv); + params.f4KVelocity = Get4KVelocity(params.fMotionVector); + + ComputeReprojectedUVs(params, params.fReprojectedHrUv, params.bIsExistingSample); + + const FfxFloat32x2 fLumaInstabilityUv_HW = ClampUv(fHrUv, RenderSize(), MaxRenderSize()); + params.fLumaInstabilityFactor = SampleLumaInstability(fLumaInstabilityUv_HW); + + const FfxFloat32x2 fFarthestDepthUv = ClampUv(params.fLrUvJittered, RenderSize() / 2, GetFarthestDepthMip1ResourceDimensions()); + params.fFarthestDepthInMeters = SampleFarthestDepthMip1(fFarthestDepthUv); + params.bIsNewSample = (params.bIsExistingSample == false || 0 == FrameIndex()); + + const FfxFloat32x4 fDilatedReactiveMasks = SampleDilatedReactiveMasks(params.fLrUv_HwSampler); + params.fReactiveMask = ffxSaturate(fDilatedReactiveMasks[REACTIVE]); + params.fDisocclusion = ffxSaturate(fDilatedReactiveMasks[DISOCCLUSION]); + params.fShadingChange = ffxSaturate(fDilatedReactiveMasks[SHADING_CHANGE]); + params.fAccumulation = ffxSaturate(fDilatedReactiveMasks[ACCUMULAION]); + params.fAccumulation *= FfxFloat32(round(params.fAccumulation * 100.0f) > 1.0f); + + // Init variable data + data = (AccumulationPassData)0; + data.fUpsampledColor = FfxFloat32x3(0.0f, 0.0f, 0.0f); + data.fHistoryColor = FfxFloat32x3(0.0f, 0.0f, 0.0f); + data.fHistoryWeight = 1.0f; + data.fUpsampledWeight = 0.0f; + data.fLock = 0.0f; + data.fLockContributionThisFrame = 0.0f; +} + +void Accumulate(FfxInt32x2 iPxHrPos) +{ + AccumulationPassCommonParams params; + AccumulationPassData data; + InitPassData(iPxHrPos, params, data); + + if (params.bIsExistingSample && !params.bIsNewSample) { + ReprojectHistoryColor(params, data); + } + + UpdateLockStatus(params, data); + + ComputeBaseAccumulationWeight(params, data); + + ComputeUpsampledColorAndWeight(params, data); + + RectifyHistory(params, data); + + Accumulate(params, data); + + data.fHistoryColor /= Exposure(); + + StoreInternalColorAndWeight(iPxHrPos, FfxFloat32x4(data.fHistoryColor, data.fLock)); + + // Output final color when RCAS is disabled +#if FFX_FSR3UPSCALER_OPTION_APPLY_SHARPENING == 0 + StoreUpscaledOutput(iPxHrPos, data.fHistoryColor); +#endif + StoreNewLocks(iPxHrPos, 0); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_accumulate.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_accumulate.h.meta new file mode 100644 index 0000000..d84af26 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_accumulate.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: cab06bfc45670ef42915e0e483e2c823 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h new file mode 100644 index 0000000..1c3fc99 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h @@ -0,0 +1,980 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#include "ffx_fsr3upscaler_resources.h" + +#if defined(FFX_GPU) +#ifdef __hlsl_dx_compiler +#pragma dxc diagnostic push +#pragma dxc diagnostic ignored "-Wambig-lit-shift" +#endif //__hlsl_dx_compiler +#include "../ffx_core.h" +#ifdef __hlsl_dx_compiler +#pragma dxc diagnostic pop +#endif //__hlsl_dx_compiler +#endif // #if defined(FFX_GPU) + +#if defined(FFX_GPU) +#ifndef FFX_PREFER_WAVE64 +#define FFX_PREFER_WAVE64 +#endif // FFX_PREFER_WAVE64 + +#pragma warning(disable: 3205) // conversion from larger type to smaller + +#define DECLARE_SRV_REGISTER(regIndex) t##regIndex +#define DECLARE_UAV_REGISTER(regIndex) u##regIndex +#define DECLARE_CB_REGISTER(regIndex) b##regIndex +#define FFX_FSR3UPSCALER_DECLARE_SRV(regIndex) register(DECLARE_SRV_REGISTER(regIndex)) +#define FFX_FSR3UPSCALER_DECLARE_UAV(regIndex) register(DECLARE_UAV_REGISTER(regIndex)) +#define FFX_FSR3UPSCALER_DECLARE_CB(regIndex) register(DECLARE_CB_REGISTER(regIndex)) + +#if defined(FSR3UPSCALER_BIND_CB_FSR3UPSCALER) +cbuffer cbFSR3Upscaler : FFX_FSR3UPSCALER_DECLARE_CB(FSR3UPSCALER_BIND_CB_FSR3UPSCALER) +{ + FfxInt32x2 iRenderSize; + FfxInt32x2 iPreviousFrameRenderSize; + + FfxInt32x2 iUpscaleSize; + FfxInt32x2 iPreviousFrameUpscaleSize; + + FfxInt32x2 iMaxRenderSize; + FfxInt32x2 iMaxUpscaleSize; + + FfxFloat32x4 fDeviceToViewDepth; + + FfxFloat32x2 fJitter; + FfxFloat32x2 fPreviousFrameJitter; + + FfxFloat32x2 fMotionVectorScale; + FfxFloat32x2 fDownscaleFactor; + + FfxFloat32x2 fMotionVectorJitterCancellation; + FfxFloat32 fTanHalfFOV; + FfxFloat32 fJitterSequenceLength; + + FfxFloat32 fDeltaTime; + FfxFloat32 fDeltaPreExposure; + FfxFloat32 fViewSpaceToMetersFactor; + FfxFloat32 fFrameIndex; +}; + +#define FFX_FSR3UPSCALER_CONSTANT_BUFFER_1_SIZE (sizeof(cbFSR3Upscaler) / 4) // Number of 32-bit values. This must be kept in sync with the cbFSR3Upscaler size. + +/* Define getter functions in the order they are defined in the CB! */ +FfxInt32x2 RenderSize() +{ + return iRenderSize; +} + +FfxInt32x2 PreviousFrameRenderSize() +{ + return iPreviousFrameRenderSize; +} + +FfxInt32x2 MaxRenderSize() +{ + return iMaxRenderSize; +} + +FfxInt32x2 UpscaleSize() +{ + return iUpscaleSize; +} + +FfxInt32x2 PreviousFrameUpscaleSize() +{ + return iPreviousFrameUpscaleSize; +} + +FfxInt32x2 MaxUpscaleSize() +{ + return iMaxUpscaleSize; +} + +FfxFloat32x2 Jitter() +{ + return fJitter; +} + +FfxFloat32x2 PreviousFrameJitter() +{ + return fPreviousFrameJitter; +} + +FfxFloat32x4 DeviceToViewSpaceTransformFactors() +{ + return fDeviceToViewDepth; +} + +FfxFloat32x2 MotionVectorScale() +{ + return fMotionVectorScale; +} + +FfxFloat32x2 DownscaleFactor() +{ + return fDownscaleFactor; +} + +FfxFloat32x2 MotionVectorJitterCancellation() +{ + return fMotionVectorJitterCancellation; +} + +FfxFloat32 TanHalfFoV() +{ + return fTanHalfFOV; +} + +FfxFloat32 JitterSequenceLength() +{ + return fJitterSequenceLength; +} + +FfxFloat32 DeltaTime() +{ + return fDeltaTime; +} + +FfxFloat32 DeltaPreExposure() +{ + return fDeltaPreExposure; +} + +FfxFloat32 ViewSpaceToMetersFactor() +{ + return fViewSpaceToMetersFactor; +} + +FfxFloat32 FrameIndex() +{ + return fFrameIndex; +} + +#endif // #if defined(FSR3UPSCALER_BIND_CB_FSR3UPSCALER) + +#define FFX_FSR3UPSCALER_ROOTSIG_STRINGIFY(p) FFX_FSR3UPSCALER_ROOTSIG_STR(p) +#define FFX_FSR3UPSCALER_ROOTSIG_STR(p) #p +#define FFX_FSR3UPSCALER_ROOTSIG [RootSignature( "DescriptorTable(UAV(u0, numDescriptors = " FFX_FSR3UPSCALER_ROOTSIG_STRINGIFY(FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_COUNT) ")), " \ + "DescriptorTable(SRV(t0, numDescriptors = " FFX_FSR3UPSCALER_ROOTSIG_STRINGIFY(FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_COUNT) ")), " \ + "CBV(b0), " \ + "StaticSampler(s0, filter = FILTER_MIN_MAG_MIP_POINT, " \ + "addressU = TEXTURE_ADDRESS_CLAMP, " \ + "addressV = TEXTURE_ADDRESS_CLAMP, " \ + "addressW = TEXTURE_ADDRESS_CLAMP, " \ + "comparisonFunc = COMPARISON_NEVER, " \ + "borderColor = STATIC_BORDER_COLOR_TRANSPARENT_BLACK), " \ + "StaticSampler(s1, filter = FILTER_MIN_MAG_MIP_LINEAR, " \ + "addressU = TEXTURE_ADDRESS_CLAMP, " \ + "addressV = TEXTURE_ADDRESS_CLAMP, " \ + "addressW = TEXTURE_ADDRESS_CLAMP, " \ + "comparisonFunc = COMPARISON_NEVER, " \ + "borderColor = STATIC_BORDER_COLOR_TRANSPARENT_BLACK)" )] + +#define FFX_FSR3UPSCALER_CONSTANT_BUFFER_2_SIZE 6 // Number of 32-bit values. This must be kept in sync with max( cbRCAS , cbSPD) size. + +#define FFX_FSR3UPSCALER_CB2_ROOTSIG [RootSignature( "DescriptorTable(UAV(u0, numDescriptors = " FFX_FSR3UPSCALER_ROOTSIG_STRINGIFY(FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_COUNT) ")), " \ + "DescriptorTable(SRV(t0, numDescriptors = " FFX_FSR3UPSCALER_ROOTSIG_STRINGIFY(FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_COUNT) ")), " \ + "CBV(b0), " \ + "CBV(b1), " \ + "StaticSampler(s0, filter = FILTER_MIN_MAG_MIP_POINT, " \ + "addressU = TEXTURE_ADDRESS_CLAMP, " \ + "addressV = TEXTURE_ADDRESS_CLAMP, " \ + "addressW = TEXTURE_ADDRESS_CLAMP, " \ + "comparisonFunc = COMPARISON_NEVER, " \ + "borderColor = STATIC_BORDER_COLOR_TRANSPARENT_BLACK), " \ + "StaticSampler(s1, filter = FILTER_MIN_MAG_MIP_LINEAR, " \ + "addressU = TEXTURE_ADDRESS_CLAMP, " \ + "addressV = TEXTURE_ADDRESS_CLAMP, " \ + "addressW = TEXTURE_ADDRESS_CLAMP, " \ + "comparisonFunc = COMPARISON_NEVER, " \ + "borderColor = STATIC_BORDER_COLOR_TRANSPARENT_BLACK)" )] +#if defined(FFX_FSR3UPSCALER_EMBED_ROOTSIG) +#define FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT FFX_FSR3UPSCALER_ROOTSIG +#define FFX_FSR3UPSCALER_EMBED_CB2_ROOTSIG_CONTENT FFX_FSR3UPSCALER_CB2_ROOTSIG +#else +#define FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT +#define FFX_FSR3UPSCALER_EMBED_CB2_ROOTSIG_CONTENT +#endif // #if FFX_FSR3UPSCALER_EMBED_ROOTSIG + +#if defined(FSR3UPSCALER_BIND_CB_AUTOREACTIVE) +cbuffer cbGenerateReactive : FFX_FSR3UPSCALER_DECLARE_CB(FSR3UPSCALER_BIND_CB_AUTOREACTIVE) +{ + FfxFloat32 fTcThreshold; // 0.1 is a good starting value, lower will result in more TC pixels + FfxFloat32 fTcScale; + FfxFloat32 fReactiveScale; + FfxFloat32 fReactiveMax; +}; + +FfxFloat32 TcThreshold() +{ + return fTcThreshold; +} + +FfxFloat32 TcScale() +{ + return fTcScale; +} + +FfxFloat32 ReactiveScale() +{ + return fReactiveScale; +} + +FfxFloat32 ReactiveMax() +{ + return fReactiveMax; +} +#endif // #if defined(FSR3UPSCALER_BIND_CB_AUTOREACTIVE) + +#if defined(FSR3UPSCALER_BIND_CB_RCAS) +cbuffer cbRCAS : FFX_FSR3UPSCALER_DECLARE_CB(FSR3UPSCALER_BIND_CB_RCAS) +{ + FfxUInt32x4 rcasConfig; +}; + +FfxUInt32x4 RCASConfig() +{ + return rcasConfig; +} +#endif // #if defined(FSR3UPSCALER_BIND_CB_RCAS) + + +#if defined(FSR3UPSCALER_BIND_CB_REACTIVE) +cbuffer cbGenerateReactive : FFX_FSR3UPSCALER_DECLARE_CB(FSR3UPSCALER_BIND_CB_REACTIVE) +{ + FfxFloat32 gen_reactive_scale; + FfxFloat32 gen_reactive_threshold; + FfxFloat32 gen_reactive_binaryValue; + FfxUInt32 gen_reactive_flags; +}; + +FfxFloat32 GenReactiveScale() +{ + return gen_reactive_scale; +} + +FfxFloat32 GenReactiveThreshold() +{ + return gen_reactive_threshold; +} + +FfxFloat32 GenReactiveBinaryValue() +{ + return gen_reactive_binaryValue; +} + +FfxUInt32 GenReactiveFlags() +{ + return gen_reactive_flags; +} +#endif // #if defined(FSR3UPSCALER_BIND_CB_REACTIVE) + +#if defined(FSR3UPSCALER_BIND_CB_SPD) +cbuffer cbSPD : FFX_FSR3UPSCALER_DECLARE_CB(FSR3UPSCALER_BIND_CB_SPD) { + + FfxUInt32 mips; + FfxUInt32 numWorkGroups; + FfxUInt32x2 workGroupOffset; + FfxUInt32x2 renderSize; +}; + +FfxUInt32 MipCount() +{ + return mips; +} + +FfxUInt32 NumWorkGroups() +{ + return numWorkGroups; +} + +FfxUInt32x2 WorkGroupOffset() +{ + return workGroupOffset; +} + +FfxUInt32x2 SPD_RenderSize() +{ + return renderSize; +} +#endif // #if defined(FSR3UPSCALER_BIND_CB_SPD) + +// Declare and sample camera buffers as regular textures, unless overridden +#if !defined(UNITY_FSR_TEX2D) +#define UNITY_FSR_TEX2D(type) Texture2D +#endif +#if !defined(UNITY_FSR_RWTEX2D) +#define UNITY_FSR_RWTEX2D(type) RWTexture2D +#endif +#if !defined(UNITY_FSR_POS) +#define UNITY_FSR_POS(pxPos) (pxPos) +#endif +#if !defined(UNITY_FSR_UV) +#define UNITY_FSR_UV(uv) (uv) +#endif +#if !defined(UNITY_FSR_GETDIMS) +#define UNITY_FSR_GETDIMS(tex, w, h) (tex).GetDimensions((w), (h)) +#endif + +SamplerState s_PointClamp : register(s0); +SamplerState s_LinearClamp : register(s1); + +#if defined(FSR3UPSCALER_BIND_SRV_SPD_MIPS) +Texture2D r_spd_mips : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_SPD_MIPS); + +FfxInt32x2 GetSPDMipDimensions(FfxUInt32 uMipLevel) +{ + FfxUInt32 uWidth; + FfxUInt32 uHeight; + FfxUInt32 uLevels; + r_spd_mips.GetDimensions(uMipLevel, uWidth, uHeight, uLevels); + + return FfxInt32x2(uWidth, uHeight); +} + +FfxFloat32x2 SampleSPDMipLevel(FfxFloat32x2 fUV, FfxUInt32 mipLevel) +{ + return r_spd_mips.SampleLevel(s_LinearClamp, fUV, mipLevel); +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_INPUT_DEPTH) +UNITY_FSR_TEX2D(FfxFloat32) r_input_depth : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_INPUT_DEPTH); + +FfxFloat32 LoadInputDepth(FfxUInt32x2 iPxPos) +{ + return r_input_depth[UNITY_FSR_POS(iPxPos)]; +} + +FfxFloat32 SampleInputDepth(FfxFloat32x2 fUV) +{ + return r_input_depth.SampleLevel(s_LinearClamp, UNITY_FSR_UV(fUV), 0).x; +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_REACTIVE_MASK) +UNITY_FSR_TEX2D(FfxFloat32) r_reactive_mask : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_REACTIVE_MASK); + +FfxFloat32 LoadReactiveMask(FfxUInt32x2 iPxPos) +{ + return r_reactive_mask[UNITY_FSR_POS(iPxPos)]; +} + +FfxInt32x2 GetReactiveMaskResourceDimensions() +{ + FfxUInt32 uWidth; + FfxUInt32 uHeight; + UNITY_FSR_GETDIMS(r_reactive_mask, uWidth, uHeight); + + return FfxInt32x2(uWidth, uHeight); +} + +FfxFloat32 SampleReactiveMask(FfxFloat32x2 fUV) +{ + return r_reactive_mask.SampleLevel(s_LinearClamp, UNITY_FSR_UV(fUV), 0).x; +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_TRANSPARENCY_AND_COMPOSITION_MASK) +UNITY_FSR_TEX2D(FfxFloat32) r_transparency_and_composition_mask : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_TRANSPARENCY_AND_COMPOSITION_MASK); + +FfxFloat32 LoadTransparencyAndCompositionMask(FfxUInt32x2 iPxPos) +{ + return r_transparency_and_composition_mask[UNITY_FSR_POS(iPxPos)]; +} + +FfxInt32x2 GetTransparencyAndCompositionMaskResourceDimensions() +{ + FfxUInt32 uWidth; + FfxUInt32 uHeight; + UNITY_FSR_GETDIMS(r_transparency_and_composition_mask, uWidth, uHeight); + + return FfxInt32x2(uWidth, uHeight); +} + +FfxFloat32 SampleTransparencyAndCompositionMask(FfxFloat32x2 fUV) +{ + return r_transparency_and_composition_mask.SampleLevel(s_LinearClamp, UNITY_FSR_UV(fUV), 0).x; +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_INPUT_COLOR) +UNITY_FSR_TEX2D(FfxFloat32x4) r_input_color_jittered : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_INPUT_COLOR); + +FfxFloat32x3 LoadInputColor(FfxUInt32x2 iPxPos) +{ + return r_input_color_jittered[UNITY_FSR_POS(iPxPos)].rgb; +} + +FfxFloat32x3 SampleInputColor(FfxFloat32x2 fUV) +{ + return r_input_color_jittered.SampleLevel(s_LinearClamp, UNITY_FSR_UV(fUV), 0).rgb; +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_INPUT_MOTION_VECTORS) +UNITY_FSR_TEX2D(FfxFloat32x4) r_input_motion_vectors : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_INPUT_MOTION_VECTORS); + +FfxFloat32x2 LoadInputMotionVector(FfxUInt32x2 iPxDilatedMotionVectorPos) +{ + FfxFloat32x2 fSrcMotionVector = r_input_motion_vectors[UNITY_FSR_POS(iPxDilatedMotionVectorPos)].xy; + + FfxFloat32x2 fUvMotionVector = fSrcMotionVector * MotionVectorScale(); + +#if FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS + fUvMotionVector -= MotionVectorJitterCancellation(); +#endif + + return fUvMotionVector; +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_INTERNAL_UPSCALED) +Texture2D r_internal_upscaled_color : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_INTERNAL_UPSCALED); + +FfxFloat32x4 LoadHistory(FfxUInt32x2 iPxHistory) +{ + return r_internal_upscaled_color[iPxHistory]; +} + +FfxFloat32x4 SampleHistory(FfxFloat32x2 fUV) +{ + return r_internal_upscaled_color.SampleLevel(s_LinearClamp, fUV, 0); +} +#endif + +#if defined(FSR3UPSCALER_BIND_UAV_LUMA_HISTORY) +RWTexture2D rw_luma_history : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_LUMA_HISTORY); + +void StoreLumaHistory(FfxUInt32x2 iPxPos, FfxFloat32x4 fLumaHistory) +{ + rw_luma_history[iPxPos] = fLumaHistory; +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_LUMA_HISTORY) +Texture2D r_luma_history : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_LUMA_HISTORY); + +FfxFloat32x4 LoadLumaHistory(FfxInt32x2 iPxPos) +{ + return r_luma_history[iPxPos]; +} + +FfxFloat32x4 SampleLumaHistory(FfxFloat32x2 fUV) +{ + return r_luma_history.SampleLevel(s_LinearClamp, fUV, 0); +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_RCAS_INPUT) +Texture2D r_rcas_input : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_RCAS_INPUT); + +FfxFloat32x4 LoadRCAS_Input(FfxInt32x2 iPxPos) +{ + return r_rcas_input[iPxPos]; +} + +FfxFloat32x3 SampleRCAS_Input(FfxFloat32x2 fUV) +{ + return r_rcas_input.SampleLevel(s_LinearClamp, fUV, 0).rgb; +} +#endif + +#if defined(FSR3UPSCALER_BIND_UAV_INTERNAL_UPSCALED) +RWTexture2D rw_internal_upscaled_color : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_INTERNAL_UPSCALED); + +void StoreReprojectedHistory(FfxUInt32x2 iPxHistory, FfxFloat32x4 fHistory) +{ + rw_internal_upscaled_color[iPxHistory] = fHistory; +} + +void StoreInternalColorAndWeight(FfxUInt32x2 iPxPos, FfxFloat32x4 fColorAndWeight) +{ + rw_internal_upscaled_color[iPxPos] = fColorAndWeight; +} +#endif + +#if defined(FSR3UPSCALER_BIND_UAV_UPSCALED_OUTPUT) +UNITY_FSR_RWTEX2D(FfxFloat32x4) rw_upscaled_output : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_UPSCALED_OUTPUT); + +void StoreUpscaledOutput(FfxUInt32x2 iPxPos, FfxFloat32x3 fColor) +{ + rw_upscaled_output[UNITY_FSR_POS(iPxPos)] = FfxFloat32x4(fColor, 1.f); +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_ACCUMULATION) +Texture2D r_accumulation : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_ACCUMULATION); + +FfxFloat32 SampleAccumulation(FfxFloat32x2 fUV) +{ + return r_accumulation.SampleLevel(s_LinearClamp, fUV, 0); +} +#endif + +#if defined(FSR3UPSCALER_BIND_UAV_ACCUMULATION) +RWTexture2D rw_accumulation : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_ACCUMULATION); + +void StoreAccumulation(FfxUInt32x2 iPxPos, FfxFloat32 fAccumulation) +{ + rw_accumulation[iPxPos] = fAccumulation; +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_SHADING_CHANGE) +Texture2D r_shading_change : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_SHADING_CHANGE); + +FfxFloat32 LoadShadingChange(FfxUInt32x2 iPxPos) +{ + return r_shading_change[iPxPos]; +} + +FfxFloat32 SampleShadingChange(FfxFloat32x2 fUV) +{ + return r_shading_change.SampleLevel(s_LinearClamp, fUV, 0); +} +#endif + +#if defined(FSR3UPSCALER_BIND_UAV_SHADING_CHANGE) +RWTexture2D rw_shading_change : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_SHADING_CHANGE); + +void StoreShadingChange(FfxUInt32x2 iPxPos, FfxFloat32 fShadingChange) +{ + rw_shading_change[iPxPos] = fShadingChange; +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_FARTHEST_DEPTH) +Texture2D r_farthest_depth : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_FARTHEST_DEPTH); + +FfxInt32x2 GetFarthestDepthResourceDimensions() +{ + FfxUInt32 uWidth; + FfxUInt32 uHeight; + r_farthest_depth.GetDimensions(uWidth, uHeight); + + return FfxInt32x2(uWidth, uHeight); +} + +FfxFloat32 LoadFarthestDepth(FfxUInt32x2 iPxPos) +{ + return r_farthest_depth[iPxPos]; +} + +FfxFloat32 SampleFarthestDepth(FfxFloat32x2 fUV) +{ + return r_farthest_depth.SampleLevel(s_LinearClamp, fUV, 0); +} +#endif + +#if defined(FSR3UPSCALER_BIND_UAV_FARTHEST_DEPTH) +RWTexture2D rw_farthest_depth : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_FARTHEST_DEPTH); + +void StoreFarthestDepth(FfxUInt32x2 iPxPos, FfxFloat32 fDepth) +{ + rw_farthest_depth[iPxPos] = fDepth; +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_FARTHEST_DEPTH_MIP1) +Texture2D r_farthest_depth_mip1 : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_FARTHEST_DEPTH_MIP1); + +FfxInt32x2 GetFarthestDepthMip1ResourceDimensions() +{ + FfxUInt32 uWidth; + FfxUInt32 uHeight; + r_farthest_depth_mip1.GetDimensions(uWidth, uHeight); + + return FfxInt32x2(uWidth, uHeight); +} + +FfxFloat32 LoadFarthestDepthMip1(FfxUInt32x2 iPxPos) +{ + return r_farthest_depth_mip1[iPxPos]; +} + +FfxFloat32 SampleFarthestDepthMip1(FfxFloat32x2 fUV) +{ + return r_farthest_depth_mip1.SampleLevel(s_LinearClamp, fUV, 0); +} +#endif + +#if defined(FSR3UPSCALER_BIND_UAV_FARTHEST_DEPTH_MIP1) +RWTexture2D rw_farthest_depth_mip1 : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_FARTHEST_DEPTH_MIP1); + +void StoreFarthestDepthMip1(FfxUInt32x2 iPxPos, FfxFloat32 fDepth) +{ + rw_farthest_depth_mip1[iPxPos] = fDepth; +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_CURRENT_LUMA) +Texture2D r_current_luma : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_CURRENT_LUMA); + +FfxFloat32 LoadCurrentLuma(FfxUInt32x2 iPxPos) +{ + return r_current_luma[iPxPos]; +} + +FfxFloat32 SampleCurrentLuma(FfxFloat32x2 uv) +{ + return r_current_luma.SampleLevel(s_LinearClamp, uv, 0); +} +#endif + +#if defined(FSR3UPSCALER_BIND_UAV_CURRENT_LUMA) +RWTexture2D rw_current_luma : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_CURRENT_LUMA); + +void StoreCurrentLuma(FfxUInt32x2 iPxPos, FfxFloat32 fLuma) +{ + rw_current_luma[iPxPos] = fLuma; +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_LUMA_INSTABILITY) +Texture2D r_luma_instability : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_LUMA_INSTABILITY); + +FfxFloat32 SampleLumaInstability(FfxFloat32x2 uv) +{ + return r_luma_instability.SampleLevel(s_LinearClamp, uv, 0); +} +#endif + +#if defined(FSR3UPSCALER_BIND_UAV_LUMA_INSTABILITY) +RWTexture2D rw_luma_instability : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_LUMA_INSTABILITY); + +void StoreLumaInstability(FfxUInt32x2 iPxPos, FfxFloat32 fLumaInstability) +{ + rw_luma_instability[iPxPos] = fLumaInstability; +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_PREVIOUS_LUMA) +Texture2D r_previous_luma : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_PREVIOUS_LUMA); + +FfxFloat32 LoadPreviousLuma(FfxUInt32x2 iPxPos) +{ + return r_previous_luma[iPxPos]; +} + +FfxFloat32 SamplePreviousLuma(FfxFloat32x2 uv) +{ + return r_previous_luma.SampleLevel(s_LinearClamp, uv, 0); +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_NEW_LOCKS) +Texture2D r_new_locks : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_NEW_LOCKS); + +FfxFloat32 LoadNewLocks(FfxUInt32x2 iPxPos) +{ + return r_new_locks[iPxPos]; +} +#endif + +#if defined(FSR3UPSCALER_BIND_UAV_NEW_LOCKS) +RWTexture2D rw_new_locks : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_NEW_LOCKS); + +FfxFloat32 LoadRwNewLocks(FfxUInt32x2 iPxPos) +{ + return rw_new_locks[iPxPos]; +} + +void StoreNewLocks(FfxUInt32x2 iPxPos, FfxFloat32 newLock) +{ + rw_new_locks[iPxPos] = newLock; +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_RECONSTRUCTED_PREV_NEAREST_DEPTH) +Texture2D r_reconstructed_previous_nearest_depth : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_RECONSTRUCTED_PREV_NEAREST_DEPTH); + +FfxFloat32 LoadReconstructedPrevDepth(FfxUInt32x2 iPxPos) +{ + return asfloat(r_reconstructed_previous_nearest_depth[iPxPos]); +} +#endif + +#if defined(FSR3UPSCALER_BIND_UAV_RECONSTRUCTED_PREV_NEAREST_DEPTH) +RWTexture2D rw_reconstructed_previous_nearest_depth : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_RECONSTRUCTED_PREV_NEAREST_DEPTH); + +void StoreReconstructedDepth(FfxUInt32x2 iPxSample, FfxFloat32 fDepth) +{ + FfxUInt32 uDepth = asuint(fDepth); + +#if FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH + InterlockedMax(rw_reconstructed_previous_nearest_depth[iPxSample], uDepth); +#else + InterlockedMin(rw_reconstructed_previous_nearest_depth[iPxSample], uDepth); // min for standard, max for inverted depth +#endif +} + +void SetReconstructedDepth(FfxUInt32x2 iPxSample, const FfxUInt32 uValue) +{ + rw_reconstructed_previous_nearest_depth[iPxSample] = uValue; +} +#endif + +#if defined(FSR3UPSCALER_BIND_UAV_DILATED_DEPTH) +RWTexture2D rw_dilated_depth : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_DILATED_DEPTH); + +void StoreDilatedDepth(FFX_PARAMETER_IN FfxUInt32x2 iPxPos, FFX_PARAMETER_IN FfxFloat32 fDepth) +{ + rw_dilated_depth[iPxPos] = fDepth; +} +#endif + +#if defined(FSR3UPSCALER_BIND_UAV_DILATED_MOTION_VECTORS) +RWTexture2D rw_dilated_motion_vectors : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_DILATED_MOTION_VECTORS); + +void StoreDilatedMotionVector(FFX_PARAMETER_IN FfxUInt32x2 iPxPos, FFX_PARAMETER_IN FfxFloat32x2 fMotionVector) +{ + rw_dilated_motion_vectors[iPxPos] = fMotionVector; +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_DILATED_MOTION_VECTORS) +Texture2D r_dilated_motion_vectors : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_DILATED_MOTION_VECTORS); + +FfxFloat32x2 LoadDilatedMotionVector(FfxUInt32x2 iPxInput) +{ + return r_dilated_motion_vectors[iPxInput]; +} + +FfxFloat32x2 SampleDilatedMotionVector(FfxFloat32x2 fUV) +{ + return r_dilated_motion_vectors.SampleLevel(s_LinearClamp, fUV, 0); +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_DILATED_DEPTH) +Texture2D r_dilated_depth : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_DILATED_DEPTH); + +FfxFloat32 LoadDilatedDepth(FfxUInt32x2 iPxInput) +{ + return r_dilated_depth[iPxInput]; +} + +FfxFloat32 SampleDilatedDepth(FfxFloat32x2 fUV) +{ + return r_dilated_depth.SampleLevel(s_LinearClamp, fUV, 0); +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_INPUT_EXPOSURE) +Texture2D r_input_exposure : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_INPUT_EXPOSURE); + +FfxFloat32 Exposure() +{ + FfxFloat32 exposure = r_input_exposure[FfxUInt32x2(0, 0)].x; + + if (exposure == 0.0f) { + exposure = 1.0f; + } + + return exposure; +} +#endif + +// BEGIN: FSR3UPSCALER_BIND_SRV_LANCZOS_LUT +#if defined(FSR3UPSCALER_BIND_SRV_LANCZOS_LUT) +Texture2D r_lanczos_lut : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_LANCZOS_LUT); +#endif + +FfxFloat32 SampleLanczos2Weight(FfxFloat32 x) +{ +#if defined(FSR3UPSCALER_BIND_SRV_LANCZOS_LUT) + return r_lanczos_lut.SampleLevel(s_LinearClamp, FfxFloat32x2(x / 2, 0.5f), 0); +#else + return 0.f; +#endif +} +// END: FSR3UPSCALER_BIND_SRV_LANCZOS_LUT + +#if defined(FSR3UPSCALER_BIND_SRV_DILATED_REACTIVE_MASKS) +Texture2D r_dilated_reactive_masks : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_DILATED_REACTIVE_MASKS); + +FfxFloat32x4 SampleDilatedReactiveMasks(FfxFloat32x2 fUV) +{ + return r_dilated_reactive_masks.SampleLevel(s_LinearClamp, fUV, 0); +} +#endif + +#if defined(FSR3UPSCALER_BIND_UAV_DILATED_REACTIVE_MASKS) +RWTexture2D rw_dilated_reactive_masks : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_DILATED_REACTIVE_MASKS); + +void StoreDilatedReactiveMasks(FFX_PARAMETER_IN FfxUInt32x2 iPxPos, FFX_PARAMETER_IN FfxFloat32x4 fDilatedReactiveMasks) +{ + rw_dilated_reactive_masks[iPxPos] = fDilatedReactiveMasks; +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_INPUT_OPAQUE_ONLY) +UNITY_FSR_TEX2D(FfxFloat32x4) r_input_opaque_only : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_INPUT_OPAQUE_ONLY); + +FfxFloat32x3 LoadOpaqueOnly(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos) +{ + return r_input_opaque_only[UNITY_FSR_POS(iPxPos)].xyz; +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_PREV_PRE_ALPHA_COLOR) +Texture2D r_input_prev_color_pre_alpha : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_PREV_PRE_ALPHA_COLOR); + +FfxFloat32x3 LoadPrevPreAlpha(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos) +{ + return r_input_prev_color_pre_alpha[iPxPos]; +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_PREV_POST_ALPHA_COLOR) +Texture2D r_input_prev_color_post_alpha : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_PREV_POST_ALPHA_COLOR); + +FfxFloat32x3 LoadPrevPostAlpha(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos) +{ + return r_input_prev_color_post_alpha[iPxPos]; +} +#endif + +#if defined(FSR3UPSCALER_BIND_UAV_AUTOREACTIVE) && \ + defined(FSR3UPSCALER_BIND_UAV_AUTOCOMPOSITION) + +RWTexture2D rw_output_autoreactive : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_AUTOREACTIVE); +RWTexture2D rw_output_autocomposition : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_AUTOCOMPOSITION); + +void StoreAutoReactive(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos, FFX_PARAMETER_IN FFX_MIN16_F2 fReactive) +{ + rw_output_autoreactive[iPxPos] = fReactive.x; + + rw_output_autocomposition[iPxPos] = fReactive.y; +} +#endif + +#if defined(FSR3UPSCALER_BIND_UAV_PREV_PRE_ALPHA_COLOR) +RWTexture2D rw_output_prev_color_pre_alpha : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_PREV_PRE_ALPHA_COLOR); + +void StorePrevPreAlpha(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos, FFX_PARAMETER_IN FFX_MIN16_F3 color) +{ + rw_output_prev_color_pre_alpha[iPxPos] = color; +} +#endif + +#if defined(FSR3UPSCALER_BIND_UAV_PREV_POST_ALPHA_COLOR) +RWTexture2D rw_output_prev_color_post_alpha : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_PREV_POST_ALPHA_COLOR); + +void StorePrevPostAlpha(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos, FFX_PARAMETER_IN FFX_MIN16_F3 color) +{ + rw_output_prev_color_post_alpha[iPxPos] = color; +} +#endif + +#if defined(FSR3UPSCALER_BIND_UAV_FRAME_INFO) +RWTexture2D rw_frame_info : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_FRAME_INFO); + +FfxFloat32x4 LoadFrameInfo() +{ + return rw_frame_info[FfxInt32x2(0, 0)]; +} + +void StoreFrameInfo(FfxFloat32x4 fInfo) +{ + rw_frame_info[FfxInt32x2(0, 0)] = fInfo; +} +#endif + +#if defined(FSR3UPSCALER_BIND_SRV_FRAME_INFO) +Texture2D r_frame_info : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_FRAME_INFO); + +FfxFloat32x4 FrameInfo() +{ + return r_frame_info[FfxInt32x2(0, 0)]; +} +#endif + +#if defined(FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_0) && \ + defined(FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_1) && \ + defined(FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_2) && \ + defined(FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_3) && \ + defined(FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_4) && \ + defined(FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_5) + +RWTexture2D rw_spd_mip0 : FFX_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_0); +RWTexture2D rw_spd_mip1 : FFX_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_1); +RWTexture2D rw_spd_mip2 : FFX_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_2); +RWTexture2D rw_spd_mip3 : FFX_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_3); +RWTexture2D rw_spd_mip4 : FFX_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_4); +globallycoherent RWTexture2D rw_spd_mip5 : FFX_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_SPD_MIPS_LEVEL_5); + +FfxFloat32x2 RWLoadPyramid(FFX_PARAMETER_IN FfxInt32x2 iPxPos, FFX_PARAMETER_IN FfxUInt32 index) +{ +#define LOAD(idx) \ + if (index == idx) \ + { \ + return rw_spd_mip##idx[iPxPos]; \ + } + LOAD(0); + LOAD(1); + LOAD(2); + LOAD(3); + LOAD(4); + LOAD(5); + + return 0; + +#undef LOAD +} + +void StorePyramid(FFX_PARAMETER_IN FfxInt32x2 iPxPos, FFX_PARAMETER_IN FfxFloat32x2 outValue, FFX_PARAMETER_IN FfxUInt32 index) +{ +#define STORE(idx) \ + if (index == idx) \ + { \ + rw_spd_mip##idx[iPxPos] = outValue; \ + } + + STORE(0); + STORE(1); + STORE(2); + STORE(3); + STORE(4); + STORE(5); + +#undef STORE +} +#endif + +#if defined FSR3UPSCALER_BIND_UAV_SPD_GLOBAL_ATOMIC +globallycoherent RWTexture2D rw_spd_global_atomic : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_SPD_GLOBAL_ATOMIC); + +void SPD_IncreaseAtomicCounter(inout FfxUInt32 spdCounter) +{ + InterlockedAdd(rw_spd_global_atomic[FfxInt32x2(0, 0)], 1, spdCounter); +} + +void SPD_ResetAtomicCounter() +{ + rw_spd_global_atomic[FfxInt32x2(0, 0)] = 0; +} +#endif + +#endif // #if defined(FFX_GPU) diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h.meta new file mode 100644 index 0000000..aca1bc0 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: aef945ff764c453428c8c4759706228d +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_common.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_common.h new file mode 100644 index 0000000..dd479b1 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_common.h @@ -0,0 +1,403 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#if !defined(FFX_FSR3UPSCALER_COMMON_H) +#define FFX_FSR3UPSCALER_COMMON_H + +#if defined(FFX_GPU) +#pragma warning(error : 3206) // treat vector truncation warnings as errors +#pragma warning(disable : 3205) // conversion from larger type to smaller +#pragma warning(disable : 3571) // in ffxPow(f, e), f could be negative + +FFX_STATIC const FfxFloat32 FSR3UPSCALER_FP16_MIN = 6.10e-05f; +FFX_STATIC const FfxFloat32 FSR3UPSCALER_FP16_MAX = 65504.0f; +FFX_STATIC const FfxFloat32 FSR3UPSCALER_EPSILON = FSR3UPSCALER_FP16_MIN; +FFX_STATIC const FfxFloat32 FSR3UPSCALER_TONEMAP_EPSILON = FSR3UPSCALER_FP16_MIN; +FFX_STATIC const FfxFloat32 FSR3UPSCALER_FP32_MAX = 3.402823466e+38f; +FFX_STATIC const FfxFloat32 FSR3UPSCALER_FP32_MIN = 1.175494351e-38f; + +// Reconstructed depth usage +FFX_STATIC const FfxFloat32 fReconstructedDepthBilinearWeightThreshold = FSR3UPSCALER_EPSILON * 10; + +FfxFloat32 ReconstructedDepthMvPxThreshold(FfxFloat32 fNearestDepthInMeters) +{ + return ffxLerp(0.25f, 0.75f, ffxSaturate(fNearestDepthInMeters / 100.0f)); +} + +// Accumulation +FFX_STATIC const FfxFloat32 fUpsampleLanczosWeightScale = 1.0f / 16.0f; +FFX_STATIC const FfxFloat32 fAverageLanczosWeightPerFrame = 0.74f * fUpsampleLanczosWeightScale; // Average lanczos weight for jitter accumulated samples +FFX_STATIC const FfxFloat32 fAccumulationMaxOnMotion = 3.0f * fUpsampleLanczosWeightScale; + +#define SHADING_CHANGE_SET_SIZE 5 +FFX_STATIC const FfxInt32 iShadingChangeMipStart = 0; +FFX_STATIC const FfxFloat32 fShadingChangeSamplePow = 1.0f / 1.0f; + + +FFX_STATIC const FfxFloat32 fLockThreshold = 1.0f; +FFX_STATIC const FfxFloat32 fLockMax = 2.0f; + +FFX_STATIC const FfxInt32 REACTIVE = 0; +FFX_STATIC const FfxInt32 DISOCCLUSION = 1; +FFX_STATIC const FfxInt32 SHADING_CHANGE = 2; +FFX_STATIC const FfxInt32 ACCUMULAION = 3; + +FFX_STATIC const FfxInt32 FRAME_INFO_EXPOSURE = 0; +FFX_STATIC const FfxInt32 FRAME_INFO_LOG_LUMA = 1; +FFX_STATIC const FfxInt32 FRAME_INFO_SCENE_AVERAGE_LUMA = 2; + +FfxBoolean TonemapFirstFrame() +{ + const FfxBoolean bEnabled = true; + return FrameIndex() == 0 && bEnabled; +} + +FfxFloat32 AverageLanczosWeightPerFrame() +{ + return 0.74f; +} + +FfxInt32x2 ShadingChangeRenderSize() +{ + return FfxInt32x2(RenderSize() * 0.5f); +} + +FfxInt32x2 ShadingChangeMaxRenderSize() +{ + return FfxInt32x2(MaxRenderSize() * 0.5f); +} + +FfxInt32x2 PreviousFrameShadingChangeRenderSize() +{ + return FfxInt32x2(PreviousFrameRenderSize() * 0.5f); +} + +#if defined(FSR3UPSCALER_BIND_SRV_FRAME_INFO) +FfxFloat32 SceneAverageLuma() +{ + return FrameInfo()[FRAME_INFO_SCENE_AVERAGE_LUMA]; +} +#endif + +// Auto exposure +FFX_STATIC const FfxFloat32 resetAutoExposureAverageSmoothing = 1e8f; + +struct AccumulationPassCommonParams +{ + FfxInt32x2 iPxHrPos; + FfxFloat32x2 fHrUv; + FfxFloat32x2 fLrUvJittered; + FfxFloat32x2 fLrUv_HwSampler; + FfxFloat32x2 fMotionVector; + FfxFloat32x2 fReprojectedHrUv; + FfxFloat32 f4KVelocity; + FfxFloat32 fDisocclusion; + FfxFloat32 fReactiveMask; + FfxFloat32 fShadingChange; + FfxFloat32 fAccumulation; + FfxFloat32 fLumaInstabilityFactor; + FfxFloat32 fFarthestDepthInMeters; + + FfxBoolean bIsExistingSample; + FfxBoolean bIsNewSample; +}; + +FfxFloat32 Get4KVelocity(FfxFloat32x2 fMotionVector) +{ + return length(fMotionVector * FfxFloat32x2(3840.0f, 2160.0f)); +} + +struct RectificationBox +{ + FfxFloat32x3 boxCenter; + FfxFloat32x3 boxVec; + FfxFloat32x3 aabbMin; + FfxFloat32x3 aabbMax; + FfxFloat32 fBoxCenterWeight; +}; + +struct AccumulationPassData +{ + RectificationBox clippingBox; + FfxFloat32x3 fUpsampledColor; + FfxFloat32 fUpsampledWeight; + FfxFloat32x3 fHistoryColor; + FfxFloat32 fHistoryWeight; + FfxFloat32 fLock; + FfxFloat32 fLockContributionThisFrame; +}; + +void RectificationBoxAddInitialSample(FFX_PARAMETER_INOUT RectificationBox rectificationBox, const FfxFloat32x3 colorSample, const FfxFloat32 fSampleWeight) +{ + rectificationBox.aabbMin = colorSample; + rectificationBox.aabbMax = colorSample; + + FfxFloat32x3 weightedSample = colorSample * fSampleWeight; + rectificationBox.boxCenter = weightedSample; + rectificationBox.boxVec = colorSample * weightedSample; + rectificationBox.fBoxCenterWeight = fSampleWeight; +} + +void RectificationBoxAddSample(FfxBoolean bInitialSample, FFX_PARAMETER_INOUT RectificationBox rectificationBox, const FfxFloat32x3 colorSample, const FfxFloat32 fSampleWeight) +{ + if (bInitialSample) { + RectificationBoxAddInitialSample(rectificationBox, colorSample, fSampleWeight); + } else { + rectificationBox.aabbMin = ffxMin(rectificationBox.aabbMin, colorSample); + rectificationBox.aabbMax = ffxMax(rectificationBox.aabbMax, colorSample); + + FfxFloat32x3 weightedSample = colorSample * fSampleWeight; + rectificationBox.boxCenter += weightedSample; + rectificationBox.boxVec += colorSample * weightedSample; + rectificationBox.fBoxCenterWeight += fSampleWeight; + } +} + +void RectificationBoxComputeVarianceBoxData(FFX_PARAMETER_INOUT RectificationBox rectificationBox) +{ + rectificationBox.fBoxCenterWeight = (abs(rectificationBox.fBoxCenterWeight) > FfxFloat32(FSR3UPSCALER_FP32_MIN) ? rectificationBox.fBoxCenterWeight : FfxFloat32(1.f)); + rectificationBox.boxCenter /= rectificationBox.fBoxCenterWeight; + rectificationBox.boxVec /= rectificationBox.fBoxCenterWeight; + FfxFloat32x3 stdDev = sqrt(abs(rectificationBox.boxVec - rectificationBox.boxCenter * rectificationBox.boxCenter)); + rectificationBox.boxVec = stdDev; +} + +FfxFloat32x3 SafeRcp3(FfxFloat32x3 v) +{ + return (all(FFX_NOT_EQUAL(v, FfxFloat32x3(0, 0, 0)))) ? (FfxFloat32x3(1, 1, 1) / v) : FfxFloat32x3(0, 0, 0); +} + +FfxFloat32 MinDividedByMax(const FfxFloat32 v0, const FfxFloat32 v1, const FfxFloat32 fOnZeroReturnValue) +{ + const FfxFloat32 m = ffxMax(v0, v1); + return m != 0 ? ffxMin(v0, v1) / m : fOnZeroReturnValue; +} + +FfxFloat32 MinDividedByMax(const FfxFloat32 v0, const FfxFloat32 v1) +{ + const FfxFloat32 m = ffxMax(v0, v1); + return m != 0 ? ffxMin(v0, v1) / m : 0; +} + +FfxFloat32x3 YCoCgToRGB(FfxFloat32x3 fYCoCg) +{ + FfxFloat32x3 fRgb; + + fRgb = FfxFloat32x3( + fYCoCg.x + fYCoCg.y - fYCoCg.z, + fYCoCg.x + fYCoCg.z, + fYCoCg.x - fYCoCg.y - fYCoCg.z); + + return fRgb; +} + +FfxFloat32x3 RGBToYCoCg(FfxFloat32x3 fRgb) +{ + FfxFloat32x3 fYCoCg; + + fYCoCg = FfxFloat32x3( + 0.25f * fRgb.r + 0.5f * fRgb.g + 0.25f * fRgb.b, + 0.5f * fRgb.r - 0.5f * fRgb.b, + -0.25f * fRgb.r + 0.5f * fRgb.g - 0.25f * fRgb.b); + + return fYCoCg; +} + +FfxFloat32 RGBToLuma(FfxFloat32x3 fLinearRgb) +{ + return dot(fLinearRgb, FfxFloat32x3(0.2126f, 0.7152f, 0.0722f)); +} + +FfxFloat32 RGBToPerceivedLuma(FfxFloat32x3 fLinearRgb) +{ + FfxFloat32 fLuminance = RGBToLuma(fLinearRgb); + + FfxFloat32 fPercievedLuminance = 0; + if (fLuminance <= 216.0f / 24389.0f) { + fPercievedLuminance = fLuminance * (24389.0f / 27.0f); + } + else { + fPercievedLuminance = ffxPow(fLuminance, 1.0f / 3.0f) * 116.0f - 16.0f; + } + + return fPercievedLuminance * 0.01f; +} + +FfxFloat32x3 Tonemap(FfxFloat32x3 fRgb) +{ + return fRgb / (ffxMax(ffxMax(0.f, fRgb.r), ffxMax(fRgb.g, fRgb.b)) + 1.f).xxx; +} + +FfxFloat32x3 InverseTonemap(FfxFloat32x3 fRgb) +{ + return fRgb / ffxMax(FSR3UPSCALER_TONEMAP_EPSILON, 1.f - ffxMax(fRgb.r, ffxMax(fRgb.g, fRgb.b))).xxx; +} + +FfxBoolean IsUvInside(FfxFloat32x2 fUv) +{ + return (fUv.x >= 0.0f && fUv.x <= 1.0f) && (fUv.y >= 0.0f && fUv.y <= 1.0f); +} + +FfxInt32x2 ClampLoad(FfxInt32x2 iPxSample, FfxInt32x2 iPxOffset, FfxInt32x2 iTextureSize) +{ + FfxInt32x2 result = iPxSample + iPxOffset; + result.x = ffxMax(0, ffxMin(result.x, iTextureSize.x - 1)); + result.y = ffxMax(0, ffxMin(result.y, iTextureSize.y - 1)); + return result; +} + +FfxFloat32x2 ClampUv(FfxFloat32x2 fUv, FfxInt32x2 iTextureSize, FfxInt32x2 iResourceSize) +{ + const FfxFloat32x2 fSampleLocation = fUv * iTextureSize; + const FfxFloat32x2 fClampedLocation = ffxMax(FfxFloat32x2(0.5f, 0.5f), ffxMin(fSampleLocation, FfxFloat32x2(iTextureSize) - FfxFloat32x2(0.5f, 0.5f))); + const FfxFloat32x2 fClampedUv = fClampedLocation / FfxFloat32x2(iResourceSize); + + return fClampedUv; +} + +FfxBoolean IsOnScreen(FfxInt32x2 pos, FfxInt32x2 size) +{ + return all(FFX_LESS_THAN(FfxUInt32x2(pos), FfxUInt32x2(size))); +} + +FfxFloat32 ComputeAutoExposureFromLavg(FfxFloat32 Lavg) +{ + Lavg = exp(Lavg); + + const FfxFloat32 S = 100.0f; //ISO arithmetic speed + const FfxFloat32 K = 12.5f; + FfxFloat32 ExposureISO100 = log2((Lavg * S) / K); + + const FfxFloat32 q = 0.65f; + FfxFloat32 Lmax = (78.0f / (q * S)) * ffxPow(2.0f, ExposureISO100); + + return 1.0f / Lmax; +} + +FfxInt32x2 ComputeHrPosFromLrPos(FfxInt32x2 iPxLrPos) +{ + FfxFloat32x2 fSrcJitteredPos = FfxFloat32x2(iPxLrPos) + 0.5f - Jitter(); + FfxFloat32x2 fLrPosInHr = (fSrcJitteredPos / RenderSize()) * UpscaleSize(); + FfxInt32x2 iPxHrPos = FfxInt32x2(floor(fLrPosInHr)); + return iPxHrPos; +} + +FfxFloat32x2 ComputeNdc(FfxFloat32x2 fPxPos, FfxInt32x2 iSize) +{ + return fPxPos / FfxFloat32x2(iSize) * FfxFloat32x2(2.0f, -2.0f) + FfxFloat32x2(-1.0f, 1.0f); +} + +FfxFloat32 GetViewSpaceDepth(FfxFloat32 fDeviceDepth) +{ + const FfxFloat32x4 fDeviceToViewDepth = DeviceToViewSpaceTransformFactors(); + + // fDeviceToViewDepth details found in ffx_fsr3upscaler.cpp + return (fDeviceToViewDepth[1] / (fDeviceDepth - fDeviceToViewDepth[0])); +} + +FfxFloat32 GetViewSpaceDepthInMeters(FfxFloat32 fDeviceDepth) +{ + return GetViewSpaceDepth(fDeviceDepth) * ViewSpaceToMetersFactor(); +} + +FfxFloat32x3 GetViewSpacePosition(FfxInt32x2 iViewportPos, FfxInt32x2 iViewportSize, FfxFloat32 fDeviceDepth) +{ + const FfxFloat32x4 fDeviceToViewDepth = DeviceToViewSpaceTransformFactors(); + + const FfxFloat32 Z = GetViewSpaceDepth(fDeviceDepth); + + const FfxFloat32x2 fNdcPos = ComputeNdc(iViewportPos, iViewportSize); + const FfxFloat32 X = fDeviceToViewDepth[2] * fNdcPos.x * Z; + const FfxFloat32 Y = fDeviceToViewDepth[3] * fNdcPos.y * Z; + + return FfxFloat32x3(X, Y, Z); +} + +FfxFloat32x3 GetViewSpacePositionInMeters(FfxInt32x2 iViewportPos, FfxInt32x2 iViewportSize, FfxFloat32 fDeviceDepth) +{ + return GetViewSpacePosition(iViewportPos, iViewportSize, fDeviceDepth) * ViewSpaceToMetersFactor(); +} + +FfxFloat32 GetMaxDistanceInMeters() +{ +#if FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH + return GetViewSpaceDepth(0.0f) * ViewSpaceToMetersFactor(); +#else + return GetViewSpaceDepth(1.0f) * ViewSpaceToMetersFactor(); +#endif +} + + +struct BilinearSamplingData +{ + FfxInt32x2 iOffsets[4]; + FfxFloat32 fWeights[4]; + FfxInt32x2 iBasePos; +}; + +BilinearSamplingData GetBilinearSamplingData(FfxFloat32x2 fUv, FfxInt32x2 iSize) +{ + BilinearSamplingData data; + + FfxFloat32x2 fPxSample = (fUv * iSize) - FfxFloat32x2(0.5f, 0.5f); + data.iBasePos = FfxInt32x2(floor(fPxSample)); + FfxFloat32x2 fPxFrac = ffxFract(fPxSample); + + data.iOffsets[0] = FfxInt32x2(0, 0); + data.iOffsets[1] = FfxInt32x2(1, 0); + data.iOffsets[2] = FfxInt32x2(0, 1); + data.iOffsets[3] = FfxInt32x2(1, 1); + + data.fWeights[0] = (1 - fPxFrac.x) * (1 - fPxFrac.y); + data.fWeights[1] = (fPxFrac.x) * (1 - fPxFrac.y); + data.fWeights[2] = (1 - fPxFrac.x) * (fPxFrac.y); + data.fWeights[3] = (fPxFrac.x) * (fPxFrac.y); + + return data; +} + +struct PlaneData +{ + FfxFloat32x3 fNormal; + FfxFloat32 fDistanceFromOrigin; +}; + +PlaneData GetPlaneFromPoints(FfxFloat32x3 fP0, FfxFloat32x3 fP1, FfxFloat32x3 fP2) +{ + PlaneData plane; + + FfxFloat32x3 v0 = fP0 - fP1; + FfxFloat32x3 v1 = fP0 - fP2; + plane.fNormal = normalize(cross(v0, v1)); + plane.fDistanceFromOrigin = -dot(fP0, plane.fNormal); + + return plane; +} + +FfxFloat32 PointToPlaneDistance(PlaneData plane, FfxFloat32x3 fPoint) +{ + return abs(dot(plane.fNormal, fPoint) + plane.fDistanceFromOrigin); +} + +#endif // #if defined(FFX_GPU) + +#endif //!defined(FFX_FSR3UPSCALER_COMMON_H) diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_common.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_common.h.meta new file mode 100644 index 0000000..c735997 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_common.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 16bb355abb68c044983f42086cf3eb7e +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_debug_view.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_debug_view.h new file mode 100644 index 0000000..6f4fa33 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_debug_view.h @@ -0,0 +1,159 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +struct FfxDebugViewport +{ + FfxInt32x2 offset; + FfxInt32x2 size; +}; + +// Macro to cull and draw debug viewport +#define DRAW_VIEWPORT(function, pos, vp) \ + { \ + if (pointIsInsideViewport(pos, vp)) \ + { \ + function(pos, vp); \ + } \ + } + +FfxFloat32x2 getTransformedUv(FfxInt32x2 iPxPos, FfxDebugViewport vp) +{ + FfxFloat32x2 fUv = (FfxFloat32x2(iPxPos - vp.offset) + 0.5f) / vp.size; + + return fUv; +} + +FfxFloat32x3 getMotionVectorColor(FfxFloat32x2 fMotionVector) +{ + return FfxFloat32x3(0.5f + fMotionVector * RenderSize() * 0.5f, 0.5f); +} + +FfxFloat32x4 getUnusedIndicationColor(FfxInt32x2 iPxPos, FfxDebugViewport vp) +{ + FfxInt32x2 basePos = iPxPos - vp.offset; + + FfxFloat32 ar = FfxFloat32(vp.size.x) / FfxFloat32(vp.size.y); + + return FfxFloat32x4(basePos.x == FfxInt32(basePos.y * ar), 0, 0, 1); +} + +void drawDilatedMotionVectors(FfxInt32x2 iPxPos, FfxDebugViewport vp) +{ + FfxFloat32x2 fUv = getTransformedUv(iPxPos, vp); + + FfxFloat32x2 fUv_HW = ClampUv(fUv, RenderSize(), MaxRenderSize()); + + FfxFloat32x2 fMotionVector = SampleDilatedMotionVector(fUv_HW); + + StoreUpscaledOutput(iPxPos, getMotionVectorColor(fMotionVector)); +} + +void drawDisocclusionMask(FfxInt32x2 iPxPos, FfxDebugViewport vp) +{ + FfxFloat32x2 fUv = getTransformedUv(iPxPos, vp); + + FfxFloat32x2 fUv_HW = ClampUv(fUv, RenderSize(), MaxRenderSize()); + + FfxFloat32 fDisocclusionFactor = ffxSaturate(SampleDilatedReactiveMasks(fUv_HW)[DISOCCLUSION]); + + StoreUpscaledOutput(iPxPos, FfxFloat32x3(0, fDisocclusionFactor, 0)); +} + +void drawDetailProtectionTakedown(FfxInt32x2 iPxPos, FfxDebugViewport vp) +{ + FfxFloat32x2 fUv = getTransformedUv(iPxPos, vp); + + FfxFloat32x2 fUv_HW = ClampUv(fUv, RenderSize(), MaxRenderSize()); + + FfxFloat32 fProtectionTakedown = ffxSaturate(SampleDilatedReactiveMasks(fUv_HW)[REACTIVE]); + + StoreUpscaledOutput(iPxPos, FfxFloat32x3(0, fProtectionTakedown, 0)); +} + +void drawReactiveness(FfxInt32x2 iPxPos, FfxDebugViewport vp) +{ + FfxFloat32x2 fUv = getTransformedUv(iPxPos, vp); + + FfxFloat32x2 fUv_HW = ClampUv(fUv, RenderSize(), MaxRenderSize()); + + FfxFloat32 fShadingChange = ffxSaturate(SampleDilatedReactiveMasks(fUv_HW)[SHADING_CHANGE]); + + StoreUpscaledOutput(iPxPos, FfxFloat32x3(0, fShadingChange, 0)); +} + +void drawProtectedAreas(FfxInt32x2 iPxPos, FfxDebugViewport vp) +{ + FfxFloat32x2 fUv = getTransformedUv(iPxPos, vp); + + FfxFloat32 fProtection = ffxSaturate(SampleHistory(fUv).w - fLockThreshold); + + StoreUpscaledOutput(iPxPos, FfxFloat32x3(fProtection, 0, 0)); +} + +void drawDilatedDepthInMeters(FfxInt32x2 iPxPos, FfxDebugViewport vp) +{ + FfxFloat32x2 fUv = getTransformedUv(iPxPos, vp); + + FfxFloat32x2 fUv_HW = ClampUv(fUv, RenderSize(), MaxRenderSize()); + + const FfxFloat32 fDilatedDepth = SampleDilatedDepth(fUv_HW); + const FfxFloat32 fDepthInMeters = GetViewSpaceDepthInMeters(fDilatedDepth); + + StoreUpscaledOutput(iPxPos, FfxFloat32x3(ffxSaturate(fDepthInMeters / 25.0f), 0, 0)); +} + +FfxBoolean pointIsInsideViewport(FfxInt32x2 iPxPos, FfxDebugViewport vp) +{ + FfxInt32x2 extent = vp.offset + vp.size; + + return (iPxPos.x >= vp.offset.x && iPxPos.x < extent.x) && (iPxPos.y >= vp.offset.y && iPxPos.y < extent.y); +} + +void DebugView(FfxInt32x2 iPxPos) +{ +#define VIEWPORT_GRID_SIZE_X 3 +#define VIEWPORT_GRID_SIZE_Y 3 + + FfxFloat32x2 fViewportScale = FfxFloat32x2(1.0f / VIEWPORT_GRID_SIZE_X, 1.0f / VIEWPORT_GRID_SIZE_Y); + FfxInt32x2 iViewportSize = FfxInt32x2(UpscaleSize() * fViewportScale); + + // compute grid [y][x] for easier placement of viewports + FfxDebugViewport vp[VIEWPORT_GRID_SIZE_Y][VIEWPORT_GRID_SIZE_X]; + for (FfxInt32 y = 0; y < VIEWPORT_GRID_SIZE_Y; y++) + { + for (FfxInt32 x = 0; x < VIEWPORT_GRID_SIZE_X; x++) + { + vp[y][x].offset = iViewportSize * FfxInt32x2(x, y); + vp[y][x].size = iViewportSize; + } + } + + // top row + DRAW_VIEWPORT(drawDilatedMotionVectors, iPxPos, vp[0][0]); + DRAW_VIEWPORT(drawProtectedAreas, iPxPos, vp[0][1]); + DRAW_VIEWPORT(drawDilatedDepthInMeters, iPxPos, vp[0][2]); + + // bottom row + DRAW_VIEWPORT(drawDisocclusionMask, iPxPos, vp[2][0]); + DRAW_VIEWPORT(drawReactiveness, iPxPos, vp[2][1]); + DRAW_VIEWPORT(drawDetailProtectionTakedown, iPxPos, vp[2][2]); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_debug_view.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_debug_view.h.meta new file mode 100644 index 0000000..3ec32db --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_debug_view.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 9c5c756c93605b6428eaae6c8aaabbe5 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_luma_instability.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_luma_instability.h new file mode 100644 index 0000000..624b7ca --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_luma_instability.h @@ -0,0 +1,115 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +struct LumaInstabilityFactorData +{ + FfxFloat32x4 fLumaHistory; + FfxFloat32 fLumaInstabilityFactor; +}; + +LumaInstabilityFactorData ComputeLumaInstabilityFactor(LumaInstabilityFactorData data, FfxFloat32 fCurrentFrameLuma, FfxFloat32 fFarthestDepthInMeters) +{ + const FfxInt32 N_MINUS_1 = 0; + const FfxInt32 N_MINUS_2 = 1; + const FfxInt32 N_MINUS_3 = 2; + const FfxInt32 N_MINUS_4 = 3; + + FfxFloat32 fLumaInstability = 0.0f; + const FfxFloat32 fDiffs0 = (fCurrentFrameLuma - data.fLumaHistory[N_MINUS_1]); + const FfxFloat32 fSimilarity0 = MinDividedByMax(fCurrentFrameLuma, data.fLumaHistory[N_MINUS_1], 1.0f); + + FfxFloat32 fMaxSimilarity = fSimilarity0; + + if (fSimilarity0 < 1.0f) { + for (int i = N_MINUS_2; i <= N_MINUS_4; i++) { + const FfxFloat32 fDiffs1 = (fCurrentFrameLuma - data.fLumaHistory[i]); + const FfxFloat32 fSimilarity1 = MinDividedByMax(fCurrentFrameLuma, data.fLumaHistory[i]); + + if (sign(fDiffs0) == sign(fDiffs1)) { + + fMaxSimilarity = ffxMax(fMaxSimilarity, fSimilarity1); + } + } + + fLumaInstability = FfxFloat32(fMaxSimilarity > fSimilarity0); + } + + // Shift history + data.fLumaHistory[N_MINUS_4] = data.fLumaHistory[N_MINUS_3]; + data.fLumaHistory[N_MINUS_3] = data.fLumaHistory[N_MINUS_2]; + data.fLumaHistory[N_MINUS_2] = data.fLumaHistory[N_MINUS_1]; + data.fLumaHistory[N_MINUS_1] = fCurrentFrameLuma; + + data.fLumaHistory /= Exposure(); + + data.fLumaInstabilityFactor = fLumaInstability * FfxFloat32(data.fLumaHistory[N_MINUS_4] != 0); + + return data; +} + +void LumaInstability(FfxInt32x2 iPxPos) +{ + LumaInstabilityFactorData data; + data.fLumaInstabilityFactor = 0.0f; + data.fLumaHistory = FfxFloat32x4(0.0f, 0.0f, 0.0f, 0.0f); + + const FfxFloat32x2 fDilatedMotionVector = LoadDilatedMotionVector(iPxPos); + const FfxFloat32x2 fUv = (iPxPos + 0.5f) / RenderSize(); + const FfxFloat32x2 fUvCurrFrameJittered = fUv + Jitter() / RenderSize(); + const FfxFloat32x2 fUvPrevFrameJittered = fUv + PreviousFrameJitter() / PreviousFrameRenderSize(); + const FfxFloat32x2 fReprojectedUv = fUvPrevFrameJittered + fDilatedMotionVector; + + if (IsUvInside(fReprojectedUv)) + { + const FfxFloat32x2 fUvReactive_HW = ClampUv(fUvCurrFrameJittered, RenderSize(), MaxRenderSize()); + + const FfxFloat32x4 fDilatedReactiveMasks = SampleDilatedReactiveMasks(fUvReactive_HW); + const FfxFloat32 fReactiveMask = ffxSaturate(fDilatedReactiveMasks[REACTIVE]); + const FfxFloat32 fDisocclusion = ffxSaturate(fDilatedReactiveMasks[DISOCCLUSION]); + const FfxFloat32 fShadingChange = ffxSaturate(fDilatedReactiveMasks[SHADING_CHANGE]); + const FfxFloat32 fAccumulation = ffxSaturate(fDilatedReactiveMasks[ACCUMULAION]); + + const FfxBoolean bAccumulationFactor = fAccumulation > 0.9f; + + const FfxBoolean bComputeInstability = bAccumulationFactor; + + if (bComputeInstability) { + + const FfxFloat32x2 fUv_HW = ClampUv(fUvCurrFrameJittered, RenderSize(), MaxRenderSize()); + const FfxFloat32 fCurrentFrameLuma = SampleCurrentLuma(fUv_HW) * Exposure(); + + const FfxFloat32x2 fReprojectedUv_HW = ClampUv(fReprojectedUv, PreviousFrameRenderSize(), MaxRenderSize()); + data.fLumaHistory = SampleLumaHistory(fReprojectedUv_HW) * DeltaPreExposure() * Exposure(); + + const FfxFloat32x2 fFarthestDepthUv_HW = ClampUv(fUvCurrFrameJittered, RenderSize() / 2, GetFarthestDepthMip1ResourceDimensions()); + const FfxFloat32 fFarthestDepthInMeters = SampleFarthestDepthMip1(fFarthestDepthUv_HW); + + data = ComputeLumaInstabilityFactor(data, fCurrentFrameLuma, fFarthestDepthInMeters); + + const FfxFloat32 fVelocityWeight = 1.0f - ffxSaturate(Get4KVelocity(fDilatedMotionVector) / 20.0f); + data.fLumaInstabilityFactor *= fVelocityWeight * (1.0f - fDisocclusion) * (1.0f - fReactiveMask) * (1.0f - fShadingChange); + } + } + + StoreLumaHistory(iPxPos, data.fLumaHistory); + StoreLumaInstability(iPxPos, data.fLumaInstabilityFactor); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_luma_instability.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_luma_instability.h.meta new file mode 100644 index 0000000..7df8e8b --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_luma_instability.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: c34dbb67d00358d4b9de0a9fb0e0ace2 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_luma_pyramid.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_luma_pyramid.h new file mode 100644 index 0000000..e8a8c49 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_luma_pyramid.h @@ -0,0 +1,192 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +FFX_GROUPSHARED FfxUInt32 spdCounter; + +void SpdIncreaseAtomicCounter(FfxUInt32 slice) +{ + SPD_IncreaseAtomicCounter(spdCounter); +} + +FfxUInt32 SpdGetAtomicCounter() +{ + return spdCounter; +} + +void SpdResetAtomicCounter(FfxUInt32 slice) +{ + SPD_ResetAtomicCounter(); +} + +#ifndef SPD_PACKED_ONLY +FFX_GROUPSHARED FfxFloat32 spdIntermediateR[16][16]; +FFX_GROUPSHARED FfxFloat32 spdIntermediateG[16][16]; +FFX_GROUPSHARED FfxFloat32 spdIntermediateB[16][16]; +FFX_GROUPSHARED FfxFloat32 spdIntermediateA[16][16]; + +FFX_STATIC const FfxInt32 LOG_LUMA = 0; +FFX_STATIC const FfxInt32 LUMA = 1; +FFX_STATIC const FfxInt32 DEPTH_IN_METERS = 2; + +FfxFloat32x4 SpdLoadSourceImage(FfxFloat32x2 iPxPos, FfxUInt32 slice) +{ + //We assume linear data. if non-linear input (sRGB, ...), + //then we should convert to linear first and back to sRGB on output. + const FfxInt32x2 iPxSamplePos = ClampLoad(FfxInt32x2(iPxPos), FfxInt32x2(0, 0), FfxInt32x2(RenderSize())); + + const FfxFloat32 fLuma = LoadCurrentLuma(iPxSamplePos); + const FfxFloat32 fLogLuma = ffxMax(FSR3UPSCALER_EPSILON, log(fLuma)); + const FfxFloat32 fFarthestDepthInMeters = LoadFarthestDepth(iPxSamplePos); + + FfxFloat32x4 fOutput = FfxFloat32x4(0.0f, 0.0f, 0.0f, 0.0f); + fOutput[LOG_LUMA] = fLogLuma; + fOutput[LUMA] = fLuma; + fOutput[DEPTH_IN_METERS] = fFarthestDepthInMeters; + + return fOutput; +} + +FfxFloat32x4 SpdLoad(FfxInt32x2 tex, FfxUInt32 slice) +{ + return FfxFloat32x4(RWLoadPyramid(tex, 5), 0, 0); +} + +FfxFloat32x4 SpdReduce4(FfxFloat32x4 v0, FfxFloat32x4 v1, FfxFloat32x4 v2, FfxFloat32x4 v3) +{ + return (v0 + v1 + v2 + v3) * 0.25f; +} + +void SpdStore(FfxInt32x2 pix, FfxFloat32x4 outValue, FfxUInt32 index, FfxUInt32 slice) +{ + if (index == 5) + { + StorePyramid(pix, outValue.xy, index); + } + else if (index == 0) { + StoreFarthestDepthMip1(pix, outValue[DEPTH_IN_METERS]); + } + + if (index == MipCount() - 1) { //accumulate on 1x1 level + + if (all(FFX_EQUAL(pix, FfxInt32x2(0, 0)))) + { + FfxFloat32x4 frameInfo = LoadFrameInfo(); + const FfxFloat32 fSceneAvgLuma = outValue[LUMA]; + const FfxFloat32 fPrevLogLuma = frameInfo[FRAME_INFO_LOG_LUMA]; + FfxFloat32 fLogLuma = outValue[LOG_LUMA]; + + if (fPrevLogLuma < resetAutoExposureAverageSmoothing) // Compare Lavg, so small or negative values + { + fLogLuma = fPrevLogLuma + (fLogLuma - fPrevLogLuma) * (1.0f - exp(-DeltaTime())); + fLogLuma = ffxMax(0.0f, fLogLuma); + } + + frameInfo[FRAME_INFO_EXPOSURE] = ComputeAutoExposureFromLavg(fLogLuma); + frameInfo[FRAME_INFO_LOG_LUMA] = fLogLuma; + frameInfo[FRAME_INFO_SCENE_AVERAGE_LUMA] = fSceneAvgLuma; + + StoreFrameInfo(frameInfo); + } + } +} + +FfxFloat32x4 SpdLoadIntermediate(FfxUInt32 x, FfxUInt32 y) +{ + return FfxFloat32x4( + spdIntermediateR[x][y], + spdIntermediateG[x][y], + spdIntermediateB[x][y], + spdIntermediateA[x][y]); +} +void SpdStoreIntermediate(FfxUInt32 x, FfxUInt32 y, FfxFloat32x4 value) +{ + spdIntermediateR[x][y] = value.x; + spdIntermediateG[x][y] = value.y; + spdIntermediateB[x][y] = value.z; + spdIntermediateA[x][y] = value.w; +} + +#endif + +// define fetch and store functions Packed +#if FFX_HALF + +FFX_GROUPSHARED FfxFloat16x2 spdIntermediateRG[16][16]; +FFX_GROUPSHARED FfxFloat16x2 spdIntermediateBA[16][16]; + +FfxFloat16x4 SpdLoadSourceImageH(FfxFloat32x2 tex, FfxUInt32 slice) +{ + return FfxFloat16x4(0, 0, 0, 0); +} + +FfxFloat16x4 SpdLoadH(FfxInt32x2 p, FfxUInt32 slice) +{ + return FfxFloat16x4(0, 0, 0, 0); +} + +void SpdStoreH(FfxInt32x2 p, FfxFloat16x4 value, FfxUInt32 mip, FfxUInt32 slice) +{ +} + +FfxFloat16x4 SpdLoadIntermediateH(FfxUInt32 x, FfxUInt32 y) +{ + return FfxFloat16x4( + spdIntermediateRG[x][y].x, + spdIntermediateRG[x][y].y, + spdIntermediateBA[x][y].x, + spdIntermediateBA[x][y].y); +} + +void SpdStoreIntermediateH(FfxUInt32 x, FfxUInt32 y, FfxFloat16x4 value) +{ + spdIntermediateRG[x][y] = value.xy; + spdIntermediateBA[x][y] = value.zw; +} + +FfxFloat16x4 SpdReduce4H(FfxFloat16x4 v0, FfxFloat16x4 v1, FfxFloat16x4 v2, FfxFloat16x4 v3) +{ + return (v0 + v1 + v2 + v3) * FfxFloat16(0.25); +} +#endif + +#include "../spd/ffx_spd.h" + +void ComputeAutoExposure(FfxUInt32x3 WorkGroupId, FfxUInt32 LocalThreadIndex) +{ +#if FFX_HALF + SpdDownsampleH( + FfxUInt32x2(WorkGroupId.xy), + FfxUInt32(LocalThreadIndex), + FfxUInt32(MipCount()), + FfxUInt32(NumWorkGroups()), + FfxUInt32(WorkGroupId.z), + FfxUInt32x2(WorkGroupOffset())); +#else + SpdDownsample( + FfxUInt32x2(WorkGroupId.xy), + FfxUInt32(LocalThreadIndex), + FfxUInt32(MipCount()), + FfxUInt32(NumWorkGroups()), + FfxUInt32(WorkGroupId.z), + FfxUInt32x2(WorkGroupOffset())); +#endif +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_luma_pyramid.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_luma_pyramid.h.meta new file mode 100644 index 0000000..72f45f7 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_luma_pyramid.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 5138e351a00fc914f973d2495f9aa07f +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_prepare_inputs.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_prepare_inputs.h new file mode 100644 index 0000000..59c765b --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_prepare_inputs.h @@ -0,0 +1,152 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +void ReconstructPrevDepth(FfxInt32x2 iPxPos, FfxFloat32 fDepth, FfxFloat32x2 fMotionVector) +{ + const FfxFloat32 fNearestDepthInMeters = ffxMin(GetViewSpaceDepthInMeters(fDepth), FSR3UPSCALER_FP16_MAX); + const FfxFloat32 fReconstructedDeptMvThreshold = ReconstructedDepthMvPxThreshold(fNearestDepthInMeters); + + // Discard small mvs + fMotionVector *= FfxFloat32(Get4KVelocity(fMotionVector) > fReconstructedDeptMvThreshold); + + const FfxFloat32x2 fUv = (iPxPos + FfxFloat32(0.5)) / RenderSize(); + const FfxFloat32x2 fReprojectedUv = fUv + fMotionVector; + const BilinearSamplingData bilinearInfo = GetBilinearSamplingData(fReprojectedUv, RenderSize()); + + // Project current depth into previous frame locations. + // Push to all pixels having some contribution if reprojection is using bilinear logic. + for (FfxInt32 iSampleIndex = 0; iSampleIndex < 4; iSampleIndex++) { + + const FfxInt32x2 iOffset = bilinearInfo.iOffsets[iSampleIndex]; + const FfxFloat32 fWeight = bilinearInfo.fWeights[iSampleIndex]; + + if (fWeight > fReconstructedDepthBilinearWeightThreshold) { + + const FfxInt32x2 iStorePos = bilinearInfo.iBasePos + iOffset; + if (IsOnScreen(iStorePos, RenderSize())) { + StoreReconstructedDepth(iStorePos, fDepth); + } + } + } +} + +struct DepthExtents +{ + FfxFloat32 fNearest; + FfxInt32x2 fNearestCoord; + FfxFloat32 fFarthest; +}; + +DepthExtents FindDepthExtents(FFX_PARAMETER_IN FfxInt32x2 iPxPos) +{ + DepthExtents extents; + const FfxInt32 iSampleCount = 9; + const FfxInt32x2 iSampleOffsets[iSampleCount] = { + FfxInt32x2(+0, +0), + FfxInt32x2(+1, +0), + FfxInt32x2(+0, +1), + FfxInt32x2(+0, -1), + FfxInt32x2(-1, +0), + FfxInt32x2(-1, +1), + FfxInt32x2(+1, +1), + FfxInt32x2(-1, -1), + FfxInt32x2(+1, -1), + }; + + // pull out the depth loads to allow SC to batch them + FfxFloat32 depth[9]; + FfxInt32 iSampleIndex = 0; + FFX_UNROLL + for (iSampleIndex = 0; iSampleIndex < iSampleCount; ++iSampleIndex) { + + FfxInt32x2 iPos = iPxPos + iSampleOffsets[iSampleIndex]; + depth[iSampleIndex] = LoadInputDepth(iPos); + } + + // find closest depth + extents.fNearestCoord = iPxPos; + extents.fNearest = depth[0]; + extents.fFarthest = depth[0]; + FFX_UNROLL + for (iSampleIndex = 1; iSampleIndex < iSampleCount; ++iSampleIndex) { + + const FfxInt32x2 iPos = iPxPos + iSampleOffsets[iSampleIndex]; + if (IsOnScreen(iPos, RenderSize())) { + + FfxFloat32 fNdDepth = depth[iSampleIndex]; +#if FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH + if (fNdDepth > extents.fNearest) { + extents.fFarthest = ffxMin(extents.fFarthest, fNdDepth); +#else + if (fNdDepth < extents.fNearest) { + extents.fFarthest = ffxMax(extents.fFarthest, fNdDepth); +#endif + extents.fNearestCoord = iPos; + extents.fNearest = fNdDepth; + } + } + } + + return extents; +} + +FfxFloat32x2 DilateMotionVector(FfxInt32x2 iPxPos, const DepthExtents depthExtents) +{ +#if FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS + const FfxInt32x2 iSamplePos = iPxPos; + const FfxInt32x2 iMotionVectorPos = depthExtents.fNearestCoord; +#else + const FfxInt32x2 iSamplePos = ComputeHrPosFromLrPos(iPxPos); + const FfxInt32x2 iMotionVectorPos = ComputeHrPosFromLrPos(depthExtents.fNearestCoord); +#endif + + const FfxFloat32x2 fDilatedMotionVector = LoadInputMotionVector(iMotionVectorPos); + + return fDilatedMotionVector; +} + +FfxFloat32 GetCurrentFrameLuma(FfxInt32x2 iPxPos) +{ + //We assume linear data. if non-linear input (sRGB, ...), + //then we should convert to linear first and back to sRGB on output. + const FfxFloat32x3 fRgb = ffxMax(FfxFloat32x3(0, 0, 0), LoadInputColor(iPxPos)); + const FfxFloat32 fLuma = RGBToLuma(fRgb); + + return fLuma; +} + +void PrepareInputs(FfxInt32x2 iPxPos) +{ + const DepthExtents depthExtents = FindDepthExtents(iPxPos); + const FfxFloat32x2 fDilatedMotionVector = DilateMotionVector(iPxPos, depthExtents); + + ReconstructPrevDepth(iPxPos, depthExtents.fNearest, fDilatedMotionVector); + + StoreDilatedMotionVector(iPxPos, fDilatedMotionVector); + StoreDilatedDepth(iPxPos, depthExtents.fNearest); + + const FfxFloat32 fFarthestDepthInMeters = ffxMin(GetViewSpaceDepthInMeters(depthExtents.fFarthest), FSR3UPSCALER_FP16_MAX); + StoreFarthestDepth(iPxPos, fFarthestDepthInMeters); + + const FfxFloat32 fLuma = GetCurrentFrameLuma(iPxPos); + StoreCurrentLuma(iPxPos, fLuma); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_prepare_inputs.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_prepare_inputs.h.meta new file mode 100644 index 0000000..9e7a5db --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_prepare_inputs.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: b08f1fd2bbf082242b930cb94652872c +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_prepare_reactivity.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_prepare_reactivity.h new file mode 100644 index 0000000..fa9571d --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_prepare_reactivity.h @@ -0,0 +1,270 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +FfxFloat32 ComputeDisocclusions(FfxFloat32x2 fUv, FfxFloat32x2 fMotionVector, FfxFloat32 fCurrentDepthViewSpace) +{ + const FfxFloat32 fNearestDepthInMeters = ffxMin(fCurrentDepthViewSpace * ViewSpaceToMetersFactor(), FSR3UPSCALER_FP16_MAX); + const FfxFloat32 fReconstructedDeptMvThreshold = ReconstructedDepthMvPxThreshold(fNearestDepthInMeters); + + fMotionVector *= FfxFloat32(Get4KVelocity(fMotionVector) > fReconstructedDeptMvThreshold); + + const FfxFloat32x2 fReprojectedUv = fUv + fMotionVector; + const BilinearSamplingData bilinearInfo = GetBilinearSamplingData(fReprojectedUv, RenderSize()); + + FfxFloat32 fDisocclusion = 0.0f; + FfxFloat32 fWeightSum = 0.0f; + FfxBoolean bPotentialDisocclusion = true; + + for (FfxInt32 iSampleIndex = 0; iSampleIndex < 4 && bPotentialDisocclusion; iSampleIndex++) + { + + const FfxInt32x2 iOffset = bilinearInfo.iOffsets[iSampleIndex]; + const FfxInt32x2 iSamplePos = ClampLoad(bilinearInfo.iBasePos, iOffset, FfxInt32x2(RenderSize())); + + if (IsOnScreen(iSamplePos, RenderSize())) { + const FfxFloat32 fWeight = bilinearInfo.fWeights[iSampleIndex]; + if (fWeight > fReconstructedDepthBilinearWeightThreshold) { + + const FfxFloat32 fPrevNearestDepthViewSpace = GetViewSpaceDepth(LoadReconstructedPrevDepth(iSamplePos)); + const FfxFloat32 fDepthDifference = fCurrentDepthViewSpace - fPrevNearestDepthViewSpace; + + bPotentialDisocclusion = bPotentialDisocclusion && (fDepthDifference > FSR3UPSCALER_FP32_MIN); + + if (bPotentialDisocclusion) { + const FfxFloat32 fHalfViewportWidth = length(FfxFloat32x2(RenderSize()) * 0.5f); + const FfxFloat32 fDepthThreshold = ffxMax(fCurrentDepthViewSpace, fPrevNearestDepthViewSpace); + + const FfxFloat32 Ksep = 1.37e-05f; + const FfxFloat32 fRequiredDepthSeparation = Ksep * fHalfViewportWidth * fDepthThreshold; + + fDisocclusion += ffxSaturate(FfxFloat32(fRequiredDepthSeparation / fDepthDifference)) * fWeight; + fWeightSum += fWeight; + } + } + } + } + + fDisocclusion = (bPotentialDisocclusion && fWeightSum > 0) ? ffxSaturate(1.0f - fDisocclusion / fWeightSum) : 0.0f; + + return fDisocclusion; +} + +FfxFloat32 ComputeMotionDivergence(FfxFloat32x2 fUv, FfxFloat32x2 fMotionVector, FfxFloat32 fCurrentDepthSample) +{ + const FfxInt32x2 iPxReprojectedPos = FfxInt32x2((fUv + fMotionVector) * RenderSize()); + const FfxFloat32 fReprojectedDepth = LoadDilatedDepth(iPxReprojectedPos); + const FfxFloat32x2 fReprojectedMotionVector = LoadDilatedMotionVector(iPxReprojectedPos); + + const FfxFloat32 fReprojectedVelocity = Get4KVelocity(fReprojectedMotionVector); + const FfxFloat32 f4KVelocity = Get4KVelocity(fMotionVector); + + const FfxFloat32 fMaxLen = max(length(fMotionVector), length(fReprojectedMotionVector)); + + const FfxFloat32 fNucleusDepthInMeters = GetViewSpaceDepthInMeters(fReprojectedDepth); + const FfxFloat32 fCurrentDepthInMeters = GetViewSpaceDepthInMeters(fCurrentDepthSample); + + const FfxFloat32 fDistanceFactor = MinDividedByMax(fNucleusDepthInMeters, fCurrentDepthInMeters); + const FfxFloat32 fVelocityFactor = ffxSaturate(f4KVelocity / 10.0f); + const FfxFloat32 fMotionVectorFieldConfidence = (1.0f - ffxSaturate(fReprojectedVelocity / f4KVelocity)) * fDistanceFactor * fVelocityFactor; + + return fMotionVectorFieldConfidence; +} + +FfxFloat32 DilateReactiveMasks(FfxInt32x2 iPxPos, FfxFloat32x2 fUv) +{ + FfxFloat32 fDilatedReactiveMasks = 0.0f; + + FFX_UNROLL + for (FfxInt32 y = -1; y <=1; y++) + { + FFX_UNROLL + for (FfxInt32 x = -1; x <= 1; x++) + { + const FfxInt32x2 sampleCoord = ClampLoad(iPxPos, FfxInt32x2(x, y), FfxInt32x2(RenderSize())); + fDilatedReactiveMasks = ffxMax(fDilatedReactiveMasks, LoadReactiveMask(sampleCoord)); + } + } + + return fDilatedReactiveMasks; +} + +FfxFloat32 DilateTransparencyAndCompositionMasks(FfxInt32x2 iPxPos, FfxFloat32x2 fUv) +{ + const FfxFloat32x2 fUvTransparencyAndCompositionMask = ClampUv(fUv, RenderSize(), GetTransparencyAndCompositionMaskResourceDimensions()); + return SampleTransparencyAndCompositionMask(fUvTransparencyAndCompositionMask); +} + +FfxFloat32 ComputeThinFeatureConfidence(FfxInt32x2 iPxPos) +{ + /* + 1 2 3 + 4 0 5 + 6 7 8 + */ + + const FfxInt32 iNucleusIndex = 0; + const FfxInt32 iSampleCount = 9; + const FfxInt32x2 iSampleOffsets[iSampleCount] = { + FfxInt32x2(+0, +0), + FfxInt32x2(-1, -1), + FfxInt32x2(+0, -1), + FfxInt32x2(+1, -1), + FfxInt32x2(-1, +0), + FfxInt32x2(+1, +0), + FfxInt32x2(-1, +1), + FfxInt32x2(+0, +1), + FfxInt32x2(+1, +1), + }; + + FfxFloat32 fSamples[iSampleCount]; + + FfxFloat32 fLumaMin = FSR3UPSCALER_FP32_MAX; + FfxFloat32 fLumaMax = FSR3UPSCALER_FP32_MIN; + + FFX_UNROLL + for (FfxInt32 iSampleIndex = 0; iSampleIndex < iSampleCount; ++iSampleIndex) { + const FfxInt32x2 iPxSamplePos = ClampLoad(iPxPos, iSampleOffsets[iSampleIndex], FfxInt32x2(RenderSize())); + fSamples[iSampleIndex] = LoadCurrentLuma(iPxSamplePos) * Exposure(); + + fLumaMin = ffxMin(fLumaMin, fSamples[iSampleIndex]); + fLumaMax = ffxMax(fLumaMax, fSamples[iSampleIndex]); + } + + const FfxFloat32 fThreshold = 0.9f; + FfxFloat32 fDissimilarLumaMin = FSR3UPSCALER_FP32_MAX; + FfxFloat32 fDissimilarLumaMax = 0; + +#define SETBIT(x) (1U << x) + + FfxUInt32 uPatternMask = SETBIT(iNucleusIndex); // Flag nucleus as similar + + const FfxUInt32 uNumRejectionMasks = 4; + const FfxUInt32 uRejectionMasks[uNumRejectionMasks] = { + SETBIT(1) | SETBIT(2) | SETBIT(4) | SETBIT(iNucleusIndex), // Upper left + SETBIT(2) | SETBIT(3) | SETBIT(5) | SETBIT(iNucleusIndex), // Upper right + SETBIT(4) | SETBIT(6) | SETBIT(7) | SETBIT(iNucleusIndex), // Lower left + SETBIT(5) | SETBIT(7) | SETBIT(8) | SETBIT(iNucleusIndex) // Lower right + }; + + FfxInt32 iBitIndex = 1; + FFX_UNROLL + for (FfxInt32 iSampleIndex = 1; iSampleIndex < iSampleCount; ++iSampleIndex, ++iBitIndex) { + + const FfxFloat32 fDifference = abs(fSamples[iSampleIndex] - fSamples[iNucleusIndex]) / (fLumaMax - fLumaMin); + + if (fDifference < fThreshold) + { + uPatternMask |= SETBIT(iBitIndex); + } + else + { + fDissimilarLumaMin = ffxMin(fDissimilarLumaMin, fSamples[iSampleIndex]); + fDissimilarLumaMax = ffxMax(fDissimilarLumaMax, fSamples[iSampleIndex]); + } + } + + const FfxBoolean bIsRidge = fSamples[iNucleusIndex] > fDissimilarLumaMax || fSamples[iNucleusIndex] < fDissimilarLumaMin; + + if (FFX_FALSE == bIsRidge) + { + return 0.0f; + } + + FFX_UNROLL + for (FfxInt32 i = 0; i < uNumRejectionMasks; i++) + { + if ((uPatternMask & uRejectionMasks[i]) == uRejectionMasks[i]) + { + return 0.0f; + } + } + + return 1.0f - fLumaMin / fLumaMax; +} + +FfxFloat32 UpdateAccumulation(FfxInt32x2 iPxPos, FfxFloat32x2 fUv, FfxFloat32x2 fMotionVector, FfxFloat32 fDisocclusion, FfxFloat32 fShadingChange) +{ + const FfxFloat32x2 fReprojectedUv = fUv + fMotionVector; + FfxFloat32 fAccumulation = 0.0f; + + if (IsUvInside(fReprojectedUv)) { + const FfxFloat32x2 fReprojectedUv_HW = ClampUv(fReprojectedUv, PreviousFrameRenderSize(), MaxRenderSize()); + fAccumulation = ffxSaturate(SampleAccumulation(fReprojectedUv_HW)); + } + + fAccumulation = ffxLerp(fAccumulation, 0.0f, fShadingChange); + fAccumulation = ffxLerp(fAccumulation, ffxMin(fAccumulation, 0.25f), fDisocclusion); + + fAccumulation *= FfxFloat32(round(fAccumulation * 100.0f) > 1.0f); + + // Update for next frame, normalize to store in unorm + const FfxFloat32 fAccumulatedFramesMax = 3.0f; + const FfxFloat32 fAccumulatedFramesToStore = ffxSaturate(fAccumulation + (1.0f / fAccumulatedFramesMax)); + StoreAccumulation(iPxPos, fAccumulatedFramesToStore); + + return fAccumulation; +} + +FfxFloat32 ComputeShadingChange(FfxFloat32x2 fUv) +{ + // NOTE: Here we re-apply jitter, will be reverted again when sampled in accumulation pass + const FfxFloat32x2 fShadingChangeUv = ClampUv(fUv - Jitter() / RenderSize(), ShadingChangeRenderSize(), ShadingChangeMaxRenderSize()); + const FfxFloat32 fShadingChange = ffxSaturate(SampleShadingChange(fShadingChangeUv)); + + return fShadingChange; +} + +void PrepareReactivity(FfxInt32x2 iPxPos) +{ + const FfxFloat32x2 fUv = (iPxPos + 0.5f) / RenderSize(); + const FfxFloat32x2 fMotionVector = LoadDilatedMotionVector(iPxPos); + + // Discard small mvs + const FfxFloat32 f4KVelocity = Get4KVelocity(fMotionVector); + + const FfxFloat32x2 fDilatedUv = fUv + fMotionVector; + const FfxFloat32 fDilatedDepth = LoadDilatedDepth(iPxPos); + const FfxFloat32 fDepthInMeters = GetViewSpaceDepthInMeters(fDilatedDepth); + + const FfxFloat32 fDisocclusion = ComputeDisocclusions(fUv, fMotionVector, GetViewSpaceDepth(fDilatedDepth)); + const FfxFloat32 fShadingChange = ffxMax(DilateReactiveMasks(iPxPos, fUv), ComputeShadingChange(fUv)); + + const FfxFloat32 fMotionDivergence = ComputeMotionDivergence(fUv, fMotionVector, fDilatedDepth); + const FfxFloat32 fDilatedTransparencyAndComposition = DilateTransparencyAndCompositionMasks(iPxPos, fUv); + const FfxFloat32 fFinalReactiveness = ffxMax(fMotionDivergence, fDilatedTransparencyAndComposition); + + const FfxFloat32 fAccumulation = UpdateAccumulation(iPxPos, fUv, fMotionVector, fDisocclusion, fShadingChange); + + FfxFloat32x4 fOutput; + fOutput[REACTIVE] = fFinalReactiveness; + fOutput[DISOCCLUSION] = fDisocclusion; + fOutput[SHADING_CHANGE] = fShadingChange; + fOutput[ACCUMULAION] = fAccumulation; + + StoreDilatedReactiveMasks(iPxPos, fOutput); + + const FfxFloat32 fLockStrength = ComputeThinFeatureConfidence(iPxPos); + if (fLockStrength > (1.0f / 100.0f)) + { + StoreNewLocks(ComputeHrPosFromLrPos(FfxInt32x2(iPxPos)), fLockStrength); + } +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_prepare_reactivity.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_prepare_reactivity.h.meta new file mode 100644 index 0000000..46fcfc6 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_prepare_reactivity.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 0070b6300195a7649a899fe0ed099c82 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_rcas.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_rcas.h new file mode 100644 index 0000000..90a85b3 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_rcas.h @@ -0,0 +1,67 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#define GROUP_SIZE 8 +#define FSR_RCAS_DENOISE 1 + +#include "../ffx_core.h" + +void WriteUpscaledOutput(FFX_MIN16_U2 iPxHrPos, FfxFloat32x3 fUpscaledColor) +{ + StoreUpscaledOutput(FFX_MIN16_I2(iPxHrPos), fUpscaledColor); +} + +#define FSR_RCAS_F 1 +FfxFloat32x4 FsrRcasLoadF(FfxInt32x2 p) +{ + FfxFloat32x4 fColor = LoadRCAS_Input(p); + + fColor.rgb *= Exposure(); + + return fColor; +} +void FsrRcasInputF(inout FfxFloat32 r, inout FfxFloat32 g, inout FfxFloat32 b) {} + +#include "../fsr1/ffx_fsr1.h" + +void CurrFilter(FFX_MIN16_U2 pos) +{ + FfxFloat32x3 c; + FsrRcasF(c.r, c.g, c.b, pos, RCASConfig()); + + c /= Exposure(); + + WriteUpscaledOutput(pos, c); +} + +void RCAS(FfxUInt32x3 LocalThreadId, FfxUInt32x3 WorkGroupId, FfxUInt32x3 Dtid) +{ + // Do remapping of local xy in workgroup for a more PS-like swizzle pattern. + FfxUInt32x2 gxy = ffxRemapForQuad(LocalThreadId.x) + FfxUInt32x2(WorkGroupId.x << 4u, WorkGroupId.y << 4u); + CurrFilter(FFX_MIN16_U2(gxy)); + gxy.x += 8u; + CurrFilter(FFX_MIN16_U2(gxy)); + gxy.y += 8u; + CurrFilter(FFX_MIN16_U2(gxy)); + gxy.x -= 8u; + CurrFilter(FFX_MIN16_U2(gxy)); +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_rcas.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_rcas.h.meta new file mode 100644 index 0000000..f21aeb7 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_rcas.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: dc2e80d2251e46c4d9af5696c3061fab +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_reproject.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_reproject.h new file mode 100644 index 0000000..153a9b7 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_reproject.h @@ -0,0 +1,64 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#ifndef FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE +#define FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE 0 // Reference +#endif + +FfxFloat32x4 WrapHistory(FfxInt32x2 iPxSample) +{ + return LoadHistory(iPxSample); +} + +DeclareCustomFetchBicubicSamples(FetchHistorySamples, WrapHistory) +DeclareCustomTextureSample(HistorySample, FFX_FSR3UPSCALER_GET_LANCZOS_SAMPLER1D(FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE), FetchHistorySamples) + +FfxFloat32x2 GetMotionVector(FfxInt32x2 iPxHrPos, FfxFloat32x2 fHrUv) +{ +#if FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS + const FfxFloat32x2 fDilatedMotionVector = LoadDilatedMotionVector(FFX_MIN16_I2(fHrUv * RenderSize())); +#else + const FfxFloat32x2 fDilatedMotionVector = LoadInputMotionVector(iPxHrPos); +#endif + + return fDilatedMotionVector; +} + +void ComputeReprojectedUVs(const AccumulationPassCommonParams params, FFX_PARAMETER_OUT FfxFloat32x2 fReprojectedHrUv, FFX_PARAMETER_OUT FfxBoolean bIsExistingSample) +{ + fReprojectedHrUv = params.fHrUv + params.fMotionVector; + + bIsExistingSample = IsUvInside(fReprojectedHrUv); +} + +void ReprojectHistoryColor(const AccumulationPassCommonParams params, FFX_PARAMETER_INOUT AccumulationPassData data) +{ + const FfxFloat32x4 fReprojectedHistory = HistorySample(params.fReprojectedHrUv, PreviousFrameUpscaleSize()); + + data.fHistoryColor = fReprojectedHistory.rgb; + data.fHistoryColor *= DeltaPreExposure(); + data.fHistoryColor *= Exposure(); + + data.fHistoryColor = RGBToYCoCg(data.fHistoryColor); + + data.fLock = fReprojectedHistory.w; +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_reproject.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_reproject.h.meta new file mode 100644 index 0000000..cb9c6d7 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_reproject.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 2bcf78cbd908b664cb9c656e28b430d4 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_resources.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_resources.h new file mode 100644 index 0000000..b3d8ddb --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_resources.h @@ -0,0 +1,100 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#ifndef FFX_FSR3UPSCALER_RESOURCES_H +#define FFX_FSR3UPSCALER_RESOURCES_H + +#if defined(FFX_CPU) || defined(FFX_GPU) +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_NULL 0 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INPUT_OPAQUE_ONLY 1 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INPUT_COLOR 2 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INPUT_MOTION_VECTORS 3 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INPUT_DEPTH 4 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INPUT_EXPOSURE 5 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INPUT_REACTIVE_MASK 6 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INPUT_TRANSPARENCY_AND_COMPOSITION_MASK 7 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_RECONSTRUCTED_PREVIOUS_NEAREST_DEPTH 8 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_DILATED_MOTION_VECTORS 9 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_DILATED_DEPTH 10 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INTERNAL_UPSCALED_COLOR 11 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_ACCUMULATION 12 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_NEW_LOCKS 13 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LUMA_HISTORY 14 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_DEBUG_OUTPUT 15 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LANCZOS_LUT 16 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SPD_ATOMIC_COUNT 17 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_UPSCALED_OUTPUT 18 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_RCAS_INPUT 19 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_ACCUMULATION_1 20 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_ACCUMULATION_2 21 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INTERNAL_UPSCALED_COLOR_1 22 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INTERNAL_UPSCALED_COLOR_2 23 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INTERNAL_DEFAULT_REACTIVITY 24 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INTERNAL_DEFAULT_TRANSPARENCY_AND_COMPOSITION 25 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_DILATED_REACTIVE_MASKS 26 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SPD_MIPS 27 // same as FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SPD_MIPS_LEVEL_0 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SPD_MIPS_LEVEL_0 27 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SPD_MIPS_LEVEL_1 28 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SPD_MIPS_LEVEL_2 29 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SPD_MIPS_LEVEL_3 30 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SPD_MIPS_LEVEL_4 31 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SPD_MIPS_LEVEL_5 32 + +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INTERNAL_DEFAULT_EXPOSURE 33 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_FRAME_INFO 34 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_AUTOREACTIVE 35 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_AUTOCOMPOSITION_DEPRECATED 36 + +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LUMA_HISTORY_1 37 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LUMA_HISTORY_2 38 + +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LUMA_1 40 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LUMA_2 41 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SHADING_CHANGE 42 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_FARTHEST_DEPTH 43 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_FARTHEST_DEPTH_MIP1 44 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_CURRENT_LUMA 45 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_PREVIOUS_LUMA 46 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LUMA_INSTABILITY 48 +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INTERMEDIATE_FP16x1 49 + + +// Shading change detection mip level setting, value must be in the range [FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_0, FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_12] +//#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_SHADING_CHANGE FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_4 +//#define FFX_FSR3UPSCALER_SHADING_CHANGE_MIP_LEVEL (FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_SHADING_CHANGE - FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE) + +#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_COUNT 60 + +#define FFX_FSR3UPSCALER_CONSTANTBUFFER_IDENTIFIER_FSR3UPSCALER 0 +#define FFX_FSR3UPSCALER_CONSTANTBUFFER_IDENTIFIER_SPD 1 +#define FFX_FSR3UPSCALER_CONSTANTBUFFER_IDENTIFIER_RCAS 2 +#define FFX_FSR3UPSCALER_CONSTANTBUFFER_IDENTIFIER_GENREACTIVE 3 +#define FFX_FSR3UPSCALER_CONSTANTBUFFER_COUNT 4 + +#define FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_APPLY_TONEMAP 1 +#define FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_APPLY_INVERSETONEMAP 2 +#define FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_APPLY_THRESHOLD 4 +#define FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_USE_COMPONENTS_MAX 8 + +#endif // #if defined(FFX_CPU) || defined(FFX_GPU) + +#endif //!defined( FFX_FSR3UPSCALER_RESOURCES_H ) diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_resources.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_resources.h.meta new file mode 100644 index 0000000..591d1d8 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_resources.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 0a6f299994574d641a6a2f864d6b78b7 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_sample.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_sample.h new file mode 100644 index 0000000..5f727b1 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_sample.h @@ -0,0 +1,602 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +#ifndef FFX_FSR3UPSCALER_SAMPLE_H +#define FFX_FSR3UPSCALER_SAMPLE_H + +// suppress warnings +#ifdef FFX_HLSL +#pragma warning(disable: 4008) // potentially divide by zero +#endif //FFX_HLSL + +struct FetchedBilinearSamples { + + FfxFloat32x4 fColor00; + FfxFloat32x4 fColor10; + + FfxFloat32x4 fColor01; + FfxFloat32x4 fColor11; +}; + +struct FetchedBicubicSamples { + + FfxFloat32x4 fColor00; + FfxFloat32x4 fColor10; + FfxFloat32x4 fColor20; + FfxFloat32x4 fColor30; + + FfxFloat32x4 fColor01; + FfxFloat32x4 fColor11; + FfxFloat32x4 fColor21; + FfxFloat32x4 fColor31; + + FfxFloat32x4 fColor02; + FfxFloat32x4 fColor12; + FfxFloat32x4 fColor22; + FfxFloat32x4 fColor32; + + FfxFloat32x4 fColor03; + FfxFloat32x4 fColor13; + FfxFloat32x4 fColor23; + FfxFloat32x4 fColor33; +}; + +#if FFX_HALF +struct FetchedBilinearSamplesMin16 { + + FFX_MIN16_F4 fColor00; + FFX_MIN16_F4 fColor10; + + FFX_MIN16_F4 fColor01; + FFX_MIN16_F4 fColor11; +}; + +struct FetchedBicubicSamplesMin16 { + + FFX_MIN16_F4 fColor00; + FFX_MIN16_F4 fColor10; + FFX_MIN16_F4 fColor20; + FFX_MIN16_F4 fColor30; + + FFX_MIN16_F4 fColor01; + FFX_MIN16_F4 fColor11; + FFX_MIN16_F4 fColor21; + FFX_MIN16_F4 fColor31; + + FFX_MIN16_F4 fColor02; + FFX_MIN16_F4 fColor12; + FFX_MIN16_F4 fColor22; + FFX_MIN16_F4 fColor32; + + FFX_MIN16_F4 fColor03; + FFX_MIN16_F4 fColor13; + FFX_MIN16_F4 fColor23; + FFX_MIN16_F4 fColor33; +}; +#else //FFX_HALF +#define FetchedBicubicSamplesMin16 FetchedBicubicSamples +#define FetchedBilinearSamplesMin16 FetchedBilinearSamples +#endif //FFX_HALF + +FfxFloat32x4 Linear(FfxFloat32x4 A, FfxFloat32x4 B, FfxFloat32 t) +{ + return A + (B - A) * t; +} + +FfxFloat32x4 Bilinear(FetchedBilinearSamples BilinearSamples, FfxFloat32x2 fPxFrac) +{ + FfxFloat32x4 fColorX0 = Linear(BilinearSamples.fColor00, BilinearSamples.fColor10, fPxFrac.x); + FfxFloat32x4 fColorX1 = Linear(BilinearSamples.fColor01, BilinearSamples.fColor11, fPxFrac.x); + FfxFloat32x4 fColorXY = Linear(fColorX0, fColorX1, fPxFrac.y); + return fColorXY; +} + +#if FFX_HALF +FFX_MIN16_F4 Linear(FFX_MIN16_F4 A, FFX_MIN16_F4 B, FFX_MIN16_F t) +{ + return A + (B - A) * t; +} + +FFX_MIN16_F4 Bilinear(FetchedBilinearSamplesMin16 BilinearSamples, FFX_MIN16_F2 fPxFrac) +{ + FFX_MIN16_F4 fColorX0 = Linear(BilinearSamples.fColor00, BilinearSamples.fColor10, fPxFrac.x); + FFX_MIN16_F4 fColorX1 = Linear(BilinearSamples.fColor01, BilinearSamples.fColor11, fPxFrac.x); + FFX_MIN16_F4 fColorXY = Linear(fColorX0, fColorX1, fPxFrac.y); + return fColorXY; +} +#endif + +FfxFloat32 Lanczos2NoClamp(FfxFloat32 x) +{ + const FfxFloat32 PI = 3.141592653589793f; // TODO: share SDK constants + return abs(x) < FSR3UPSCALER_EPSILON ? 1.f : (sin(PI * x) / (PI * x)) * (sin(0.5f * PI * x) / (0.5f * PI * x)); +} + +FfxFloat32 Lanczos2(FfxFloat32 x) +{ + x = ffxMin(abs(x), 2.0f); + return Lanczos2NoClamp(x); +} + +#if FFX_HALF + +#if 0 +FFX_MIN16_F Lanczos2NoClamp(FFX_MIN16_F x) +{ + const FFX_MIN16_F PI = FFX_MIN16_F(3.141592653589793f); // TODO: share SDK constants + return abs(x) < FFX_MIN16_F(FSR3UPSCALER_EPSILON) ? FFX_MIN16_F(1.f) : (sin(PI * x) / (PI * x)) * (sin(FFX_MIN16_F(0.5f) * PI * x) / (FFX_MIN16_F(0.5f) * PI * x)); +} +#endif + +FFX_MIN16_F Lanczos2(FFX_MIN16_F x) +{ + x = ffxMin(abs(x), FFX_MIN16_F(2.0f)); + return FFX_MIN16_F(Lanczos2NoClamp(x)); +} +#endif //FFX_HALF + +// FSR1 lanczos approximation. Input is x*x and must be <= 4. +FfxFloat32 Lanczos2ApproxSqNoClamp(FfxFloat32 x2) +{ + FfxFloat32 a = (2.0f / 5.0f) * x2 - 1; + FfxFloat32 b = (1.0f / 4.0f) * x2 - 1; + return ((25.0f / 16.0f) * a * a - (25.0f / 16.0f - 1)) * (b * b); +} + +#if FFX_HALF +FFX_MIN16_F Lanczos2ApproxSqNoClamp(FFX_MIN16_F x2) +{ + FFX_MIN16_F a = FFX_MIN16_F(2.0f / 5.0f) * x2 - FFX_MIN16_F(1); + FFX_MIN16_F b = FFX_MIN16_F(1.0f / 4.0f) * x2 - FFX_MIN16_F(1); + return (FFX_MIN16_F(25.0f / 16.0f) * a * a - FFX_MIN16_F(25.0f / 16.0f - 1)) * (b * b); +} +#endif //FFX_HALF + +FfxFloat32 Lanczos2ApproxSq(FfxFloat32 x2) +{ + x2 = ffxMin(x2, 4.0f); + return Lanczos2ApproxSqNoClamp(x2); +} + +#if FFX_HALF +FFX_MIN16_F Lanczos2ApproxSq(FFX_MIN16_F x2) +{ + x2 = ffxMin(x2, FFX_MIN16_F(4.0f)); + return Lanczos2ApproxSqNoClamp(x2); +} +#endif //FFX_HALF + +FfxFloat32 Lanczos2ApproxNoClamp(FfxFloat32 x) +{ + return Lanczos2ApproxSqNoClamp(x * x); +} + +#if FFX_HALF +FFX_MIN16_F Lanczos2ApproxNoClamp(FFX_MIN16_F x) +{ + return Lanczos2ApproxSqNoClamp(x * x); +} +#endif //FFX_HALF + +FfxFloat32 Lanczos2Approx(FfxFloat32 x) +{ + return Lanczos2ApproxSq(x * x); +} + +#if FFX_HALF +FFX_MIN16_F Lanczos2Approx(FFX_MIN16_F x) +{ + return Lanczos2ApproxSq(x * x); +} +#endif //FFX_HALF + +FfxFloat32 Lanczos2_UseLUT(FfxFloat32 x) +{ + return SampleLanczos2Weight(abs(x)); +} + +#if FFX_HALF +FFX_MIN16_F Lanczos2_UseLUT(FFX_MIN16_F x) +{ + return FFX_MIN16_F(SampleLanczos2Weight(abs(x))); +} +#endif //FFX_HALF + +FfxFloat32x4 Lanczos2_UseLUT(FfxFloat32x4 fColor0, FfxFloat32x4 fColor1, FfxFloat32x4 fColor2, FfxFloat32x4 fColor3, FfxFloat32 t) +{ + FfxFloat32 fWeight0 = Lanczos2_UseLUT(-1.f - t); + FfxFloat32 fWeight1 = Lanczos2_UseLUT(-0.f - t); + FfxFloat32 fWeight2 = Lanczos2_UseLUT(+1.f - t); + FfxFloat32 fWeight3 = Lanczos2_UseLUT(+2.f - t); + return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); +} +#if FFX_HALF +FFX_MIN16_F4 Lanczos2_UseLUT(FFX_MIN16_F4 fColor0, FFX_MIN16_F4 fColor1, FFX_MIN16_F4 fColor2, FFX_MIN16_F4 fColor3, FFX_MIN16_F t) +{ + FFX_MIN16_F fWeight0 = Lanczos2_UseLUT(FFX_MIN16_F(-1.f) - t); + FFX_MIN16_F fWeight1 = Lanczos2_UseLUT(FFX_MIN16_F(-0.f) - t); + FFX_MIN16_F fWeight2 = Lanczos2_UseLUT(FFX_MIN16_F(+1.f) - t); + FFX_MIN16_F fWeight3 = Lanczos2_UseLUT(FFX_MIN16_F(+2.f) - t); + return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); +} +#endif + +FfxFloat32x4 Lanczos2(FfxFloat32x4 fColor0, FfxFloat32x4 fColor1, FfxFloat32x4 fColor2, FfxFloat32x4 fColor3, FfxFloat32 t) +{ + FfxFloat32 fWeight0 = Lanczos2(-1.f - t); + FfxFloat32 fWeight1 = Lanczos2(-0.f - t); + FfxFloat32 fWeight2 = Lanczos2(+1.f - t); + FfxFloat32 fWeight3 = Lanczos2(+2.f - t); + return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); +} + +FfxFloat32x4 Lanczos2(FetchedBicubicSamples Samples, FfxFloat32x2 fPxFrac) +{ + FfxFloat32x4 fColorX0 = Lanczos2(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); + FfxFloat32x4 fColorX1 = Lanczos2(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); + FfxFloat32x4 fColorX2 = Lanczos2(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); + FfxFloat32x4 fColorX3 = Lanczos2(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); + FfxFloat32x4 fColorXY = Lanczos2(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); + + // Deringing + + // TODO: only use 4 by checking jitter + const FfxInt32 iDeringingSampleCount = 4; + const FfxFloat32x4 fDeringingSamples[4] = { + Samples.fColor11, + Samples.fColor21, + Samples.fColor12, + Samples.fColor22, + }; + + FfxFloat32x4 fDeringingMin = fDeringingSamples[0]; + FfxFloat32x4 fDeringingMax = fDeringingSamples[0]; + + FFX_UNROLL + for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) { + + fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); + fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); + } + + fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); + + return fColorXY; +} + +#if FFX_HALF +FFX_MIN16_F4 Lanczos2(FFX_MIN16_F4 fColor0, FFX_MIN16_F4 fColor1, FFX_MIN16_F4 fColor2, FFX_MIN16_F4 fColor3, FFX_MIN16_F t) +{ + FFX_MIN16_F fWeight0 = Lanczos2(FFX_MIN16_F(-1.f) - t); + FFX_MIN16_F fWeight1 = Lanczos2(FFX_MIN16_F(-0.f) - t); + FFX_MIN16_F fWeight2 = Lanczos2(FFX_MIN16_F(+1.f) - t); + FFX_MIN16_F fWeight3 = Lanczos2(FFX_MIN16_F(+2.f) - t); + return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); +} + +FFX_MIN16_F4 Lanczos2(FetchedBicubicSamplesMin16 Samples, FFX_MIN16_F2 fPxFrac) +{ + FFX_MIN16_F4 fColorX0 = Lanczos2(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); + FFX_MIN16_F4 fColorX1 = Lanczos2(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); + FFX_MIN16_F4 fColorX2 = Lanczos2(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); + FFX_MIN16_F4 fColorX3 = Lanczos2(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); + FFX_MIN16_F4 fColorXY = Lanczos2(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); + + // Deringing + + // TODO: only use 4 by checking jitter + const FfxInt32 iDeringingSampleCount = 4; + const FFX_MIN16_F4 fDeringingSamples[4] = { + Samples.fColor11, + Samples.fColor21, + Samples.fColor12, + Samples.fColor22, + }; + + FFX_MIN16_F4 fDeringingMin = fDeringingSamples[0]; + FFX_MIN16_F4 fDeringingMax = fDeringingSamples[0]; + + FFX_UNROLL + for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) + { + fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); + fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); + } + + fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); + + return fColorXY; +} +#endif //FFX_HALF + + +FfxFloat32x4 Lanczos2LUT(FetchedBicubicSamples Samples, FfxFloat32x2 fPxFrac) +{ + FfxFloat32x4 fColorX0 = Lanczos2_UseLUT(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); + FfxFloat32x4 fColorX1 = Lanczos2_UseLUT(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); + FfxFloat32x4 fColorX2 = Lanczos2_UseLUT(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); + FfxFloat32x4 fColorX3 = Lanczos2_UseLUT(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); + FfxFloat32x4 fColorXY = Lanczos2_UseLUT(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); + + // Deringing + + // TODO: only use 4 by checking jitter + const FfxInt32 iDeringingSampleCount = 4; + const FfxFloat32x4 fDeringingSamples[4] = { + Samples.fColor11, + Samples.fColor21, + Samples.fColor12, + Samples.fColor22, + }; + + FfxFloat32x4 fDeringingMin = fDeringingSamples[0]; + FfxFloat32x4 fDeringingMax = fDeringingSamples[0]; + + FFX_UNROLL + for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) { + + fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); + fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); + } + + fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); + + return fColorXY; +} + +#if FFX_HALF +FFX_MIN16_F4 Lanczos2LUT(FetchedBicubicSamplesMin16 Samples, FFX_MIN16_F2 fPxFrac) +{ + FFX_MIN16_F4 fColorX0 = Lanczos2_UseLUT(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); + FFX_MIN16_F4 fColorX1 = Lanczos2_UseLUT(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); + FFX_MIN16_F4 fColorX2 = Lanczos2_UseLUT(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); + FFX_MIN16_F4 fColorX3 = Lanczos2_UseLUT(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); + FFX_MIN16_F4 fColorXY = Lanczos2_UseLUT(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); + + // Deringing + + // TODO: only use 4 by checking jitter + const FfxInt32 iDeringingSampleCount = 4; + const FFX_MIN16_F4 fDeringingSamples[4] = { + Samples.fColor11, + Samples.fColor21, + Samples.fColor12, + Samples.fColor22, + }; + + FFX_MIN16_F4 fDeringingMin = fDeringingSamples[0]; + FFX_MIN16_F4 fDeringingMax = fDeringingSamples[0]; + + FFX_UNROLL + for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) + { + fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); + fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); + } + + fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); + + return fColorXY; +} +#endif //FFX_HALF + + + +FfxFloat32x4 Lanczos2Approx(FfxFloat32x4 fColor0, FfxFloat32x4 fColor1, FfxFloat32x4 fColor2, FfxFloat32x4 fColor3, FfxFloat32 t) +{ + FfxFloat32 fWeight0 = Lanczos2ApproxNoClamp(-1.f - t); + FfxFloat32 fWeight1 = Lanczos2ApproxNoClamp(-0.f - t); + FfxFloat32 fWeight2 = Lanczos2ApproxNoClamp(+1.f - t); + FfxFloat32 fWeight3 = Lanczos2ApproxNoClamp(+2.f - t); + return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); +} + +#if FFX_HALF +FFX_MIN16_F4 Lanczos2Approx(FFX_MIN16_F4 fColor0, FFX_MIN16_F4 fColor1, FFX_MIN16_F4 fColor2, FFX_MIN16_F4 fColor3, FFX_MIN16_F t) +{ + FFX_MIN16_F fWeight0 = Lanczos2ApproxNoClamp(FFX_MIN16_F(-1.f) - t); + FFX_MIN16_F fWeight1 = Lanczos2ApproxNoClamp(FFX_MIN16_F(-0.f) - t); + FFX_MIN16_F fWeight2 = Lanczos2ApproxNoClamp(FFX_MIN16_F(+1.f) - t); + FFX_MIN16_F fWeight3 = Lanczos2ApproxNoClamp(FFX_MIN16_F(+2.f) - t); + return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); +} +#endif //FFX_HALF + +FfxFloat32x4 Lanczos2Approx(FetchedBicubicSamples Samples, FfxFloat32x2 fPxFrac) +{ + FfxFloat32x4 fColorX0 = Lanczos2Approx(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); + FfxFloat32x4 fColorX1 = Lanczos2Approx(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); + FfxFloat32x4 fColorX2 = Lanczos2Approx(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); + FfxFloat32x4 fColorX3 = Lanczos2Approx(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); + FfxFloat32x4 fColorXY = Lanczos2Approx(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); + + // Deringing + + // TODO: only use 4 by checking jitter + const FfxInt32 iDeringingSampleCount = 4; + const FfxFloat32x4 fDeringingSamples[4] = { + Samples.fColor11, + Samples.fColor21, + Samples.fColor12, + Samples.fColor22, + }; + + FfxFloat32x4 fDeringingMin = fDeringingSamples[0]; + FfxFloat32x4 fDeringingMax = fDeringingSamples[0]; + + FFX_UNROLL + for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) + { + fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); + fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); + } + + fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); + + return fColorXY; +} + +#if FFX_HALF +FFX_MIN16_F4 Lanczos2Approx(FetchedBicubicSamplesMin16 Samples, FFX_MIN16_F2 fPxFrac) +{ + FFX_MIN16_F4 fColorX0 = Lanczos2Approx(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); + FFX_MIN16_F4 fColorX1 = Lanczos2Approx(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); + FFX_MIN16_F4 fColorX2 = Lanczos2Approx(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); + FFX_MIN16_F4 fColorX3 = Lanczos2Approx(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); + FFX_MIN16_F4 fColorXY = Lanczos2Approx(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); + + // Deringing + + // TODO: only use 4 by checking jitter + const FfxInt32 iDeringingSampleCount = 4; + const FFX_MIN16_F4 fDeringingSamples[4] = { + Samples.fColor11, + Samples.fColor21, + Samples.fColor12, + Samples.fColor22, + }; + + FFX_MIN16_F4 fDeringingMin = fDeringingSamples[0]; + FFX_MIN16_F4 fDeringingMax = fDeringingSamples[0]; + + FFX_UNROLL + for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) + { + fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); + fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); + } + + fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); + + return fColorXY; +} +#endif + +// Clamp by offset direction. Assuming iPxSample is already in range and iPxOffset is compile time constant. +FfxInt32x2 ClampCoord(FfxInt32x2 iPxSample, FfxInt32x2 iPxOffset, FfxInt32x2 iTextureSize) +{ + FfxInt32x2 result = iPxSample + iPxOffset; + result.x = ffxMax(1, ffxMin(result.x, iTextureSize.x - 2)); + result.y = ffxMax(1, ffxMin(result.y, iTextureSize.y - 2)); + return result; +} +#if FFX_HALF +FFX_MIN16_I2 ClampCoord(FFX_MIN16_I2 iPxSample, FFX_MIN16_I2 iPxOffset, FFX_MIN16_I2 iTextureSize) +{ + FFX_MIN16_I2 result = iPxSample + iPxOffset; + result.x = ffxMax(FFX_MIN16_I(1), ffxMin(result.x, iTextureSize.x - FFX_MIN16_I(2))); + result.y = ffxMax(FFX_MIN16_I(1), ffxMin(result.y, iTextureSize.y - FFX_MIN16_I(2))); + return result; +} +#endif //FFX_HALF + + +#define DeclareCustomFetchBicubicSamplesWithType(SampleType, TextureType, AddrType, Name, LoadTexture) \ + SampleType Name(AddrType iPxSample, AddrType iTextureSize) \ + { \ + SampleType Samples; \ + \ + Samples.fColor00 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(-1, -1), iTextureSize))); \ + Samples.fColor10 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, -1), iTextureSize))); \ + Samples.fColor20 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, -1), iTextureSize))); \ + Samples.fColor30 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+2, -1), iTextureSize))); \ + \ + Samples.fColor01 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(-1, +0), iTextureSize))); \ + Samples.fColor11 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, +0), iTextureSize))); \ + Samples.fColor21 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, +0), iTextureSize))); \ + Samples.fColor31 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+2, +0), iTextureSize))); \ + \ + Samples.fColor02 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(-1, +1), iTextureSize))); \ + Samples.fColor12 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, +1), iTextureSize))); \ + Samples.fColor22 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, +1), iTextureSize))); \ + Samples.fColor32 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+2, +1), iTextureSize))); \ + \ + Samples.fColor03 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(-1, +2), iTextureSize))); \ + Samples.fColor13 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, +2), iTextureSize))); \ + Samples.fColor23 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, +2), iTextureSize))); \ + Samples.fColor33 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+2, +2), iTextureSize))); \ + \ + return Samples; \ + } + +#define DeclareCustomFetchBicubicSamples(Name, LoadTexture) \ + DeclareCustomFetchBicubicSamplesWithType(FetchedBicubicSamples, FfxFloat32x4, FfxInt32x2, Name, LoadTexture) + +#define DeclareCustomFetchBicubicSamplesMin16(Name, LoadTexture) \ + DeclareCustomFetchBicubicSamplesWithType(FetchedBicubicSamplesMin16, FFX_MIN16_F4, FfxInt32x2, Name, LoadTexture) + +#define DeclareCustomFetchBilinearSamplesWithType(SampleType, TextureType,AddrType, Name, LoadTexture) \ + SampleType Name(AddrType iPxSample, AddrType iTextureSize) \ + { \ + SampleType Samples; \ + Samples.fColor00 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, +0), iTextureSize))); \ + Samples.fColor10 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, +0), iTextureSize))); \ + Samples.fColor01 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, +1), iTextureSize))); \ + Samples.fColor11 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, +1), iTextureSize))); \ + return Samples; \ + } + +#define DeclareCustomFetchBilinearSamples(Name, LoadTexture) \ + DeclareCustomFetchBilinearSamplesWithType(FetchedBilinearSamples, FfxFloat32x4, FfxInt32x2, Name, LoadTexture) + +#define DeclareCustomFetchBilinearSamplesMin16(Name, LoadTexture) \ + DeclareCustomFetchBilinearSamplesWithType(FetchedBilinearSamplesMin16, FFX_MIN16_F4, FfxInt32x2, Name, LoadTexture) + +// BE CAREFUL: there is some precision issues and (3253, 125) leading to (3252.9989778, 125.001102) +// is common, so iPxSample can "jitter" +#define DeclareCustomTextureSample(Name, InterpolateSamples, FetchSamples) \ + FfxFloat32x4 Name(FfxFloat32x2 fUvSample, FfxInt32x2 iTextureSize) \ + { \ + FfxFloat32x2 fPxSample = (fUvSample * FfxFloat32x2(iTextureSize)) - FfxFloat32x2(0.5f, 0.5f); \ + FfxFloat32x2 fPxFrac = ffxFract(fPxSample); \ + /* Clamp base coords */ \ + fPxSample.x = ffxMax(0.0f, ffxMin(FfxFloat32(iTextureSize.x-1), fPxSample.x)); \ + fPxSample.y = ffxMax(0.0f, ffxMin(FfxFloat32(iTextureSize.y-1), fPxSample.y)); \ + /* */ \ + FfxInt32x2 iPxSample = FfxInt32x2(floor(fPxSample)); \ + FfxFloat32x4 fColorXY = FfxFloat32x4(InterpolateSamples(FetchSamples(iPxSample, iTextureSize), fPxFrac)); \ + return fColorXY; \ + } + +#define DeclareCustomTextureSampleMin16(Name, InterpolateSamples, FetchSamples) \ + FFX_MIN16_F4 Name(FfxFloat32x2 fUvSample, FfxInt32x2 iTextureSize) \ + { \ + FfxFloat32x2 fPxSample = (fUvSample * FfxFloat32x2(iTextureSize)) - FfxFloat32x2(0.5f, 0.5f); \ + FFX_MIN16_F2 fPxFrac = FFX_MIN16_F2(ffxFract(fPxSample)); \ + /* Clamp base coords */ \ + fPxSample.x = ffxMax(0.0f, ffxMin(FfxFloat32(iTextureSize.x), fPxSample.x)); \ + fPxSample.y = ffxMax(0.0f, ffxMin(FfxFloat32(iTextureSize.y), fPxSample.y)); \ + /* */ \ + FfxInt32x2 iPxSample = FfxInt32x2(floor(fPxSample)); \ + FFX_MIN16_F4 fColorXY = FFX_MIN16_F4(InterpolateSamples(FetchSamples(iPxSample, iTextureSize), fPxFrac)); \ + return fColorXY; \ + } + +#define FFX_FSR3UPSCALER_CONCAT_ID(x, y) x ## y +#define FFX_FSR3UPSCALER_CONCAT(x, y) FFX_FSR3UPSCALER_CONCAT_ID(x, y) +#define FFX_FSR3UPSCALER_SAMPLER_1D_0 Lanczos2 +#define FFX_FSR3UPSCALER_SAMPLER_1D_1 Lanczos2LUT +#define FFX_FSR3UPSCALER_SAMPLER_1D_2 Lanczos2Approx + +#define FFX_FSR3UPSCALER_GET_LANCZOS_SAMPLER1D(x) FFX_FSR3UPSCALER_CONCAT(FFX_FSR3UPSCALER_SAMPLER_1D_, x) + +#endif //!defined( FFX_FSR3UPSCALER_SAMPLE_H ) diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_sample.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_sample.h.meta new file mode 100644 index 0000000..1e2fde8 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_sample.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 0c406ac147d40644fab8636a35029185 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_shading_change.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_shading_change.h new file mode 100644 index 0000000..2eb23aa --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_shading_change.h @@ -0,0 +1,68 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +FFX_STATIC const FfxInt32 s_MipLevelsToUse = 3; + +struct ShadingChangeLumaInfo +{ + FfxFloat32 fSamples[s_MipLevelsToUse]; +}; + +ShadingChangeLumaInfo ComputeShadingChangeLuma(FfxInt32x2 iPxPos, FfxFloat32x2 fUv, const FfxInt32x2 iCurrentSize) +{ + ShadingChangeLumaInfo info; + + const FfxFloat32x2 fMipUv = ClampUv(fUv, ShadingChangeRenderSize(), GetSPDMipDimensions(0)); + + FFX_UNROLL + for (FfxInt32 iMipLevel = iShadingChangeMipStart; iMipLevel < s_MipLevelsToUse; iMipLevel++) { + + const FfxFloat32x2 fSample = SampleSPDMipLevel(fMipUv, iMipLevel); + + info.fSamples[iMipLevel] = abs(fSample.x * fSample.y); + } + + return info; +} + +void ShadingChange(FfxInt32x2 iPxPos) +{ + if (IsOnScreen(FfxInt32x2(iPxPos), ShadingChangeRenderSize())) { + + const FfxFloat32x2 fUv = (iPxPos + 0.5f) / ShadingChangeRenderSize(); + const FfxFloat32x2 fUvJittered = fUv + Jitter() / RenderSize(); + + const ShadingChangeLumaInfo info = ComputeShadingChangeLuma(iPxPos, fUvJittered, ShadingChangeRenderSize()); + + const FfxFloat32 fScale = 1.0f + iShadingChangeMipStart / s_MipLevelsToUse; + FfxFloat32 fShadingChange = 0.0f; + FFX_UNROLL + for (int iMipLevel = iShadingChangeMipStart; iMipLevel < s_MipLevelsToUse; iMipLevel++) + { + if (info.fSamples[iMipLevel] > 0) { + fShadingChange = ffxMax(fShadingChange, info.fSamples[iMipLevel]) * fScale; + } + } + + StoreShadingChange(iPxPos, ffxSaturate(fShadingChange)); + } +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_shading_change.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_shading_change.h.meta new file mode 100644 index 0000000..b2c234a --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_shading_change.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: d20f9266c072698448f3b51d3d478520 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_shading_change_pyramid.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_shading_change_pyramid.h new file mode 100644 index 0000000..651c5b3 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_shading_change_pyramid.h @@ -0,0 +1,299 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +FFX_GROUPSHARED FfxUInt32 spdCounter; + +void SpdIncreaseAtomicCounter(FfxUInt32 slice) +{ + SPD_IncreaseAtomicCounter(spdCounter); +} + +FfxUInt32 SpdGetAtomicCounter() +{ + return spdCounter; +} + +void SpdResetAtomicCounter(FfxUInt32 slice) +{ + SPD_ResetAtomicCounter(); +} + +#ifndef SPD_PACKED_ONLY +FFX_GROUPSHARED FfxFloat32 spdIntermediateR[16][16]; +FFX_GROUPSHARED FfxFloat32 spdIntermediateG[16][16]; +FFX_GROUPSHARED FfxFloat32 spdIntermediateB[16][16]; +FFX_GROUPSHARED FfxFloat32 spdIntermediateA[16][16]; + +FFX_STATIC const FfxInt32 DIFFERENCE = 0; +FFX_STATIC const FfxInt32 SIGN_SUM = 1; +FFX_STATIC const FfxInt32 MIP0_INDICATOR = 2; + +FfxFloat32x2 Sort2(FfxFloat32x2 v) +{ + return FfxFloat32x2(ffxMin(v.x, v.y), ffxMax(v.x, v.y)); +} + +struct SampleSet +{ + FfxFloat32 fSamples[SHADING_CHANGE_SET_SIZE]; +}; + +#define CompareSwap(i, j) \ +{ \ +FfxFloat32 fTmp = ffxMin(fSet.fSamples[i], fSet.fSamples[j]);\ +fSet.fSamples[j] = ffxMax(fSet.fSamples[i], fSet.fSamples[j]);\ +fSet.fSamples[i] = fTmp;\ +} + +#if SHADING_CHANGE_SET_SIZE == 5 +FFX_STATIC const FfxInt32x2 iSampleOffsets[5] = {FfxInt32x2(+0, +0), FfxInt32x2(-1, +0), FfxInt32x2(+1, +0), FfxInt32x2(+0, -1), FfxInt32x2(+0, +1)}; + +void SortSet(FFX_PARAMETER_INOUT SampleSet fSet) +{ + CompareSwap(0, 3); + CompareSwap(1, 4); + CompareSwap(0, 2); + CompareSwap(1, 3); + CompareSwap(0, 1); + CompareSwap(2, 4); + CompareSwap(1, 2); + CompareSwap(3, 4); + CompareSwap(2, 3); +} +#endif + +FfxFloat32 ComputeMinimumDifference(FfxInt32x2 iPxPos, SampleSet fSet0, SampleSet fSet1) +{ + FfxFloat32 fMinDiff = FSR3UPSCALER_FP16_MAX - 1; + FfxInt32 a = 0; + FfxInt32 b = 0; + + SortSet(fSet0); + SortSet(fSet1); + + const FfxFloat32 fMax = ffxMin(fSet0.fSamples[SHADING_CHANGE_SET_SIZE-1], fSet1.fSamples[SHADING_CHANGE_SET_SIZE-1]); + + if (fMax > FSR3UPSCALER_FP32_MIN) { + + FFX_UNROLL + for (FfxInt32 i = 0; i < SHADING_CHANGE_SET_SIZE && (fMinDiff < FSR3UPSCALER_FP16_MAX); i++) { + + FfxFloat32 fDiff = fSet0.fSamples[a] - fSet1.fSamples[b]; + + if (abs(fDiff) > FSR3UPSCALER_FP16_MIN) { + + fDiff = sign(fDiff) * (1.0f - MinDividedByMax(fSet0.fSamples[a], fSet1.fSamples[b])); + + fMinDiff = (abs(fDiff) < abs(fMinDiff)) ? fDiff : fMinDiff; + + a += FfxInt32(fSet0.fSamples[a] < fSet1.fSamples[b]); + b += FfxInt32(fSet0.fSamples[a] >= fSet1.fSamples[b]); + } + else + { + fMinDiff = FSR3UPSCALER_FP16_MAX; + } + } + } + + return fMinDiff * FfxFloat32(fMinDiff < (FSR3UPSCALER_FP16_MAX - 1)); +} + +SampleSet GetCurrentLumaBilinearSamples(FfxFloat32x2 fUv) +{ + const FfxFloat32x2 fUvJittered = fUv + Jitter() / RenderSize(); + const FfxInt32x2 iBasePos = FfxInt32x2(floor(fUvJittered * RenderSize())); + + SampleSet fSet; + + FFX_UNROLL + for (FfxInt32 iSampleIndex = 0; iSampleIndex < SHADING_CHANGE_SET_SIZE; iSampleIndex++) { + const FfxInt32x2 iSamplePos = ClampLoad(iBasePos, iSampleOffsets[iSampleIndex], RenderSize()); + fSet.fSamples[iSampleIndex] = LoadCurrentLuma(iSamplePos) * Exposure(); + fSet.fSamples[iSampleIndex] = ffxPow(fSet.fSamples[iSampleIndex], fShadingChangeSamplePow); + fSet.fSamples[iSampleIndex] = ffxMax(fSet.fSamples[iSampleIndex], FSR3UPSCALER_EPSILON); + } + + return fSet; +} + +struct PreviousLumaBilinearSamplesData +{ + SampleSet fSet; + FfxBoolean bIsExistingSample; +}; + +PreviousLumaBilinearSamplesData GetPreviousLumaBilinearSamples(FfxFloat32x2 fUv, FfxFloat32x2 fMotionVector) +{ + PreviousLumaBilinearSamplesData data = (PreviousLumaBilinearSamplesData)0; + + const FfxFloat32x2 fUvJittered = fUv + PreviousFrameJitter() / PreviousFrameRenderSize(); + const FfxFloat32x2 fReprojectedUv = fUvJittered + fMotionVector; + + data.bIsExistingSample = IsUvInside(fReprojectedUv); + + if (data.bIsExistingSample) { + + const FfxInt32x2 iBasePos = FfxInt32x2(floor(fReprojectedUv * PreviousFrameRenderSize())); + + FFX_UNROLL + for (FfxInt32 iSampleIndex = 0; iSampleIndex < SHADING_CHANGE_SET_SIZE; iSampleIndex++) { + + const FfxInt32x2 iSamplePos = ClampLoad(iBasePos, iSampleOffsets[iSampleIndex], PreviousFrameRenderSize()); + data.fSet.fSamples[iSampleIndex] = LoadPreviousLuma(iSamplePos) * DeltaPreExposure() * Exposure(); + data.fSet.fSamples[iSampleIndex] = ffxPow(data.fSet.fSamples[iSampleIndex], fShadingChangeSamplePow); + data.fSet.fSamples[iSampleIndex] = ffxMax(data.fSet.fSamples[iSampleIndex], FSR3UPSCALER_EPSILON); + } + } + + return data; +} + +FfxFloat32 ComputeDiff(FfxInt32x2 iPxPos, FfxFloat32x2 fUv, FfxFloat32x2 fMotionVector) +{ + FfxFloat32 fMinDiff = 0.0f; + + const SampleSet fCurrentSamples = GetCurrentLumaBilinearSamples(fUv); + const PreviousLumaBilinearSamplesData previousData = GetPreviousLumaBilinearSamples(fUv, fMotionVector); + + if (previousData.bIsExistingSample) { + fMinDiff = ComputeMinimumDifference(iPxPos, fCurrentSamples, previousData.fSet); + } + + return fMinDiff; +} + +FfxFloat32x4 SpdLoadSourceImage(FfxFloat32x2 iPxPos, FfxUInt32 slice) +{ + const FfxInt32x2 iPxSamplePos = ClampLoad(FfxInt32x2(iPxPos), FfxInt32x2(0, 0), FfxInt32x2(RenderSize())); + const FfxFloat32x2 fDilatedMotionVector = LoadDilatedMotionVector(iPxSamplePos); + const FfxFloat32x2 fUv = (iPxSamplePos + 0.5f) / RenderSize(); + + const FfxFloat32 fScaledAndSignedLumaDiff = ComputeDiff(iPxSamplePos, fUv, fDilatedMotionVector); + + FfxFloat32x4 fOutput = FfxFloat32x4(0.0f, 0.0f, 0.0f, 0.0f); + fOutput[DIFFERENCE] = fScaledAndSignedLumaDiff; + fOutput[SIGN_SUM] = (fScaledAndSignedLumaDiff != 0.0f) ? sign(fScaledAndSignedLumaDiff) : 0.0f; + fOutput[MIP0_INDICATOR] = 1.0f; + + return fOutput; +} + +FfxFloat32x4 SpdLoad(FfxInt32x2 tex, FfxUInt32 slice) +{ + return FfxFloat32x4(RWLoadPyramid(tex, 5), 0, 0); +} + +FfxFloat32x4 SpdReduce4(FfxFloat32x4 v0, FfxFloat32x4 v1, FfxFloat32x4 v2, FfxFloat32x4 v3) +{ + return (v0 + v1 + v2 + v3) * 0.25f; +} + +void SpdStore(FfxInt32x2 pix, FfxFloat32x4 outValue, FfxUInt32 index, FfxUInt32 slice) +{ + if (index >= iShadingChangeMipStart) + { + StorePyramid(pix, outValue.xy, index); + } +} + +FfxFloat32x4 SpdLoadIntermediate(FfxUInt32 x, FfxUInt32 y) +{ + return FfxFloat32x4( + spdIntermediateR[x][y], + spdIntermediateG[x][y], + spdIntermediateB[x][y], + spdIntermediateA[x][y]); +} +void SpdStoreIntermediate(FfxUInt32 x, FfxUInt32 y, FfxFloat32x4 value) +{ + spdIntermediateR[x][y] = value.x; + spdIntermediateG[x][y] = value.y; + spdIntermediateB[x][y] = value.z; + spdIntermediateA[x][y] = value.w; +} + +#endif + +// define fetch and store functions Packed +#if FFX_HALF + +FFX_GROUPSHARED FfxFloat16x2 spdIntermediateRG[16][16]; +FFX_GROUPSHARED FfxFloat16x2 spdIntermediateBA[16][16]; + +FfxFloat16x4 SpdLoadSourceImageH(FfxFloat32x2 tex, FfxUInt32 slice) +{ + return FfxFloat16x4(0, 0, 0, 0); +} + +FfxFloat16x4 SpdLoadH(FfxInt32x2 p, FfxUInt32 slice) +{ + return FfxFloat16x4(0, 0, 0, 0); +} + +void SpdStoreH(FfxInt32x2 p, FfxFloat16x4 value, FfxUInt32 mip, FfxUInt32 slice) +{ +} + +FfxFloat16x4 SpdLoadIntermediateH(FfxUInt32 x, FfxUInt32 y) +{ + return FfxFloat16x4( + spdIntermediateRG[x][y].x, + spdIntermediateRG[x][y].y, + spdIntermediateBA[x][y].x, + spdIntermediateBA[x][y].y); +} + +void SpdStoreIntermediateH(FfxUInt32 x, FfxUInt32 y, FfxFloat16x4 value) +{ + spdIntermediateRG[x][y] = value.xy; + spdIntermediateBA[x][y] = value.zw; +} + +FfxFloat16x4 SpdReduce4H(FfxFloat16x4 v0, FfxFloat16x4 v1, FfxFloat16x4 v2, FfxFloat16x4 v3) +{ + return (v0 + v1 + v2 + v3) * FfxFloat16(0.25); +} +#endif + +#include "../spd/ffx_spd.h" + +void ComputeShadingChangePyramid(FfxUInt32x3 WorkGroupId, FfxUInt32 LocalThreadIndex) +{ +#if FFX_HALF + SpdDownsampleH( + FfxUInt32x2(WorkGroupId.xy), + FfxUInt32(LocalThreadIndex), + FfxUInt32(MipCount()), + FfxUInt32(NumWorkGroups()), + FfxUInt32(WorkGroupId.z), + FfxUInt32x2(WorkGroupOffset())); +#else + SpdDownsample( + FfxUInt32x2(WorkGroupId.xy), + FfxUInt32(LocalThreadIndex), + FfxUInt32(MipCount()), + FfxUInt32(NumWorkGroups()), + FfxUInt32(WorkGroupId.z), + FfxUInt32x2(WorkGroupOffset())); +#endif +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_shading_change_pyramid.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_shading_change_pyramid.h.meta new file mode 100644 index 0000000..94175f1 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_shading_change_pyramid.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: fac4b69395377df409c7677a54515cd8 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_tcr_autogen.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_tcr_autogen.h similarity index 100% rename from Assets/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_tcr_autogen.h rename to Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_tcr_autogen.h diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_tcr_autogen.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_tcr_autogen.h.meta new file mode 100644 index 0000000..67b7d76 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_tcr_autogen.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 21f8219a8876f77478e60862240a41f8 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_upsample.h b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_upsample.h new file mode 100644 index 0000000..2d587f0 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_upsample.h @@ -0,0 +1,184 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +void Deringing(RectificationBox clippingBox, FFX_PARAMETER_INOUT FfxFloat32x3 fColor) +{ + fColor = clamp(fColor, clippingBox.aabbMin, clippingBox.aabbMax); +} + +#ifndef FFX_FSR3UPSCALER_OPTION_UPSAMPLE_USE_LANCZOS_TYPE +#define FFX_FSR3UPSCALER_OPTION_UPSAMPLE_USE_LANCZOS_TYPE 2 // Approximate +#endif + +FfxFloat32 GetUpsampleLanczosWeight(FfxFloat32x2 fSrcSampleOffset, FfxFloat32 fKernelWeight) +{ + FfxFloat32x2 fSrcSampleOffsetBiased = fSrcSampleOffset * fKernelWeight.xx; +#if FFX_FSR3UPSCALER_OPTION_UPSAMPLE_USE_LANCZOS_TYPE == 0 // LANCZOS_TYPE_REFERENCE + FfxFloat32 fSampleWeight = Lanczos2(length(fSrcSampleOffsetBiased)); +#elif FFX_FSR3UPSCALER_OPTION_UPSAMPLE_USE_LANCZOS_TYPE == 1 // LANCZOS_TYPE_LUT + FfxFloat32 fSampleWeight = Lanczos2_UseLUT(length(fSrcSampleOffsetBiased)); +#elif FFX_FSR3UPSCALER_OPTION_UPSAMPLE_USE_LANCZOS_TYPE == 2 // LANCZOS_TYPE_APPROXIMATE + FfxFloat32 fSampleWeight = Lanczos2ApproxSq(dot(fSrcSampleOffsetBiased, fSrcSampleOffsetBiased)); +#else +#error "Invalid Lanczos type" +#endif + return fSampleWeight; +} + +FfxFloat32 ComputeMaxKernelWeight(const AccumulationPassCommonParams params, FFX_PARAMETER_INOUT AccumulationPassData data) { + + const FfxFloat32 fKernelSizeBias = 1.0f + (1.0f / FfxFloat32x2(DownscaleFactor()) - 1.0f).x; + + return ffxMin(FfxFloat32(1.99f), fKernelSizeBias); +} + +FfxFloat32x3 LoadPreparedColor(FfxInt32x2 iSamplePos) +{ + const FfxFloat32x3 fRgb = ffxMax(FfxFloat32x3(0, 0, 0), LoadInputColor(iSamplePos)) * Exposure(); + const FfxFloat32x3 fPreparedYCoCg = RGBToYCoCg(fRgb); + + return fPreparedYCoCg; +} + +void ComputeUpsampledColorAndWeight(const AccumulationPassCommonParams params, FFX_PARAMETER_INOUT AccumulationPassData data) +{ + // We compute a sliced lanczos filter with 2 lobes (other slices are accumulated temporaly) + const FfxFloat32x2 fDstOutputPos = FfxFloat32x2(params.iPxHrPos) + FFX_BROADCAST_FLOAT32X2(0.5f); + const FfxFloat32x2 fSrcOutputPos = fDstOutputPos * DownscaleFactor(); + const FfxInt32x2 iSrcInputPos = FfxInt32x2(floor(fSrcOutputPos)); + const FfxFloat32x2 fSrcUnjitteredPos = (FfxFloat32x2(iSrcInputPos) + FfxFloat32x2(0.5f, 0.5f)) - Jitter(); // This is the un-jittered position of the sample at offset 0,0 + const FfxFloat32x2 fBaseSampleOffset = FfxFloat32x2(fSrcUnjitteredPos - fSrcOutputPos); + + FfxInt32x2 offsetTL; + offsetTL.x = (fSrcUnjitteredPos.x > fSrcOutputPos.x) ? FfxInt32(-2) : FfxInt32(-1); + offsetTL.y = (fSrcUnjitteredPos.y > fSrcOutputPos.y) ? FfxInt32(-2) : FfxInt32(-1); + + //Load samples + // If fSrcUnjitteredPos.y > fSrcOutputPos.y, indicates offsetTL.y = -2, sample offset Y will be [-2, 1], clipbox will be rows [1, 3]. + // Flip row# for sampling offset in this case, so first 0~2 rows in the sampled array can always be used for computing the clipbox. + // This reduces branch or cmove on sampled colors, but moving this overhead to sample position / weight calculation time which apply to less values. + const FfxBoolean bFlipRow = fSrcUnjitteredPos.y > fSrcOutputPos.y; + const FfxBoolean bFlipCol = fSrcUnjitteredPos.x > fSrcOutputPos.x; + const FfxFloat32x2 fOffsetTL = FfxFloat32x2(offsetTL); + + const FfxBoolean bIsInitialSample = (params.fAccumulation == 0.0f); + + FfxFloat32x3 fSamples[9]; + FfxInt32 iSampleIndex = 0; + + FFX_UNROLL + for (FfxInt32 row = 0; row < 3; row++) { + FFX_UNROLL + for (FfxInt32 col = 0; col < 3; col++) { + const FfxInt32x2 iSampleColRow = FfxInt32x2(bFlipCol ? (3 - col) : col, bFlipRow ? (3 - row) : row); + const FfxInt32x2 iSrcSamplePos = FfxInt32x2(iSrcInputPos) + offsetTL + iSampleColRow; + const FfxInt32x2 iSampleCoord = ClampLoad(iSrcSamplePos, FfxInt32x2(0, 0), FfxInt32x2(RenderSize())); + + fSamples[iSampleIndex] = LoadPreparedColor(iSampleCoord); + + ++iSampleIndex; + } + } + +#if FFX_FSR3UPSCALER_OPTION_HDR_COLOR_INPUT + if (bIsInitialSample) + { + for (iSampleIndex = 0; iSampleIndex < 9; ++iSampleIndex) + { + //YCoCg -> RGB -> Tonemap -> YCoCg (Use RGB tonemapper to avoid color desaturation) + fSamples[iSampleIndex] = RGBToYCoCg(Tonemap(YCoCgToRGB(fSamples[iSampleIndex]))); + } + } +#endif + + // Identify how much of each upsampled color to be used for this frame + const FfxFloat32 fKernelBiasMax = ComputeMaxKernelWeight(params, data); + const FfxFloat32 fKernelBiasMin = ffxMax(1.0f, ((1.0f + fKernelBiasMax) * 0.3f)); + + const FfxFloat32 fKernelBiasWeight = + ffxMin(1.0f - params.fDisocclusion * 0.5f, + ffxMin(1.0f - params.fShadingChange, + ffxSaturate(data.fHistoryWeight * 5.0f) + )); + + const FfxFloat32 fKernelBias = ffxLerp(fKernelBiasMin, fKernelBiasMax, fKernelBiasWeight); + + + iSampleIndex = 0; + + FFX_UNROLL + for (FfxInt32 row = 0; row < 3; row++) + { + FFX_UNROLL + for (FfxInt32 col = 0; col < 3; col++) + { + const FfxInt32x2 sampleColRow = FfxInt32x2(bFlipCol ? (3 - col) : col, bFlipRow ? (3 - row) : row); + const FfxFloat32x2 fOffset = fOffsetTL + FfxFloat32x2(sampleColRow); + const FfxFloat32x2 fSrcSampleOffset = fBaseSampleOffset + fOffset; + + const FfxInt32x2 iSrcSamplePos = FfxInt32x2(iSrcInputPos) + FfxInt32x2(offsetTL) + sampleColRow; + const FfxFloat32 fOnScreenFactor = FfxFloat32(IsOnScreen(FfxInt32x2(iSrcSamplePos), FfxInt32x2(RenderSize()))); + + if (!bIsInitialSample) + { + const FfxFloat32 fSampleWeight = fOnScreenFactor * FfxFloat32(GetUpsampleLanczosWeight(fSrcSampleOffset, fKernelBias)); + + data.fUpsampledColor += fSamples[iSampleIndex] * fSampleWeight; + data.fUpsampledWeight += fSampleWeight; + } + + // Update rectification box + { + const FfxFloat32 fRectificationCurveBias = -2.3f; + const FfxFloat32 fSrcSampleOffsetSq = dot(fSrcSampleOffset, fSrcSampleOffset); + const FfxFloat32 fBoxSampleWeight = exp(fRectificationCurveBias * fSrcSampleOffsetSq) * fOnScreenFactor; + + const FfxBoolean bInitialSample = (row == 0) && (col == 0); + RectificationBoxAddSample(bInitialSample, data.clippingBox, fSamples[iSampleIndex], fBoxSampleWeight); + } + ++iSampleIndex; + } + } + + RectificationBoxComputeVarianceBoxData(data.clippingBox); + + data.fUpsampledWeight *= FfxFloat32(data.fUpsampledWeight > FSR3UPSCALER_EPSILON); + + if (data.fUpsampledWeight > FSR3UPSCALER_EPSILON) { + // Normalize for deringing (we need to compare colors) + data.fUpsampledColor = data.fUpsampledColor / data.fUpsampledWeight; + data.fUpsampledWeight *= fAverageLanczosWeightPerFrame; + + Deringing(data.clippingBox, data.fUpsampledColor); + } + + // Initial samples using tonemapped upsampling + if (bIsInitialSample) { +#if FFX_FSR3UPSCALER_OPTION_HDR_COLOR_INPUT + data.fUpsampledColor = RGBToYCoCg(InverseTonemap(YCoCgToRGB(data.clippingBox.boxCenter))); +#else + data.fUpsampledColor = data.clippingBox.boxCenter; +#endif + data.fUpsampledWeight = 1.0f; + data.fHistoryWeight = 0.0f; + } +} diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_upsample.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_upsample.h.meta new file mode 100644 index 0000000..7dd7a51 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/fsr3upscaler/ffx_fsr3upscaler_upsample.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: 06167cdbc8ec55e4ab6913f7b5f1abf8 +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/spd.meta b/Packages/fidelityfx.fsr/Shaders/shaders/spd.meta new file mode 100644 index 0000000..c564f28 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/spd.meta @@ -0,0 +1,8 @@ +fileFormatVersion: 2 +guid: bd7cab5c46d4cd740ae57c03c8789ca4 +folderAsset: yes +DefaultImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/spd/ffx_spd.h b/Packages/fidelityfx.fsr/Shaders/shaders/spd/ffx_spd.h new file mode 100644 index 0000000..6d4f997 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/spd/ffx_spd.h @@ -0,0 +1,1007 @@ +// This file is part of the FidelityFX SDK. +// +// Copyright (C) 2024 Advanced Micro Devices, Inc. +// +// Permission is hereby granted, free of charge, to any person obtaining a copy +// of this software and associated documentation files(the "Software"), to deal +// in the Software without restriction, including without limitation the rights +// to use, copy, modify, merge, publish, distribute, sublicense, and /or sell +// copies of the Software, and to permit persons to whom the Software is +// furnished to do so, subject to the following conditions : +// +// The above copyright notice and this permission notice shall be included in +// all copies or substantial portions of the Software. +// +// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +// THE SOFTWARE. + +/// @defgroup FfxGPUSpd FidelityFX SPD +/// FidelityFX Single Pass Downsampler 2.0 GPU documentation +/// +/// @ingroup FfxGPUEffects + +/// Setup required constant values for SPD (CPU). +/// +/// @param [out] dispatchThreadGroupCountXY CPU side: dispatch thread group count xy. z is number of slices of the input texture +/// @param [out] workGroupOffset GPU side: pass in as constant +/// @param [out] numWorkGroupsAndMips GPU side: pass in as constant +/// @param [in] rectInfo left, top, width, height +/// @param [in] mips optional: if -1, calculate based on rect width and height +/// +/// @ingroup FfxGPUSpd +#if defined(FFX_CPU) +FFX_STATIC void ffxSpdSetup(FfxUInt32x2 dispatchThreadGroupCountXY, + FfxUInt32x2 workGroupOffset, + FfxUInt32x2 numWorkGroupsAndMips, + FfxUInt32x4 rectInfo, + FfxInt32 mips) +{ + // determines the offset of the first tile to downsample based on + // left (rectInfo[0]) and top (rectInfo[1]) of the subregion. + workGroupOffset[0] = rectInfo[0] / 64; + workGroupOffset[1] = rectInfo[1] / 64; + + FfxUInt32 endIndexX = (rectInfo[0] + rectInfo[2] - 1) / 64; // rectInfo[0] = left, rectInfo[2] = width + FfxUInt32 endIndexY = (rectInfo[1] + rectInfo[3] - 1) / 64; // rectInfo[1] = top, rectInfo[3] = height + + // we only need to dispatch as many thread groups as tiles we need to downsample + // number of tiles per slice depends on the subregion to downsample + dispatchThreadGroupCountXY[0] = endIndexX + 1 - workGroupOffset[0]; + dispatchThreadGroupCountXY[1] = endIndexY + 1 - workGroupOffset[1]; + + // number of thread groups per slice + numWorkGroupsAndMips[0] = (dispatchThreadGroupCountXY[0]) * (dispatchThreadGroupCountXY[1]); + + if (mips >= 0) + { + numWorkGroupsAndMips[1] = FfxUInt32(mips); + } + else + { + // calculate based on rect width and height + FfxUInt32 resolution = ffxMax(rectInfo[2], rectInfo[3]); + numWorkGroupsAndMips[1] = FfxUInt32((ffxMin(floor(log2(FfxFloat32(resolution))), FfxFloat32(12)))); + } +} + +/// Setup required constant values for SPD (CPU). +/// +/// @param [out] dispatchThreadGroupCountXY CPU side: dispatch thread group count xy. z is number of slices of the input texture +/// @param [out] workGroupOffset GPU side: pass in as constant +/// @param [out] numWorkGroupsAndMips GPU side: pass in as constant +/// @param [in] rectInfo left, top, width, height +/// +/// @ingroup FfxGPUSpd +FFX_STATIC void ffxSpdSetup(FfxUInt32x2 dispatchThreadGroupCountXY, + FfxUInt32x2 workGroupOffset, + FfxUInt32x2 numWorkGroupsAndMips, + FfxUInt32x4 rectInfo) +{ + ffxSpdSetup(dispatchThreadGroupCountXY, workGroupOffset, numWorkGroupsAndMips, rectInfo, -1); +} +#endif // #if defined(FFX_CPU) + + +//============================================================================================================================== +// NON-PACKED VERSION +//============================================================================================================================== +#if defined(FFX_GPU) +#if defined(FFX_SPD_PACKED_ONLY) +// Avoid compiler errors by including default implementations of these callbacks. +FfxFloat32x4 SpdLoadSourceImage(FfxInt32x2 p, FfxUInt32 slice) +{ + return FfxFloat32x4(0.0, 0.0, 0.0, 0.0); +} + +FfxFloat32x4 SpdLoad(FfxInt32x2 p, FfxUInt32 slice) +{ + return FfxFloat32x4(0.0, 0.0, 0.0, 0.0); +} +void SpdStore(FfxInt32x2 p, FfxFloat32x4 value, FfxUInt32 mip, FfxUInt32 slice) +{ +} +FfxFloat32x4 SpdLoadIntermediate(FfxUInt32 x, FfxUInt32 y) +{ + return FfxFloat32x4(0.0, 0.0, 0.0, 0.0); +} +void SpdStoreIntermediate(FfxUInt32 x, FfxUInt32 y, FfxFloat32x4 value) +{ +} +FfxFloat32x4 SpdReduce4(FfxFloat32x4 v0, FfxFloat32x4 v1, FfxFloat32x4 v2, FfxFloat32x4 v3) +{ + return FfxFloat32x4(0.0, 0.0, 0.0, 0.0); +} +#endif // #if FFX_SPD_PACKED_ONLY + +//_____________________________________________________________/\_______________________________________________________________ + +void ffxSpdWorkgroupShuffleBarrier() +{ + FFX_GROUP_MEMORY_BARRIER; +} + +// Only last active workgroup should proceed +bool SpdExitWorkgroup(FfxUInt32 numWorkGroups, FfxUInt32 localInvocationIndex, FfxUInt32 slice) +{ + // global atomic counter + if (localInvocationIndex == 0) + { + SpdIncreaseAtomicCounter(slice); + } + + ffxSpdWorkgroupShuffleBarrier(); + return (SpdGetAtomicCounter() != (numWorkGroups - 1)); +} + +// User defined: FfxFloat32x4 SpdReduce4(FfxFloat32x4 v0, FfxFloat32x4 v1, FfxFloat32x4 v2, FfxFloat32x4 v3); +FfxFloat32x4 SpdReduceQuad(FfxFloat32x4 v) +{ +#if defined(FFX_GLSL) && !defined(FFX_SPD_NO_WAVE_OPERATIONS) + + FfxFloat32x4 v0 = v; + FfxFloat32x4 v1 = subgroupQuadSwapHorizontal(v); + FfxFloat32x4 v2 = subgroupQuadSwapVertical(v); + FfxFloat32x4 v3 = subgroupQuadSwapDiagonal(v); + return SpdReduce4(v0, v1, v2, v3); + +#elif defined(FFX_HLSL) && !defined(FFX_SPD_NO_WAVE_OPERATIONS) + + // requires SM6.0 + FfxFloat32x4 v0 = v; + FfxFloat32x4 v1 = QuadReadAcrossX(v); + FfxFloat32x4 v2 = QuadReadAcrossY(v); + FfxFloat32x4 v3 = QuadReadAcrossDiagonal(v); + return SpdReduce4(v0, v1, v2, v3); +/* + // if SM6.0 is not available, you can use the AMD shader intrinsics + // the AMD shader intrinsics are available in AMD GPU Services (AGS) library: + // https://gpuopen.com/amd-gpu-services-ags-library/ + // works for DX11 + FfxFloat32x4 v0 = v; + FfxFloat32x4 v1; + v1.x = AmdExtD3DShaderIntrinsics_SwizzleF(v.x, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); + v1.y = AmdExtD3DShaderIntrinsics_SwizzleF(v.y, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); + v1.z = AmdExtD3DShaderIntrinsics_SwizzleF(v.z, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); + v1.w = AmdExtD3DShaderIntrinsics_SwizzleF(v.w, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); + FfxFloat32x4 v2; + v2.x = AmdExtD3DShaderIntrinsics_SwizzleF(v.x, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); + v2.y = AmdExtD3DShaderIntrinsics_SwizzleF(v.y, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); + v2.z = AmdExtD3DShaderIntrinsics_SwizzleF(v.z, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); + v2.w = AmdExtD3DShaderIntrinsics_SwizzleF(v.w, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); + FfxFloat32x4 v3; + v3.x = AmdExtD3DShaderIntrinsics_SwizzleF(v.x, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); + v3.y = AmdExtD3DShaderIntrinsics_SwizzleF(v.y, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); + v3.z = AmdExtD3DShaderIntrinsics_SwizzleF(v.z, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); + v3.w = AmdExtD3DShaderIntrinsics_SwizzleF(v.w, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); + return SpdReduce4(v0, v1, v2, v3); + */ +#endif + return v; +} + +FfxFloat32x4 SpdReduceIntermediate(FfxUInt32x2 i0, FfxUInt32x2 i1, FfxUInt32x2 i2, FfxUInt32x2 i3) +{ + FfxFloat32x4 v0 = SpdLoadIntermediate(i0.x, i0.y); + FfxFloat32x4 v1 = SpdLoadIntermediate(i1.x, i1.y); + FfxFloat32x4 v2 = SpdLoadIntermediate(i2.x, i2.y); + FfxFloat32x4 v3 = SpdLoadIntermediate(i3.x, i3.y); + return SpdReduce4(v0, v1, v2, v3); +} + +FfxFloat32x4 SpdReduceLoad4(FfxUInt32x2 i0, FfxUInt32x2 i1, FfxUInt32x2 i2, FfxUInt32x2 i3, FfxUInt32 slice) +{ + FfxFloat32x4 v0 = SpdLoad(FfxInt32x2(i0), slice); + FfxFloat32x4 v1 = SpdLoad(FfxInt32x2(i1), slice); + FfxFloat32x4 v2 = SpdLoad(FfxInt32x2(i2), slice); + FfxFloat32x4 v3 = SpdLoad(FfxInt32x2(i3), slice); + return SpdReduce4(v0, v1, v2, v3); +} + +FfxFloat32x4 SpdReduceLoad4(FfxUInt32x2 base, FfxUInt32 slice) +{ + return SpdReduceLoad4(FfxUInt32x2(base + FfxUInt32x2(0, 0)), FfxUInt32x2(base + FfxUInt32x2(0, 1)), FfxUInt32x2(base + FfxUInt32x2(1, 0)), FfxUInt32x2(base + FfxUInt32x2(1, 1)), slice); +} + +FfxFloat32x4 SpdReduceLoadSourceImage4(FfxUInt32x2 i0, FfxUInt32x2 i1, FfxUInt32x2 i2, FfxUInt32x2 i3, FfxUInt32 slice) +{ + FfxFloat32x4 v0 = SpdLoadSourceImage(FfxInt32x2(i0), slice); + FfxFloat32x4 v1 = SpdLoadSourceImage(FfxInt32x2(i1), slice); + FfxFloat32x4 v2 = SpdLoadSourceImage(FfxInt32x2(i2), slice); + FfxFloat32x4 v3 = SpdLoadSourceImage(FfxInt32x2(i3), slice); + return SpdReduce4(v0, v1, v2, v3); +} + +FfxFloat32x4 SpdReduceLoadSourceImage(FfxUInt32x2 base, FfxUInt32 slice) +{ +#if defined(SPD_LINEAR_SAMPLER) + return SpdLoadSourceImage(FfxInt32x2(base), slice); +#else + return SpdReduceLoadSourceImage4(FfxUInt32x2(base + FfxUInt32x2(0, 0)), FfxUInt32x2(base + FfxUInt32x2(0, 1)), FfxUInt32x2(base + FfxUInt32x2(1, 0)), FfxUInt32x2(base + FfxUInt32x2(1, 1)), slice); +#endif +} + +void SpdDownsampleMips_0_1_Intrinsics(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) +{ + FfxFloat32x4 v[4]; + + FfxInt32x2 tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2); + FfxInt32x2 pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y); + v[0] = SpdReduceLoadSourceImage(tex, slice); + SpdStore(pix, v[0], 0, slice); + + tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2); + pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y); + v[1] = SpdReduceLoadSourceImage(tex, slice); + SpdStore(pix, v[1], 0, slice); + + tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2 + 32); + pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y + 16); + v[2] = SpdReduceLoadSourceImage(tex, slice); + SpdStore(pix, v[2], 0, slice); + + tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2 + 32); + pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y + 16); + v[3] = SpdReduceLoadSourceImage(tex, slice); + SpdStore(pix, v[3], 0, slice); + + if (mip <= 1) + return; + + v[0] = SpdReduceQuad(v[0]); + v[1] = SpdReduceQuad(v[1]); + v[2] = SpdReduceQuad(v[2]); + v[3] = SpdReduceQuad(v[3]); + + if ((localInvocationIndex % 4) == 0) + { + SpdStore(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2, y / 2), v[0], 1, slice); + SpdStoreIntermediate(x / 2, y / 2, v[0]); + + SpdStore(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2 + 8, y / 2), v[1], 1, slice); + SpdStoreIntermediate(x / 2 + 8, y / 2, v[1]); + + SpdStore(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2, y / 2 + 8), v[2], 1, slice); + SpdStoreIntermediate(x / 2, y / 2 + 8, v[2]); + + SpdStore(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2 + 8, y / 2 + 8), v[3], 1, slice); + SpdStoreIntermediate(x / 2 + 8, y / 2 + 8, v[3]); + } +} + +void SpdDownsampleMips_0_1_LDS(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) +{ + FfxFloat32x4 v[4]; + + FfxInt32x2 tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2); + FfxInt32x2 pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y); + v[0] = SpdReduceLoadSourceImage(tex, slice); + SpdStore(pix, v[0], 0, slice); + + tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2); + pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y); + v[1] = SpdReduceLoadSourceImage(tex, slice); + SpdStore(pix, v[1], 0, slice); + + tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2 + 32); + pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y + 16); + v[2] = SpdReduceLoadSourceImage(tex, slice); + SpdStore(pix, v[2], 0, slice); + + tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2 + 32); + pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y + 16); + v[3] = SpdReduceLoadSourceImage(tex, slice); + SpdStore(pix, v[3], 0, slice); + + if (mip <= 1) + return; + + for (FfxUInt32 i = 0; i < 4; i++) + { + SpdStoreIntermediate(x, y, v[i]); + ffxSpdWorkgroupShuffleBarrier(); + if (localInvocationIndex < 64) + { + v[i] = SpdReduceIntermediate(FfxUInt32x2(x * 2 + 0, y * 2 + 0), FfxUInt32x2(x * 2 + 1, y * 2 + 0), FfxUInt32x2(x * 2 + 0, y * 2 + 1), FfxUInt32x2(x * 2 + 1, y * 2 + 1)); + SpdStore(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x + (i % 2) * 8, y + (i / 2) * 8), v[i], 1, slice); + } + ffxSpdWorkgroupShuffleBarrier(); + } + + if (localInvocationIndex < 64) + { + SpdStoreIntermediate(x + 0, y + 0, v[0]); + SpdStoreIntermediate(x + 8, y + 0, v[1]); + SpdStoreIntermediate(x + 0, y + 8, v[2]); + SpdStoreIntermediate(x + 8, y + 8, v[3]); + } +} + +void SpdDownsampleMips_0_1(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) +{ +#if defined(FFX_SPD_NO_WAVE_OPERATIONS) + SpdDownsampleMips_0_1_LDS(x, y, workGroupID, localInvocationIndex, mip, slice); +#else + SpdDownsampleMips_0_1_Intrinsics(x, y, workGroupID, localInvocationIndex, mip, slice); +#endif +} + + +void SpdDownsampleMip_2(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) +{ +#if defined(FFX_SPD_NO_WAVE_OPERATIONS) + if (localInvocationIndex < 64) + { + FfxFloat32x4 v = SpdReduceIntermediate(FfxUInt32x2(x * 2 + 0, y * 2 + 0), FfxUInt32x2(x * 2 + 1, y * 2 + 0), FfxUInt32x2(x * 2 + 0, y * 2 + 1), FfxUInt32x2(x * 2 + 1, y * 2 + 1)); + SpdStore(FfxInt32x2(workGroupID.xy * 8) + FfxInt32x2(x, y), v, mip, slice); + // store to LDS, try to reduce bank conflicts + // x 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 + // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 + // 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 x + // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 + // x 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 + // ... + // x 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 + SpdStoreIntermediate(x * 2 + y % 2, y * 2, v); + } +#else + FfxFloat32x4 v = SpdLoadIntermediate(x, y); + v = SpdReduceQuad(v); + // quad index 0 stores result + if (localInvocationIndex % 4 == 0) + { + SpdStore(FfxInt32x2(workGroupID.xy * 8) + FfxInt32x2(x / 2, y / 2), v, mip, slice); + SpdStoreIntermediate(x + (y / 2) % 2, y, v); + } +#endif +} + +void SpdDownsampleMip_3(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) +{ +#if defined(FFX_SPD_NO_WAVE_OPERATIONS) + if (localInvocationIndex < 16) + { + // x 0 x 0 + // 0 0 0 0 + // 0 x 0 x + // 0 0 0 0 + FfxFloat32x4 v = + SpdReduceIntermediate(FfxUInt32x2(x * 4 + 0 + 0, y * 4 + 0), FfxUInt32x2(x * 4 + 2 + 0, y * 4 + 0), FfxUInt32x2(x * 4 + 0 + 1, y * 4 + 2), FfxUInt32x2(x * 4 + 2 + 1, y * 4 + 2)); + SpdStore(FfxInt32x2(workGroupID.xy * 4) + FfxInt32x2(x, y), v, mip, slice); + // store to LDS + // x 0 0 0 x 0 0 0 x 0 0 0 x 0 0 0 + // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 + // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 + // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 + // 0 x 0 0 0 x 0 0 0 x 0 0 0 x 0 0 + // ... + // 0 0 x 0 0 0 x 0 0 0 x 0 0 0 x 0 + // ... + // 0 0 0 x 0 0 0 x 0 0 0 x 0 0 0 x + // ... + SpdStoreIntermediate(x * 4 + y, y * 4, v); + } +#else + if (localInvocationIndex < 64) + { + FfxFloat32x4 v = SpdLoadIntermediate(x * 2 + y % 2, y * 2); + v = SpdReduceQuad(v); + // quad index 0 stores result + if (localInvocationIndex % 4 == 0) + { + SpdStore(FfxInt32x2(workGroupID.xy * 4) + FfxInt32x2(x / 2, y / 2), v, mip, slice); + SpdStoreIntermediate(x * 2 + y / 2, y * 2, v); + } + } +#endif +} + +void SpdDownsampleMip_4(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) +{ +#if defined(FFX_SPD_NO_WAVE_OPERATIONS) + if (localInvocationIndex < 4) + { + // x 0 0 0 x 0 0 0 + // ... + // 0 x 0 0 0 x 0 0 + FfxFloat32x4 v = SpdReduceIntermediate(FfxUInt32x2(x * 8 + 0 + 0 + y * 2, y * 8 + 0), + FfxUInt32x2(x * 8 + 4 + 0 + y * 2, y * 8 + 0), + FfxUInt32x2(x * 8 + 0 + 1 + y * 2, y * 8 + 4), + FfxUInt32x2(x * 8 + 4 + 1 + y * 2, y * 8 + 4)); + SpdStore(FfxInt32x2(workGroupID.xy * 2) + FfxInt32x2(x, y), v, mip, slice); + // store to LDS + // x x x x 0 ... + // 0 ... + SpdStoreIntermediate(x + y * 2, 0, v); + } +#else + if (localInvocationIndex < 16) + { + FfxFloat32x4 v = SpdLoadIntermediate(x * 4 + y, y * 4); + v = SpdReduceQuad(v); + // quad index 0 stores result + if (localInvocationIndex % 4 == 0) + { + SpdStore(FfxInt32x2(workGroupID.xy * 2) + FfxInt32x2(x / 2, y / 2), v, mip, slice); + SpdStoreIntermediate(x / 2 + y, 0, v); + } + } +#endif +} + +void SpdDownsampleMip_5(FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) +{ +#if defined(FFX_SPD_NO_WAVE_OPERATIONS) + if (localInvocationIndex < 1) + { + // x x x x 0 ... + // 0 ... + FfxFloat32x4 v = SpdReduceIntermediate(FfxUInt32x2(0, 0), FfxUInt32x2(1, 0), FfxUInt32x2(2, 0), FfxUInt32x2(3, 0)); + SpdStore(FfxInt32x2(workGroupID.xy), v, mip, slice); + } +#else + if (localInvocationIndex < 4) + { + FfxFloat32x4 v = SpdLoadIntermediate(localInvocationIndex, 0); + v = SpdReduceQuad(v); + // quad index 0 stores result + if (localInvocationIndex % 4 == 0) + { + SpdStore(FfxInt32x2(workGroupID.xy), v, mip, slice); + } + } +#endif +} + +void SpdDownsampleMips_6_7(FfxUInt32 x, FfxUInt32 y, FfxUInt32 mips, FfxUInt32 slice) +{ + FfxInt32x2 tex = FfxInt32x2(x * 4 + 0, y * 4 + 0); + FfxInt32x2 pix = FfxInt32x2(x * 2 + 0, y * 2 + 0); + FfxFloat32x4 v0 = SpdReduceLoad4(tex, slice); + SpdStore(pix, v0, 6, slice); + + tex = FfxInt32x2(x * 4 + 2, y * 4 + 0); + pix = FfxInt32x2(x * 2 + 1, y * 2 + 0); + FfxFloat32x4 v1 = SpdReduceLoad4(tex, slice); + SpdStore(pix, v1, 6, slice); + + tex = FfxInt32x2(x * 4 + 0, y * 4 + 2); + pix = FfxInt32x2(x * 2 + 0, y * 2 + 1); + FfxFloat32x4 v2 = SpdReduceLoad4(tex, slice); + SpdStore(pix, v2, 6, slice); + + tex = FfxInt32x2(x * 4 + 2, y * 4 + 2); + pix = FfxInt32x2(x * 2 + 1, y * 2 + 1); + FfxFloat32x4 v3 = SpdReduceLoad4(tex, slice); + SpdStore(pix, v3, 6, slice); + + if (mips <= 7) + return; + // no barrier needed, working on values only from the same thread + + FfxFloat32x4 v = SpdReduce4(v0, v1, v2, v3); + SpdStore(FfxInt32x2(x, y), v, 7, slice); + SpdStoreIntermediate(x, y, v); +} + +void SpdDownsampleNextFour(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 baseMip, FfxUInt32 mips, FfxUInt32 slice) +{ + if (mips <= baseMip) + return; + ffxSpdWorkgroupShuffleBarrier(); + SpdDownsampleMip_2(x, y, workGroupID, localInvocationIndex, baseMip, slice); + + if (mips <= baseMip + 1) + return; + ffxSpdWorkgroupShuffleBarrier(); + SpdDownsampleMip_3(x, y, workGroupID, localInvocationIndex, baseMip + 1, slice); + + if (mips <= baseMip + 2) + return; + ffxSpdWorkgroupShuffleBarrier(); + SpdDownsampleMip_4(x, y, workGroupID, localInvocationIndex, baseMip + 2, slice); + + if (mips <= baseMip + 3) + return; + ffxSpdWorkgroupShuffleBarrier(); + SpdDownsampleMip_5(workGroupID, localInvocationIndex, baseMip + 3, slice); +} + +/// Downsamples a 64x64 tile based on the work group id. +/// If after downsampling it's the last active thread group, computes the remaining MIP levels. +/// +/// @param [in] workGroupID index of the work group / thread group +/// @param [in] localInvocationIndex index of the thread within the thread group in 1D +/// @param [in] mips the number of total MIP levels to compute for the input texture +/// @param [in] numWorkGroups the total number of dispatched work groups / thread groups for this slice +/// @param [in] slice the slice of the input texture +/// +/// @ingroup FfxGPUSpd +void SpdDownsample(FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mips, FfxUInt32 numWorkGroups, FfxUInt32 slice) +{ + // compute MIP level 0 and 1 + FfxUInt32x2 sub_xy = ffxRemapForWaveReduction(localInvocationIndex % 64); + FfxUInt32 x = sub_xy.x + 8 * ((localInvocationIndex >> 6) % 2); + FfxUInt32 y = sub_xy.y + 8 * ((localInvocationIndex >> 7)); + SpdDownsampleMips_0_1(x, y, workGroupID, localInvocationIndex, mips, slice); + + // compute MIP level 2, 3, 4, 5 + SpdDownsampleNextFour(x, y, workGroupID, localInvocationIndex, 2, mips, slice); + + if (mips <= 6) + return; + + // increase the global atomic counter for the given slice and check if it's the last remaining thread group: + // terminate if not, continue if yes. + if (SpdExitWorkgroup(numWorkGroups, localInvocationIndex, slice)) + return; + + // reset the global atomic counter back to 0 for the next spd dispatch + SpdResetAtomicCounter(slice); + + // After mip 5 there is only a single workgroup left that downsamples the remaining up to 64x64 texels. + // compute MIP level 6 and 7 + SpdDownsampleMips_6_7(x, y, mips, slice); + + // compute MIP level 8, 9, 10, 11 + SpdDownsampleNextFour(x, y, FfxUInt32x2(0, 0), localInvocationIndex, 8, mips, slice); +} +/// Downsamples a 64x64 tile based on the work group id and work group offset. +/// If after downsampling it's the last active thread group, computes the remaining MIP levels. +/// +/// @param [in] workGroupID index of the work group / thread group +/// @param [in] localInvocationIndex index of the thread within the thread group in 1D +/// @param [in] mips the number of total MIP levels to compute for the input texture +/// @param [in] numWorkGroups the total number of dispatched work groups / thread groups for this slice +/// @param [in] slice the slice of the input texture +/// @param [in] workGroupOffset the work group offset. it's (0,0) in case the entire input texture is downsampled. +/// +/// @ingroup FfxGPUSpd +void SpdDownsample(FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mips, FfxUInt32 numWorkGroups, FfxUInt32 slice, FfxUInt32x2 workGroupOffset) +{ + SpdDownsample(workGroupID + workGroupOffset, localInvocationIndex, mips, numWorkGroups, slice); +} + +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// +//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// + +//============================================================================================================================== +// PACKED VERSION +//============================================================================================================================== + +#if FFX_HALF + +FfxFloat16x4 SpdReduceQuadH(FfxFloat16x4 v) +{ +#if defined(FFX_GLSL) && !defined(FFX_SPD_NO_WAVE_OPERATIONS) + FfxFloat16x4 v0 = v; + FfxFloat16x4 v1 = subgroupQuadSwapHorizontal(v); + FfxFloat16x4 v2 = subgroupQuadSwapVertical(v); + FfxFloat16x4 v3 = subgroupQuadSwapDiagonal(v); + return SpdReduce4H(v0, v1, v2, v3); +#elif defined(FFX_HLSL) && !defined(FFX_SPD_NO_WAVE_OPERATIONS) + // requires SM6.0 + FfxFloat16x4 v0 = v; + FfxFloat16x4 v1 = QuadReadAcrossX(v); + FfxFloat16x4 v2 = QuadReadAcrossY(v); + FfxFloat16x4 v3 = QuadReadAcrossDiagonal(v); + return SpdReduce4H(v0, v1, v2, v3); +/* + // if SM6.0 is not available, you can use the AMD shader intrinsics + // the AMD shader intrinsics are available in AMD GPU Services (AGS) library: + // https://gpuopen.com/amd-gpu-services-ags-library/ + // works for DX11 + FfxFloat16x4 v0 = v; + FfxFloat16x4 v1; + v1.x = AmdExtD3DShaderIntrinsics_SwizzleF(v.x, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); + v1.y = AmdExtD3DShaderIntrinsics_SwizzleF(v.y, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); + v1.z = AmdExtD3DShaderIntrinsics_SwizzleF(v.z, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); + v1.w = AmdExtD3DShaderIntrinsics_SwizzleF(v.w, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); + FfxFloat16x4 v2; + v2.x = AmdExtD3DShaderIntrinsics_SwizzleF(v.x, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); + v2.y = AmdExtD3DShaderIntrinsics_SwizzleF(v.y, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); + v2.z = AmdExtD3DShaderIntrinsics_SwizzleF(v.z, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); + v2.w = AmdExtD3DShaderIntrinsics_SwizzleF(v.w, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); + FfxFloat16x4 v3; + v3.x = AmdExtD3DShaderIntrinsics_SwizzleF(v.x, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); + v3.y = AmdExtD3DShaderIntrinsics_SwizzleF(v.y, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); + v3.z = AmdExtD3DShaderIntrinsics_SwizzleF(v.z, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); + v3.w = AmdExtD3DShaderIntrinsics_SwizzleF(v.w, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); + return SpdReduce4H(v0, v1, v2, v3); + */ +#endif + return FfxFloat16x4(0.0, 0.0, 0.0, 0.0); +} + +FfxFloat16x4 SpdReduceIntermediateH(FfxUInt32x2 i0, FfxUInt32x2 i1, FfxUInt32x2 i2, FfxUInt32x2 i3) +{ + FfxFloat16x4 v0 = SpdLoadIntermediateH(i0.x, i0.y); + FfxFloat16x4 v1 = SpdLoadIntermediateH(i1.x, i1.y); + FfxFloat16x4 v2 = SpdLoadIntermediateH(i2.x, i2.y); + FfxFloat16x4 v3 = SpdLoadIntermediateH(i3.x, i3.y); + return SpdReduce4H(v0, v1, v2, v3); +} + +FfxFloat16x4 SpdReduceLoad4H(FfxUInt32x2 i0, FfxUInt32x2 i1, FfxUInt32x2 i2, FfxUInt32x2 i3, FfxUInt32 slice) +{ + FfxFloat16x4 v0 = SpdLoadH(FfxInt32x2(i0), slice); + FfxFloat16x4 v1 = SpdLoadH(FfxInt32x2(i1), slice); + FfxFloat16x4 v2 = SpdLoadH(FfxInt32x2(i2), slice); + FfxFloat16x4 v3 = SpdLoadH(FfxInt32x2(i3), slice); + return SpdReduce4H(v0, v1, v2, v3); +} + +FfxFloat16x4 SpdReduceLoad4H(FfxUInt32x2 base, FfxUInt32 slice) +{ + return SpdReduceLoad4H(FfxUInt32x2(base + FfxUInt32x2(0, 0)), FfxUInt32x2(base + FfxUInt32x2(0, 1)), FfxUInt32x2(base + FfxUInt32x2(1, 0)), FfxUInt32x2(base + FfxUInt32x2(1, 1)), slice); +} + +FfxFloat16x4 SpdReduceLoadSourceImage4H(FfxUInt32x2 i0, FfxUInt32x2 i1, FfxUInt32x2 i2, FfxUInt32x2 i3, FfxUInt32 slice) +{ + FfxFloat16x4 v0 = SpdLoadSourceImageH(FfxInt32x2(i0), slice); + FfxFloat16x4 v1 = SpdLoadSourceImageH(FfxInt32x2(i1), slice); + FfxFloat16x4 v2 = SpdLoadSourceImageH(FfxInt32x2(i2), slice); + FfxFloat16x4 v3 = SpdLoadSourceImageH(FfxInt32x2(i3), slice); + return SpdReduce4H(v0, v1, v2, v3); +} + +FfxFloat16x4 SpdReduceLoadSourceImageH(FfxUInt32x2 base, FfxUInt32 slice) +{ +#if defined(SPD_LINEAR_SAMPLER) + return SpdLoadSourceImageH(FfxInt32x2(base), slice); +#else + return SpdReduceLoadSourceImage4H(FfxUInt32x2(base + FfxUInt32x2(0, 0)), FfxUInt32x2(base + FfxUInt32x2(0, 1)), FfxUInt32x2(base + FfxUInt32x2(1, 0)), FfxUInt32x2(base + FfxUInt32x2(1, 1)), slice); +#endif +} + +void SpdDownsampleMips_0_1_IntrinsicsH(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mips, FfxUInt32 slice) +{ + FfxFloat16x4 v[4]; + + FfxInt32x2 tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2); + FfxInt32x2 pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y); + v[0] = SpdReduceLoadSourceImageH(tex, slice); + SpdStoreH(pix, v[0], 0, slice); + + tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2); + pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y); + v[1] = SpdReduceLoadSourceImageH(tex, slice); + SpdStoreH(pix, v[1], 0, slice); + + tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2 + 32); + pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y + 16); + v[2] = SpdReduceLoadSourceImageH(tex, slice); + SpdStoreH(pix, v[2], 0, slice); + + tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2 + 32); + pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y + 16); + v[3] = SpdReduceLoadSourceImageH(tex, slice); + SpdStoreH(pix, v[3], 0, slice); + + if (mips <= 1) + return; + + v[0] = SpdReduceQuadH(v[0]); + v[1] = SpdReduceQuadH(v[1]); + v[2] = SpdReduceQuadH(v[2]); + v[3] = SpdReduceQuadH(v[3]); + + if ((localInvocationIndex % 4) == 0) + { + SpdStoreH(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2, y / 2), v[0], 1, slice); + SpdStoreIntermediateH(x / 2, y / 2, v[0]); + + SpdStoreH(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2 + 8, y / 2), v[1], 1, slice); + SpdStoreIntermediateH(x / 2 + 8, y / 2, v[1]); + + SpdStoreH(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2, y / 2 + 8), v[2], 1, slice); + SpdStoreIntermediateH(x / 2, y / 2 + 8, v[2]); + + SpdStoreH(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2 + 8, y / 2 + 8), v[3], 1, slice); + SpdStoreIntermediateH(x / 2 + 8, y / 2 + 8, v[3]); + } +} + +void SpdDownsampleMips_0_1_LDSH(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mips, FfxUInt32 slice) +{ + FfxFloat16x4 v[4]; + + FfxInt32x2 tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2); + FfxInt32x2 pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y); + v[0] = SpdReduceLoadSourceImageH(tex, slice); + SpdStoreH(pix, v[0], 0, slice); + + tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2); + pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y); + v[1] = SpdReduceLoadSourceImageH(tex, slice); + SpdStoreH(pix, v[1], 0, slice); + + tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2 + 32); + pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y + 16); + v[2] = SpdReduceLoadSourceImageH(tex, slice); + SpdStoreH(pix, v[2], 0, slice); + + tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2 + 32); + pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y + 16); + v[3] = SpdReduceLoadSourceImageH(tex, slice); + SpdStoreH(pix, v[3], 0, slice); + + if (mips <= 1) + return; + + for (FfxInt32 i = 0; i < 4; i++) + { + SpdStoreIntermediateH(x, y, v[i]); + ffxSpdWorkgroupShuffleBarrier(); + if (localInvocationIndex < 64) + { + v[i] = SpdReduceIntermediateH(FfxUInt32x2(x * 2 + 0, y * 2 + 0), FfxUInt32x2(x * 2 + 1, y * 2 + 0), FfxUInt32x2(x * 2 + 0, y * 2 + 1), FfxUInt32x2(x * 2 + 1, y * 2 + 1)); + SpdStoreH(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x + (i % 2) * 8, y + (i / 2) * 8), v[i], 1, slice); + } + ffxSpdWorkgroupShuffleBarrier(); + } + + if (localInvocationIndex < 64) + { + SpdStoreIntermediateH(x + 0, y + 0, v[0]); + SpdStoreIntermediateH(x + 8, y + 0, v[1]); + SpdStoreIntermediateH(x + 0, y + 8, v[2]); + SpdStoreIntermediateH(x + 8, y + 8, v[3]); + } +} + +void SpdDownsampleMips_0_1H(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mips, FfxUInt32 slice) +{ +#if defined(FFX_SPD_NO_WAVE_OPERATIONS) + SpdDownsampleMips_0_1_LDSH(x, y, workGroupID, localInvocationIndex, mips, slice); +#else + SpdDownsampleMips_0_1_IntrinsicsH(x, y, workGroupID, localInvocationIndex, mips, slice); +#endif +} + + +void SpdDownsampleMip_2H(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) +{ +#if defined(FFX_SPD_NO_WAVE_OPERATIONS) + if (localInvocationIndex < 64) + { + FfxFloat16x4 v = SpdReduceIntermediateH(FfxUInt32x2(x * 2 + 0, y * 2 + 0), FfxUInt32x2(x * 2 + 1, y * 2 + 0), FfxUInt32x2(x * 2 + 0, y * 2 + 1), FfxUInt32x2(x * 2 + 1, y * 2 + 1)); + SpdStoreH(FfxInt32x2(workGroupID.xy * 8) + FfxInt32x2(x, y), v, mip, slice); + // store to LDS, try to reduce bank conflicts + // x 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 + // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 + // 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 x + // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 + // x 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 + // ... + // x 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 + SpdStoreIntermediateH(x * 2 + y % 2, y * 2, v); + } +#else + FfxFloat16x4 v = SpdLoadIntermediateH(x, y); + v = SpdReduceQuadH(v); + // quad index 0 stores result + if (localInvocationIndex % 4 == 0) + { + SpdStoreH(FfxInt32x2(workGroupID.xy * 8) + FfxInt32x2(x / 2, y / 2), v, mip, slice); + SpdStoreIntermediateH(x + (y / 2) % 2, y, v); + } +#endif +} + +void SpdDownsampleMip_3H(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) +{ +#if defined(FFX_SPD_NO_WAVE_OPERATIONS) + if (localInvocationIndex < 16) + { + // x 0 x 0 + // 0 0 0 0 + // 0 x 0 x + // 0 0 0 0 + FfxFloat16x4 v = + SpdReduceIntermediateH(FfxUInt32x2(x * 4 + 0 + 0, y * 4 + 0), FfxUInt32x2(x * 4 + 2 + 0, y * 4 + 0), FfxUInt32x2(x * 4 + 0 + 1, y * 4 + 2), FfxUInt32x2(x * 4 + 2 + 1, y * 4 + 2)); + SpdStoreH(FfxInt32x2(workGroupID.xy * 4) + FfxInt32x2(x, y), v, mip, slice); + // store to LDS + // x 0 0 0 x 0 0 0 x 0 0 0 x 0 0 0 + // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 + // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 + // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 + // 0 x 0 0 0 x 0 0 0 x 0 0 0 x 0 0 + // ... + // 0 0 x 0 0 0 x 0 0 0 x 0 0 0 x 0 + // ... + // 0 0 0 x 0 0 0 x 0 0 0 x 0 0 0 x + // ... + SpdStoreIntermediateH(x * 4 + y, y * 4, v); + } +#else + if (localInvocationIndex < 64) + { + FfxFloat16x4 v = SpdLoadIntermediateH(x * 2 + y % 2, y * 2); + v = SpdReduceQuadH(v); + // quad index 0 stores result + if (localInvocationIndex % 4 == 0) + { + SpdStoreH(FfxInt32x2(workGroupID.xy * 4) + FfxInt32x2(x / 2, y / 2), v, mip, slice); + SpdStoreIntermediateH(x * 2 + y / 2, y * 2, v); + } + } +#endif +} + +void SpdDownsampleMip_4H(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) +{ +#if defined(FFX_SPD_NO_WAVE_OPERATIONS) + if (localInvocationIndex < 4) + { + // x 0 0 0 x 0 0 0 + // ... + // 0 x 0 0 0 x 0 0 + FfxFloat16x4 v = SpdReduceIntermediateH(FfxUInt32x2(x * 8 + 0 + 0 + y * 2, y * 8 + 0), + FfxUInt32x2(x * 8 + 4 + 0 + y * 2, y * 8 + 0), + FfxUInt32x2(x * 8 + 0 + 1 + y * 2, y * 8 + 4), + FfxUInt32x2(x * 8 + 4 + 1 + y * 2, y * 8 + 4)); + SpdStoreH(FfxInt32x2(workGroupID.xy * 2) + FfxInt32x2(x, y), v, mip, slice); + // store to LDS + // x x x x 0 ... + // 0 ... + SpdStoreIntermediateH(x + y * 2, 0, v); + } +#else + if (localInvocationIndex < 16) + { + FfxFloat16x4 v = SpdLoadIntermediateH(x * 4 + y, y * 4); + v = SpdReduceQuadH(v); + // quad index 0 stores result + if (localInvocationIndex % 4 == 0) + { + SpdStoreH(FfxInt32x2(workGroupID.xy * 2) + FfxInt32x2(x / 2, y / 2), v, mip, slice); + SpdStoreIntermediateH(x / 2 + y, 0, v); + } + } +#endif +} + +void SpdDownsampleMip_5H(FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) +{ +#if defined(FFX_SPD_NO_WAVE_OPERATIONS) + if (localInvocationIndex < 1) + { + // x x x x 0 ... + // 0 ... + FfxFloat16x4 v = SpdReduceIntermediateH(FfxUInt32x2(0, 0), FfxUInt32x2(1, 0), FfxUInt32x2(2, 0), FfxUInt32x2(3, 0)); + SpdStoreH(FfxInt32x2(workGroupID.xy), v, mip, slice); + } +#else + if (localInvocationIndex < 4) + { + FfxFloat16x4 v = SpdLoadIntermediateH(localInvocationIndex, 0); + v = SpdReduceQuadH(v); + // quad index 0 stores result + if (localInvocationIndex % 4 == 0) + { + SpdStoreH(FfxInt32x2(workGroupID.xy), v, mip, slice); + } + } +#endif +} + +void SpdDownsampleMips_6_7H(FfxUInt32 x, FfxUInt32 y, FfxUInt32 mips, FfxUInt32 slice) +{ + FfxInt32x2 tex = FfxInt32x2(x * 4 + 0, y * 4 + 0); + FfxInt32x2 pix = FfxInt32x2(x * 2 + 0, y * 2 + 0); + FfxFloat16x4 v0 = SpdReduceLoad4H(tex, slice); + SpdStoreH(pix, v0, 6, slice); + + tex = FfxInt32x2(x * 4 + 2, y * 4 + 0); + pix = FfxInt32x2(x * 2 + 1, y * 2 + 0); + FfxFloat16x4 v1 = SpdReduceLoad4H(tex, slice); + SpdStoreH(pix, v1, 6, slice); + + tex = FfxInt32x2(x * 4 + 0, y * 4 + 2); + pix = FfxInt32x2(x * 2 + 0, y * 2 + 1); + FfxFloat16x4 v2 = SpdReduceLoad4H(tex, slice); + SpdStoreH(pix, v2, 6, slice); + + tex = FfxInt32x2(x * 4 + 2, y * 4 + 2); + pix = FfxInt32x2(x * 2 + 1, y * 2 + 1); + FfxFloat16x4 v3 = SpdReduceLoad4H(tex, slice); + SpdStoreH(pix, v3, 6, slice); + + if (mips < 8) + return; + // no barrier needed, working on values only from the same thread + + FfxFloat16x4 v = SpdReduce4H(v0, v1, v2, v3); + SpdStoreH(FfxInt32x2(x, y), v, 7, slice); + SpdStoreIntermediateH(x, y, v); +} + +void SpdDownsampleNextFourH(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 baseMip, FfxUInt32 mips, FfxUInt32 slice) +{ + if (mips <= baseMip) + return; + ffxSpdWorkgroupShuffleBarrier(); + SpdDownsampleMip_2H(x, y, workGroupID, localInvocationIndex, baseMip, slice); + + if (mips <= baseMip + 1) + return; + ffxSpdWorkgroupShuffleBarrier(); + SpdDownsampleMip_3H(x, y, workGroupID, localInvocationIndex, baseMip + 1, slice); + + if (mips <= baseMip + 2) + return; + ffxSpdWorkgroupShuffleBarrier(); + SpdDownsampleMip_4H(x, y, workGroupID, localInvocationIndex, baseMip + 2, slice); + + if (mips <= baseMip + 3) + return; + ffxSpdWorkgroupShuffleBarrier(); + SpdDownsampleMip_5H(workGroupID, localInvocationIndex, baseMip + 3, slice); +} + +/// Downsamples a 64x64 tile based on the work group id and work group offset. +/// If after downsampling it's the last active thread group, computes the remaining MIP levels. +/// Uses half types. +/// +/// @param [in] workGroupID index of the work group / thread group +/// @param [in] localInvocationIndex index of the thread within the thread group in 1D +/// @param [in] mips the number of total MIP levels to compute for the input texture +/// @param [in] numWorkGroups the total number of dispatched work groups / thread groups for this slice +/// @param [in] slice the slice of the input texture +/// +/// @ingroup FfxGPUSpd +void SpdDownsampleH(FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mips, FfxUInt32 numWorkGroups, FfxUInt32 slice) +{ + FfxUInt32x2 sub_xy = ffxRemapForWaveReduction(localInvocationIndex % 64); + FfxUInt32 x = sub_xy.x + 8 * ((localInvocationIndex >> 6) % 2); + FfxUInt32 y = sub_xy.y + 8 * ((localInvocationIndex >> 7)); + + // compute MIP level 0 and 1 + SpdDownsampleMips_0_1H(x, y, workGroupID, localInvocationIndex, mips, slice); + + // compute MIP level 2, 3, 4, 5 + SpdDownsampleNextFourH(x, y, workGroupID, localInvocationIndex, 2, mips, slice); + + if (mips < 7) + return; + + // increase the global atomic counter for the given slice and check if it's the last remaining thread group: + // terminate if not, continue if yes. + if (SpdExitWorkgroup(numWorkGroups, localInvocationIndex, slice)) + return; + + // reset the global atomic counter back to 0 for the next spd dispatch + SpdResetAtomicCounter(slice); + + // After mip 5 there is only a single workgroup left that downsamples the remaining up to 64x64 texels. + // compute MIP level 6 and 7 + SpdDownsampleMips_6_7H(x, y, mips, slice); + + // compute MIP level 8, 9, 10, 11 + SpdDownsampleNextFourH(x, y, FfxUInt32x2(0, 0), localInvocationIndex, 8, mips, slice); +} + +/// Downsamples a 64x64 tile based on the work group id and work group offset. +/// If after downsampling it's the last active thread group, computes the remaining MIP levels. +/// Uses half types. +/// +/// @param [in] workGroupID index of the work group / thread group +/// @param [in] localInvocationIndex index of the thread within the thread group in 1D +/// @param [in] mips the number of total MIP levels to compute for the input texture +/// @param [in] numWorkGroups the total number of dispatched work groups / thread groups for this slice +/// @param [in] slice the slice of the input texture +/// @param [in] workGroupOffset the work group offset. it's (0,0) in case the entire input texture is downsampled. +/// +/// @ingroup FfxGPUSpd +void SpdDownsampleH(FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mips, FfxUInt32 numWorkGroups, FfxUInt32 slice, FfxUInt32x2 workGroupOffset) +{ + SpdDownsampleH(workGroupID + workGroupOffset, localInvocationIndex, mips, numWorkGroups, slice); +} + +#endif // #if FFX_HALF +#endif // #if defined(FFX_GPU) diff --git a/Packages/fidelityfx.fsr/Shaders/shaders/spd/ffx_spd.h.meta b/Packages/fidelityfx.fsr/Shaders/shaders/spd/ffx_spd.h.meta new file mode 100644 index 0000000..2739093 --- /dev/null +++ b/Packages/fidelityfx.fsr/Shaders/shaders/spd/ffx_spd.h.meta @@ -0,0 +1,60 @@ +fileFormatVersion: 2 +guid: fe8e621d12c641c4297675f3015697fc +PluginImporter: + externalObjects: {} + serializedVersion: 2 + iconMap: {} + executionOrder: {} + defineConstraints: [] + isPreloaded: 0 + isOverridable: 1 + isExplicitlyReferenced: 0 + validateReferences: 1 + platformData: + - first: + : Any + second: + enabled: 0 + settings: + Exclude Editor: 1 + Exclude GameCoreScarlett: 1 + Exclude GameCoreXboxOne: 1 + Exclude Linux64: 1 + Exclude OSXUniversal: 1 + Exclude PS4: 1 + Exclude PS5: 1 + Exclude WebGL: 1 + Exclude Win: 1 + Exclude Win64: 1 + - first: + Any: + second: + enabled: 0 + settings: {} + - first: + Editor: Editor + second: + enabled: 0 + settings: + DefaultValueInitialized: true + - first: + Standalone: Linux64 + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win + second: + enabled: 0 + settings: + CPU: None + - first: + Standalone: Win64 + second: + enabled: 0 + settings: + CPU: None + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/fidelityfx.fsr/package.json b/Packages/fidelityfx.fsr/package.json new file mode 100644 index 0000000..d672c3f --- /dev/null +++ b/Packages/fidelityfx.fsr/package.json @@ -0,0 +1,12 @@ +{ + "name": "fidelityfx.fsr", + "version": "1.0.0", + "displayName": "FidelityFX FSR", + "description": "FidelityFX Super Resolution 2/3 Upscaler core assets", + "unity": "2020.1", + "documentationUrl": "https://github.com/ndepoel/FSR2Unity", + "author": { + "name": "Nico de Poel", + "email": "ndepoel@gmail.com.com" + } +} \ No newline at end of file diff --git a/Packages/fidelityfx.fsr/package.json.meta b/Packages/fidelityfx.fsr/package.json.meta new file mode 100644 index 0000000..c3ad2ad --- /dev/null +++ b/Packages/fidelityfx.fsr/package.json.meta @@ -0,0 +1,7 @@ +fileFormatVersion: 2 +guid: 202b84e3e446cc544a82954c2235ec60 +PackageManifestImporter: + externalObjects: {} + userData: + assetBundleName: + assetBundleVariant: diff --git a/Packages/packages-lock.json b/Packages/packages-lock.json index 4441e75..0daceeb 100644 --- a/Packages/packages-lock.json +++ b/Packages/packages-lock.json @@ -40,13 +40,13 @@ "url": "https://packages.unity.com" }, "com.unity.postprocessing": { - "version": "3.2.2", + "version": "file:com.unity.postprocessing", "depth": 0, - "source": "registry", + "source": "embedded", "dependencies": { - "com.unity.modules.physics": "1.0.0" - }, - "url": "https://packages.unity.com" + "com.unity.modules.physics": "1.0.0", + "fidelityfx.fsr": "1.0.0" + } }, "com.unity.test-framework": { "version": "1.1.31", @@ -99,6 +99,12 @@ }, "url": "https://packages.unity.com" }, + "fidelityfx.fsr": { + "version": "file:fidelityfx.fsr", + "depth": 0, + "source": "embedded", + "dependencies": {} + }, "com.unity.modules.ai": { "version": "1.0.0", "depth": 0, diff --git a/README.md b/README.md index c2fc3fd..fb3694d 100644 --- a/README.md +++ b/README.md @@ -1,14 +1,14 @@ -# FSR 3.0 Upscaler for Unity +# FSR 3.1 Upscaler for Unity -FidelityFX Super Resolution 3 (FSR3) Upscaler is an advanced image upscaling and temporal reconstruction technique developed by AMD and [released as open source](https://github.com/GPUOpen-LibrariesAndSDKs/FidelityFX-SDK/tree/release-FSR3-3.0.3) through the GPUOpen initiative. It provides both high-quality anti-aliasing and the ability for games to output at high resolutions while rendering at a lower internal resolution, improving performance. +FidelityFX Super Resolution 3 (FSR3) Upscaler is an advanced image upscaling and temporal reconstruction technique developed by AMD and [released as open source](https://github.com/GPUOpen-LibrariesAndSDKs/FidelityFX-SDK) through the GPUOpen initiative. It provides both high-quality anti-aliasing and the ability for games to output at high resolutions while rendering at a lower internal resolution, improving performance. This project aims to bring FSR3 upscaling to Unity as an alternative to Unity's existing FXAA, SMAA and TAA anti-aliasing solutions. Compared to Unity's TAA implementation, FSR3 Upscaler offers a sharper and more stable image, with a better sub-pixel detail resolve and better performance due to its reliance on upscaling. -![Comparison](images/comparison.png) +![Comparison](images/comparison-fsr3.1.png) -Rather than attempting to integrate AMD's open source C++ libraries for FSR3, this project instead reimplements the C++ backend in C# while adapting FSR3's HLSL shader code to work within Unity. This allows for maximum compatibility with the platforms and graphics APIs that Unity supports, including ones that normally wouldn't be supported by FSR3. +Rather than attempting to integrate AMD's native plugin libraries for FSR3, this project instead reimplements the C++ backend in C# using Unity's scripting APIs while adapting FSR3's HLSL shader code to work within Unity. This allows for maximum compatibility with the platforms and graphics APIs that Unity supports, including ones that normally wouldn't be supported by FSR3. -Focus of this project lies initially on making FSR3 Upscaler work with the traditional Unity built-in render pipeline. However the core FSR3 Upscaler classes are render pipeline-agnostic, so there is every possibility for URP and HDRP to be supported as well in the future. +Focus of this project lies initially on making FSR3 Upscaler work with the traditional Unity built-in render pipeline. However the core FSR3 Upscaler classes and shaders are render pipeline-agnostic, so there is every possibility for URP and HDRP integrations to be built as well. ## Table of Contents @@ -32,7 +32,7 @@ Focus of this project lies initially on making FSR3 Upscaler work with the tradi FSR3 Upscaler for Unity requires Unity 2020.1 or higher. This version of Unity added support for the `multi_compile` keyword in compute shaders, which the FSR3 Upscaler implementation makes use of. -Other than the above, requirements are no different from standard FSR3 Upscaler: it requires a GPU with compute shader support, typed UAV load and R16G16B16A16_UNORM support. In practice, almost every GPU made in the past decade will be able to run it. +Other than the above, requirements are no different from the standard FSR3 Upscaler libraries: it requires a GPU with compute shader support, typed UAV load and R16G16B16A16_UNORM support. In practice, almost every GPU made in the past decade will be able to run it. Platforms tested and confirmed working: - Windows @@ -171,6 +171,9 @@ Dynamic resolution works really well in combination with FSR3 Upscaler. Any run- - Texture mipmap bias adjustment is not working on MacOS Metal. This causes blurry textures as the internal render resolution is lowered. This is a Unity issue of some sort. Workaround: no known workaround yet. +- Using FP16 mode on Xbox with GDK October 2023 or later causes random jittering artifacts. + This seems to be a shader compilation issue introduced by this GDK update. + Workaround: disable FP16 on Xbox consoles. ## Details on implementation diff --git a/com.unity.postprocessing/PostProcessing/Editor/PostProcessLayerEditor.cs b/com.unity.postprocessing/PostProcessing/Editor/PostProcessLayerEditor.cs deleted file mode 100644 index 691483d..0000000 --- a/com.unity.postprocessing/PostProcessing/Editor/PostProcessLayerEditor.cs +++ /dev/null @@ -1,472 +0,0 @@ -using System; -using System.Collections.Generic; -using System.Linq; -using UnityEngine; -using UnityEngine.Rendering.PostProcessing; -using UnityEditorInternal; -using System.IO; - -namespace UnityEditor.Rendering.PostProcessing -{ - using SerializedBundleRef = PostProcessLayer.SerializedBundleRef; - using EXRFlags = Texture2D.EXRFlags; - - [CanEditMultipleObjects, CustomEditor(typeof(PostProcessLayer))] - sealed class PostProcessLayerEditor : BaseEditor - { - SerializedProperty m_StopNaNPropagation; -#pragma warning disable 414 - SerializedProperty m_DirectToCameraTarget; -#pragma warning restore 414 - SerializedProperty m_VolumeTrigger; - SerializedProperty m_VolumeLayer; - - SerializedProperty m_AntialiasingMode; - SerializedProperty m_TaaJitterSpread; - SerializedProperty m_TaaSharpness; - SerializedProperty m_TaaStationaryBlending; - SerializedProperty m_TaaMotionBlending; - SerializedProperty m_SmaaQuality; - SerializedProperty m_FxaaFastMode; - SerializedProperty m_FxaaKeepAlpha; - - SerializedProperty m_FsrQualityMode; - SerializedProperty m_FsrPerformSharpen; - SerializedProperty m_FsrSharpness; - SerializedProperty m_FsrEnableFP16; - SerializedProperty m_FsrExposureSource; - SerializedProperty m_FsrExposureTexture; - SerializedProperty m_FsrPreExposure; - SerializedProperty m_FsrAutoReactive; - SerializedProperty m_FsrAutoReactiveParams; - SerializedProperty m_FsrReactiveMaskTexture; - SerializedProperty m_FsrAutoTcr; - SerializedProperty m_FsrAutoTcrParams; - SerializedProperty m_FsrTcrMaskTexture; - - SerializedProperty m_FogEnabled; - SerializedProperty m_FogExcludeSkybox; - - SerializedProperty m_ShowToolkit; - SerializedProperty m_ShowCustomSorter; - - Dictionary m_CustomLists; - -#if UNITY_2017_3_OR_NEWER - Camera m_TargetCameraComponent; -#endif - - static GUIContent[] s_AntialiasingMethodNames = - { - new GUIContent("No Anti-aliasing"), - new GUIContent("Fast Approximate Anti-aliasing (FXAA)"), - new GUIContent("Subpixel Morphological Anti-aliasing (SMAA)"), - new GUIContent("Temporal Anti-aliasing (TAA)"), - new GUIContent("FidelityFX Super Resolution 3 (FSR3) Upscaler") - }; - - enum ExportMode - { - FullFrame, - DisablePost, - BreakBeforeColorGradingLinear, - BreakBeforeColorGradingLog - } - - void OnEnable() - { - m_StopNaNPropagation = FindProperty(x => x.stopNaNPropagation); - m_DirectToCameraTarget = FindProperty(x => x.finalBlitToCameraTarget); - m_VolumeTrigger = FindProperty(x => x.volumeTrigger); - m_VolumeLayer = FindProperty(x => x.volumeLayer); - - m_AntialiasingMode = FindProperty(x => x.antialiasingMode); - m_TaaJitterSpread = FindProperty(x => x.temporalAntialiasing.jitterSpread); - m_TaaSharpness = FindProperty(x => x.temporalAntialiasing.sharpness); - m_TaaStationaryBlending = FindProperty(x => x.temporalAntialiasing.stationaryBlending); - m_TaaMotionBlending = FindProperty(x => x.temporalAntialiasing.motionBlending); - m_SmaaQuality = FindProperty(x => x.subpixelMorphologicalAntialiasing.quality); - m_FxaaFastMode = FindProperty(x => x.fastApproximateAntialiasing.fastMode); - m_FxaaKeepAlpha = FindProperty(x => x.fastApproximateAntialiasing.keepAlpha); - - m_FsrQualityMode = FindProperty(x => x.superResolution.qualityMode); - m_FsrPerformSharpen = FindProperty(x => x.superResolution.performSharpenPass); - m_FsrSharpness = FindProperty(x => x.superResolution.sharpness); - m_FsrEnableFP16 = FindProperty(x => x.superResolution.enableFP16); - m_FsrExposureSource = FindProperty(x => x.superResolution.exposureSource); - m_FsrExposureTexture = FindProperty(x => x.superResolution.exposure); - m_FsrPreExposure = FindProperty(x => x.superResolution.preExposure); - m_FsrAutoReactive = FindProperty(x => x.superResolution.autoGenerateReactiveMask); - m_FsrAutoReactiveParams = FindProperty(x => x.superResolution.generateReactiveParameters); - m_FsrReactiveMaskTexture = FindProperty(x => x.superResolution.reactiveMask); - m_FsrAutoTcr = FindProperty(x => x.superResolution.autoGenerateTransparencyAndComposition); - m_FsrAutoTcrParams = FindProperty(x => x.superResolution.generateTransparencyAndCompositionParameters); - m_FsrTcrMaskTexture = FindProperty(x => x.superResolution.transparencyAndCompositionMask); - - m_FogEnabled = FindProperty(x => x.fog.enabled); - m_FogExcludeSkybox = FindProperty(x => x.fog.excludeSkybox); - - m_ShowToolkit = serializedObject.FindProperty("m_ShowToolkit"); - m_ShowCustomSorter = serializedObject.FindProperty("m_ShowCustomSorter"); - -#if UNITY_2017_3_OR_NEWER - m_TargetCameraComponent = m_Target.GetComponent(); -#endif - } - - void OnDisable() - { - m_CustomLists = null; - } - - public override void OnInspectorGUI() - { - serializedObject.Update(); - - var camera = m_Target.GetComponent(); - - DoVolumeBlending(); - DoAntialiasing(); - DoFog(camera); - - EditorGUILayout.PropertyField(m_StopNaNPropagation, EditorUtilities.GetContent("Stop NaN Propagation|Automatically replaces NaN/Inf in shaders by a black pixel to avoid breaking some effects. This will slightly affect performances and should only be used if you experience NaN issues that you can't fix. Has no effect on GLES2 platforms.")); - -#if UNITY_2019_1_OR_NEWER - if (!RuntimeUtilities.scriptableRenderPipelineActive) - EditorGUILayout.PropertyField(m_DirectToCameraTarget, EditorUtilities.GetContent("Directly to Camera Target|Use the final blit to the camera render target for postprocessing. This has less overhead but breaks compatibility with legacy image effect that use OnRenderImage.")); -#endif - - EditorGUILayout.Space(); - - DoToolkit(); - DoCustomEffectSorter(); - - EditorUtilities.DrawSplitter(); - EditorGUILayout.Space(); - - serializedObject.ApplyModifiedProperties(); - } - - void DoVolumeBlending() - { - EditorGUILayout.LabelField(EditorUtilities.GetContent("Volume blending"), EditorStyles.boldLabel); - EditorGUI.indentLevel++; - { - // The layout system sort of break alignement when mixing inspector fields with - // custom layouted fields, do the layout manually instead - var indentOffset = EditorGUI.indentLevel * 15f; - var lineRect = GUILayoutUtility.GetRect(1, EditorGUIUtility.singleLineHeight); - var labelRect = new Rect(lineRect.x, lineRect.y, EditorGUIUtility.labelWidth - indentOffset, lineRect.height); - var fieldRect = new Rect(labelRect.xMax, lineRect.y, lineRect.width - labelRect.width - 60f, lineRect.height); - var buttonRect = new Rect(fieldRect.xMax, lineRect.y, 60f, lineRect.height); - - EditorGUI.PrefixLabel(labelRect, EditorUtilities.GetContent("Trigger|A transform that will act as a trigger for volume blending.")); - m_VolumeTrigger.objectReferenceValue = (Transform)EditorGUI.ObjectField(fieldRect, m_VolumeTrigger.objectReferenceValue, typeof(Transform), true); - if (GUI.Button(buttonRect, EditorUtilities.GetContent("This|Assigns the current GameObject as a trigger."), EditorStyles.miniButton)) - m_VolumeTrigger.objectReferenceValue = m_Target.transform; - - if (m_VolumeTrigger.objectReferenceValue == null) - EditorGUILayout.HelpBox("No trigger has been set, the camera will only be affected by global volumes.", MessageType.Info); - - EditorGUILayout.PropertyField(m_VolumeLayer, EditorUtilities.GetContent("Layer|This camera will only be affected by volumes in the selected scene-layers.")); - - int mask = m_VolumeLayer.intValue; - if (mask == 0) - EditorGUILayout.HelpBox("No layer has been set, the trigger will never be affected by volumes.", MessageType.Warning); - else if (mask == -1 || ((mask & 1) != 0)) - EditorGUILayout.HelpBox("Do not use \"Everything\" or \"Default\" as a layer mask as it will slow down the volume blending process! Put post-processing volumes in their own dedicated layer for best performances.", MessageType.Warning); - } - EditorGUI.indentLevel--; - - EditorGUILayout.Space(); - } - - void DoAntialiasing() - { - EditorGUILayout.LabelField(EditorUtilities.GetContent("Anti-aliasing"), EditorStyles.boldLabel); - EditorGUI.indentLevel++; - { - m_AntialiasingMode.intValue = EditorGUILayout.Popup(EditorUtilities.GetContent("Mode|The anti-aliasing method to use. FXAA is fast but low quality. SMAA works well for non-HDR scenes. TAA is a bit slower but higher quality and works well with HDR."), m_AntialiasingMode.intValue, s_AntialiasingMethodNames); - - if (m_AntialiasingMode.intValue == (int)PostProcessLayer.Antialiasing.TemporalAntialiasing) - { -#if !UNITY_2017_3_OR_NEWER - if (RuntimeUtilities.isSinglePassStereoSelected) - EditorGUILayout.HelpBox("TAA requires Unity 2017.3+ for Single-pass stereo rendering support.", MessageType.Warning); -#endif -#if UNITY_2017_3_OR_NEWER - if (m_TargetCameraComponent != null && RuntimeUtilities.IsDynamicResolutionEnabled(m_TargetCameraComponent)) - EditorGUILayout.HelpBox("TAA is not supported with Dynamic Resolution.", MessageType.Warning); -#endif - - EditorGUILayout.PropertyField(m_TaaJitterSpread); - EditorGUILayout.PropertyField(m_TaaStationaryBlending); - EditorGUILayout.PropertyField(m_TaaMotionBlending); - EditorGUILayout.PropertyField(m_TaaSharpness); - } - else if (m_AntialiasingMode.intValue == (int)PostProcessLayer.Antialiasing.SubpixelMorphologicalAntialiasing) - { - if (RuntimeUtilities.isSinglePassStereoSelected) - EditorGUILayout.HelpBox("SMAA doesn't work with Single-pass stereo rendering.", MessageType.Warning); - - EditorGUILayout.PropertyField(m_SmaaQuality); - - if (m_SmaaQuality.intValue != (int)SubpixelMorphologicalAntialiasing.Quality.Low && EditorUtilities.isTargetingConsolesOrMobiles) - EditorGUILayout.HelpBox("For performance reasons it is recommended to use Low Quality on mobile and console platforms.", MessageType.Warning); - } - else if (m_AntialiasingMode.intValue == (int)PostProcessLayer.Antialiasing.FastApproximateAntialiasing) - { - EditorGUILayout.PropertyField(m_FxaaFastMode); - EditorGUILayout.PropertyField(m_FxaaKeepAlpha); - - if (!m_FxaaFastMode.boolValue && EditorUtilities.isTargetingConsolesOrMobiles) - EditorGUILayout.HelpBox("For performance reasons it is recommended to use Fast Mode on mobile and console platforms.", MessageType.Warning); - } - else if (m_AntialiasingMode.intValue == (int)PostProcessLayer.Antialiasing.SuperResolution) - { - EditorGUILayout.PropertyField(m_FsrQualityMode); - EditorGUILayout.PropertyField(m_FsrPerformSharpen); - EditorGUILayout.PropertyField(m_FsrSharpness); - EditorGUILayout.PropertyField(m_FsrEnableFP16); - EditorGUILayout.PropertyField(m_FsrExposureSource); - if (m_FsrExposureSource.intValue == (int)SuperResolution.ExposureSource.Manual) EditorGUILayout.PropertyField(m_FsrExposureTexture); - EditorGUILayout.PropertyField(m_FsrPreExposure); - EditorGUILayout.PropertyField(m_FsrAutoReactive); - EditorGUILayout.PropertyField(m_FsrAutoReactive.boolValue ? m_FsrAutoReactiveParams : m_FsrReactiveMaskTexture); - EditorGUILayout.PropertyField(m_FsrAutoTcr); - EditorGUILayout.PropertyField(m_FsrAutoTcr.boolValue ? m_FsrAutoTcrParams : m_FsrTcrMaskTexture); - } - } - EditorGUI.indentLevel--; - - EditorGUILayout.Space(); - } - - void DoFog(Camera camera) - { - if (camera == null || camera.actualRenderingPath != RenderingPath.DeferredShading) - return; - - EditorGUILayout.LabelField(EditorUtilities.GetContent("Deferred Fog"), EditorStyles.boldLabel); - EditorGUI.indentLevel++; - { - EditorGUILayout.PropertyField(m_FogEnabled); - - if (m_FogEnabled.boolValue) - { - EditorGUILayout.PropertyField(m_FogExcludeSkybox); - EditorGUILayout.HelpBox("This adds fog compatibility to the deferred rendering path; actual fog settings should be set in the Lighting panel.", MessageType.Info); - } - } - EditorGUI.indentLevel--; - - EditorGUILayout.Space(); - } - - void DoToolkit() - { - EditorUtilities.DrawSplitter(); - m_ShowToolkit.boolValue = EditorUtilities.DrawHeader("Toolkit", m_ShowToolkit.boolValue); - - if (m_ShowToolkit.boolValue) - { - GUILayout.Space(2); - - if (GUILayout.Button(EditorUtilities.GetContent("Export frame to EXR..."), EditorStyles.miniButton)) - { - var menu = new GenericMenu(); - menu.AddItem(EditorUtilities.GetContent("Full Frame (as displayed)"), false, () => ExportFrameToExr(ExportMode.FullFrame)); - menu.AddItem(EditorUtilities.GetContent("Disable post-processing"), false, () => ExportFrameToExr(ExportMode.DisablePost)); - menu.AddItem(EditorUtilities.GetContent("Break before Color Grading (Linear)"), false, () => ExportFrameToExr(ExportMode.BreakBeforeColorGradingLinear)); - menu.AddItem(EditorUtilities.GetContent("Break before Color Grading (Log)"), false, () => ExportFrameToExr(ExportMode.BreakBeforeColorGradingLog)); - menu.ShowAsContext(); - } - - if (GUILayout.Button(EditorUtilities.GetContent("Select all layer volumes|Selects all the volumes that will influence this layer."), EditorStyles.miniButton)) - { - var volumes = RuntimeUtilities.GetAllSceneObjects() - .Where(x => (m_VolumeLayer.intValue & (1 << x.gameObject.layer)) != 0) - .Select(x => x.gameObject) - .Cast() - .ToArray(); - - if (volumes.Length > 0) - Selection.objects = volumes; - } - - if (GUILayout.Button(EditorUtilities.GetContent("Select all active volumes|Selects all volumes currently affecting the layer."), EditorStyles.miniButton)) - { - var volumes = new List(); - PostProcessManager.instance.GetActiveVolumes(m_Target, volumes); - - if (volumes.Count > 0) - { - Selection.objects = volumes - .Select(x => x.gameObject) - .Cast() - .ToArray(); - } - } - - GUILayout.Space(3); - } - } - - void DoCustomEffectSorter() - { - EditorUtilities.DrawSplitter(); - m_ShowCustomSorter.boolValue = EditorUtilities.DrawHeader("Custom Effect Sorting", m_ShowCustomSorter.boolValue); - - if (m_ShowCustomSorter.boolValue) - { - bool isInPrefab = false; - - // Init lists if needed - if (m_CustomLists == null) - { - // In some cases the editor will refresh before components which means - // components might not have been fully initialized yet. In this case we also - // need to make sure that we're not in a prefab as sorteBundles isn't a - // serializable object and won't exist until put on a scene. - if (m_Target.sortedBundles == null) - { - isInPrefab = string.IsNullOrEmpty(m_Target.gameObject.scene.name); - - if (!isInPrefab) - { - // sortedBundles will be initialized and ready to use on the next frame - Repaint(); - } - } - else - { - // Create a reorderable list for each injection event - m_CustomLists = new Dictionary(); - foreach (var evt in Enum.GetValues(typeof(PostProcessEvent)).Cast()) - { - var bundles = m_Target.sortedBundles[evt]; - var listName = ObjectNames.NicifyVariableName(evt.ToString()); - - var list = new ReorderableList(bundles, typeof(SerializedBundleRef), true, true, false, false); - - list.drawHeaderCallback = (rect) => - { - EditorGUI.LabelField(rect, listName); - }; - - list.drawElementCallback = (rect, index, isActive, isFocused) => - { - var sbr = (SerializedBundleRef)list.list[index]; - EditorGUI.LabelField(rect, sbr.bundle.attribute.menuItem); - }; - - list.onReorderCallback = (l) => - { - EditorUtility.SetDirty(m_Target); - }; - - m_CustomLists.Add(evt, list); - } - } - } - - GUILayout.Space(5); - - if (isInPrefab) - { - EditorGUILayout.HelpBox("Not supported in prefabs.", MessageType.Info); - GUILayout.Space(3); - return; - } - - bool anyList = false; - if (m_CustomLists != null) - { - foreach (var kvp in m_CustomLists) - { - var list = kvp.Value; - - // Skip empty lists to avoid polluting the inspector - if (list.count == 0) - continue; - - list.DoLayoutList(); - anyList = true; - } - } - - if (!anyList) - { - EditorGUILayout.HelpBox("No custom effect loaded.", MessageType.Info); - GUILayout.Space(3); - } - } - } - - void ExportFrameToExr(ExportMode mode) - { - string path = EditorUtility.SaveFilePanel("Export EXR...", "", "Frame", "exr"); - - if (string.IsNullOrEmpty(path)) - return; - - EditorUtility.DisplayProgressBar("Export EXR", "Rendering...", 0f); - - var camera = m_Target.GetComponent(); - var w = camera.pixelWidth; - var h = camera.pixelHeight; - - var texOut = new Texture2D(w, h, TextureFormat.RGBAFloat, false, true); - var target = RenderTexture.GetTemporary(w, h, 24, RenderTextureFormat.ARGBFloat, RenderTextureReadWrite.Linear); - - var lastActive = RenderTexture.active; - var lastTargetSet = camera.targetTexture; - var lastPostFXState = m_Target.enabled; - var lastBreakColorGradingState = m_Target.breakBeforeColorGrading; - - if (mode == ExportMode.DisablePost) - m_Target.enabled = false; - else if (mode == ExportMode.BreakBeforeColorGradingLinear || mode == ExportMode.BreakBeforeColorGradingLog) - m_Target.breakBeforeColorGrading = true; - - camera.targetTexture = target; - camera.Render(); - camera.targetTexture = lastTargetSet; - - EditorUtility.DisplayProgressBar("Export EXR", "Reading...", 0.25f); - - m_Target.enabled = lastPostFXState; - m_Target.breakBeforeColorGrading = lastBreakColorGradingState; - - if (mode == ExportMode.BreakBeforeColorGradingLog) - { - // Convert to log - var material = new Material(Shader.Find("Hidden/PostProcessing/Editor/ConvertToLog")); - var newTarget = RenderTexture.GetTemporary(w, h, 0, RenderTextureFormat.ARGBFloat, RenderTextureReadWrite.Linear); - Graphics.Blit(target, newTarget, material, 0); - RenderTexture.ReleaseTemporary(target); - DestroyImmediate(material); - target = newTarget; - } - - RenderTexture.active = target; - texOut.ReadPixels(new Rect(0, 0, w, h), 0, 0); - texOut.Apply(); - RenderTexture.active = lastActive; - - EditorUtility.DisplayProgressBar("Export EXR", "Encoding...", 0.5f); - - var bytes = texOut.EncodeToEXR(EXRFlags.OutputAsFloat | EXRFlags.CompressZIP); - - EditorUtility.DisplayProgressBar("Export EXR", "Saving...", 0.75f); - - File.WriteAllBytes(path, bytes); - - EditorUtility.ClearProgressBar(); - AssetDatabase.Refresh(); - - RenderTexture.ReleaseTemporary(target); - DestroyImmediate(texOut); - } - } -} diff --git a/com.unity.postprocessing/PostProcessing/Editor/Unity.Postprocessing.Editor.asmdef b/com.unity.postprocessing/PostProcessing/Editor/Unity.Postprocessing.Editor.asmdef deleted file mode 100644 index a96f35c..0000000 --- a/com.unity.postprocessing/PostProcessing/Editor/Unity.Postprocessing.Editor.asmdef +++ /dev/null @@ -1,36 +0,0 @@ -{ - "name": "Unity.Postprocessing.Editor", - "rootNamespace": "", - "references": [ - "Unity.Postprocessing.Runtime", - "Unity.XR.Management", - "Unity.XR.Management.Editor" - ], - "includePlatforms": [ - "Editor" - ], - "excludePlatforms": [], - "allowUnsafeCode": false, - "overrideReferences": false, - "precompiledReferences": [], - "autoReferenced": true, - "defineConstraints": [], - "versionDefines": [ - { - "name": "com.unity.xr.management", - "expression": "4.0.1", - "define": "XR_MANAGEMENT_4_0_1_OR_NEWER" - }, - { - "name": "com.unity.modules.vr", - "expression": "1.0.0", - "define": "ENABLE_VR_MODULE" - }, - { - "name": "com.unity.modules.xr", - "expression": "1.0.0", - "define": "ENABLE_XR_MODULE" - } - ], - "noEngineReferences": false -} diff --git a/com.unity.postprocessing/PostProcessing/PostProcessResources.asset b/com.unity.postprocessing/PostProcessing/PostProcessResources.asset deleted file mode 100644 index 835a2eb..0000000 --- a/com.unity.postprocessing/PostProcessing/PostProcessResources.asset +++ /dev/null @@ -1,138 +0,0 @@ -%YAML 1.1 -%TAG !u! tag:unity3d.com,2011: ---- !u!114 &11400000 -MonoBehaviour: - m_ObjectHideFlags: 0 - m_CorrespondingSourceObject: {fileID: 0} - m_PrefabInstance: {fileID: 0} - m_PrefabAsset: {fileID: 0} - m_GameObject: {fileID: 0} - m_Enabled: 1 - m_EditorHideFlags: 0 - m_Script: {fileID: 11500000, guid: 30f4b897495c7ad40b2d47143e02aaba, type: 3} - m_Name: PostProcessResources - m_EditorClassIdentifier: - blueNoise64: - - {fileID: 2800000, guid: 50b54341495978843a6f85583ed4417d, type: 3} - - {fileID: 2800000, guid: 3c2f1fb7e4b66e74191b7c328ada52d9, type: 3} - - {fileID: 2800000, guid: a469f920b21fc7c4fb5b950917ce2fb2, type: 3} - - {fileID: 2800000, guid: 373f9bf6b0841af4ebf26d25e4a3f4e2, type: 3} - - {fileID: 2800000, guid: 6fa5cf178eaaa5f42b820f636bb6e0bd, type: 3} - - {fileID: 2800000, guid: a1ae041906217ae44a774d4ca139af50, type: 3} - - {fileID: 2800000, guid: 79b86f3419b87f3429164a956da8cfab, type: 3} - - {fileID: 2800000, guid: 3ac02e7e783571c468f9c086d2384ba7, type: 3} - - {fileID: 2800000, guid: c55042318a938344ab23cd7f09dd0076, type: 3} - - {fileID: 2800000, guid: 71583cfd8899717428d5b1a95fa39cda, type: 3} - - {fileID: 2800000, guid: afe1e502240079342a0a980484b6da8b, type: 3} - - {fileID: 2800000, guid: 771903fe7b4674445829e52e91cff019, type: 3} - - {fileID: 2800000, guid: 980acadb960f8424c94307ec0e585b4e, type: 3} - - {fileID: 2800000, guid: 68613e6a221be1a4b9f31d7fa1c2d1bf, type: 3} - - {fileID: 2800000, guid: f6439b54b28f3884eb67579dec0b6f21, type: 3} - - {fileID: 2800000, guid: 2ee161d8945169243b5698fec114e1b7, type: 3} - - {fileID: 2800000, guid: 153f7d6dfbe713d4884df0f1e243ba92, type: 3} - - {fileID: 2800000, guid: bf95b6fdc179b0e4f890c841406193fc, type: 3} - - {fileID: 2800000, guid: 74aca53eb7273624baffc2bf5e5cc173, type: 3} - - {fileID: 2800000, guid: 729a3ae164bcb3b4380459386adcf331, type: 3} - - {fileID: 2800000, guid: 6dda07f1420a968449cf4c6620c44d9f, type: 3} - - {fileID: 2800000, guid: b7f000750830ddb4bbc80065b9314ce9, type: 3} - - {fileID: 2800000, guid: df01d03f056c6f445b4b8a0ae054207c, type: 3} - - {fileID: 2800000, guid: bfe953600e8fb1849a804ee08ace7b4c, type: 3} - - {fileID: 2800000, guid: 32c6a5f7143b86c44bd5cdee2ff3f8ad, type: 3} - - {fileID: 2800000, guid: f4b8ab78b57749d4e96d36f6d8a395d0, type: 3} - - {fileID: 2800000, guid: 09f6c01f98a3ded4daf1afc52a3c260f, type: 3} - - {fileID: 2800000, guid: bdd06fb88ef36ed4a85dd506352c2d80, type: 3} - - {fileID: 2800000, guid: 02c0a84bd64c6f044954d8bde9b46ec8, type: 3} - - {fileID: 2800000, guid: aa80dc44aa4fe4c43bb9d51d90cf2958, type: 3} - - {fileID: 2800000, guid: 0fa10b21877c61b4db40ba5708815f81, type: 3} - - {fileID: 2800000, guid: 6b0a189df0bd4d5448eaefb4e673ace8, type: 3} - - {fileID: 2800000, guid: 87a5e40cc271ea648b583616f6ebe7fe, type: 3} - - {fileID: 2800000, guid: b71bb466b71fd13449dd736f63caeb67, type: 3} - - {fileID: 2800000, guid: 319b8e66db3faa4438cf6982e9c89b2f, type: 3} - - {fileID: 2800000, guid: 0a79c155edf9b2d429d4736abee5acdb, type: 3} - - {fileID: 2800000, guid: 351e95d0e20a54849bd4ce5f9b498934, type: 3} - - {fileID: 2800000, guid: 1d6958e30e40a254dbe5a54c573eeb3c, type: 3} - - {fileID: 2800000, guid: 9660a4ca1ca8425408ac25c641932977, type: 3} - - {fileID: 2800000, guid: 547dbd5f858c74047ba3f213e4408307, type: 3} - - {fileID: 2800000, guid: 1a9ce5640cde5934aae0022f020464a6, type: 3} - - {fileID: 2800000, guid: cd9006dc442cc244e89b3f492384d46a, type: 3} - - {fileID: 2800000, guid: b266511438fae724f9d3ce6bd26583e8, type: 3} - - {fileID: 2800000, guid: 71bc1b6b66e8b784b972199b7e90204e, type: 3} - - {fileID: 2800000, guid: 15e54aa23a938444389469d53765d741, type: 3} - - {fileID: 2800000, guid: b9960364038cbfa4aa49d7b2032d3110, type: 3} - - {fileID: 2800000, guid: 8ecbbcae4cc747a4abbc4adce795d25e, type: 3} - - {fileID: 2800000, guid: 1378a33cdd085d64c9da863d2484ff21, type: 3} - - {fileID: 2800000, guid: aff59c63d25d43f4c938f248837c30fb, type: 3} - - {fileID: 2800000, guid: 3f7c3687170b90e4a8d2ee6b142670f4, type: 3} - - {fileID: 2800000, guid: d8c290e38ff0425409d0ae6a98c95e41, type: 3} - - {fileID: 2800000, guid: d5a51525b27e3ee4aadbeb39cbcf0750, type: 3} - - {fileID: 2800000, guid: d2e8e90fac2e6a341a38e1c3963c218d, type: 3} - - {fileID: 2800000, guid: c94b57b5a32a22d43ade66e09f6a4bd2, type: 3} - - {fileID: 2800000, guid: 936dea238abb0864ab3985a995e16a29, type: 3} - - {fileID: 2800000, guid: 5e542d0126a2c7848b66bffc428905fd, type: 3} - - {fileID: 2800000, guid: 70f23eaf7d8ae9147aa542d20e93733b, type: 3} - - {fileID: 2800000, guid: e138166e7a7c70f49943be7edda35d35, type: 3} - - {fileID: 2800000, guid: 85a45a6d8b2ffb84987d2b028ecfb220, type: 3} - - {fileID: 2800000, guid: d96974690c77f50489eb60ec84bd8dac, type: 3} - - {fileID: 2800000, guid: 404fa8def46b1c447817e1ebdaa7144e, type: 3} - - {fileID: 2800000, guid: 119591e0bb084e848835d237546b3882, type: 3} - - {fileID: 2800000, guid: a03c400b0e3959f428ee99dfc6cfc263, type: 3} - - {fileID: 2800000, guid: 4a11d65ce13d5f542a0ff136cc2f3fba, type: 3} - blueNoise256: - - {fileID: 2800000, guid: 6017f374382d64245a0a4aab668e6f38, type: 3} - - {fileID: 2800000, guid: 0f8fa14b3731cda4e947062e734d5e1e, type: 3} - - {fileID: 2800000, guid: 1abfe0e165ca1e9428b455ffc9a2d9ef, type: 3} - - {fileID: 2800000, guid: c072b653e98a06e40857d76ca8c7eecd, type: 3} - - {fileID: 2800000, guid: b52d5033b68309943a2386c270a90f44, type: 3} - - {fileID: 2800000, guid: acde5141d5f4f7a4188394bd52c4dc38, type: 3} - - {fileID: 2800000, guid: 999434725cbc2be4eb54043b36efd4a8, type: 3} - - {fileID: 2800000, guid: 70d0a1182b29d6347ac70374c3593bba, type: 3} - smaaLuts: - area: {fileID: 2800000, guid: 73ec4ae984a0a0f44a2be737e41a6f2f, type: 3} - search: {fileID: 2800000, guid: d99701099481a2f489610e977df6dcbc, type: 3} - shaders: - bloom: {fileID: 4800000, guid: c1e1d3119c6fd4646aea0b4b74cacc1a, type: 3} - copy: {fileID: 4800000, guid: cdbdb71de5f9c454b980f6d0e87f0afb, type: 3} - copyStd: {fileID: 4800000, guid: 4bf4cff0d0bac3d43894e2e8839feb40, type: 3} - copyStdFromTexArray: {fileID: 4800000, guid: 02d2da9bc88d25c4d878c1ed4e0b3854, type: 3} - copyStdFromDoubleWide: {fileID: 4800000, guid: e8ce9961912f3214586fe8709b9012c1, type: 3} - discardAlpha: {fileID: 4800000, guid: 5ab0816423f0dfe45841cab3b05ec9ef, type: 3} - depthOfField: {fileID: 4800000, guid: 0ef78d24e85a44f4da9d5b5eaa00e50b, type: 3} - finalPass: {fileID: 4800000, guid: f75014305794b3948a3c6d5ccd550e05, type: 3} - grainBaker: {fileID: 4800000, guid: 0d8afcb51cc9f0349a6d190da929b838, type: 3} - motionBlur: {fileID: 4800000, guid: 2c459b89a7c8b1a4fbefe0d81341651c, type: 3} - temporalAntialiasing: {fileID: 4800000, guid: 51bcf79c50dc92e47ba87821b61100c3, type: 3} - subpixelMorphologicalAntialiasing: {fileID: 4800000, guid: 81af42a93ade3dd46a9b583d4eec76d6, type: 3} - texture2dLerp: {fileID: 4800000, guid: 34a819c9e33402547a81619693adc8d5, type: 3} - uber: {fileID: 4800000, guid: 382151503e2a43a4ebb7366d1632731d, type: 3} - lut2DBaker: {fileID: 4800000, guid: 7ad194cbe7d006f4bace915156972026, type: 3} - lightMeter: {fileID: 4800000, guid: b34a29e523cb9d545881e193a079f2df, type: 3} - gammaHistogram: {fileID: 4800000, guid: f7ea35cfb33fcad4ab8f2429ec103bef, type: 3} - waveform: {fileID: 4800000, guid: 3020ac7ece79a7f4eb789a236f8bd6c5, type: 3} - vectorscope: {fileID: 4800000, guid: a71093f2a4fe26a40805c22739e10e4a, type: 3} - debugOverlays: {fileID: 4800000, guid: b958ad1c92bd3d64c9e61318b8681dab, type: 3} - deferredFog: {fileID: 4800000, guid: 4117fce9491711c4094d33a048e36e73, type: 3} - scalableAO: {fileID: 4800000, guid: d7640629310e79646af0f46eb55ae466, type: 3} - multiScaleAO: {fileID: 4800000, guid: 67f9497810829eb4791ec19e95781e51, type: 3} - screenSpaceReflections: {fileID: 4800000, guid: f997a3dc9254c44459323cced085150c, type: 3} - computeShaders: - autoExposure: {fileID: 7200000, guid: 34845e0ca016b7448842e965db5890a5, type: 3} - exposureHistogram: {fileID: 7200000, guid: 8c2fcbdf9bc58664f89917f7b9d79501, type: 3} - lut3DBaker: {fileID: 7200000, guid: 42496b74c071f5749950ca1abe33e945, type: 3} - texture3dLerp: {fileID: 7200000, guid: 31e9175024adfd44aba2530ff9b77494, type: 3} - gammaHistogram: {fileID: 7200000, guid: 18183ebfeeab97749b43e38b928604a7, type: 3} - waveform: {fileID: 7200000, guid: 92c63830cd50c0b4fbb8233613839958, type: 3} - vectorscope: {fileID: 7200000, guid: e1efca7c36fd01840aae0dd10378de5c, type: 3} - multiScaleAODownsample1: {fileID: 7200000, guid: 4c63bc487e6c29a4a99f85a6c47b292b, type: 3} - multiScaleAODownsample2: {fileID: 7200000, guid: e4d3e4779e48a374f91d48d4c0aedb7b, type: 3} - multiScaleAORender: {fileID: 7200000, guid: 34a460e8a2e66c243a9c12024e5a798d, type: 3} - multiScaleAOUpsample: {fileID: 7200000, guid: 600d6212b59bb40409d19d750b5fd1e9, type: 3} - gaussianDownsample: {fileID: 7200000, guid: 6dba4103d23a7904fbc49099355aff3e, type: 3} - superResolution: - computeLuminancePyramidPass: {fileID: 7200000, guid: d253be05abcdc80428503d3e4cce3a36, type: 3} - reconstructPreviousDepthPass: {fileID: 7200000, guid: 4f59e5b9179d74844ae06a30ae1e0629, type: 3} - depthClipPass: {fileID: 7200000, guid: 20e44016ed34b0d4b8de499d1b566c69, type: 3} - lockPass: {fileID: 7200000, guid: a135306e6d1857e43a86ef20db2a47fe, type: 3} - accumulatePass: {fileID: 7200000, guid: c9b45f0ae7673694ba57a4aadfe212e9, type: 3} - sharpenPass: {fileID: 7200000, guid: 7aaf5cfff022de2499e9b0412f947f6c, type: 3} - autoGenReactivePass: {fileID: 7200000, guid: 5716b91fdaa4e9e439df6b96a796fe6e, type: 3} - tcrAutoGenPass: {fileID: 7200000, guid: 75cdc6ef23f08ed498d4da511923fcea, type: 3} diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Effects/SuperResolution.cs b/com.unity.postprocessing/PostProcessing/Runtime/Effects/SuperResolution.cs deleted file mode 100644 index 90c4f59..0000000 --- a/com.unity.postprocessing/PostProcessing/Runtime/Effects/SuperResolution.cs +++ /dev/null @@ -1,334 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -using System; -using System.Collections; -using System.Collections.Generic; -using UnityEngine; -using UnityEngine.Experimental.Rendering; -using FidelityFX; - -namespace UnityEngine.Rendering.PostProcessing -{ - [UnityEngine.Scripting.Preserve] - [Serializable] - public class SuperResolution - { - public Func callbacksFactory { get; set; } = (context) => new Fsr3UpscalerCallbacksBase(); - - [Tooltip("Standard scaling ratio presets.")] - public Fsr3Upscaler.QualityMode qualityMode = Fsr3Upscaler.QualityMode.Quality; - - [Tooltip("Apply RCAS sharpening to the image after upscaling.")] - public bool performSharpenPass = true; - [Tooltip("Strength of the sharpening effect.")] - [Range(0, 1)] public float sharpness = 0.8f; - - [Tooltip("Allow the use of half precision compute operations, potentially improving performance if the platform supports it.")] - public bool enableFP16 = false; - - [Tooltip("Choose where to get the exposure value from. Use auto-exposure from either FSR3 or Unity, provide a manual exposure texture, or use a default value.")] - public ExposureSource exposureSource = ExposureSource.Auto; - [Tooltip("Value by which the input signal will be divided, to get back to the original signal produced by the game.")] - public float preExposure = 1.0f; - [Tooltip("Optional 1x1 texture containing the exposure value for the current frame.")] - public Texture exposure = null; - - public enum ExposureSource - { - Default, - Auto, - Unity, - Manual, - } - - [Tooltip("Optional texture to control the influence of the current frame on the reconstructed output. If unset, either an auto-generated or a default cleared reactive mask will be used.")] - public Texture reactiveMask = null; - [Tooltip("Optional texture for marking areas of specialist rendering which should be accounted for during the upscaling process. If unset, a default cleared mask will be used.")] - public Texture transparencyAndCompositionMask = null; - [Tooltip("Automatically generate a reactive mask based on the difference between opaque-only render output and the final render output including alpha transparencies.")] - public bool autoGenerateReactiveMask = true; - [Tooltip("Parameters to control the process of auto-generating a reactive mask.")] - public GenerateReactiveParameters generateReactiveParameters = new GenerateReactiveParameters(); - - [Serializable] - public class GenerateReactiveParameters - { - [Tooltip("A value to scale the output")] - [Range(0, 2)] public float scale = 0.5f; - [Tooltip("A threshold value to generate a binary reactive mask")] - [Range(0, 1)] public float cutoffThreshold = 0.2f; - [Tooltip("A value to set for the binary reactive mask")] - [Range(0, 1)] public float binaryValue = 0.9f; - [Tooltip("Flags to determine how to generate the reactive mask")] - public Fsr3Upscaler.GenerateReactiveFlags flags = Fsr3Upscaler.GenerateReactiveFlags.ApplyTonemap | Fsr3Upscaler.GenerateReactiveFlags.ApplyThreshold | Fsr3Upscaler.GenerateReactiveFlags.UseComponentsMax; - } - - [Tooltip("(Experimental) Automatically generate and use Reactive mask and Transparency & composition mask internally.")] - public bool autoGenerateTransparencyAndComposition = false; - [Tooltip("Parameters to control the process of auto-generating transparency and composition masks.")] - public GenerateTcrParameters generateTransparencyAndCompositionParameters = new GenerateTcrParameters(); - - [Serializable] - public class GenerateTcrParameters - { - [Tooltip("Setting this value too small will cause visual instability. Larger values can cause ghosting.")] - [Range(0, 1)] public float autoTcThreshold = 0.05f; - [Tooltip("Smaller values will increase stability at hard edges of translucent objects.")] - [Range(0, 2)] public float autoTcScale = 1.0f; - [Tooltip("Larger values result in more reactive pixels.")] - [Range(0, 10)] public float autoReactiveScale = 5.0f; - [Tooltip("Maximum value reactivity can reach.")] - [Range(0, 1)] public float autoReactiveMax = 0.9f; - } - - public Vector2 jitter { get; private set; } - public Vector2Int renderSize => _maxRenderSize; - public Vector2Int displaySize => _displaySize; - public RenderTargetIdentifier colorOpaqueOnly { get; set; } - - private Fsr3UpscalerContext _fsrContext; - private Vector2Int _maxRenderSize; - private Vector2Int _displaySize; - private bool _resetHistory; - - private IFsr3UpscalerCallbacks _callbacks; - - private readonly Fsr3Upscaler.DispatchDescription _dispatchDescription = new Fsr3Upscaler.DispatchDescription(); - private readonly Fsr3Upscaler.GenerateReactiveDescription _genReactiveDescription = new Fsr3Upscaler.GenerateReactiveDescription(); - - private Fsr3Upscaler.QualityMode _prevQualityMode; - private ExposureSource _prevExposureSource; - private Vector2Int _prevDisplaySize; - - private Rect _originalRect; - - public bool IsSupported() - { - return SystemInfo.supportsComputeShaders && SystemInfo.supportsMotionVectors; - } - - public DepthTextureMode GetCameraFlags() - { - return DepthTextureMode.Depth | DepthTextureMode.MotionVectors; - } - - public void Release() - { - DestroyFsrContext(); - } - - public void ResetHistory() - { - _resetHistory = true; - } - - public void ConfigureJitteredProjectionMatrix(PostProcessRenderContext context) - { - ApplyJitter(context.camera); - } - - public void ConfigureCameraViewport(PostProcessRenderContext context) - { - var camera = context.camera; - _originalRect = camera.rect; - - // Determine the desired rendering and display resolutions - _displaySize = new Vector2Int(camera.pixelWidth, camera.pixelHeight); - Fsr3Upscaler.GetRenderResolutionFromQualityMode(out int maxRenderWidth, out int maxRenderHeight, _displaySize.x, _displaySize.y, qualityMode); - _maxRenderSize = new Vector2Int(maxRenderWidth, maxRenderHeight); - - // Render to a smaller portion of the screen by manipulating the camera's viewport rect - camera.aspect = (float)_displaySize.x / _displaySize.y; - camera.rect = new Rect(0, 0, _originalRect.width * _maxRenderSize.x / _displaySize.x, _originalRect.height * _maxRenderSize.y / _displaySize.y); - } - - public void ResetCameraViewport(PostProcessRenderContext context) - { - context.camera.rect = _originalRect; - } - - public void Render(PostProcessRenderContext context) - { - var cmd = context.command; - cmd.BeginSample("FSR3 Upscaler"); - - // Monitor for any resolution changes and recreate the FSR3 Upscaler context if necessary - // We can't create an FSR3 Upscaler context without info from the post-processing context, so delay the initial setup until here - if (_fsrContext == null || _displaySize.x != _prevDisplaySize.x || _displaySize.y != _prevDisplaySize.y || qualityMode != _prevQualityMode || exposureSource != _prevExposureSource) - { - DestroyFsrContext(); - CreateFsrContext(context); - } - - SetupDispatchDescription(context); - - if (autoGenerateReactiveMask) - { - SetupAutoReactiveDescription(context); - - var scaledRenderSize = _genReactiveDescription.RenderSize; - cmd.GetTemporaryRT(Fsr3ShaderIDs.UavAutoReactive, scaledRenderSize.x, scaledRenderSize.y, 0, default, GraphicsFormat.R8_UNorm, 1, true); - _fsrContext.GenerateReactiveMask(_genReactiveDescription, cmd); - _dispatchDescription.Reactive = new ResourceView(Fsr3ShaderIDs.UavAutoReactive); - } - - _fsrContext.Dispatch(_dispatchDescription, cmd); - - cmd.EndSample("FSR3 Upscaler"); - - _resetHistory = false; - } - - private void CreateFsrContext(PostProcessRenderContext context) - { - _prevQualityMode = qualityMode; - _prevExposureSource = exposureSource; - _prevDisplaySize = _displaySize; - - // Initialize FSR3 Upscaler context - Fsr3Upscaler.InitializationFlags flags = 0; - if (context.camera.allowHDR) flags |= Fsr3Upscaler.InitializationFlags.EnableHighDynamicRange; - if (enableFP16) flags |= Fsr3Upscaler.InitializationFlags.EnableFP16Usage; - if (exposureSource == ExposureSource.Auto) flags |= Fsr3Upscaler.InitializationFlags.EnableAutoExposure; - if (RuntimeUtilities.IsDynamicResolutionEnabled(context.camera)) flags |= Fsr3Upscaler.InitializationFlags.EnableDynamicResolution; - - _callbacks = callbacksFactory(context); - _fsrContext = Fsr3Upscaler.CreateContext(_displaySize, _maxRenderSize, context.resources.computeShaders.superResolution, flags); - - // Apply a mipmap bias so that textures retain their sharpness - float biasOffset = Fsr3Upscaler.GetMipmapBiasOffset(_maxRenderSize.x, _displaySize.x); - if (!float.IsNaN(biasOffset) && !float.IsInfinity(biasOffset)) - { - _callbacks.ApplyMipmapBias(biasOffset); - } - } - - private void DestroyFsrContext() - { - if (_fsrContext != null) - { - _fsrContext.Destroy(); - _fsrContext = null; - } - - if (_callbacks != null) - { - // Undo the current mipmap bias offset - _callbacks.UndoMipmapBias(); - _callbacks = null; - } - } - - private void ApplyJitter(Camera camera) - { - var scaledRenderSize = GetScaledRenderSize(camera); - - // Perform custom jittering of the camera's projection matrix according to FSR3's recipe - int jitterPhaseCount = Fsr3Upscaler.GetJitterPhaseCount(scaledRenderSize.x, _displaySize.x); - Fsr3Upscaler.GetJitterOffset(out float jitterX, out float jitterY, Time.frameCount, jitterPhaseCount); - - _dispatchDescription.JitterOffset = new Vector2(jitterX, jitterY); - - jitterX = 2.0f * jitterX / scaledRenderSize.x; - jitterY = 2.0f * jitterY / scaledRenderSize.y; - - var jitterTranslationMatrix = Matrix4x4.Translate(new Vector3(jitterX, jitterY, 0)); - camera.nonJitteredProjectionMatrix = camera.projectionMatrix; - camera.projectionMatrix = jitterTranslationMatrix * camera.nonJitteredProjectionMatrix; - camera.useJitteredProjectionMatrixForTransparentRendering = true; - - jitter = new Vector2(jitterX, jitterY); - } - - private void SetupDispatchDescription(PostProcessRenderContext context) - { - var camera = context.camera; - - // Set up the main FSR3 Upscaler dispatch parameters - _dispatchDescription.Color = new ResourceView(context.source); - _dispatchDescription.Depth = new ResourceView(BuiltinRenderTextureType.CameraTarget, RenderTextureSubElement.Depth); - _dispatchDescription.MotionVectors = new ResourceView(BuiltinRenderTextureType.MotionVectors); - _dispatchDescription.Exposure = ResourceView.Unassigned; - _dispatchDescription.Reactive = ResourceView.Unassigned; - _dispatchDescription.TransparencyAndComposition = ResourceView.Unassigned; - - if (exposureSource == ExposureSource.Manual && exposure != null) _dispatchDescription.Exposure = new ResourceView(exposure); - if (exposureSource == ExposureSource.Unity) _dispatchDescription.Exposure = new ResourceView(context.autoExposureTexture); - if (reactiveMask != null) _dispatchDescription.Reactive = new ResourceView(reactiveMask); - if (transparencyAndCompositionMask != null) _dispatchDescription.TransparencyAndComposition = new ResourceView(transparencyAndCompositionMask); - - var scaledRenderSize = GetScaledRenderSize(context.camera); - - _dispatchDescription.Output = new ResourceView(context.destination); - _dispatchDescription.PreExposure = preExposure; - _dispatchDescription.EnableSharpening = performSharpenPass; - _dispatchDescription.Sharpness = sharpness; - _dispatchDescription.MotionVectorScale.x = -scaledRenderSize.x; - _dispatchDescription.MotionVectorScale.y = -scaledRenderSize.y; - _dispatchDescription.RenderSize = scaledRenderSize; - _dispatchDescription.InputResourceSize = scaledRenderSize; - _dispatchDescription.FrameTimeDelta = Time.unscaledDeltaTime; - _dispatchDescription.CameraNear = camera.nearClipPlane; - _dispatchDescription.CameraFar = camera.farClipPlane; - _dispatchDescription.CameraFovAngleVertical = camera.fieldOfView * Mathf.Deg2Rad; - _dispatchDescription.ViewSpaceToMetersFactor = 1.0f; // 1 unit is 1 meter in Unity - _dispatchDescription.Reset = _resetHistory; - - // Set up the parameters for the optional experimental auto-TCR feature - _dispatchDescription.EnableAutoReactive = autoGenerateTransparencyAndComposition; - if (autoGenerateTransparencyAndComposition) - { - _dispatchDescription.ColorOpaqueOnly = new ResourceView(colorOpaqueOnly); - _dispatchDescription.AutoTcThreshold = generateTransparencyAndCompositionParameters.autoTcThreshold; - _dispatchDescription.AutoTcScale = generateTransparencyAndCompositionParameters.autoTcScale; - _dispatchDescription.AutoReactiveScale = generateTransparencyAndCompositionParameters.autoReactiveScale; - _dispatchDescription.AutoReactiveMax = generateTransparencyAndCompositionParameters.autoReactiveMax; - } - - if (SystemInfo.usesReversedZBuffer) - { - // Swap the near and far clip plane distances as FSR3 expects this when using inverted depth - (_dispatchDescription.CameraNear, _dispatchDescription.CameraFar) = (_dispatchDescription.CameraFar, _dispatchDescription.CameraNear); - } - } - - private void SetupAutoReactiveDescription(PostProcessRenderContext context) - { - // Set up the parameters to auto-generate a reactive mask - _genReactiveDescription.ColorOpaqueOnly = new ResourceView(colorOpaqueOnly); - _genReactiveDescription.ColorPreUpscale = new ResourceView(context.source); - _genReactiveDescription.OutReactive = new ResourceView(Fsr3ShaderIDs.UavAutoReactive); - _genReactiveDescription.RenderSize = GetScaledRenderSize(context.camera); - _genReactiveDescription.Scale = generateReactiveParameters.scale; - _genReactiveDescription.CutoffThreshold = generateReactiveParameters.cutoffThreshold; - _genReactiveDescription.BinaryValue = generateReactiveParameters.binaryValue; - _genReactiveDescription.Flags = generateReactiveParameters.flags; - } - - internal Vector2Int GetScaledRenderSize(Camera camera) - { - if (!RuntimeUtilities.IsDynamicResolutionEnabled(camera)) - return _maxRenderSize; - - return new Vector2Int(Mathf.CeilToInt(_maxRenderSize.x * ScalableBufferManager.widthScaleFactor), Mathf.CeilToInt(_maxRenderSize.y * ScalableBufferManager.heightScaleFactor)); - } - } -} diff --git a/com.unity.postprocessing/PostProcessing/Runtime/FSR3.meta b/com.unity.postprocessing/PostProcessing/Runtime/FSR3.meta deleted file mode 100644 index e12cfa8..0000000 --- a/com.unity.postprocessing/PostProcessing/Runtime/FSR3.meta +++ /dev/null @@ -1,8 +0,0 @@ -fileFormatVersion: 2 -guid: bfeca308812376e4a8e5e49e0d96c5c6 -folderAsset: yes -DefaultImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3ShaderIDs.cs b/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3ShaderIDs.cs deleted file mode 100644 index 3a28843..0000000 --- a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3ShaderIDs.cs +++ /dev/null @@ -1,80 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -using UnityEngine; - -namespace FidelityFX -{ - internal static class Fsr3ShaderIDs - { - // Shader resource views, i.e. read-only bindings - internal static readonly int SrvInputColor = Shader.PropertyToID("r_input_color_jittered"); - internal static readonly int SrvOpaqueOnly = Shader.PropertyToID("r_input_opaque_only"); - internal static readonly int SrvInputMotionVectors = Shader.PropertyToID("r_input_motion_vectors"); - internal static readonly int SrvInputDepth = Shader.PropertyToID("r_input_depth"); - internal static readonly int SrvInputExposure = Shader.PropertyToID("r_input_exposure"); - internal static readonly int SrvAutoExposure = Shader.PropertyToID("r_auto_exposure"); - internal static readonly int SrvReactiveMask = Shader.PropertyToID("r_reactive_mask"); - internal static readonly int SrvTransparencyAndCompositionMask = Shader.PropertyToID("r_transparency_and_composition_mask"); - internal static readonly int SrvReconstructedPrevNearestDepth = Shader.PropertyToID("r_reconstructed_previous_nearest_depth"); - internal static readonly int SrvDilatedMotionVectors = Shader.PropertyToID("r_dilated_motion_vectors"); - internal static readonly int SrvPrevDilatedMotionVectors = Shader.PropertyToID("r_previous_dilated_motion_vectors"); - internal static readonly int SrvDilatedDepth = Shader.PropertyToID("r_dilated_depth"); - internal static readonly int SrvInternalUpscaled = Shader.PropertyToID("r_internal_upscaled_color"); - internal static readonly int SrvLockStatus = Shader.PropertyToID("r_lock_status"); - internal static readonly int SrvLockInputLuma = Shader.PropertyToID("r_lock_input_luma"); - internal static readonly int SrvPreparedInputColor = Shader.PropertyToID("r_prepared_input_color"); - internal static readonly int SrvLumaHistory = Shader.PropertyToID("r_luma_history"); - internal static readonly int SrvRcasInput = Shader.PropertyToID("r_rcas_input"); - internal static readonly int SrvLanczosLut = Shader.PropertyToID("r_lanczos_lut"); - internal static readonly int SrvSceneLuminanceMips = Shader.PropertyToID("r_imgMips"); - internal static readonly int SrvUpscaleMaximumBiasLut = Shader.PropertyToID("r_upsample_maximum_bias_lut"); - internal static readonly int SrvDilatedReactiveMasks = Shader.PropertyToID("r_dilated_reactive_masks"); - internal static readonly int SrvPrevColorPreAlpha = Shader.PropertyToID("r_input_prev_color_pre_alpha"); - internal static readonly int SrvPrevColorPostAlpha = Shader.PropertyToID("r_input_prev_color_post_alpha"); - - // Unordered access views, i.e. random read/write bindings - internal static readonly int UavReconstructedPrevNearestDepth = Shader.PropertyToID("rw_reconstructed_previous_nearest_depth"); - internal static readonly int UavDilatedMotionVectors = Shader.PropertyToID("rw_dilated_motion_vectors"); - internal static readonly int UavDilatedDepth = Shader.PropertyToID("rw_dilated_depth"); - internal static readonly int UavInternalUpscaled = Shader.PropertyToID("rw_internal_upscaled_color"); - internal static readonly int UavLockStatus = Shader.PropertyToID("rw_lock_status"); - internal static readonly int UavLockInputLuma = Shader.PropertyToID("rw_lock_input_luma"); - internal static readonly int UavNewLocks = Shader.PropertyToID("rw_new_locks"); - internal static readonly int UavPreparedInputColor = Shader.PropertyToID("rw_prepared_input_color"); - internal static readonly int UavLumaHistory = Shader.PropertyToID("rw_luma_history"); - internal static readonly int UavUpscaledOutput = Shader.PropertyToID("rw_upscaled_output"); - internal static readonly int UavExposureMipLumaChange = Shader.PropertyToID("rw_img_mip_shading_change"); - internal static readonly int UavExposureMip5 = Shader.PropertyToID("rw_img_mip_5"); - internal static readonly int UavDilatedReactiveMasks = Shader.PropertyToID("rw_dilated_reactive_masks"); - internal static readonly int UavAutoExposure = Shader.PropertyToID("rw_auto_exposure"); - internal static readonly int UavSpdAtomicCount = Shader.PropertyToID("rw_spd_global_atomic"); - internal static readonly int UavAutoReactive = Shader.PropertyToID("rw_output_autoreactive"); - internal static readonly int UavAutoComposition = Shader.PropertyToID("rw_output_autocomposition"); - internal static readonly int UavPrevColorPreAlpha = Shader.PropertyToID("rw_output_prev_color_pre_alpha"); - internal static readonly int UavPrevColorPostAlpha = Shader.PropertyToID("rw_output_prev_color_post_alpha"); - - // Constant buffer bindings - internal static readonly int CbFsr3Upscaler = Shader.PropertyToID("cbFSR3Upscaler"); - internal static readonly int CbSpd = Shader.PropertyToID("cbSPD"); - internal static readonly int CbRcas = Shader.PropertyToID("cbRCAS"); - internal static readonly int CbGenReactive = Shader.PropertyToID("cbGenerateReactive"); - } -} diff --git a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3ShaderIDs.cs.meta b/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3ShaderIDs.cs.meta deleted file mode 100644 index 9012617..0000000 --- a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3ShaderIDs.cs.meta +++ /dev/null @@ -1,3 +0,0 @@ -fileFormatVersion: 2 -guid: a0e0bcc2967836742b7864d1cafafbf0 -timeCreated: 1679060863 \ No newline at end of file diff --git a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3Upscaler.cs b/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3Upscaler.cs deleted file mode 100644 index c636c93..0000000 --- a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3Upscaler.cs +++ /dev/null @@ -1,333 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -using System; -using System.Runtime.InteropServices; -using UnityEngine; -using UnityEngine.Rendering; - -namespace FidelityFX -{ - /// - /// A collection of helper functions and data structures required by the FSR3 Upscaler process. - /// - public static class Fsr3Upscaler - { - /// - /// Creates a new FSR3 Upscaler context with standard parameters that are appropriate for the current platform. - /// - public static Fsr3UpscalerContext CreateContext(Vector2Int displaySize, Vector2Int maxRenderSize, Fsr3UpscalerShaders shaders, InitializationFlags flags = 0) - { - if (SystemInfo.usesReversedZBuffer) - flags |= InitializationFlags.EnableDepthInverted; - else - flags &= ~InitializationFlags.EnableDepthInverted; - -#if UNITY_EDITOR || DEVELOPMENT_BUILD - flags |= InitializationFlags.EnableDebugChecking; -#endif - - Debug.Log($"Setting up FSR3 Upscaler with render size: {maxRenderSize.x}x{maxRenderSize.y}, display size: {displaySize.x}x{displaySize.y}, flags: {flags}"); - - var contextDescription = new ContextDescription - { - Flags = flags, - DisplaySize = displaySize, - MaxRenderSize = maxRenderSize, - Shaders = shaders, - }; - - var context = new Fsr3UpscalerContext(); - context.Create(contextDescription); - return context; - } - - public static float GetUpscaleRatioFromQualityMode(QualityMode qualityMode) - { - switch (qualityMode) - { - case QualityMode.NativeAA: - return 1.0f; - case QualityMode.UltraQuality: - return 1.2f; - case QualityMode.Quality: - return 1.5f; - case QualityMode.Balanced: - return 1.7f; - case QualityMode.Performance: - return 2.0f; - case QualityMode.UltraPerformance: - return 3.0f; - default: - return 1.0f; - } - } - - public static void GetRenderResolutionFromQualityMode( - out int renderWidth, out int renderHeight, - int displayWidth, int displayHeight, QualityMode qualityMode) - { - float ratio = GetUpscaleRatioFromQualityMode(qualityMode); - renderWidth = Mathf.RoundToInt(displayWidth / ratio); - renderHeight = Mathf.RoundToInt(displayHeight / ratio); - } - - public static float GetMipmapBiasOffset(int renderWidth, int displayWidth) - { - return Mathf.Log((float)renderWidth / displayWidth, 2.0f) - 1.0f; - } - - public static int GetJitterPhaseCount(int renderWidth, int displayWidth) - { - const float basePhaseCount = 8.0f; - int jitterPhaseCount = (int)(basePhaseCount * Mathf.Pow((float)displayWidth / renderWidth, 2.0f)); - return jitterPhaseCount; - } - - public static void GetJitterOffset(out float outX, out float outY, int index, int phaseCount) - { - outX = Halton((index % phaseCount) + 1, 2) - 0.5f; - outY = Halton((index % phaseCount) + 1, 3) - 0.5f; - } - - // Calculate halton number for index and base. - private static float Halton(int index, int @base) - { - float f = 1.0f, result = 0.0f; - - for (int currentIndex = index; currentIndex > 0;) { - - f /= @base; - result += f * (currentIndex % @base); - currentIndex = (int)Mathf.Floor((float)currentIndex / @base); - } - - return result; - } - - public static float Lanczos2(float value) - { - return Mathf.Abs(value) < Mathf.Epsilon ? 1.0f : Mathf.Sin(Mathf.PI * value) / (Mathf.PI * value) * (Mathf.Sin(0.5f * Mathf.PI * value) / (0.5f * Mathf.PI * value)); - } - -#if !UNITY_2021_1_OR_NEWER - internal static void SetBufferData(this CommandBuffer commandBuffer, ComputeBuffer computeBuffer, Array data) - { - commandBuffer.SetComputeBufferData(computeBuffer, data); - } -#endif - - public enum QualityMode - { - NativeAA = 0, - UltraQuality = 1, - Quality = 2, - Balanced = 3, - Performance = 4, - UltraPerformance = 5, - } - - [Flags] - public enum InitializationFlags - { - EnableHighDynamicRange = 1 << 0, - EnableDisplayResolutionMotionVectors = 1 << 1, - EnableMotionVectorsJitterCancellation = 1 << 2, - EnableDepthInverted = 1 << 3, - EnableDepthInfinite = 1 << 4, - EnableAutoExposure = 1 << 5, - EnableDynamicResolution = 1 << 6, - EnableFP16Usage = 1 << 7, - EnableDebugChecking = 1 << 8, - } - - /// - /// A structure encapsulating the parameters required to initialize FidelityFX Super Resolution 3 upscaling. - /// - public struct ContextDescription - { - public InitializationFlags Flags; - public Vector2Int MaxRenderSize; - public Vector2Int DisplaySize; - public Fsr3UpscalerShaders Shaders; - } - - /// - /// A structure encapsulating the parameters for dispatching the various passes of FidelityFX Super Resolution 3. - /// - public class DispatchDescription - { - public ResourceView Color; - public ResourceView Depth; - public ResourceView MotionVectors; - public ResourceView Exposure; // optional - public ResourceView Reactive; // optional - public ResourceView TransparencyAndComposition; // optional - public ResourceView Output; - public Vector2 JitterOffset; - public Vector2 MotionVectorScale; - public Vector2Int RenderSize; - public Vector2Int InputResourceSize; - public bool EnableSharpening; - public float Sharpness; - public float FrameTimeDelta; // in seconds - public float PreExposure; - public bool Reset; - public float CameraNear; - public float CameraFar; - public float CameraFovAngleVertical; - public float ViewSpaceToMetersFactor; - - // EXPERIMENTAL reactive mask generation parameters - public bool EnableAutoReactive; - public ResourceView ColorOpaqueOnly; - public float AutoTcThreshold = 0.05f; - public float AutoTcScale = 1.0f; - public float AutoReactiveScale = 5.0f; - public float AutoReactiveMax = 0.9f; - } - - /// - /// A structure encapsulating the parameters for automatic generation of a reactive mask. - /// The default values for Scale, CutoffThreshold, BinaryValue and Flags were taken from the FSR3 demo project. - /// - public class GenerateReactiveDescription - { - public ResourceView ColorOpaqueOnly; - public ResourceView ColorPreUpscale; - public ResourceView OutReactive; - public Vector2Int RenderSize; - public float Scale = 0.5f; - public float CutoffThreshold = 0.2f; - public float BinaryValue = 0.9f; - public GenerateReactiveFlags Flags = GenerateReactiveFlags.ApplyTonemap | GenerateReactiveFlags.ApplyThreshold | GenerateReactiveFlags.UseComponentsMax; - } - - [Flags] - public enum GenerateReactiveFlags - { - ApplyTonemap = 1 << 0, - ApplyInverseTonemap = 1 << 1, - ApplyThreshold = 1 << 2, - UseComponentsMax = 1 << 3, - } - - [Serializable, StructLayout(LayoutKind.Sequential)] - internal struct UpscalerConstants - { - public Vector2Int renderSize; - public Vector2Int maxRenderSize; - public Vector2Int displaySize; - public Vector2Int inputColorResourceDimensions; - public Vector2Int lumaMipDimensions; - public int lumaMipLevelToUse; - public int frameIndex; - - public Vector4 deviceToViewDepth; - public Vector2 jitterOffset; - public Vector2 motionVectorScale; - public Vector2 downscaleFactor; - public Vector2 motionVectorJitterCancellation; - public float preExposure; - public float previousFramePreExposure; - public float tanHalfFOV; - public float jitterPhaseCount; - public float deltaTime; - public float dynamicResChangeFactor; - public float viewSpaceToMetersFactor; - - public int dummy; - } - - [Serializable, StructLayout(LayoutKind.Sequential)] - internal struct SpdConstants - { - public uint mips; - public uint numWorkGroups; - public uint workGroupOffsetX, workGroupOffsetY; - public uint renderSizeX, renderSizeY; - } - - [Serializable, StructLayout(LayoutKind.Sequential)] - internal struct GenerateReactiveConstants - { - public float scale; - public float threshold; - public float binaryValue; - public uint flags; - } - - [Serializable, StructLayout(LayoutKind.Sequential)] - internal struct GenerateReactiveConstants2 - { - public float autoTcThreshold; - public float autoTcScale; - public float autoReactiveScale; - public float autoReactiveMax; - } - - [Serializable, StructLayout(LayoutKind.Sequential)] - internal struct RcasConstants - { - public RcasConstants(uint sharpness, uint halfSharp) - { - this.sharpness = sharpness; - this.halfSharp = halfSharp; - dummy0 = dummy1 = 0; - } - - public readonly uint sharpness; - public readonly uint halfSharp; - public readonly uint dummy0; - public readonly uint dummy1; - } - } - - /// - /// An immutable structure wrapping all of the necessary information to bind a specific buffer or attachment of a render target to a compute shader. - /// - public readonly struct ResourceView - { - /// - /// This value is the equivalent of not setting any value at all; all struct fields will have their default values. - /// It does not refer to a valid texture, therefore any variable set to this value should be checked for IsValid and reassigned before being bound to a shader. - /// - public static readonly ResourceView Unassigned = new ResourceView(default); - - /// - /// This value contains a valid texture reference that can be bound to a shader, however it is just an empty placeholder texture. - /// Binding this to a shader can be seen as setting the texture variable inside the shader to null. - /// - public static readonly ResourceView None = new ResourceView(BuiltinRenderTextureType.None); - - public ResourceView(in RenderTargetIdentifier renderTarget, RenderTextureSubElement subElement = RenderTextureSubElement.Default, int mipLevel = 0) - { - RenderTarget = renderTarget; - SubElement = subElement; - MipLevel = mipLevel; - } - - public bool IsValid => !RenderTarget.Equals(default); - - public readonly RenderTargetIdentifier RenderTarget; - public readonly RenderTextureSubElement SubElement; - public readonly int MipLevel; - } -} diff --git a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3Upscaler.cs.meta b/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3Upscaler.cs.meta deleted file mode 100644 index 6717df4..0000000 --- a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3Upscaler.cs.meta +++ /dev/null @@ -1,3 +0,0 @@ -fileFormatVersion: 2 -guid: df8b18c192f2dc145b4b43e68fd3407d -timeCreated: 1673441954 \ No newline at end of file diff --git a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerAssets.cs b/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerAssets.cs deleted file mode 100644 index 3e4e24f..0000000 --- a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerAssets.cs +++ /dev/null @@ -1,151 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -using UnityEngine; - -namespace FidelityFX -{ - /// - /// Scriptable object containing all shader resources required by FidelityFX Super Resolution 3 (FSR3) Upscaler. - /// These can be stored in an asset file and referenced from a scene or prefab, avoiding the need to load the shaders from a Resources folder. - /// - public class Fsr3UpscalerAssets : ScriptableObject - { - public Fsr3UpscalerShaders shaders; - -#if UNITY_EDITOR - private void Reset() - { - shaders = new Fsr3UpscalerShaders - { - computeLuminancePyramidPass = FindComputeShader("ffx_fsr3upscaler_compute_luminance_pyramid_pass"), - reconstructPreviousDepthPass = FindComputeShader("ffx_fsr3upscaler_reconstruct_previous_depth_pass"), - depthClipPass = FindComputeShader("ffx_fsr3upscaler_depth_clip_pass"), - lockPass = FindComputeShader("ffx_fsr3upscaler_lock_pass"), - accumulatePass = FindComputeShader("ffx_fsr3upscaler_accumulate_pass"), - sharpenPass = FindComputeShader("ffx_fsr3upscaler_rcas_pass"), - autoGenReactivePass = FindComputeShader("ffx_fsr3upscaler_autogen_reactive_pass"), - tcrAutoGenPass = FindComputeShader("ffx_fsr3upscaler_tcr_autogen_pass"), - }; - } - - private static ComputeShader FindComputeShader(string name) - { - string[] assetGuids = UnityEditor.AssetDatabase.FindAssets($"t:ComputeShader {name}"); - if (assetGuids == null || assetGuids.Length == 0) - return null; - - string assetPath = UnityEditor.AssetDatabase.GUIDToAssetPath(assetGuids[0]); - return UnityEditor.AssetDatabase.LoadAssetAtPath(assetPath); - } -#endif - } - - /// - /// All the compute shaders used by the FSR3 Upscaler. - /// - [System.Serializable] - public class Fsr3UpscalerShaders - { - /// - /// The compute shader used by the luminance pyramid computation pass. - /// - public ComputeShader computeLuminancePyramidPass; - - /// - /// The compute shader used by the previous depth reconstruction pass. - /// - public ComputeShader reconstructPreviousDepthPass; - - /// - /// The compute shader used by the depth clip pass. - /// - public ComputeShader depthClipPass; - - /// - /// The compute shader used by the lock pass. - /// - public ComputeShader lockPass; - - /// - /// The compute shader used by the accumulation pass. - /// - public ComputeShader accumulatePass; - - /// - /// The compute shader used by the RCAS sharpening pass. - /// - public ComputeShader sharpenPass; - - /// - /// The compute shader used to auto-generate a reactive mask. - /// - public ComputeShader autoGenReactivePass; - - /// - /// The compute shader used to auto-generate a transparency & composition mask. - /// - public ComputeShader tcrAutoGenPass; - - /// - /// Returns a copy of this class and its contents. - /// - public Fsr3UpscalerShaders Clone() - { - return (Fsr3UpscalerShaders)MemberwiseClone(); - } - - /// - /// Returns a copy of this class with clones of all its shaders. - /// This can be useful if you're running multiple FSR3 Upscaler instances with different shader configurations. - /// Be sure to clean up these clones through Dispose once you're done with them. - /// - public Fsr3UpscalerShaders DeepCopy() - { - return new Fsr3UpscalerShaders - { - computeLuminancePyramidPass = Object.Instantiate(computeLuminancePyramidPass), - reconstructPreviousDepthPass = Object.Instantiate(reconstructPreviousDepthPass), - depthClipPass = Object.Instantiate(depthClipPass), - lockPass = Object.Instantiate(lockPass), - accumulatePass = Object.Instantiate(accumulatePass), - sharpenPass = Object.Instantiate(sharpenPass), - autoGenReactivePass = Object.Instantiate(autoGenReactivePass), - tcrAutoGenPass = Object.Instantiate(tcrAutoGenPass), - }; - } - - /// - /// Destroy all the shaders within this instance. - /// Use this only on clones created through DeepCopy. - /// - public void Dispose() - { - Object.Destroy(computeLuminancePyramidPass); - Object.Destroy(reconstructPreviousDepthPass); - Object.Destroy(depthClipPass); - Object.Destroy(lockPass); - Object.Destroy(accumulatePass); - Object.Destroy(sharpenPass); - Object.Destroy(autoGenReactivePass); - Object.Destroy(tcrAutoGenPass); - } - } -} diff --git a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerAssets.cs.meta b/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerAssets.cs.meta deleted file mode 100644 index de48032..0000000 --- a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerAssets.cs.meta +++ /dev/null @@ -1,11 +0,0 @@ -fileFormatVersion: 2 -guid: aaeb3d821f826d44b84289a2dd23f90e -MonoImporter: - externalObjects: {} - serializedVersion: 2 - defaultReferences: [] - executionOrder: 0 - icon: {instanceID: 0} - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerCallbacks.cs b/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerCallbacks.cs deleted file mode 100644 index 5b2c89c..0000000 --- a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerCallbacks.cs +++ /dev/null @@ -1,81 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -using UnityEngine; - -namespace FidelityFX -{ - /// - /// A collection of callbacks required by the FSR3 Upscaler process. - /// This allows some customization by the game dev on how to integrate FSR3 upscaling into their own game setup. - /// - public interface IFsr3UpscalerCallbacks - { - /// - /// Apply a mipmap bias to in-game textures to prevent them from becoming blurry as the internal rendering resolution lowers. - /// This will need to be customized on a per-game basis, as there is no clear universal way to determine what are "in-game" textures. - /// The default implementation will simply apply a mipmap bias to all 2D textures, which will include things like UI textures and which might miss things like terrain texture arrays. - /// - /// Depending on how your game organizes its assets, you will want to create a filter that more specifically selects the textures that need to have this mipmap bias applied. - /// You may also want to store the bias offset value and apply it to any assets that are loaded in on demand. - /// - void ApplyMipmapBias(float biasOffset); - - void UndoMipmapBias(); - } - - /// - /// Default implementation of IFsr3UpscalerCallbacks using simple Resources calls. - /// These are fine for testing but a proper game will want to extend and override these methods. - /// - public class Fsr3UpscalerCallbacksBase: IFsr3UpscalerCallbacks - { - protected float CurrentBiasOffset = 0; - - public virtual void ApplyMipmapBias(float biasOffset) - { - if (float.IsNaN(biasOffset) || float.IsInfinity(biasOffset)) - return; - - CurrentBiasOffset += biasOffset; - - if (Mathf.Approximately(CurrentBiasOffset, 0f)) - { - CurrentBiasOffset = 0f; - } - - foreach (var texture in Resources.FindObjectsOfTypeAll()) - { - if (texture.mipmapCount <= 1) - continue; - - texture.mipMapBias += biasOffset; - } - } - - public virtual void UndoMipmapBias() - { - if (CurrentBiasOffset == 0f) - return; - - ApplyMipmapBias(-CurrentBiasOffset); - } - } -} diff --git a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerCallbacks.cs.meta b/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerCallbacks.cs.meta deleted file mode 100644 index ae2ee1a..0000000 --- a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerCallbacks.cs.meta +++ /dev/null @@ -1,11 +0,0 @@ -fileFormatVersion: 2 -guid: e1b555daa29ec3043a8cf89b4db31a26 -MonoImporter: - externalObjects: {} - serializedVersion: 2 - defaultReferences: [] - executionOrder: 0 - icon: {instanceID: 0} - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerContext.cs b/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerContext.cs deleted file mode 100644 index da02d9f..0000000 --- a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerContext.cs +++ /dev/null @@ -1,610 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -using System; -using System.Runtime.InteropServices; -using UnityEngine; -using UnityEngine.Rendering; - -namespace FidelityFX -{ - /// - /// This class loosely matches the FfxFsr3UpscalerContext struct from the original FSR3 codebase. - /// It manages the various resources and compute passes required by the FSR3 Upscaler process. - /// Note that this class does not know anything about Unity render pipelines; all it knows is CommandBuffers and RenderTargetIdentifiers. - /// This should make it suitable for integration with any of the available Unity render pipelines. - /// - public class Fsr3UpscalerContext - { - private const int MaxQueuedFrames = 16; - - private Fsr3Upscaler.ContextDescription _contextDescription; - private CommandBuffer _commandBuffer; - - private Fsr3UpscalerPass _depthClipPass; - private Fsr3UpscalerPass _reconstructPreviousDepthPass; - private Fsr3UpscalerPass _lockPass; - private Fsr3UpscalerPass _accumulatePass; - private Fsr3UpscalerPass _sharpenPass; - private Fsr3UpscalerPass _computeLuminancePyramidPass; - private Fsr3UpscalerPass _generateReactivePass; - private Fsr3UpscalerPass _tcrAutogeneratePass; - - private readonly Fsr3UpscalerResources _resources = new Fsr3UpscalerResources(); - - private ComputeBuffer _upscalerConstantsBuffer; - private readonly Fsr3Upscaler.UpscalerConstants[] _upscalerConstantsArray = { new Fsr3Upscaler.UpscalerConstants() }; - private ref Fsr3Upscaler.UpscalerConstants UpscalerConsts => ref _upscalerConstantsArray[0]; - - private ComputeBuffer _spdConstantsBuffer; - private readonly Fsr3Upscaler.SpdConstants[] _spdConstantsArray = { new Fsr3Upscaler.SpdConstants() }; - private ref Fsr3Upscaler.SpdConstants SpdConsts => ref _spdConstantsArray[0]; - - private ComputeBuffer _rcasConstantsBuffer; - private readonly Fsr3Upscaler.RcasConstants[] _rcasConstantsArray = new Fsr3Upscaler.RcasConstants[1]; - private ref Fsr3Upscaler.RcasConstants RcasConsts => ref _rcasConstantsArray[0]; - - private ComputeBuffer _generateReactiveConstantsBuffer; - private readonly Fsr3Upscaler.GenerateReactiveConstants[] _generateReactiveConstantsArray = { new Fsr3Upscaler.GenerateReactiveConstants() }; - private ref Fsr3Upscaler.GenerateReactiveConstants GenReactiveConsts => ref _generateReactiveConstantsArray[0]; - - private ComputeBuffer _tcrAutogenerateConstantsBuffer; - private readonly Fsr3Upscaler.GenerateReactiveConstants2[] _tcrAutogenerateConstantsArray = { new Fsr3Upscaler.GenerateReactiveConstants2() }; - private ref Fsr3Upscaler.GenerateReactiveConstants2 TcrAutoGenConsts => ref _tcrAutogenerateConstantsArray[0]; - - private bool _firstExecution; - private Vector2 _previousJitterOffset; - private int _resourceFrameIndex; - - public void Create(Fsr3Upscaler.ContextDescription contextDescription) - { - _contextDescription = contextDescription; - _commandBuffer = new CommandBuffer { name = "FSR3 Upscaler" }; - - _upscalerConstantsBuffer = CreateConstantBuffer(); - _spdConstantsBuffer = CreateConstantBuffer(); - _rcasConstantsBuffer = CreateConstantBuffer(); - _generateReactiveConstantsBuffer = CreateConstantBuffer(); - _tcrAutogenerateConstantsBuffer = CreateConstantBuffer(); - - // Set defaults - _firstExecution = true; - _resourceFrameIndex = 0; - - UpscalerConsts.displaySize = _contextDescription.DisplaySize; - - _resources.Create(_contextDescription); - CreatePasses(); - } - - private void CreatePasses() - { - _computeLuminancePyramidPass = new Fsr3UpscalerComputeLuminancePyramidPass(_contextDescription, _resources, _upscalerConstantsBuffer, _spdConstantsBuffer); - _reconstructPreviousDepthPass = new Fsr3UpscalerReconstructPreviousDepthPass(_contextDescription, _resources, _upscalerConstantsBuffer); - _depthClipPass = new Fsr3UpscalerDepthClipPass(_contextDescription, _resources, _upscalerConstantsBuffer); - _lockPass = new Fsr3UpscalerLockPass(_contextDescription, _resources, _upscalerConstantsBuffer); - _accumulatePass = new Fsr3UpscalerAccumulatePass(_contextDescription, _resources, _upscalerConstantsBuffer); - _sharpenPass = new Fsr3UpscalerSharpenPass(_contextDescription, _resources, _upscalerConstantsBuffer, _rcasConstantsBuffer); - _generateReactivePass = new Fsr3UpscalerGenerateReactivePass(_contextDescription, _resources, _generateReactiveConstantsBuffer); - _tcrAutogeneratePass = new Fsr3UpscalerTcrAutogeneratePass(_contextDescription, _resources, _upscalerConstantsBuffer, _tcrAutogenerateConstantsBuffer); - } - - public void Destroy() - { - DestroyPass(ref _tcrAutogeneratePass); - DestroyPass(ref _generateReactivePass); - DestroyPass(ref _computeLuminancePyramidPass); - DestroyPass(ref _sharpenPass); - DestroyPass(ref _accumulatePass); - DestroyPass(ref _lockPass); - DestroyPass(ref _reconstructPreviousDepthPass); - DestroyPass(ref _depthClipPass); - - _resources.Destroy(); - - DestroyConstantBuffer(ref _tcrAutogenerateConstantsBuffer); - DestroyConstantBuffer(ref _generateReactiveConstantsBuffer); - DestroyConstantBuffer(ref _rcasConstantsBuffer); - DestroyConstantBuffer(ref _spdConstantsBuffer); - DestroyConstantBuffer(ref _upscalerConstantsBuffer); - - _commandBuffer.Dispose(); - _commandBuffer = null; - } - - public void Dispatch(Fsr3Upscaler.DispatchDescription dispatchParams) - { - _commandBuffer.Clear(); - Dispatch(dispatchParams, _commandBuffer); - Graphics.ExecuteCommandBuffer(_commandBuffer); - } - - public void Dispatch(Fsr3Upscaler.DispatchDescription dispatchParams, CommandBuffer commandBuffer) - { - if ((_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDebugChecking) != 0) - { - DebugCheckDispatch(dispatchParams); - } - - if (_firstExecution) - { - commandBuffer.SetRenderTarget(_resources.LockStatus[0]); - commandBuffer.ClearRenderTarget(false, true, Color.clear); - commandBuffer.SetRenderTarget(_resources.LockStatus[1]); - commandBuffer.ClearRenderTarget(false, true, Color.clear); - } - - int frameIndex = _resourceFrameIndex % 2; - bool resetAccumulation = dispatchParams.Reset || _firstExecution; - _firstExecution = false; - - // If auto exposure is enabled use the auto exposure SRV, otherwise what the app sends - if ((_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableAutoExposure) != 0) - dispatchParams.Exposure = new ResourceView(_resources.AutoExposure); - else if (!dispatchParams.Exposure.IsValid) - dispatchParams.Exposure = new ResourceView(_resources.DefaultExposure); - - if (dispatchParams.EnableAutoReactive) - { - // Create the auto-TCR resources only when we need them - if (_resources.AutoReactive == null) - _resources.CreateTcrAutogenResources(_contextDescription); - - if (resetAccumulation) - { - RenderTargetIdentifier opaqueOnly = dispatchParams.ColorOpaqueOnly.IsValid ? dispatchParams.ColorOpaqueOnly.RenderTarget : Fsr3ShaderIDs.SrvOpaqueOnly; - commandBuffer.Blit(_resources.PrevPreAlpha[frameIndex ^ 1], opaqueOnly); - } - } - else if (_resources.AutoReactive != null) - { - // Destroy the auto-TCR resources if we don't use the feature - _resources.DestroyTcrAutogenResources(); - } - - if (!dispatchParams.Reactive.IsValid) dispatchParams.Reactive = new ResourceView(_resources.DefaultReactive); - if (!dispatchParams.TransparencyAndComposition.IsValid) dispatchParams.TransparencyAndComposition = new ResourceView(_resources.DefaultReactive); - Fsr3UpscalerResources.CreateAliasableResources(commandBuffer, _contextDescription, dispatchParams); - - SetupConstants(dispatchParams, resetAccumulation); - - // Reactive mask bias - const int threadGroupWorkRegionDim = 8; - int dispatchSrcX = (UpscalerConsts.renderSize.x + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; - int dispatchSrcY = (UpscalerConsts.renderSize.y + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; - int dispatchDstX = (_contextDescription.DisplaySize.x + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; - int dispatchDstY = (_contextDescription.DisplaySize.y + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; - - // Clear reconstructed depth for max depth store - if (resetAccumulation) - { - commandBuffer.SetRenderTarget(_resources.LockStatus[frameIndex ^ 1]); - commandBuffer.ClearRenderTarget(false, true, Color.clear); - - commandBuffer.SetRenderTarget(_resources.InternalUpscaled[frameIndex ^ 1]); - commandBuffer.ClearRenderTarget(false, true, Color.clear); - - commandBuffer.SetRenderTarget(_resources.SceneLuminance); - commandBuffer.ClearRenderTarget(false, true, Color.clear); - - // Auto exposure always used to track luma changes in locking logic - commandBuffer.SetRenderTarget(_resources.AutoExposure); - commandBuffer.ClearRenderTarget(false, true, new Color(0f, 1e8f, 0f, 0f)); - - // Reset atomic counter to 0 - commandBuffer.SetRenderTarget(_resources.SpdAtomicCounter); - commandBuffer.ClearRenderTarget(false, true, Color.clear); - } - - // FSR3: need to clear here since we need the content of this surface for frame interpolation, so clearing in the lock pass is not an option - bool depthInverted = (_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDepthInverted) == Fsr3Upscaler.InitializationFlags.EnableDepthInverted; - commandBuffer.SetRenderTarget(Fsr3ShaderIDs.UavReconstructedPrevNearestDepth); - commandBuffer.ClearRenderTarget(false, true, depthInverted ? Color.clear : Color.white); - - // Auto exposure - SetupSpdConstants(dispatchParams, out var dispatchThreadGroupCount); - - // Initialize constant buffers data - commandBuffer.SetBufferData(_upscalerConstantsBuffer, _upscalerConstantsArray); - commandBuffer.SetBufferData(_spdConstantsBuffer, _spdConstantsArray); - - // Auto reactive - if (dispatchParams.EnableAutoReactive) - { - GenerateTransparencyCompositionReactive(dispatchParams, commandBuffer, frameIndex); - dispatchParams.Reactive = new ResourceView(_resources.AutoReactive); - dispatchParams.TransparencyAndComposition = new ResourceView(_resources.AutoComposition); - } - - // Compute luminance pyramid - _computeLuminancePyramidPass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchThreadGroupCount.x, dispatchThreadGroupCount.y); - - // Reconstruct previous depth - _reconstructPreviousDepthPass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchSrcX, dispatchSrcY); - - // Depth clip - _depthClipPass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchSrcX, dispatchSrcY); - - // Create locks - _lockPass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchSrcX, dispatchSrcY); - - // Accumulate - _accumulatePass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchDstX, dispatchDstY); - - if (dispatchParams.EnableSharpening) - { - // Compute the constants - SetupRcasConstants(dispatchParams); - commandBuffer.SetBufferData(_rcasConstantsBuffer, _rcasConstantsArray); - - // Dispatch RCAS - const int threadGroupWorkRegionDimRcas = 16; - int threadGroupsX = (Screen.width + threadGroupWorkRegionDimRcas - 1) / threadGroupWorkRegionDimRcas; - int threadGroupsY = (Screen.height + threadGroupWorkRegionDimRcas - 1) / threadGroupWorkRegionDimRcas; - _sharpenPass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, threadGroupsX, threadGroupsY); - } - - _resourceFrameIndex = (_resourceFrameIndex + 1) % MaxQueuedFrames; - - Fsr3UpscalerResources.DestroyAliasableResources(commandBuffer); - } - - public void GenerateReactiveMask(Fsr3Upscaler.GenerateReactiveDescription dispatchParams) - { - _commandBuffer.Clear(); - GenerateReactiveMask(dispatchParams, _commandBuffer); - Graphics.ExecuteCommandBuffer(_commandBuffer); - } - - public void GenerateReactiveMask(Fsr3Upscaler.GenerateReactiveDescription dispatchParams, CommandBuffer commandBuffer) - { - const int threadGroupWorkRegionDim = 8; - int dispatchSrcX = (dispatchParams.RenderSize.x + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; - int dispatchSrcY = (dispatchParams.RenderSize.y + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; - - GenReactiveConsts.scale = dispatchParams.Scale; - GenReactiveConsts.threshold = dispatchParams.CutoffThreshold; - GenReactiveConsts.binaryValue = dispatchParams.BinaryValue; - GenReactiveConsts.flags = (uint)dispatchParams.Flags; - commandBuffer.SetBufferData(_generateReactiveConstantsBuffer, _generateReactiveConstantsArray); - - ((Fsr3UpscalerGenerateReactivePass)_generateReactivePass).ScheduleDispatch(commandBuffer, dispatchParams, dispatchSrcX, dispatchSrcY); - } - - private void GenerateTransparencyCompositionReactive(Fsr3Upscaler.DispatchDescription dispatchParams, CommandBuffer commandBuffer, int frameIndex) - { - const int threadGroupWorkRegionDim = 8; - int dispatchSrcX = (dispatchParams.RenderSize.x + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; - int dispatchSrcY = (dispatchParams.RenderSize.y + (threadGroupWorkRegionDim - 1)) / threadGroupWorkRegionDim; - - TcrAutoGenConsts.autoTcThreshold = dispatchParams.AutoTcThreshold; - TcrAutoGenConsts.autoTcScale = dispatchParams.AutoTcScale; - TcrAutoGenConsts.autoReactiveScale = dispatchParams.AutoReactiveScale; - TcrAutoGenConsts.autoReactiveMax = dispatchParams.AutoReactiveMax; - commandBuffer.SetBufferData(_tcrAutogenerateConstantsBuffer, _tcrAutogenerateConstantsArray); - - _tcrAutogeneratePass.ScheduleDispatch(commandBuffer, dispatchParams, frameIndex, dispatchSrcX, dispatchSrcY); - } - - private void SetupConstants(Fsr3Upscaler.DispatchDescription dispatchParams, bool resetAccumulation) - { - ref Fsr3Upscaler.UpscalerConstants constants = ref UpscalerConsts; - - constants.jitterOffset = dispatchParams.JitterOffset; - constants.renderSize = dispatchParams.RenderSize; - constants.maxRenderSize = _contextDescription.MaxRenderSize; - constants.inputColorResourceDimensions = dispatchParams.InputResourceSize; - - // Compute the horizontal FOV for the shader from the vertical one - float aspectRatio = (float)dispatchParams.RenderSize.x / dispatchParams.RenderSize.y; - float cameraAngleHorizontal = Mathf.Atan(Mathf.Tan(dispatchParams.CameraFovAngleVertical / 2.0f) * aspectRatio) * 2.0f; - constants.tanHalfFOV = Mathf.Tan(cameraAngleHorizontal * 0.5f); - constants.viewSpaceToMetersFactor = (dispatchParams.ViewSpaceToMetersFactor > 0.0f) ? dispatchParams.ViewSpaceToMetersFactor : 1.0f; - - // Compute params to enable device depth to view space depth computation in shader - constants.deviceToViewDepth = SetupDeviceDepthToViewSpaceDepthParams(dispatchParams); - - // To be updated if resource is larger than the actual image size - constants.downscaleFactor = new Vector2((float)constants.renderSize.x / _contextDescription.DisplaySize.x, (float)constants.renderSize.y / _contextDescription.DisplaySize.y); - constants.previousFramePreExposure = constants.preExposure; - constants.preExposure = (dispatchParams.PreExposure != 0) ? dispatchParams.PreExposure : 1.0f; - - // Motion vector data - Vector2Int motionVectorsTargetSize = (_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDisplayResolutionMotionVectors) != 0 ? constants.displaySize : constants.renderSize; - constants.motionVectorScale = dispatchParams.MotionVectorScale / motionVectorsTargetSize; - - // Compute jitter cancellation - if ((_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableMotionVectorsJitterCancellation) != 0) - { - constants.motionVectorJitterCancellation = (_previousJitterOffset - constants.jitterOffset) / motionVectorsTargetSize; - _previousJitterOffset = constants.jitterOffset; - } - - int jitterPhaseCount = Fsr3Upscaler.GetJitterPhaseCount(dispatchParams.RenderSize.x, _contextDescription.DisplaySize.x); - if (resetAccumulation || constants.jitterPhaseCount == 0) - { - constants.jitterPhaseCount = jitterPhaseCount; - } - else - { - int jitterPhaseCountDelta = (int)(jitterPhaseCount - constants.jitterPhaseCount); - if (jitterPhaseCountDelta > 0) - constants.jitterPhaseCount++; - else if (jitterPhaseCountDelta < 0) - constants.jitterPhaseCount--; - } - - // Convert delta time to seconds and clamp to [0, 1] - constants.deltaTime = Mathf.Clamp01(dispatchParams.FrameTimeDelta); - - if (resetAccumulation) - constants.frameIndex = 0; - else - constants.frameIndex++; - - // Shading change usage of the SPD mip levels - constants.lumaMipLevelToUse = Fsr3UpscalerPass.ShadingChangeMipLevel; - - float mipDiv = 2 << constants.lumaMipLevelToUse; - constants.lumaMipDimensions.x = (int)(constants.maxRenderSize.x / mipDiv); - constants.lumaMipDimensions.y = (int)(constants.maxRenderSize.y / mipDiv); - } - - private Vector4 SetupDeviceDepthToViewSpaceDepthParams(Fsr3Upscaler.DispatchDescription dispatchParams) - { - bool inverted = (_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDepthInverted) != 0; - bool infinite = (_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDepthInfinite) != 0; - - // make sure it has no impact if near and far plane values are swapped in dispatch params - // the flags "inverted" and "infinite" will decide what transform to use - float min = Mathf.Min(dispatchParams.CameraNear, dispatchParams.CameraFar); - float max = Mathf.Max(dispatchParams.CameraNear, dispatchParams.CameraFar); - - if (inverted) - { - (min, max) = (max, min); - } - - float q = max / (min - max); - float d = -1.0f; - - Vector4 matrixElemC = new Vector4(q, -1.0f - Mathf.Epsilon, q, 0.0f + Mathf.Epsilon); - Vector4 matrixElemE = new Vector4(q * min, -min - Mathf.Epsilon, q * min, max); - - // Revert x and y coords - float aspect = (float)dispatchParams.RenderSize.x / dispatchParams.RenderSize.y; - float cotHalfFovY = Mathf.Cos(0.5f * dispatchParams.CameraFovAngleVertical) / Mathf.Sin(0.5f * dispatchParams.CameraFovAngleVertical); - - int matrixIndex = (inverted ? 2 : 0) + (infinite ? 1 : 0); - return new Vector4( - d * matrixElemC[matrixIndex], - matrixElemE[matrixIndex], - aspect / cotHalfFovY, - 1.0f / cotHalfFovY); - } - - private void SetupRcasConstants(Fsr3Upscaler.DispatchDescription dispatchParams) - { - int sharpnessIndex = Mathf.RoundToInt(Mathf.Clamp01(dispatchParams.Sharpness) * (RcasConfigs.Length - 1)); - RcasConsts = RcasConfigs[sharpnessIndex]; - } - - private void SetupSpdConstants(Fsr3Upscaler.DispatchDescription dispatchParams, out Vector2Int dispatchThreadGroupCount) - { - RectInt rectInfo = new RectInt(0, 0, dispatchParams.RenderSize.x, dispatchParams.RenderSize.y); - SpdSetup(rectInfo, out dispatchThreadGroupCount, out var workGroupOffset, out var numWorkGroupsAndMips); - - // Downsample - ref Fsr3Upscaler.SpdConstants spdConstants = ref SpdConsts; - spdConstants.numWorkGroups = (uint)numWorkGroupsAndMips.x; - spdConstants.mips = (uint)numWorkGroupsAndMips.y; - spdConstants.workGroupOffsetX = (uint)workGroupOffset.x; - spdConstants.workGroupOffsetY = (uint)workGroupOffset.y; - spdConstants.renderSizeX = (uint)dispatchParams.RenderSize.x; - spdConstants.renderSizeY = (uint)dispatchParams.RenderSize.y; - } - - private static void SpdSetup(RectInt rectInfo, out Vector2Int dispatchThreadGroupCount, out Vector2Int workGroupOffset, out Vector2Int numWorkGroupsAndMips, int mips = -1) - { - workGroupOffset = new Vector2Int(rectInfo.x / 64, rectInfo.y / 64); - - int endIndexX = (rectInfo.x + rectInfo.width - 1) / 64; - int endIndexY = (rectInfo.y + rectInfo.height - 1) / 64; - - dispatchThreadGroupCount = new Vector2Int(endIndexX + 1 - workGroupOffset.x, endIndexY + 1 - workGroupOffset.y); - - numWorkGroupsAndMips = new Vector2Int(dispatchThreadGroupCount.x * dispatchThreadGroupCount.y, mips); - if (mips < 0) - { - float resolution = Math.Max(rectInfo.width, rectInfo.height); - numWorkGroupsAndMips.y = Math.Min(Mathf.FloorToInt(Mathf.Log(resolution, 2.0f)), 12); - } - } - - private void DebugCheckDispatch(Fsr3Upscaler.DispatchDescription dispatchParams) - { - if (!dispatchParams.Color.IsValid) - { - Debug.LogError("Color resource is null"); - } - - if (!dispatchParams.Depth.IsValid) - { - Debug.LogError("Depth resource is null"); - } - - if (!dispatchParams.MotionVectors.IsValid) - { - Debug.LogError("MotionVectors resource is null"); - } - - if (!dispatchParams.Output.IsValid) - { - Debug.LogError("Output resource is null"); - } - - if (dispatchParams.Exposure.IsValid && (_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableAutoExposure) != 0) - { - Debug.LogWarning("Exposure resource provided, however auto exposure flag is present"); - } - - if (Mathf.Abs(dispatchParams.JitterOffset.x) > 1.0f || Mathf.Abs(dispatchParams.JitterOffset.y) > 1.0f) - { - Debug.LogWarning("JitterOffset contains value outside of expected range [-1.0, 1.0]"); - } - - if (dispatchParams.MotionVectorScale.x > _contextDescription.MaxRenderSize.x || dispatchParams.MotionVectorScale.y > _contextDescription.MaxRenderSize.y) - { - Debug.LogWarning("MotionVectorScale contains scale value greater than MaxRenderSize"); - } - - if (dispatchParams.MotionVectorScale.x == 0.0f || dispatchParams.MotionVectorScale.y == 0.0f) - { - Debug.LogWarning("MotionVectorScale contains zero scale value"); - } - - if (dispatchParams.RenderSize.x > _contextDescription.MaxRenderSize.x || dispatchParams.RenderSize.y > _contextDescription.MaxRenderSize.y) - { - Debug.LogWarning("RenderSize is greater than context MaxRenderSize"); - } - - if (dispatchParams.RenderSize.x == 0 || dispatchParams.RenderSize.y == 0) - { - Debug.LogWarning("RenderSize contains zero dimension"); - } - - if (dispatchParams.FrameTimeDelta > 1.0f) - { - Debug.LogWarning("FrameTimeDelta is greater than 1.0f - this value should be seconds (~0.0166 for 60fps)"); - } - - if (dispatchParams.PreExposure == 0.0f) - { - Debug.LogError("PreExposure provided as 0.0f which is invalid"); - } - - bool infiniteDepth = (_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDepthInfinite) != 0; - bool inverseDepth = (_contextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDepthInverted) != 0; - - if (inverseDepth) - { - if (dispatchParams.CameraNear < dispatchParams.CameraFar) - { - Debug.LogWarning("EnableDepthInverted flag is present yet CameraNear is less than CameraFar"); - } - - if (infiniteDepth) - { - if (dispatchParams.CameraNear < float.MaxValue) - { - Debug.LogWarning("EnableDepthInfinite and EnableDepthInverted present, yet CameraNear != float.MaxValue"); - } - } - - if (dispatchParams.CameraFar < 0.075f) - { - Debug.LogWarning("EnableDepthInverted present, CameraFar value is very low which may result in depth separation artefacting"); - } - } - else - { - if (dispatchParams.CameraNear > dispatchParams.CameraFar) - { - Debug.LogWarning("CameraNear is greater than CameraFar in non-inverted-depth context"); - } - - if (infiniteDepth) - { - if (dispatchParams.CameraFar < float.MaxValue) - { - Debug.LogWarning("EnableDepthInfinite present, yet CameraFar != float.MaxValue"); - } - } - - if (dispatchParams.CameraNear < 0.075f) - { - Debug.LogWarning("CameraNear value is very low which may result in depth separation artefacting"); - } - } - - if (dispatchParams.CameraFovAngleVertical <= 0.0f) - { - Debug.LogError("CameraFovAngleVertical is 0.0f - this value should be > 0.0f"); - } - - if (dispatchParams.CameraFovAngleVertical > Mathf.PI) - { - Debug.LogError("CameraFovAngleVertical is greater than 180 degrees/PI"); - } - } - - /// - /// The FSR3 C++ codebase uses floats bitwise converted to ints to pass sharpness parameters to the RCAS shader. - /// This is not possible in C# without enabling unsafe code compilation, so to avoid that we instead use a table of precomputed values. - /// - private static readonly Fsr3Upscaler.RcasConstants[] RcasConfigs = new [] - { - new Fsr3Upscaler.RcasConstants(1048576000u, 872428544u), - new Fsr3Upscaler.RcasConstants(1049178080u, 877212745u), - new Fsr3Upscaler.RcasConstants(1049823372u, 882390168u), - new Fsr3Upscaler.RcasConstants(1050514979u, 887895276u), - new Fsr3Upscaler.RcasConstants(1051256227u, 893859143u), - new Fsr3Upscaler.RcasConstants(1052050675u, 900216232u), - new Fsr3Upscaler.RcasConstants(1052902144u, 907032080u), - new Fsr3Upscaler.RcasConstants(1053814727u, 914306687u), - new Fsr3Upscaler.RcasConstants(1054792807u, 922105590u), - new Fsr3Upscaler.RcasConstants(1055841087u, 930494326u), - new Fsr3Upscaler.RcasConstants(1056964608u, 939538432u), - new Fsr3Upscaler.RcasConstants(1057566688u, 944322633u), - new Fsr3Upscaler.RcasConstants(1058211980u, 949500056u), - new Fsr3Upscaler.RcasConstants(1058903587u, 955005164u), - new Fsr3Upscaler.RcasConstants(1059644835u, 960969031u), - new Fsr3Upscaler.RcasConstants(1060439283u, 967326120u), - new Fsr3Upscaler.RcasConstants(1061290752u, 974141968u), - new Fsr3Upscaler.RcasConstants(1062203335u, 981416575u), - new Fsr3Upscaler.RcasConstants(1063181415u, 989215478u), - new Fsr3Upscaler.RcasConstants(1064229695u, 997604214u), - new Fsr3Upscaler.RcasConstants(1065353216u, 1006648320), - }; - - private static ComputeBuffer CreateConstantBuffer() where TConstants: struct - { - return new ComputeBuffer(1, Marshal.SizeOf(), ComputeBufferType.Constant); - } - - private static void DestroyConstantBuffer(ref ComputeBuffer bufferRef) - { - if (bufferRef == null) - return; - - bufferRef.Release(); - bufferRef = null; - } - - private static void DestroyPass(ref Fsr3UpscalerPass pass) - { - if (pass == null) - return; - - pass.Dispose(); - pass = null; - } - } -} diff --git a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerContext.cs.meta b/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerContext.cs.meta deleted file mode 100644 index 673b2ef..0000000 --- a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerContext.cs.meta +++ /dev/null @@ -1,3 +0,0 @@ -fileFormatVersion: 2 -guid: 14c8dc4c7c3e4ac418e50a859cec0b2f -timeCreated: 1673442225 \ No newline at end of file diff --git a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerPass.cs b/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerPass.cs deleted file mode 100644 index e1dc225..0000000 --- a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerPass.cs +++ /dev/null @@ -1,372 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -using System; -using System.Runtime.InteropServices; -using UnityEngine; -using UnityEngine.Rendering; - -namespace FidelityFX -{ - /// - /// Base class for all of the compute passes that make up the FSR3 Upscaler process. - /// This loosely matches the FfxPipelineState struct from the original FSR3 codebase, wrapped in an object-oriented blanket. - /// These classes are responsible for loading compute shaders, managing temporary resources, binding resources to shader kernels and dispatching said shaders. - /// - internal abstract class Fsr3UpscalerPass: IDisposable - { - internal const int ShadingChangeMipLevel = 4; // This matches the FFX_FSR3UPSCALER_SHADING_CHANGE_MIP_LEVEL define - - protected readonly Fsr3Upscaler.ContextDescription ContextDescription; - protected readonly Fsr3UpscalerResources Resources; - protected readonly ComputeBuffer Constants; - - protected ComputeShader ComputeShader; - protected int KernelIndex; - - protected Fsr3UpscalerPass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants) - { - ContextDescription = contextDescription; - Resources = resources; - Constants = constants; - } - - public virtual void Dispose() - { - } - - public abstract void ScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY); - - protected void InitComputeShader(string passName, ComputeShader shader) - { - InitComputeShader(passName, shader, ContextDescription.Flags); - } - - private void InitComputeShader(string passName, ComputeShader shader, Fsr3Upscaler.InitializationFlags flags) - { - if (shader == null) - { - throw new MissingReferenceException($"Shader for FSR3 Upscaler '{passName}' could not be loaded! Please ensure it is included in the project correctly."); - } - - ComputeShader = shader; - KernelIndex = ComputeShader.FindKernel("CS"); - - bool useLut = false; -#if UNITY_2022_1_OR_NEWER // This will also work in 2020.3.43+ and 2021.3.14+ - if (SystemInfo.computeSubGroupSize == 64) - { - useLut = true; - } -#endif - - // This matches the permutation rules from the CreatePipeline* functions - if ((flags & Fsr3Upscaler.InitializationFlags.EnableHighDynamicRange) != 0) ComputeShader.EnableKeyword("FFX_FSR3UPSCALER_OPTION_HDR_COLOR_INPUT"); - if ((flags & Fsr3Upscaler.InitializationFlags.EnableDisplayResolutionMotionVectors) == 0) ComputeShader.EnableKeyword("FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS"); - if ((flags & Fsr3Upscaler.InitializationFlags.EnableMotionVectorsJitterCancellation) != 0) ComputeShader.EnableKeyword("FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS"); - if ((flags & Fsr3Upscaler.InitializationFlags.EnableDepthInverted) != 0) ComputeShader.EnableKeyword("FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH"); - if (useLut) ComputeShader.EnableKeyword("FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE"); - if ((flags & Fsr3Upscaler.InitializationFlags.EnableFP16Usage) != 0) ComputeShader.EnableKeyword("FFX_HALF"); - - // Inform the shader which render pipeline we're currently using - var pipeline = GraphicsSettings.currentRenderPipeline; - if (pipeline != null && pipeline.GetType().Name.Contains("HDRenderPipeline")) - { - ComputeShader.EnableKeyword("UNITY_FSR3UPSCALER_HDRP"); - } - } - } - - internal class Fsr3UpscalerComputeLuminancePyramidPass : Fsr3UpscalerPass - { - private readonly ComputeBuffer _spdConstants; - - public Fsr3UpscalerComputeLuminancePyramidPass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants, ComputeBuffer spdConstants) - : base(contextDescription, resources, constants) - { - _spdConstants = spdConstants; - - InitComputeShader("compute_luminance_pyramid_pass", contextDescription.Shaders.computeLuminancePyramidPass); - } - - public override void ScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) - { - ref var color = ref dispatchParams.Color; - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputColor, color.RenderTarget, color.MipLevel, color.SubElement); - - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavSpdAtomicCount, Resources.SpdAtomicCounter); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavExposureMipLumaChange, Resources.SceneLuminance, ShadingChangeMipLevel); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavExposureMip5, Resources.SceneLuminance, 5); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavAutoExposure, Resources.AutoExposure); - - commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); - commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbSpd, _spdConstants, 0, Marshal.SizeOf()); - - commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); - } - } - - internal class Fsr3UpscalerReconstructPreviousDepthPass : Fsr3UpscalerPass - { - public Fsr3UpscalerReconstructPreviousDepthPass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants) - : base(contextDescription, resources, constants) - { - InitComputeShader("reconstruct_previous_depth_pass", contextDescription.Shaders.reconstructPreviousDepthPass); - } - - public override void ScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) - { - ref var color = ref dispatchParams.Color; - ref var depth = ref dispatchParams.Depth; - ref var motionVectors = ref dispatchParams.MotionVectors; - ref var exposure = ref dispatchParams.Exposure; - - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputColor, color.RenderTarget, color.MipLevel, color.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputDepth, depth.RenderTarget, depth.MipLevel, depth.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputMotionVectors, motionVectors.RenderTarget, motionVectors.MipLevel, motionVectors.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputExposure, exposure.RenderTarget, exposure.MipLevel, exposure.SubElement); - - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavDilatedMotionVectors, Resources.DilatedMotionVectors[frameIndex]); - - commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); - - commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); - } - } - - internal class Fsr3UpscalerDepthClipPass : Fsr3UpscalerPass - { - public Fsr3UpscalerDepthClipPass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants) - : base(contextDescription, resources, constants) - { - InitComputeShader("depth_clip_pass", contextDescription.Shaders.depthClipPass); - } - - public override void ScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) - { - ref var color = ref dispatchParams.Color; - ref var depth = ref dispatchParams.Depth; - ref var motionVectors = ref dispatchParams.MotionVectors; - ref var exposure = ref dispatchParams.Exposure; - ref var reactive = ref dispatchParams.Reactive; - ref var tac = ref dispatchParams.TransparencyAndComposition; - - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputColor, color.RenderTarget, color.MipLevel, color.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputDepth, depth.RenderTarget, depth.MipLevel, depth.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputMotionVectors, motionVectors.RenderTarget, motionVectors.MipLevel, motionVectors.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputExposure, exposure.RenderTarget, exposure.MipLevel, exposure.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvReactiveMask, reactive.RenderTarget, reactive.MipLevel, reactive.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvTransparencyAndCompositionMask, tac.RenderTarget, tac.MipLevel, tac.SubElement); - - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvReconstructedPrevNearestDepth, Fsr3ShaderIDs.UavReconstructedPrevNearestDepth); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvDilatedMotionVectors, Resources.DilatedMotionVectors[frameIndex]); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvDilatedDepth, Fsr3ShaderIDs.UavDilatedDepth); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvPrevDilatedMotionVectors, Resources.DilatedMotionVectors[frameIndex ^ 1]); - - commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); - - commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); - } - } - - internal class Fsr3UpscalerLockPass : Fsr3UpscalerPass - { - public Fsr3UpscalerLockPass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants) - : base(contextDescription, resources, constants) - { - InitComputeShader("lock_pass", contextDescription.Shaders.lockPass); - } - - public override void ScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) - { - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvLockInputLuma, Fsr3ShaderIDs.UavLockInputLuma); - commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); - - commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); - } - } - - internal class Fsr3UpscalerAccumulatePass : Fsr3UpscalerPass - { - private const string SharpeningKeyword = "FFX_FSR3UPSCALER_OPTION_APPLY_SHARPENING"; - -#if UNITY_2021_2_OR_NEWER - private readonly LocalKeyword _sharpeningKeyword; -#endif - - public Fsr3UpscalerAccumulatePass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants) - : base(contextDescription, resources, constants) - { - InitComputeShader("accumulate_pass", contextDescription.Shaders.accumulatePass); -#if UNITY_2021_2_OR_NEWER - _sharpeningKeyword = new LocalKeyword(ComputeShader, SharpeningKeyword); -#endif - } - - public override void ScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) - { -#if UNITY_2021_2_OR_NEWER - if (dispatchParams.EnableSharpening) - commandBuffer.EnableKeyword(ComputeShader, _sharpeningKeyword); - else - commandBuffer.DisableKeyword(ComputeShader, _sharpeningKeyword); -#else - if (dispatchParams.EnableSharpening) - commandBuffer.EnableShaderKeyword(SharpeningKeyword); - else - commandBuffer.DisableShaderKeyword(SharpeningKeyword); -#endif - - if ((ContextDescription.Flags & Fsr3Upscaler.InitializationFlags.EnableDisplayResolutionMotionVectors) == 0) - { - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvDilatedMotionVectors, Resources.DilatedMotionVectors[frameIndex]); - } - else - { - ref var motionVectors = ref dispatchParams.MotionVectors; - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputMotionVectors, motionVectors.RenderTarget, motionVectors.MipLevel, motionVectors.SubElement); - } - - ref var exposure = ref dispatchParams.Exposure; - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputExposure, exposure.RenderTarget, exposure.MipLevel, exposure.SubElement); - - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvDilatedReactiveMasks, Fsr3ShaderIDs.UavDilatedReactiveMasks); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInternalUpscaled, Resources.InternalUpscaled[frameIndex ^ 1]); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvLockStatus, Resources.LockStatus[frameIndex ^ 1]); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvPreparedInputColor, Fsr3ShaderIDs.UavPreparedInputColor); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvLanczosLut, Resources.LanczosLut); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvUpscaleMaximumBiasLut, Resources.MaximumBiasLut); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvSceneLuminanceMips, Resources.SceneLuminance); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvAutoExposure, Resources.AutoExposure); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvLumaHistory, Resources.LumaHistory[frameIndex ^ 1]); - - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavInternalUpscaled, Resources.InternalUpscaled[frameIndex]); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavLockStatus, Resources.LockStatus[frameIndex]); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavLumaHistory, Resources.LumaHistory[frameIndex]); - - ref var output = ref dispatchParams.Output; - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavUpscaledOutput, output.RenderTarget, output.MipLevel, output.SubElement); - - commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); - - commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); - } - } - - internal class Fsr3UpscalerSharpenPass : Fsr3UpscalerPass - { - private readonly ComputeBuffer _rcasConstants; - - public Fsr3UpscalerSharpenPass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants, ComputeBuffer rcasConstants) - : base(contextDescription, resources, constants) - { - _rcasConstants = rcasConstants; - - InitComputeShader("rcas_pass", contextDescription.Shaders.sharpenPass); - } - - public override void ScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) - { - ref var exposure = ref dispatchParams.Exposure; - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputExposure, exposure.RenderTarget, exposure.MipLevel, exposure.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvRcasInput, Resources.InternalUpscaled[frameIndex]); - - ref var output = ref dispatchParams.Output; - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavUpscaledOutput, output.RenderTarget, output.MipLevel, output.SubElement); - - commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); - commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbRcas, _rcasConstants, 0, Marshal.SizeOf()); - - commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); - } - } - - internal class Fsr3UpscalerGenerateReactivePass : Fsr3UpscalerPass - { - private readonly ComputeBuffer _generateReactiveConstants; - - public Fsr3UpscalerGenerateReactivePass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer generateReactiveConstants) - : base(contextDescription, resources, null) - { - _generateReactiveConstants = generateReactiveConstants; - - InitComputeShader("autogen_reactive_pass", contextDescription.Shaders.autoGenReactivePass); - } - - public override void ScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) - { - } - - public void ScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.GenerateReactiveDescription dispatchParams, int dispatchX, int dispatchY) - { - ref var opaqueOnly = ref dispatchParams.ColorOpaqueOnly; - ref var color = ref dispatchParams.ColorPreUpscale; - ref var reactive = ref dispatchParams.OutReactive; - - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvOpaqueOnly, opaqueOnly.RenderTarget, opaqueOnly.MipLevel, opaqueOnly.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputColor, color.RenderTarget, color.MipLevel, color.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavAutoReactive, reactive.RenderTarget, reactive.MipLevel, reactive.SubElement); - - commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbGenReactive, _generateReactiveConstants, 0, Marshal.SizeOf()); - - commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); - } - } - - internal class Fsr3UpscalerTcrAutogeneratePass : Fsr3UpscalerPass - { - private readonly ComputeBuffer _tcrAutogenerateConstants; - - public Fsr3UpscalerTcrAutogeneratePass(Fsr3Upscaler.ContextDescription contextDescription, Fsr3UpscalerResources resources, ComputeBuffer constants, ComputeBuffer tcrAutogenerateConstants) - : base(contextDescription, resources, constants) - { - _tcrAutogenerateConstants = tcrAutogenerateConstants; - - InitComputeShader("tcr_autogen_pass", contextDescription.Shaders.tcrAutoGenPass); - } - - public override void ScheduleDispatch(CommandBuffer commandBuffer, Fsr3Upscaler.DispatchDescription dispatchParams, int frameIndex, int dispatchX, int dispatchY) - { - ref var color = ref dispatchParams.Color; - ref var motionVectors = ref dispatchParams.MotionVectors; - ref var opaqueOnly = ref dispatchParams.ColorOpaqueOnly; - ref var reactive = ref dispatchParams.Reactive; - ref var tac = ref dispatchParams.TransparencyAndComposition; - - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvOpaqueOnly, opaqueOnly.RenderTarget, opaqueOnly.MipLevel, opaqueOnly.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputColor, color.RenderTarget, color.MipLevel, color.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvInputMotionVectors, motionVectors.RenderTarget, motionVectors.MipLevel, motionVectors.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvPrevColorPreAlpha, Resources.PrevPreAlpha[frameIndex ^ 1]); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvPrevColorPostAlpha, Resources.PrevPostAlpha[frameIndex ^ 1]); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvReactiveMask, reactive.RenderTarget, reactive.MipLevel, reactive.SubElement); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.SrvTransparencyAndCompositionMask, tac.RenderTarget, tac.MipLevel, tac.SubElement); - - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavAutoReactive, Resources.AutoReactive); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavAutoComposition, Resources.AutoComposition); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavPrevColorPreAlpha, Resources.PrevPreAlpha[frameIndex]); - commandBuffer.SetComputeTextureParam(ComputeShader, KernelIndex, Fsr3ShaderIDs.UavPrevColorPostAlpha, Resources.PrevPostAlpha[frameIndex]); - - commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbFsr3Upscaler, Constants, 0, Marshal.SizeOf()); - commandBuffer.SetComputeConstantBufferParam(ComputeShader, Fsr3ShaderIDs.CbGenReactive, _tcrAutogenerateConstants, 0, Marshal.SizeOf()); - - commandBuffer.DispatchCompute(ComputeShader, KernelIndex, dispatchX, dispatchY, 1); - } - } -} diff --git a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerPass.cs.meta b/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerPass.cs.meta deleted file mode 100644 index 8ffd58e..0000000 --- a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerPass.cs.meta +++ /dev/null @@ -1,3 +0,0 @@ -fileFormatVersion: 2 -guid: cfd077da533b192458b0b548668776e7 -timeCreated: 1676885169 \ No newline at end of file diff --git a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerResources.cs b/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerResources.cs deleted file mode 100644 index 344a371..0000000 --- a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerResources.cs +++ /dev/null @@ -1,258 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -using System; -using UnityEngine; -using UnityEngine.Experimental.Rendering; -using UnityEngine.Rendering; - -namespace FidelityFX -{ - /// - /// Helper class for bundling and managing persistent resources required by the FSR3 Upscaler process. - /// This includes lookup tables, default fallback resources and double-buffered resources that get swapped between frames. - /// - internal class Fsr3UpscalerResources - { - public Texture2D DefaultExposure; - public Texture2D DefaultReactive; - public Texture2D LanczosLut; - public Texture2D MaximumBiasLut; - public RenderTexture SpdAtomicCounter; - public RenderTexture AutoExposure; - public RenderTexture SceneLuminance; - public RenderTexture AutoReactive; - public RenderTexture AutoComposition; - public readonly RenderTexture[] DilatedMotionVectors = new RenderTexture[2]; - public readonly RenderTexture[] LockStatus = new RenderTexture[2]; - public readonly RenderTexture[] InternalUpscaled = new RenderTexture[2]; - public readonly RenderTexture[] LumaHistory = new RenderTexture[2]; - public readonly RenderTexture[] PrevPreAlpha = new RenderTexture[2]; - public readonly RenderTexture[] PrevPostAlpha = new RenderTexture[2]; - - public void Create(Fsr3Upscaler.ContextDescription contextDescription) - { - // Generate the data for the LUT - const int lanczos2LutWidth = 128; - float[] lanczos2Weights = new float[lanczos2LutWidth]; - for (int currentLanczosWidthIndex = 0; currentLanczosWidthIndex < lanczos2LutWidth; ++currentLanczosWidthIndex) - { - float x = 2.0f * currentLanczosWidthIndex / (lanczos2LutWidth - 1); - float y = Fsr3Upscaler.Lanczos2(x); - lanczos2Weights[currentLanczosWidthIndex] = y; - } - - float[] maximumBias = new float[MaximumBiasTextureWidth * MaximumBiasTextureHeight]; - for (int i = 0; i < maximumBias.Length; ++i) - { - maximumBias[i] = MaximumBias[i] / 2.0f; - } - - // Resource FSR3UPSCALER_LanczosLutData: FFX_RESOURCE_USAGE_READ_ONLY, FFX_SURFACE_FORMAT_R16_SNORM, FFX_RESOURCE_FLAGS_NONE - // R16_SNorm textures are not supported by Unity on most platforms, strangely enough. So instead we use R32_SFloat and upload pre-normalized float data. - LanczosLut = new Texture2D(lanczos2LutWidth, 1, GraphicsFormat.R32_SFloat, TextureCreationFlags.None) { name = "FSR3UPSCALER_LanczosLutData" }; - LanczosLut.SetPixelData(lanczos2Weights, 0); - LanczosLut.Apply(); - - // Resource FSR3UPSCALER_MaximumUpsampleBias: FFX_RESOURCE_USAGE_READ_ONLY, FFX_SURFACE_FORMAT_R16_SNORM, FFX_RESOURCE_FLAGS_NONE - MaximumBiasLut = new Texture2D(MaximumBiasTextureWidth, MaximumBiasTextureHeight, GraphicsFormat.R32_SFloat, TextureCreationFlags.None) { name = "FSR3UPSCALER_MaximumUpsampleBias" }; - MaximumBiasLut.SetPixelData(maximumBias, 0); - MaximumBiasLut.Apply(); - - // Resource FSR3UPSCALER_DefaultExposure: FFX_RESOURCE_USAGE_READ_ONLY, FFX_SURFACE_FORMAT_R32G32_FLOAT, FFX_RESOURCE_FLAGS_NONE - DefaultExposure = new Texture2D(1, 1, GraphicsFormat.R32G32_SFloat, TextureCreationFlags.None) { name = "FSR3UPSCALER_DefaultExposure" }; - DefaultExposure.SetPixel(0, 0, Color.clear); - DefaultExposure.Apply(); - - // Resource FSR3UPSCALER_DefaultReactivityMask: FFX_RESOURCE_USAGE_READ_ONLY, FFX_SURFACE_FORMAT_R8_UNORM, FFX_RESOURCE_FLAGS_NONE - DefaultReactive = new Texture2D(1, 1, GraphicsFormat.R8_UNorm, TextureCreationFlags.None) { name = "FSR3UPSCALER_DefaultReactivityMask" }; - DefaultReactive.SetPixel(0, 0, Color.clear); - DefaultReactive.Apply(); - - // Resource FSR3UPSCALER_SpdAtomicCounter: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R32_UINT, FFX_RESOURCE_FLAGS_ALIASABLE - // Despite what the original FSR3 codebase says, this resource really isn't aliasable. Resetting this counter to 0 every frame breaks auto-exposure on MacOS Metal. - SpdAtomicCounter = new RenderTexture(1, 1, 0, GraphicsFormat.R32_UInt) { name = "FSR3UPSCALER_SpdAtomicCounter", enableRandomWrite = true }; - SpdAtomicCounter.Create(); - - // Resource FSR3UPSCALER_AutoExposure: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R32G32_FLOAT, FFX_RESOURCE_FLAGS_NONE - AutoExposure = new RenderTexture(1, 1, 0, GraphicsFormat.R32G32_SFloat) { name = "FSR3UPSCALER_AutoExposure", enableRandomWrite = true }; - AutoExposure.Create(); - - // Resource FSR3UPSCALER_ExposureMips: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16_FLOAT, FFX_RESOURCE_FLAGS_ALIASABLE - // This is a rather special case: it's an aliasable resource, but because we require a mipmap chain and bind specific mip levels per shader, we can't easily use temporary RTs for this. - int w = contextDescription.MaxRenderSize.x / 2, h = contextDescription.MaxRenderSize.y / 2; - int mipCount = 1 + Mathf.FloorToInt(Mathf.Log(Math.Max(w, h), 2.0f)); - SceneLuminance = new RenderTexture(w, h, 0, GraphicsFormat.R16_SFloat, mipCount) { name = "FSR3UPSCALER_ExposureMips", enableRandomWrite = true, useMipMap = true, autoGenerateMips = false }; - SceneLuminance.Create(); - - // Resources FSR3UPSCALER_InternalDilatedVelocity1/2: FFX_RESOURCE_USAGE_RENDERTARGET | FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16G16_FLOAT, FFX_RESOURCE_FLAGS_NONE - CreateDoubleBufferedResource(DilatedMotionVectors, "FSR3UPSCALER_InternalDilatedVelocity", contextDescription.MaxRenderSize, GraphicsFormat.R16G16_SFloat); - - // Resources FSR3UPSCALER_LockStatus1/2: FFX_RESOURCE_USAGE_RENDERTARGET | FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16G16_FLOAT, FFX_RESOURCE_FLAGS_NONE - CreateDoubleBufferedResource(LockStatus, "FSR3UPSCALER_LockStatus", contextDescription.DisplaySize, GraphicsFormat.R16G16_SFloat); - - // Resources FSR3UPSCALER_InternalUpscaled1/2: FFX_RESOURCE_USAGE_RENDERTARGET | FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16G16B16A16_FLOAT, FFX_RESOURCE_FLAGS_NONE - CreateDoubleBufferedResource(InternalUpscaled, "FSR3UPSCALER_InternalUpscaled", contextDescription.DisplaySize, GraphicsFormat.R16G16B16A16_SFloat); - - // Resources FSR3UPSCALER_LumaHistory1/2: FFX_RESOURCE_USAGE_RENDERTARGET | FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R8G8B8A8_UNORM, FFX_RESOURCE_FLAGS_NONE - CreateDoubleBufferedResource(LumaHistory, "FSR3UPSCALER_LumaHistory", contextDescription.DisplaySize, GraphicsFormat.R8G8B8A8_UNorm); - } - - public void CreateTcrAutogenResources(Fsr3Upscaler.ContextDescription contextDescription) - { - // Resource FSR3UPSCALER_AutoReactive: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R8_UNORM, FFX_RESOURCE_FLAGS_NONE - AutoReactive = new RenderTexture(contextDescription.MaxRenderSize.x, contextDescription.MaxRenderSize.y, 0, GraphicsFormat.R8_UNorm) { name = "FSR3UPSCALER_AutoReactive", enableRandomWrite = true }; - AutoReactive.Create(); - - // Resource FSR3UPSCALER_AutoComposition: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R8_UNORM, FFX_RESOURCE_FLAGS_NONE - AutoComposition = new RenderTexture(contextDescription.MaxRenderSize.x, contextDescription.MaxRenderSize.y, 0, GraphicsFormat.R8_UNorm) { name = "FSR3UPSCALER_AutoComposition", enableRandomWrite = true }; - AutoComposition.Create(); - - // Resources FSR3UPSCALER_PrevPreAlpha0/1: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R11G11B10_FLOAT, FFX_RESOURCE_FLAGS_NONE - CreateDoubleBufferedResource(PrevPreAlpha, "FSR3UPSCALER_PrevPreAlpha", contextDescription.MaxRenderSize, GraphicsFormat.B10G11R11_UFloatPack32); - - // Resources FSR3UPSCALER_PrevPostAlpha0/1: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R11G11B10_FLOAT, FFX_RESOURCE_FLAGS_NONE - CreateDoubleBufferedResource(PrevPostAlpha, "FSR3UPSCALER_PrevPostAlpha", contextDescription.MaxRenderSize, GraphicsFormat.B10G11R11_UFloatPack32); - } - - // Set up shared aliasable resources, i.e. temporary render textures - // These do not need to persist between frames, but they do need to be available between passes - public static void CreateAliasableResources(CommandBuffer commandBuffer, Fsr3Upscaler.ContextDescription contextDescription, Fsr3Upscaler.DispatchDescription dispatchParams) - { - Vector2Int displaySize = contextDescription.DisplaySize; - Vector2Int maxRenderSize = contextDescription.MaxRenderSize; - - // FSR3UPSCALER_ReconstructedPrevNearestDepth: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R32_UINT, FFX_RESOURCE_FLAGS_ALIASABLE - commandBuffer.GetTemporaryRT(Fsr3ShaderIDs.UavReconstructedPrevNearestDepth, maxRenderSize.x, maxRenderSize.y, 0, default, GraphicsFormat.R32_UInt, 1, true); - - // FSR3UPSCALER_DilatedDepth: FFX_RESOURCE_USAGE_RENDERTARGET | FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R32_FLOAT, FFX_RESOURCE_FLAGS_ALIASABLE - commandBuffer.GetTemporaryRT(Fsr3ShaderIDs.UavDilatedDepth, maxRenderSize.x, maxRenderSize.y, 0, default, GraphicsFormat.R32_SFloat, 1, true); - - // FSR3UPSCALER_LockInputLuma: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16_FLOAT, FFX_RESOURCE_FLAGS_ALIASABLE - commandBuffer.GetTemporaryRT(Fsr3ShaderIDs.UavLockInputLuma, maxRenderSize.x, maxRenderSize.y, 0, default, GraphicsFormat.R16_SFloat, 1, true); - - // FSR3UPSCALER_DilatedReactiveMasks: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R8G8_UNORM, FFX_RESOURCE_FLAGS_ALIASABLE - commandBuffer.GetTemporaryRT(Fsr3ShaderIDs.UavDilatedReactiveMasks, maxRenderSize.x, maxRenderSize.y, 0, default, GraphicsFormat.R8G8_UNorm, 1, true); - - // FSR3UPSCALER_PreparedInputColor: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R16G16B16A16_FLOAT, FFX_RESOURCE_FLAGS_ALIASABLE - commandBuffer.GetTemporaryRT(Fsr3ShaderIDs.UavPreparedInputColor, maxRenderSize.x, maxRenderSize.y, 0, default, GraphicsFormat.R16G16B16A16_SFloat, 1, true); - - // FSR3UPSCALER_NewLocks: FFX_RESOURCE_USAGE_UAV, FFX_SURFACE_FORMAT_R8_UNORM, FFX_RESOURCE_FLAGS_ALIASABLE - commandBuffer.GetTemporaryRT(Fsr3ShaderIDs.UavNewLocks, displaySize.x, displaySize.y, 0, default, GraphicsFormat.R8_UNorm, 1, true); - } - - public static void DestroyAliasableResources(CommandBuffer commandBuffer) - { - // Release all of the aliasable resources used this frame - commandBuffer.ReleaseTemporaryRT(Fsr3ShaderIDs.UavReconstructedPrevNearestDepth); - commandBuffer.ReleaseTemporaryRT(Fsr3ShaderIDs.UavDilatedDepth); - commandBuffer.ReleaseTemporaryRT(Fsr3ShaderIDs.UavLockInputLuma); - commandBuffer.ReleaseTemporaryRT(Fsr3ShaderIDs.UavDilatedReactiveMasks); - commandBuffer.ReleaseTemporaryRT(Fsr3ShaderIDs.UavPreparedInputColor); - commandBuffer.ReleaseTemporaryRT(Fsr3ShaderIDs.UavNewLocks); - } - - private static void CreateDoubleBufferedResource(RenderTexture[] resource, string name, Vector2Int size, GraphicsFormat format) - { - for (int i = 0; i < 2; ++i) - { - resource[i] = new RenderTexture(size.x, size.y, 0, format) { name = name + (i + 1), enableRandomWrite = true }; - resource[i].Create(); - } - } - - public void Destroy() - { - DestroyTcrAutogenResources(); - - DestroyResource(LumaHistory); - DestroyResource(InternalUpscaled); - DestroyResource(LockStatus); - DestroyResource(DilatedMotionVectors); - DestroyResource(ref SceneLuminance); - DestroyResource(ref AutoExposure); - DestroyResource(ref DefaultReactive); - DestroyResource(ref DefaultExposure); - DestroyResource(ref MaximumBiasLut); - DestroyResource(ref LanczosLut); - } - - public void DestroyTcrAutogenResources() - { - DestroyResource(PrevPostAlpha); - DestroyResource(PrevPreAlpha); - DestroyResource(ref AutoComposition); - DestroyResource(ref AutoReactive); - } - - private static void DestroyResource(ref Texture2D resource) - { - if (resource == null) - return; - -#if UNITY_EDITOR - if (Application.isPlaying && !UnityEditor.EditorApplication.isPaused) - UnityEngine.Object.Destroy(resource); - else - UnityEngine.Object.DestroyImmediate(resource); -#else - UnityEngine.Object.Destroy(resource); -#endif - resource = null; - } - - private static void DestroyResource(ref RenderTexture resource) - { - if (resource == null) - return; - - resource.Release(); - resource = null; - } - - private static void DestroyResource(RenderTexture[] resource) - { - for (int i = 0; i < resource.Length; ++i) - DestroyResource(ref resource[i]); - } - - private const int MaximumBiasTextureWidth = 16; - private const int MaximumBiasTextureHeight = 16; - private static readonly float[] MaximumBias = - { - 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.876f, 1.809f, 1.772f, 1.753f, 1.748f, - 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.869f, 1.801f, 1.764f, 1.745f, 1.739f, - 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.976f, 1.841f, 1.774f, 1.737f, 1.716f, 1.71f, - 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.914f, 1.784f, 1.716f, 1.673f, 1.649f, 1.641f, - 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.793f, 1.676f, 1.604f, 1.562f, 1.54f, 1.533f, - 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.802f, 1.619f, 1.536f, 1.492f, 1.467f, 1.454f, 1.449f, - 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.812f, 1.575f, 1.496f, 1.456f, 1.432f, 1.416f, 1.408f, 1.405f, - 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.555f, 1.479f, 1.438f, 1.413f, 1.398f, 1.387f, 1.381f, 1.379f, - 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.812f, 1.555f, 1.474f, 1.43f, 1.404f, 1.387f, 1.376f, 1.368f, 1.363f, 1.362f, - 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 1.802f, 1.575f, 1.479f, 1.43f, 1.401f, 1.382f, 1.369f, 1.36f, 1.354f, 1.351f, 1.35f, - 2.0f, 2.0f, 1.976f, 1.914f, 1.793f, 1.619f, 1.496f, 1.438f, 1.404f, 1.382f, 1.367f, 1.357f, 1.349f, 1.344f, 1.341f, 1.34f, - 1.876f, 1.869f, 1.841f, 1.784f, 1.676f, 1.536f, 1.456f, 1.413f, 1.387f, 1.369f, 1.357f, 1.347f, 1.341f, 1.336f, 1.333f, 1.332f, - 1.809f, 1.801f, 1.774f, 1.716f, 1.604f, 1.492f, 1.432f, 1.398f, 1.376f, 1.36f, 1.349f, 1.341f, 1.335f, 1.33f, 1.328f, 1.327f, - 1.772f, 1.764f, 1.737f, 1.673f, 1.562f, 1.467f, 1.416f, 1.387f, 1.368f, 1.354f, 1.344f, 1.336f, 1.33f, 1.326f, 1.323f, 1.323f, - 1.753f, 1.745f, 1.716f, 1.649f, 1.54f, 1.454f, 1.408f, 1.381f, 1.363f, 1.351f, 1.341f, 1.333f, 1.328f, 1.323f, 1.321f, 1.32f, - 1.748f, 1.739f, 1.71f, 1.641f, 1.533f, 1.449f, 1.405f, 1.379f, 1.362f, 1.35f, 1.34f, 1.332f, 1.327f, 1.323f, 1.32f, 1.319f, - }; - } -} diff --git a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerResources.cs.meta b/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerResources.cs.meta deleted file mode 100644 index bf03a07..0000000 --- a/com.unity.postprocessing/PostProcessing/Runtime/FSR3/Fsr3UpscalerResources.cs.meta +++ /dev/null @@ -1,3 +0,0 @@ -fileFormatVersion: 2 -guid: ec6c0c34c7b11f041885ddee4aa72818 -timeCreated: 1677236102 \ No newline at end of file diff --git a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessResources.cs b/com.unity.postprocessing/PostProcessing/Runtime/PostProcessResources.cs deleted file mode 100644 index 35293ef..0000000 --- a/com.unity.postprocessing/PostProcessing/Runtime/PostProcessResources.cs +++ /dev/null @@ -1,293 +0,0 @@ -using System; -using FidelityFX; - -namespace UnityEngine.Rendering.PostProcessing -{ - /// - /// This asset is used to store references to shaders and other resources we might need at - /// runtime without having to use a `Resources` folder. This allows for better memory management, - /// better dependency tracking and better interoperability with asset bundles. - /// - public sealed class PostProcessResources : ScriptableObject - { - /// - /// All the shaders used by post-processing. - /// - [Serializable] - public sealed class Shaders - { - /// - /// The shader used for the bloom effect. - /// - public Shader bloom; - - /// - /// The shader used for internal blit copies. - /// - public Shader copy; - - /// - /// The shader used for built-in blit copies in the built-in pipeline. - /// - public Shader copyStd; - - /// - /// The shader used for built-in blit copies in the built-in pipeline when using a - /// texture array (for stereo rendering). - /// - public Shader copyStdFromTexArray; - - /// - /// The shader used for built-in blit copies in the built-in pipeline when using a - /// double-wide texture (for stereo rendering). - /// - public Shader copyStdFromDoubleWide; - - /// - /// The shader used to kill the alpha. - /// - public Shader discardAlpha; - - /// - /// The shader used for the depth of field effect. - /// - public Shader depthOfField; - - /// - /// The shader used for the final pass. - /// - public Shader finalPass; - - /// - /// The shader used to generate the grain texture. - /// - public Shader grainBaker; - - /// - /// The shader used for the motion blur effect. - /// - public Shader motionBlur; - - /// - /// The shader used for the temporal anti-aliasing effect. - /// - public Shader temporalAntialiasing; - - /// - /// The shader used for the sub-pixel morphological anti-aliasing effect. - /// - public Shader subpixelMorphologicalAntialiasing; - - /// - /// The shader use by the volume manager to interpolate between two 2D textures. - /// - public Shader texture2dLerp; - - /// - /// The uber shader that combine several effects into one. - /// - public Shader uber; - - /// - /// The shader used to bake the 2D lookup table for color grading. - /// - public Shader lut2DBaker; - - /// - /// The shader used to draw the light meter monitor. - /// - public Shader lightMeter; - - /// - /// The shader used to draw the histogram monitor. - /// - public Shader gammaHistogram; - - /// - /// The shader used to draw the waveform monitor. - /// - public Shader waveform; - - /// - /// The shader used to draw the vectorscope monitor. - /// - public Shader vectorscope; - - /// - /// The shader used to draw debug overlays. - /// - public Shader debugOverlays; - - /// - /// The shader used for the deferred fog effect. - /// - public Shader deferredFog; - - /// - /// The shader used for the scalable ambient occlusion effect. - /// - public Shader scalableAO; - - /// - /// The shader used for the multi-scale ambient occlusion effect. - /// - public Shader multiScaleAO; - - /// - /// The shader used for the screen-space reflection effect. - /// - public Shader screenSpaceReflections; - - /// - /// Returns a copy of this class and its content. - /// - /// A copy of this class and its content. - public Shaders Clone() - { - return (Shaders)MemberwiseClone(); - } - } - - /// - /// All the compute shaders used by post-processing. - /// - [Serializable] - public sealed class ComputeShaders - { - /// - /// The compute shader used for the auto-exposure effect. - /// - public ComputeShader autoExposure; - - /// - /// The compute shader used to compute an histogram of the current frame. - /// - public ComputeShader exposureHistogram; - - /// - /// The compute shader used to bake the 3D lookup table for color grading. - /// - public ComputeShader lut3DBaker; - - /// - /// The compute shader used by the volume manager to interpolate between two 3D textures. - /// - public ComputeShader texture3dLerp; - - /// - /// The compute shader used to compute the histogram monitor. - /// - public ComputeShader gammaHistogram; - - /// - /// The compute shader used to compute the waveform monitor. - /// - public ComputeShader waveform; - - /// - /// The compute shader used to compute the vectorscope monitor. - /// - public ComputeShader vectorscope; - - /// - /// The compute shader used for the first downsampling pass of MSVO. - /// - public ComputeShader multiScaleAODownsample1; - - /// - /// The compute shader used for the second downsampling pass of MSVO. - /// - public ComputeShader multiScaleAODownsample2; - - /// - /// The compute shader used for the render pass of MSVO. - /// - public ComputeShader multiScaleAORender; - - /// - /// The compute shader used for the upsampling pass of MSVO. - /// - public ComputeShader multiScaleAOUpsample; - - /// - /// The compute shader used to a fast gaussian downsample. - /// - public ComputeShader gaussianDownsample; - - /// - /// Compute shaders used by the FidelityFX Super Resolution 3 (FSR3) Upscaler. - /// - public Fsr3UpscalerShaders superResolution; - - /// - /// Returns a copy of this class and its content. - /// - /// A copy of this class and its content. - public ComputeShaders Clone() - { - return (ComputeShaders)MemberwiseClone(); - } - } - - /// - /// A set of textures needed by the sub-pixel morphological anti-aliasing effect. - /// - [Serializable] - public sealed class SMAALuts - { - /// - /// The area lookup table. - /// - public Texture2D area; - - /// - /// The search lookup table. - /// - public Texture2D search; - } - - /// - /// A set of 64x64, single-channel blue noise textures. - /// - public Texture2D[] blueNoise64; - - /// - /// A set of 256x256, single-channel blue noise textures. - /// - public Texture2D[] blueNoise256; - - /// - /// Lookup tables used by the sub-pixel morphological anti-aliasing effect. - /// - public SMAALuts smaaLuts; - - /// - /// All the shaders used by post-processing. - /// - public Shaders shaders; - - /// - /// All the compute shaders used by post-processing. - /// - public ComputeShaders computeShaders; - -#if UNITY_EDITOR - /// - /// A delegate used to track resource changes. - /// - public delegate void ChangeHandler(); - - /// - /// Set this callback to be notified of resource changes. - /// - public ChangeHandler changeHandler; - - void OnValidate() - { - if (changeHandler != null) - changeHandler(); - } - -#endif - } -} diff --git a/com.unity.postprocessing/PostProcessing/Runtime/Unity.Postprocessing.Runtime.asmdef b/com.unity.postprocessing/PostProcessing/Runtime/Unity.Postprocessing.Runtime.asmdef deleted file mode 100644 index 91449de..0000000 --- a/com.unity.postprocessing/PostProcessing/Runtime/Unity.Postprocessing.Runtime.asmdef +++ /dev/null @@ -1,29 +0,0 @@ -{ - "name": "Unity.Postprocessing.Runtime", - "references": [], - "optionalUnityReferences": [], - "includePlatforms": [], - "excludePlatforms": [], - "versionDefines": [ - { - "name": "com.unity.render-pipelines.lightweight", - "expression": "1.0.0", - "define": "LWRP_1_0_0_OR_NEWER" - }, - { - "name": "com.unity.render-pipelines.universal", - "expression": "1.0.0", - "define": "UNIVERSAL_1_0_0_OR_NEWER" - }, - { - "name": "com.unity.modules.vr", - "expression": "1.0.0", - "define": "ENABLE_VR_MODULE" - }, - { - "name": "com.unity.modules.xr", - "expression": "1.0.0", - "define": "ENABLE_XR_MODULE" - } - ] -} diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3.meta deleted file mode 100644 index 3e30dae..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3.meta +++ /dev/null @@ -1,8 +0,0 @@ -fileFormatVersion: 2 -guid: 37c436021e67c00459f44b59099c024a -folderAsset: yes -DefaultImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_accumulate_pass.compute b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_accumulate_pass.compute deleted file mode 100644 index 5cbfb80..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_accumulate_pass.compute +++ /dev/null @@ -1,41 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -#pragma kernel CS - -#pragma multi_compile_local __ FFX_HALF -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_HDR_COLOR_INPUT -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_APPLY_SHARPENING - -#pragma multi_compile_local __ UNITY_FSR3UPSCALER_HDRP - -#include "ffx_fsr3upscaler_unity_common.cginc" - -// Ensure the correct value is defined for this keyword, as it is used to select one of multiple sampler functions -#ifdef FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE -#undef FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE -#define FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE 1 -#endif - -#include "shaders/ffx_fsr3upscaler_accumulate_pass.hlsl" diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_accumulate_pass.compute.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_accumulate_pass.compute.meta deleted file mode 100644 index dbe5282..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_accumulate_pass.compute.meta +++ /dev/null @@ -1,8 +0,0 @@ -fileFormatVersion: 2 -guid: c9b45f0ae7673694ba57a4aadfe212e9 -ComputeShaderImporter: - externalObjects: {} - preprocessorOverride: 0 - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_autogen_reactive_pass.compute b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_autogen_reactive_pass.compute deleted file mode 100644 index e13c001..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_autogen_reactive_pass.compute +++ /dev/null @@ -1,32 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -#pragma kernel CS - -#pragma multi_compile_local __ FFX_HALF -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - -#pragma multi_compile_local __ UNITY_FSR3UPSCALER_HDRP - -#include "ffx_fsr3upscaler_unity_common.cginc" - -#include "shaders/ffx_fsr3upscaler_autogen_reactive_pass.hlsl" diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_autogen_reactive_pass.compute.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_autogen_reactive_pass.compute.meta deleted file mode 100644 index 1df041b..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_autogen_reactive_pass.compute.meta +++ /dev/null @@ -1,8 +0,0 @@ -fileFormatVersion: 2 -guid: 5716b91fdaa4e9e439df6b96a796fe6e -ComputeShaderImporter: - externalObjects: {} - preprocessorOverride: 0 - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_compute_luminance_pyramid_pass.compute b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_compute_luminance_pyramid_pass.compute deleted file mode 100644 index d5903c0..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_compute_luminance_pyramid_pass.compute +++ /dev/null @@ -1,42 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -#pragma kernel CS - -#pragma multi_compile_local __ FFX_HALF -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - -#pragma multi_compile_local __ UNITY_FSR3UPSCALER_HDRP - -#include "ffx_fsr3upscaler_unity_common.cginc" - -// Wave operations require shader model 6.0; this can only be enabled when using DXC on D3D12 -// These pragmas are commented out by default as Unity will sometimes ignore the #if's and try to enable these features anyway. -// Uncomment the below lines if you intend to try wave operations on DX12 with the DXC compiler. -//#if defined(UNITY_COMPILER_DXC) && defined(SHADER_API_D3D12) -//#pragma require WaveBasic // Required for WaveGetLaneIndex -//#pragma require WaveBallot // Required for WaveReadLaneAt -//#else -#define FFX_SPD_NO_WAVE_OPERATIONS -//#endif - -#include "shaders/ffx_fsr3upscaler_compute_luminance_pyramid_pass.hlsl" diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_compute_luminance_pyramid_pass.compute.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_compute_luminance_pyramid_pass.compute.meta deleted file mode 100644 index 9e002c0..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_compute_luminance_pyramid_pass.compute.meta +++ /dev/null @@ -1,8 +0,0 @@ -fileFormatVersion: 2 -guid: d253be05abcdc80428503d3e4cce3a36 -ComputeShaderImporter: - externalObjects: {} - preprocessorOverride: 0 - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_depth_clip_pass.compute b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_depth_clip_pass.compute deleted file mode 100644 index 0ccd388..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_depth_clip_pass.compute +++ /dev/null @@ -1,32 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -#pragma kernel CS - -#pragma multi_compile_local __ FFX_HALF -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - -#pragma multi_compile_local __ UNITY_FSR3UPSCALER_HDRP - -#include "ffx_fsr3upscaler_unity_common.cginc" - -#include "shaders/ffx_fsr3upscaler_depth_clip_pass.hlsl" diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_depth_clip_pass.compute.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_depth_clip_pass.compute.meta deleted file mode 100644 index d695f48..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_depth_clip_pass.compute.meta +++ /dev/null @@ -1,8 +0,0 @@ -fileFormatVersion: 2 -guid: 20e44016ed34b0d4b8de499d1b566c69 -ComputeShaderImporter: - externalObjects: {} - preprocessorOverride: 0 - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_lock_pass.compute b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_lock_pass.compute deleted file mode 100644 index e38ad99..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_lock_pass.compute +++ /dev/null @@ -1,30 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -#pragma kernel CS - -#pragma multi_compile_local __ FFX_HALF -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - -#include "ffx_fsr3upscaler_unity_common.cginc" - -#include "shaders/ffx_fsr3upscaler_lock_pass.hlsl" diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_lock_pass.compute.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_lock_pass.compute.meta deleted file mode 100644 index c01e009..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_lock_pass.compute.meta +++ /dev/null @@ -1,8 +0,0 @@ -fileFormatVersion: 2 -guid: a135306e6d1857e43a86ef20db2a47fe -ComputeShaderImporter: - externalObjects: {} - preprocessorOverride: 0 - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_rcas_pass.compute b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_rcas_pass.compute deleted file mode 100644 index be7bbb5..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_rcas_pass.compute +++ /dev/null @@ -1,31 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -#pragma kernel CS - -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - -#pragma multi_compile_local __ UNITY_FSR3UPSCALER_HDRP - -#include "ffx_fsr3upscaler_unity_common.cginc" - -#include "shaders/ffx_fsr3upscaler_rcas_pass.hlsl" diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_rcas_pass.compute.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_rcas_pass.compute.meta deleted file mode 100644 index cd12641..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_rcas_pass.compute.meta +++ /dev/null @@ -1,8 +0,0 @@ -fileFormatVersion: 2 -guid: 7aaf5cfff022de2499e9b0412f947f6c -ComputeShaderImporter: - externalObjects: {} - preprocessorOverride: 0 - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_reconstruct_previous_depth_pass.compute b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_reconstruct_previous_depth_pass.compute deleted file mode 100644 index ee2f276..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_reconstruct_previous_depth_pass.compute +++ /dev/null @@ -1,33 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -#pragma kernel CS - -#pragma multi_compile_local __ FFX_HALF -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_HDR_COLOR_INPUT -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - -#pragma multi_compile_local __ UNITY_FSR3UPSCALER_HDRP - -#include "ffx_fsr3upscaler_unity_common.cginc" - -#include "shaders/ffx_fsr3upscaler_reconstruct_previous_depth_pass.hlsl" diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_reconstruct_previous_depth_pass.compute.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_reconstruct_previous_depth_pass.compute.meta deleted file mode 100644 index 1053c34..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_reconstruct_previous_depth_pass.compute.meta +++ /dev/null @@ -1,8 +0,0 @@ -fileFormatVersion: 2 -guid: 4f59e5b9179d74844ae06a30ae1e0629 -ComputeShaderImporter: - externalObjects: {} - preprocessorOverride: 0 - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_tcr_autogen_pass.compute b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_tcr_autogen_pass.compute deleted file mode 100644 index 6338918..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_tcr_autogen_pass.compute +++ /dev/null @@ -1,32 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -#pragma kernel CS - -#pragma multi_compile_local __ FFX_HALF -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS -#pragma multi_compile_local __ FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - -#pragma multi_compile_local __ UNITY_FSR3UPSCALER_HDRP - -#include "ffx_fsr3upscaler_unity_common.cginc" - -#include "shaders/ffx_fsr3upscaler_tcr_autogen_pass.hlsl" diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_tcr_autogen_pass.compute.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_tcr_autogen_pass.compute.meta deleted file mode 100644 index ad42fbb..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_tcr_autogen_pass.compute.meta +++ /dev/null @@ -1,8 +0,0 @@ -fileFormatVersion: 2 -guid: 75cdc6ef23f08ed498d4da511923fcea -ComputeShaderImporter: - externalObjects: {} - preprocessorOverride: 0 - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_unity_common.cginc b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_unity_common.cginc deleted file mode 100644 index 758bb0c..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_unity_common.cginc +++ /dev/null @@ -1,82 +0,0 @@ -// Copyright (c) 2023 Nico de Poel -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -// Suppress a few warnings produced by FFX's HLSL code -#pragma warning(disable: 3078) // Loop control variable conflicts -#pragma warning(disable: 3203) // Signed/unsigned mismatch - -#define FFX_GPU // Compiling for GPU -#define FFX_HLSL // Compile for plain HLSL - -// Use the DXC shader compiler on modern graphics APIs to enable a few advanced features -// The DXC-related pragmas are disabled by default, as DXC doesn't support all platforms yet and will break on some platforms when enabled. -// Consider this to be an experimental feature. If you want to benefit from 16-bit floating point and wave operations, and don't care about supporting older graphics APIs, then it's worth a try. -//#if defined(SHADER_API_D3D12) || defined(SHADER_API_VULKAN) || defined(SHADER_API_METAL) -//#pragma use_dxc // Using DXC will currently break DX11 support since DX11 and DX12 share the same shader bytecode in Unity. -//#endif - -// Enable half precision data types on platforms that support it -//#if defined(UNITY_COMPILER_DXC) && defined(FFX_HALF) -//#pragma require Native16Bit -//#endif - -// Hack to work around the lack of texture atomics on Metal -#if defined(SHADER_API_METAL) -#define InterlockedAdd(dest, val, orig) { (orig) = (dest); (dest) += (val); } -#define InterlockedMin(dest, val) { (dest) = min((dest), (val)); } -#define InterlockedMax(dest, val) { (dest) = max((dest), (val)); } -#endif - -// Workaround for HDRP using texture arrays for its camera buffers on some platforms -// The below defines are copied from: Packages/com.unity.render-pipelines.high-definition/Runtime/ShaderLibrary/TextureXR.hlsl -#if defined(UNITY_FSR3UPSCALER_HDRP) - // Must be in sync with C# with property useTexArray in TextureXR.cs - #if ((defined(SHADER_API_D3D11) || defined(SHADER_API_D3D12)) && !defined(SHADER_API_XBOXONE) && !defined(SHADER_API_GAMECORE)) || defined(SHADER_API_PSSL) || defined(SHADER_API_VULKAN) - #define UNITY_TEXTURE2D_X_ARRAY_SUPPORTED - #endif - - // Control if TEXTURE2D_X macros will expand to texture arrays - #if defined(UNITY_TEXTURE2D_X_ARRAY_SUPPORTED) && !defined(DISABLE_TEXTURE2D_X_ARRAY) - #define USE_TEXTURE2D_X_AS_ARRAY - #endif - - // Early defines for single-pass instancing - #if defined(STEREO_INSTANCING_ON) && defined(UNITY_TEXTURE2D_X_ARRAY_SUPPORTED) - #define UNITY_STEREO_INSTANCING_ENABLED - #endif - - // Helper macros to handle XR single-pass with Texture2DArray - #if defined(USE_TEXTURE2D_X_AS_ARRAY) - - // Only single-pass stereo instancing used array indexing - #if defined(UNITY_STEREO_INSTANCING_ENABLED) - #define SLICE_ARRAY_INDEX unity_StereoEyeIndex - #else - #define SLICE_ARRAY_INDEX 0 - #endif - - // Declare and sample camera buffers as texture arrays - #define UNITY_FSR3_TEX2D(type) Texture2DArray - #define UNITY_FSR3_RWTEX2D(type) RWTexture2DArray - #define UNITY_FSR3_POS(pxPos) FfxUInt32x3(pxPos, SLICE_ARRAY_INDEX) - #define UNITY_FSR3_UV(uv) FfxFloat32x3(uv, SLICE_ARRAY_INDEX) - - #endif -#endif diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_unity_common.cginc.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_unity_common.cginc.meta deleted file mode 100644 index 5a68b6c..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/ffx_fsr3upscaler_unity_common.cginc.meta +++ /dev/null @@ -1,7 +0,0 @@ -fileFormatVersion: 2 -guid: 3ce00ba677bb7e14bb91772fd68bfe6b -ShaderIncludeImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders.meta deleted file mode 100644 index 8a4ff2b..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders.meta +++ /dev/null @@ -1,8 +0,0 @@ -fileFormatVersion: 2 -guid: 8364d4f86c613ec4d999d062f5f773b8 -folderAsset: yes -DefaultImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_accumulate_pass.hlsl b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_accumulate_pass.hlsl deleted file mode 100644 index d2f1b32..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_accumulate_pass.hlsl +++ /dev/null @@ -1,79 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#define FSR3UPSCALER_BIND_SRV_INPUT_EXPOSURE 0 -#define FSR3UPSCALER_BIND_SRV_DILATED_REACTIVE_MASKS 1 -#if FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS -#define FSR3UPSCALER_BIND_SRV_DILATED_MOTION_VECTORS 2 -#else -#define FSR3UPSCALER_BIND_SRV_INPUT_MOTION_VECTORS 2 -#endif -#define FSR3UPSCALER_BIND_SRV_INTERNAL_UPSCALED 3 -#define FSR3UPSCALER_BIND_SRV_LOCK_STATUS 4 -#define FSR3UPSCALER_BIND_SRV_PREPARED_INPUT_COLOR 5 -#define FSR3UPSCALER_BIND_SRV_LANCZOS_LUT 6 -#define FSR3UPSCALER_BIND_SRV_UPSCALE_MAXIMUM_BIAS_LUT 7 -#define FSR3UPSCALER_BIND_SRV_SCENE_LUMINANCE_MIPS 8 -#define FSR3UPSCALER_BIND_SRV_AUTO_EXPOSURE 9 -#define FSR3UPSCALER_BIND_SRV_LUMA_HISTORY 10 - -#define FSR3UPSCALER_BIND_UAV_INTERNAL_UPSCALED 0 -#define FSR3UPSCALER_BIND_UAV_LOCK_STATUS 1 -#define FSR3UPSCALER_BIND_UAV_UPSCALED_OUTPUT 2 -#define FSR3UPSCALER_BIND_UAV_NEW_LOCKS 3 -#define FSR3UPSCALER_BIND_UAV_LUMA_HISTORY 4 - -#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 - -#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" -#include "fsr3upscaler/ffx_fsr3upscaler_common.h" -#include "fsr3upscaler/ffx_fsr3upscaler_sample.h" -#include "fsr3upscaler/ffx_fsr3upscaler_upsample.h" -#include "fsr3upscaler/ffx_fsr3upscaler_postprocess_lock_status.h" -#include "fsr3upscaler/ffx_fsr3upscaler_reproject.h" -#include "fsr3upscaler/ffx_fsr3upscaler_accumulate.h" - -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 8 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 8 -#endif // FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#ifndef FFX_FSR3UPSCALER_NUM_THREADS -#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] -#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS - -FFX_PREFER_WAVE64 -FFX_FSR3UPSCALER_NUM_THREADS -FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT -void CS(uint2 uGroupId : SV_GroupID, uint2 uGroupThreadId : SV_GroupThreadID) -{ - const uint GroupRows = (uint(DisplaySize().y) + FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT - 1) / FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT; - uGroupId.y = GroupRows - uGroupId.y - 1; - - uint2 uDispatchThreadId = uGroupId * uint2(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT) + uGroupThreadId; - - Accumulate(uDispatchThreadId); -} diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_accumulate_pass.hlsl.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_accumulate_pass.hlsl.meta deleted file mode 100644 index 80f209e..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_accumulate_pass.hlsl.meta +++ /dev/null @@ -1,7 +0,0 @@ -fileFormatVersion: 2 -guid: 507ab779c38eddb429cdcedf9c108d1b -ShaderIncludeImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_autogen_reactive_pass.hlsl b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_autogen_reactive_pass.hlsl deleted file mode 100644 index 0d6e2eb..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_autogen_reactive_pass.hlsl +++ /dev/null @@ -1,77 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#define FSR3UPSCALER_BIND_SRV_INPUT_OPAQUE_ONLY 0 -#define FSR3UPSCALER_BIND_SRV_INPUT_COLOR 1 - -#define FSR3UPSCALER_BIND_UAV_AUTOREACTIVE 0 - -#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 -#define FSR3UPSCALER_BIND_CB_REACTIVE 1 - -#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" -#include "fsr3upscaler/ffx_fsr3upscaler_common.h" - -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 8 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 8 -#endif // FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#ifndef FFX_FSR3UPSCALER_NUM_THREADS -#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] -#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS - -FFX_FSR3UPSCALER_NUM_THREADS -FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT -void CS(uint2 uGroupId : SV_GroupID, uint2 uGroupThreadId : SV_GroupThreadID) -{ - uint2 uDispatchThreadId = uGroupId * uint2(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT) + uGroupThreadId; - - float3 ColorPreAlpha = LoadOpaqueOnly( FFX_MIN16_I2(uDispatchThreadId) ).rgb; - float3 ColorPostAlpha = LoadInputColor(uDispatchThreadId).rgb; - - if (GenReactiveFlags() & FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_APPLY_TONEMAP) - { - ColorPreAlpha = Tonemap(ColorPreAlpha); - ColorPostAlpha = Tonemap(ColorPostAlpha); - } - - if (GenReactiveFlags() & FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_APPLY_INVERSETONEMAP) - { - ColorPreAlpha = InverseTonemap(ColorPreAlpha); - ColorPostAlpha = InverseTonemap(ColorPostAlpha); - } - - float out_reactive_value = 0.f; - float3 delta = abs(ColorPostAlpha - ColorPreAlpha); - - out_reactive_value = (GenReactiveFlags() & FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_USE_COMPONENTS_MAX) ? max(delta.x, max(delta.y, delta.z)) : length(delta); - out_reactive_value *= GenReactiveScale(); - - out_reactive_value = (GenReactiveFlags() & FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_APPLY_THRESHOLD) ? (out_reactive_value < GenReactiveThreshold() ? 0 : GenReactiveBinaryValue()) : out_reactive_value; - - rw_output_autoreactive[uDispatchThreadId] = out_reactive_value; -} diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_autogen_reactive_pass.hlsl.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_autogen_reactive_pass.hlsl.meta deleted file mode 100644 index c55f004..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_autogen_reactive_pass.hlsl.meta +++ /dev/null @@ -1,7 +0,0 @@ -fileFormatVersion: 2 -guid: 52cdb7a7c30cb614984908593ed19082 -ShaderIncludeImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_compute_luminance_pyramid_pass.hlsl b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_compute_luminance_pyramid_pass.hlsl deleted file mode 100644 index 93b7332..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_compute_luminance_pyramid_pass.hlsl +++ /dev/null @@ -1,55 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#define FSR3UPSCALER_BIND_SRV_INPUT_COLOR 0 - -#define FSR3UPSCALER_BIND_UAV_SPD_GLOBAL_ATOMIC 0 -#define FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_LUMA_CHANGE 1 -#define FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_5 2 -#define FSR3UPSCALER_BIND_UAV_AUTO_EXPOSURE 3 - -#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 -#define FSR3UPSCALER_BIND_CB_SPD 1 - -#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" -#include "fsr3upscaler/ffx_fsr3upscaler_common.h" -#include "fsr3upscaler/ffx_fsr3upscaler_compute_luminance_pyramid.h" - -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 256 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 1 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#ifndef FFX_FSR3UPSCALER_NUM_THREADS -#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] -#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS - -FFX_FSR3UPSCALER_NUM_THREADS -FFX_FSR3UPSCALER_EMBED_CB2_ROOTSIG_CONTENT -void CS(uint3 WorkGroupId : SV_GroupID, uint LocalThreadIndex : SV_GroupIndex) -{ - ComputeAutoExposure(WorkGroupId, LocalThreadIndex); -} diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_compute_luminance_pyramid_pass.hlsl.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_compute_luminance_pyramid_pass.hlsl.meta deleted file mode 100644 index 508b43e..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_compute_luminance_pyramid_pass.hlsl.meta +++ /dev/null @@ -1,7 +0,0 @@ -fileFormatVersion: 2 -guid: 2d149b52ba0f5bb468a94a71dbbcb66f -ShaderIncludeImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_depth_clip_pass.hlsl b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_depth_clip_pass.hlsl deleted file mode 100644 index 70cc7ba..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_depth_clip_pass.hlsl +++ /dev/null @@ -1,67 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#define FSR3UPSCALER_BIND_SRV_RECONSTRUCTED_PREV_NEAREST_DEPTH 0 -#define FSR3UPSCALER_BIND_SRV_DILATED_MOTION_VECTORS 1 -#define FSR3UPSCALER_BIND_SRV_DILATED_DEPTH 2 -#define FSR3UPSCALER_BIND_SRV_REACTIVE_MASK 3 -#define FSR3UPSCALER_BIND_SRV_TRANSPARENCY_AND_COMPOSITION_MASK 4 -#define FSR3UPSCALER_BIND_SRV_PREVIOUS_DILATED_MOTION_VECTORS 5 -#define FSR3UPSCALER_BIND_SRV_INPUT_MOTION_VECTORS 6 -#define FSR3UPSCALER_BIND_SRV_INPUT_COLOR 7 -#define FSR3UPSCALER_BIND_SRV_INPUT_DEPTH 8 -#define FSR3UPSCALER_BIND_SRV_INPUT_EXPOSURE 9 - -#define FSR3UPSCALER_BIND_UAV_DILATED_REACTIVE_MASKS 0 -#define FSR3UPSCALER_BIND_UAV_PREPARED_INPUT_COLOR 1 - -#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 - -#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" -#include "fsr3upscaler/ffx_fsr3upscaler_common.h" -#include "fsr3upscaler/ffx_fsr3upscaler_sample.h" -#include "fsr3upscaler/ffx_fsr3upscaler_depth_clip.h" - -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 8 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 8 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#ifndef FFX_FSR3UPSCALER_NUM_THREADS -#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] -#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS - -FFX_PREFER_WAVE64 -FFX_FSR3UPSCALER_NUM_THREADS -FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT -void CS( - int2 iGroupId : SV_GroupID, - int2 iDispatchThreadId : SV_DispatchThreadID, - int2 iGroupThreadId : SV_GroupThreadID, - int iGroupIndex : SV_GroupIndex) -{ - DepthClip(iDispatchThreadId); -} diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_depth_clip_pass.hlsl.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_depth_clip_pass.hlsl.meta deleted file mode 100644 index cde3a5e..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_depth_clip_pass.hlsl.meta +++ /dev/null @@ -1,7 +0,0 @@ -fileFormatVersion: 2 -guid: da435b71cf57e2247b80ae0f0f86d1f8 -ShaderIncludeImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_lock_pass.hlsl b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_lock_pass.hlsl deleted file mode 100644 index 26b28f0..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_lock_pass.hlsl +++ /dev/null @@ -1,56 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#define FSR3UPSCALER_BIND_SRV_LOCK_INPUT_LUMA 0 - -#define FSR3UPSCALER_BIND_UAV_NEW_LOCKS 0 -#define FSR3UPSCALER_BIND_UAV_RECONSTRUCTED_PREV_NEAREST_DEPTH 1 - -#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 - -#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" -#include "fsr3upscaler/ffx_fsr3upscaler_common.h" -#include "fsr3upscaler/ffx_fsr3upscaler_sample.h" -#include "fsr3upscaler/ffx_fsr3upscaler_lock.h" - -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 8 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 8 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#ifndef FFX_FSR3UPSCALER_NUM_THREADS -#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] -#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS - -FFX_PREFER_WAVE64 -FFX_FSR3UPSCALER_NUM_THREADS -FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT -void CS(uint2 uGroupId : SV_GroupID, uint2 uGroupThreadId : SV_GroupThreadID) -{ - uint2 uDispatchThreadId = uGroupId * uint2(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT) + uGroupThreadId; - - ComputeLock(uDispatchThreadId); -} diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_lock_pass.hlsl.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_lock_pass.hlsl.meta deleted file mode 100644 index 45c99dc..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_lock_pass.hlsl.meta +++ /dev/null @@ -1,7 +0,0 @@ -fileFormatVersion: 2 -guid: 98d2cbbda5e90dd4ebd1d70abbb63a09 -ShaderIncludeImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_rcas_pass.hlsl b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_rcas_pass.hlsl deleted file mode 100644 index bebdeb3..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_rcas_pass.hlsl +++ /dev/null @@ -1,53 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#define FSR3UPSCALER_BIND_SRV_INPUT_EXPOSURE 0 -#define FSR3UPSCALER_BIND_SRV_RCAS_INPUT 1 - -#define FSR3UPSCALER_BIND_UAV_UPSCALED_OUTPUT 0 - -#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 -#define FSR3UPSCALER_BIND_CB_RCAS 1 - -#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" -#include "fsr3upscaler/ffx_fsr3upscaler_common.h" -#include "fsr3upscaler/ffx_fsr3upscaler_rcas.h" - -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 64 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 1 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#ifndef FFX_FSR3UPSCALER_NUM_THREADS -#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] -#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS - -FFX_FSR3UPSCALER_NUM_THREADS -FFX_FSR3UPSCALER_EMBED_CB2_ROOTSIG_CONTENT -void CS(uint3 LocalThreadId : SV_GroupThreadID, uint3 WorkGroupId : SV_GroupID, uint3 Dtid : SV_DispatchThreadID) -{ - RCAS(LocalThreadId, WorkGroupId, Dtid); -} diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_rcas_pass.hlsl.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_rcas_pass.hlsl.meta deleted file mode 100644 index fb9bfe2..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_rcas_pass.hlsl.meta +++ /dev/null @@ -1,7 +0,0 @@ -fileFormatVersion: 2 -guid: 9a15fc73170a9bc478801c8fa4d8d574 -ShaderIncludeImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_reconstruct_previous_depth_pass.hlsl b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_reconstruct_previous_depth_pass.hlsl deleted file mode 100644 index f277fd1..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_reconstruct_previous_depth_pass.hlsl +++ /dev/null @@ -1,64 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#define FSR3UPSCALER_BIND_SRV_INPUT_MOTION_VECTORS 0 -#define FSR3UPSCALER_BIND_SRV_INPUT_DEPTH 1 -#define FSR3UPSCALER_BIND_SRV_INPUT_COLOR 2 -#define FSR3UPSCALER_BIND_SRV_INPUT_EXPOSURE 3 - -#define FSR3UPSCALER_BIND_UAV_RECONSTRUCTED_PREV_NEAREST_DEPTH 0 -#define FSR3UPSCALER_BIND_UAV_DILATED_MOTION_VECTORS 1 -#define FSR3UPSCALER_BIND_UAV_DILATED_DEPTH 2 -#define FSR3UPSCALER_BIND_UAV_LOCK_INPUT_LUMA 3 - -#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 - -#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" -#include "fsr3upscaler/ffx_fsr3upscaler_common.h" -#include "fsr3upscaler/ffx_fsr3upscaler_sample.h" -#include "fsr3upscaler/ffx_fsr3upscaler_reconstruct_dilated_velocity_and_previous_depth.h" - -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 8 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 8 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#ifndef FFX_FSR3UPSCALER_NUM_THREADS -#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] -#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS - -FFX_PREFER_WAVE64 -FFX_FSR3UPSCALER_NUM_THREADS -FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT -void CS( - int2 iGroupId : SV_GroupID, - int2 iDispatchThreadId : SV_DispatchThreadID, - int2 iGroupThreadId : SV_GroupThreadID, - int iGroupIndex : SV_GroupIndex -) -{ - ReconstructAndDilate(iDispatchThreadId); -} diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_reconstruct_previous_depth_pass.hlsl.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_reconstruct_previous_depth_pass.hlsl.meta deleted file mode 100644 index 6489d6d..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_reconstruct_previous_depth_pass.hlsl.meta +++ /dev/null @@ -1,7 +0,0 @@ -fileFormatVersion: 2 -guid: bafb3726a76b97a49bb343d8a4323754 -ShaderIncludeImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_tcr_autogen_pass.hlsl b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_tcr_autogen_pass.hlsl deleted file mode 100644 index 6180885..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_tcr_autogen_pass.hlsl +++ /dev/null @@ -1,90 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2022-2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -#define FSR3UPSCALER_BIND_SRV_INPUT_OPAQUE_ONLY 0 -#define FSR3UPSCALER_BIND_SRV_INPUT_COLOR 1 -#define FSR3UPSCALER_BIND_SRV_INPUT_MOTION_VECTORS 2 -#define FSR3UPSCALER_BIND_SRV_PREV_PRE_ALPHA_COLOR 3 -#define FSR3UPSCALER_BIND_SRV_PREV_POST_ALPHA_COLOR 4 -#define FSR3UPSCALER_BIND_SRV_REACTIVE_MASK 4 -#define FSR3UPSCALER_BIND_SRV_TRANSPARENCY_AND_COMPOSITION_MASK 5 - -#define FSR3UPSCALER_BIND_UAV_AUTOREACTIVE 0 -#define FSR3UPSCALER_BIND_UAV_AUTOCOMPOSITION 1 -#define FSR3UPSCALER_BIND_UAV_PREV_PRE_ALPHA_COLOR 2 -#define FSR3UPSCALER_BIND_UAV_PREV_POST_ALPHA_COLOR 3 - -#define FSR3UPSCALER_BIND_CB_FSR3UPSCALER 0 -#define FSR3UPSCALER_BIND_CB_AUTOREACTIVE 1 - -#include "fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h" -#include "fsr3upscaler/ffx_fsr3upscaler_common.h" -#include "fsr3upscaler/ffx_fsr3upscaler_tcr_autogen.h" - -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH 8 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#define FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT 8 -#endif // FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT -#ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#define FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH 1 -#endif // #ifndef FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH -#ifndef FFX_FSR3UPSCALER_NUM_THREADS -#define FFX_FSR3UPSCALER_NUM_THREADS [numthreads(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT, FFX_FSR3UPSCALER_THREAD_GROUP_DEPTH)] -#endif // #ifndef FFX_FSR3UPSCALER_NUM_THREADS - -FFX_FSR3UPSCALER_NUM_THREADS -FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT -void CS(uint2 uGroupId : SV_GroupID, uint2 uGroupThreadId : SV_GroupThreadID) -{ - FFX_MIN16_I2 uDispatchThreadId = FFX_MIN16_I2(uGroupId * uint2(FFX_FSR3UPSCALER_THREAD_GROUP_WIDTH, FFX_FSR3UPSCALER_THREAD_GROUP_HEIGHT) + uGroupThreadId); - - // ToDo: take into account jitter (i.e. add delta of previous jitter and current jitter to previous UV - // fetch pre- and post-alpha color values - FFX_MIN16_F2 fUv = ( FFX_MIN16_F2(uDispatchThreadId) + FFX_MIN16_F2(0.5f, 0.5f) ) / FFX_MIN16_F2( RenderSize() ); - FFX_MIN16_F2 fPrevUV = fUv + FFX_MIN16_F2( LoadInputMotionVector(uDispatchThreadId) ); - FFX_MIN16_I2 iPrevIdx = FFX_MIN16_I2(fPrevUV * FFX_MIN16_F2(RenderSize()) - 0.5f); - - FFX_MIN16_F3 colorPreAlpha = FFX_MIN16_F3( LoadOpaqueOnly( uDispatchThreadId ) ); - FFX_MIN16_F3 colorPostAlpha = FFX_MIN16_F3( LoadInputColor( uDispatchThreadId ) ); - - FFX_MIN16_F2 outReactiveMask = 0; - - outReactiveMask.y = ComputeTransparencyAndComposition(uDispatchThreadId, iPrevIdx); - - if (outReactiveMask.y > 0.5f) - { - outReactiveMask.x = ComputeReactive(uDispatchThreadId, iPrevIdx); - outReactiveMask.x *= FFX_MIN16_F(fReactiveScale); - outReactiveMask.x = outReactiveMask.x < fReactiveMax ? outReactiveMask.x : FFX_MIN16_F( fReactiveMax ); - } - - outReactiveMask.y *= FFX_MIN16_F(fTcScale ); - - outReactiveMask.x = max( outReactiveMask.x, FFX_MIN16_F( LoadReactiveMask(uDispatchThreadId) ) ); - outReactiveMask.y = max( outReactiveMask.y, FFX_MIN16_F( LoadTransparencyAndCompositionMask(uDispatchThreadId) ) ); - - StoreAutoReactive(uDispatchThreadId, outReactiveMask); - - StorePrevPreAlpha(uDispatchThreadId, colorPreAlpha); - StorePrevPostAlpha(uDispatchThreadId, colorPostAlpha); -} diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_tcr_autogen_pass.hlsl.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_tcr_autogen_pass.hlsl.meta deleted file mode 100644 index 02c5f46..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/ffx_fsr3upscaler_tcr_autogen_pass.hlsl.meta +++ /dev/null @@ -1,7 +0,0 @@ -fileFormatVersion: 2 -guid: 712d171118b59fc4fb28d0d487060d42 -ShaderIncludeImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler.meta deleted file mode 100644 index 2626a2e..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler.meta +++ /dev/null @@ -1,8 +0,0 @@ -fileFormatVersion: 2 -guid: ae9c6d015ae76544f9e8117e79ea862b -folderAsset: yes -DefaultImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_common_types.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_common_types.h deleted file mode 100644 index f0b62ab..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_common_types.h +++ /dev/null @@ -1,616 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#ifndef FFX_COMMON_TYPES_H -#define FFX_COMMON_TYPES_H - -#if defined(FFX_CPU) -#define FFX_PARAMETER_IN -#define FFX_PARAMETER_OUT -#define FFX_PARAMETER_INOUT -#define FFX_PARAMETER_UNIFORM -#elif defined(FFX_HLSL) -#define FFX_PARAMETER_IN in -#define FFX_PARAMETER_OUT out -#define FFX_PARAMETER_INOUT inout -#define FFX_PARAMETER_UNIFORM uniform -#elif defined(FFX_GLSL) -#define FFX_PARAMETER_IN in -#define FFX_PARAMETER_OUT out -#define FFX_PARAMETER_INOUT inout -#define FFX_PARAMETER_UNIFORM const //[cacao_placeholder] until a better fit is found! -#endif // #if defined(FFX_CPU) - -#if defined(FFX_CPU) -/// A typedef for a boolean value. -/// -/// @ingroup CPUTypes -typedef bool FfxBoolean; - -/// A typedef for a unsigned 8bit integer. -/// -/// @ingroup CPUTypes -typedef uint8_t FfxUInt8; - -/// A typedef for a unsigned 16bit integer. -/// -/// @ingroup CPUTypes -typedef uint16_t FfxUInt16; - -/// A typedef for a unsigned 32bit integer. -/// -/// @ingroup CPUTypes -typedef uint32_t FfxUInt32; - -/// A typedef for a unsigned 64bit integer. -/// -/// @ingroup CPUTypes -typedef uint64_t FfxUInt64; - -/// A typedef for a signed 8bit integer. -/// -/// @ingroup CPUTypes -typedef int8_t FfxInt8; - -/// A typedef for a signed 16bit integer. -/// -/// @ingroup CPUTypes -typedef int16_t FfxInt16; - -/// A typedef for a signed 32bit integer. -/// -/// @ingroup CPUTypes -typedef int32_t FfxInt32; - -/// A typedef for a signed 64bit integer. -/// -/// @ingroup CPUTypes -typedef int64_t FfxInt64; - -/// A typedef for a floating point value. -/// -/// @ingroup CPUTypes -typedef float FfxFloat32; - -/// A typedef for a 2-dimensional floating point value. -/// -/// @ingroup CPUTypes -typedef float FfxFloat32x2[2]; - -/// A typedef for a 3-dimensional floating point value. -/// -/// @ingroup CPUTypes -typedef float FfxFloat32x3[3]; - -/// A typedef for a 4-dimensional floating point value. -/// -/// @ingroup CPUTypes -typedef float FfxFloat32x4[4]; - -/// A typedef for a 2-dimensional 32bit unsigned integer. -/// -/// @ingroup CPUTypes -typedef uint32_t FfxUInt32x2[2]; - -/// A typedef for a 3-dimensional 32bit unsigned integer. -/// -/// @ingroup CPUTypes -typedef uint32_t FfxUInt32x3[3]; - -/// A typedef for a 4-dimensional 32bit unsigned integer. -/// -/// @ingroup CPUTypes -typedef uint32_t FfxUInt32x4[4]; -#endif // #if defined(FFX_CPU) - -#if defined(FFX_HLSL) - -#define FfxFloat32Mat4 matrix -#define FfxFloat32Mat3 matrix - -/// A typedef for a boolean value. -/// -/// @ingroup HLSLTypes -typedef bool FfxBoolean; - -#if FFX_HLSL_SM>=62 - -/// @defgroup HLSL62Types HLSL 6.2 And Above Types -/// HLSL 6.2 and above type defines for all commonly used variables -/// -/// @ingroup HLSLTypes - -/// A typedef for a floating point value. -/// -/// @ingroup HLSL62Types -typedef float32_t FfxFloat32; - -/// A typedef for a 2-dimensional floating point value. -/// -/// @ingroup HLSL62Types -typedef float32_t2 FfxFloat32x2; - -/// A typedef for a 3-dimensional floating point value. -/// -/// @ingroup HLSL62Types -typedef float32_t3 FfxFloat32x3; - -/// A typedef for a 4-dimensional floating point value. -/// -/// @ingroup HLSL62Types -typedef float32_t4 FfxFloat32x4; - -/// A [cacao_placeholder] typedef for matrix type until confirmed. -typedef float4x4 FfxFloat32x4x4; -typedef float3x3 FfxFloat32x3x3; -typedef float2x2 FfxFloat32x2x2; - -/// A typedef for a unsigned 32bit integer. -/// -/// @ingroup HLSL62Types -typedef uint32_t FfxUInt32; - -/// A typedef for a 2-dimensional 32bit unsigned integer. -/// -/// @ingroup HLSL62Types -typedef uint32_t2 FfxUInt32x2; - -/// A typedef for a 3-dimensional 32bit unsigned integer. -/// -/// @ingroup HLSL62Types -typedef uint32_t3 FfxUInt32x3; - -/// A typedef for a 4-dimensional 32bit unsigned integer. -/// -/// @ingroup HLSL62Types -typedef uint32_t4 FfxUInt32x4; - -/// A typedef for a signed 32bit integer. -/// -/// @ingroup HLSL62Types -typedef int32_t FfxInt32; - -/// A typedef for a 2-dimensional signed 32bit integer. -/// -/// @ingroup HLSL62Types -typedef int32_t2 FfxInt32x2; - -/// A typedef for a 3-dimensional signed 32bit integer. -/// -/// @ingroup HLSL62Types -typedef int32_t3 FfxInt32x3; - -/// A typedef for a 4-dimensional signed 32bit integer. -/// -/// @ingroup HLSL62Types -typedef int32_t4 FfxInt32x4; - -#else // #if FFX_HLSL_SM>=62 - -/// @defgroup HLSLBaseTypes HLSL 6.1 And Below Types -/// HLSL 6.1 and below type defines for all commonly used variables -/// -/// @ingroup HLSLTypes - -#define FfxFloat32 float -#define FfxFloat32x2 float2 -#define FfxFloat32x3 float3 -#define FfxFloat32x4 float4 - -/// A [cacao_placeholder] typedef for matrix type until confirmed. -#define FfxFloat32x4x4 float4x4 -#define FfxFloat32x3x3 float3x3 -#define FfxFloat32x2x2 float2x2 - -/// A typedef for a unsigned 32bit integer. -/// -/// @ingroup GPU -typedef uint FfxUInt32; -typedef uint2 FfxUInt32x2; -typedef uint3 FfxUInt32x3; -typedef uint4 FfxUInt32x4; - -typedef int FfxInt32; -typedef int2 FfxInt32x2; -typedef int3 FfxInt32x3; -typedef int4 FfxInt32x4; - -#endif // #if FFX_HLSL_SM>=62 - -#if FFX_HALF - -#if FFX_HLSL_SM >= 62 - -typedef float16_t FfxFloat16; -typedef float16_t2 FfxFloat16x2; -typedef float16_t3 FfxFloat16x3; -typedef float16_t4 FfxFloat16x4; - -/// A typedef for an unsigned 16bit integer. -/// -/// @ingroup HLSLTypes -typedef uint16_t FfxUInt16; -typedef uint16_t2 FfxUInt16x2; -typedef uint16_t3 FfxUInt16x3; -typedef uint16_t4 FfxUInt16x4; - -/// A typedef for a signed 16bit integer. -/// -/// @ingroup HLSLTypes -typedef int16_t FfxInt16; -typedef int16_t2 FfxInt16x2; -typedef int16_t3 FfxInt16x3; -typedef int16_t4 FfxInt16x4; -#elif SHADER_API_PSSL -#pragma argument(realtypes) // Enable true 16-bit types - -typedef half FfxFloat16; -typedef half2 FfxFloat16x2; -typedef half3 FfxFloat16x3; -typedef half4 FfxFloat16x4; - -/// A typedef for an unsigned 16bit integer. -/// -/// @ingroup GPU -typedef ushort FfxUInt16; -typedef ushort2 FfxUInt16x2; -typedef ushort3 FfxUInt16x3; -typedef ushort4 FfxUInt16x4; - -/// A typedef for a signed 16bit integer. -/// -/// @ingroup GPU -typedef short FfxInt16; -typedef short2 FfxInt16x2; -typedef short3 FfxInt16x3; -typedef short4 FfxInt16x4; -#else // #if FFX_HLSL_SM>=62 -typedef min16float FfxFloat16; -typedef min16float2 FfxFloat16x2; -typedef min16float3 FfxFloat16x3; -typedef min16float4 FfxFloat16x4; - -/// A typedef for an unsigned 16bit integer. -/// -/// @ingroup HLSLTypes -typedef min16uint FfxUInt16; -typedef min16uint2 FfxUInt16x2; -typedef min16uint3 FfxUInt16x3; -typedef min16uint4 FfxUInt16x4; - -/// A typedef for a signed 16bit integer. -/// -/// @ingroup HLSLTypes -typedef min16int FfxInt16; -typedef min16int2 FfxInt16x2; -typedef min16int3 FfxInt16x3; -typedef min16int4 FfxInt16x4; -#endif // #if FFX_HLSL_SM>=62 - -#endif // FFX_HALF - -#endif // #if defined(FFX_HLSL) - -#if defined(FFX_GLSL) - -#define FfxFloat32Mat4 mat4 -#define FfxFloat32Mat3 mat3 - -/// A typedef for a boolean value. -/// -/// @ingroup GLSLTypes -#define FfxBoolean bool -#define FfxFloat32 float -#define FfxFloat32x2 vec2 -#define FfxFloat32x3 vec3 -#define FfxFloat32x4 vec4 -#define FfxUInt32 uint -#define FfxUInt32x2 uvec2 -#define FfxUInt32x3 uvec3 -#define FfxUInt32x4 uvec4 -#define FfxInt32 int -#define FfxInt32x2 ivec2 -#define FfxInt32x3 ivec3 -#define FfxInt32x4 ivec4 - -/// A [cacao_placeholder] typedef for matrix type until confirmed. -#define FfxFloat32x4x4 mat4 -#define FfxFloat32x3x3 mat3 -#define FfxFloat32x2x2 mat2 - -#if FFX_HALF -#define FfxFloat16 float16_t -#define FfxFloat16x2 f16vec2 -#define FfxFloat16x3 f16vec3 -#define FfxFloat16x4 f16vec4 -#define FfxUInt16 uint16_t -#define FfxUInt16x2 u16vec2 -#define FfxUInt16x3 u16vec3 -#define FfxUInt16x4 u16vec4 -#define FfxInt16 int16_t -#define FfxInt16x2 i16vec2 -#define FfxInt16x3 i16vec3 -#define FfxInt16x4 i16vec4 -#endif // FFX_HALF -#endif // #if defined(FFX_GLSL) - -// Global toggles: -// #define FFX_HALF (1) -// #define FFX_HLSL_SM (62) - -#if FFX_HALF && !defined(SHADER_API_PSSL) - -#if FFX_HLSL_SM >= 62 - -#define FFX_MIN16_SCALAR( TypeName, BaseComponentType ) typedef BaseComponentType##16_t TypeName; -#define FFX_MIN16_VECTOR( TypeName, BaseComponentType, COL ) typedef vector TypeName; -#define FFX_MIN16_MATRIX( TypeName, BaseComponentType, ROW, COL ) typedef matrix TypeName; - -#define FFX_16BIT_SCALAR( TypeName, BaseComponentType ) typedef BaseComponentType##16_t TypeName; -#define FFX_16BIT_VECTOR( TypeName, BaseComponentType, COL ) typedef vector TypeName; -#define FFX_16BIT_MATRIX( TypeName, BaseComponentType, ROW, COL ) typedef matrix TypeName; - -#else //FFX_HLSL_SM>=62 - -#define FFX_MIN16_SCALAR( TypeName, BaseComponentType ) typedef min16##BaseComponentType TypeName; -#define FFX_MIN16_VECTOR( TypeName, BaseComponentType, COL ) typedef vector TypeName; -#define FFX_MIN16_MATRIX( TypeName, BaseComponentType, ROW, COL ) typedef matrix TypeName; - -#define FFX_16BIT_SCALAR( TypeName, BaseComponentType ) FFX_MIN16_SCALAR( TypeName, BaseComponentType ); -#define FFX_16BIT_VECTOR( TypeName, BaseComponentType, COL ) FFX_MIN16_VECTOR( TypeName, BaseComponentType, COL ); -#define FFX_16BIT_MATRIX( TypeName, BaseComponentType, ROW, COL ) FFX_MIN16_MATRIX( TypeName, BaseComponentType, ROW, COL ); - -#endif //FFX_HLSL_SM>=62 - -#else //FFX_HALF - -#define FFX_MIN16_SCALAR( TypeName, BaseComponentType ) typedef BaseComponentType TypeName; -#define FFX_MIN16_VECTOR( TypeName, BaseComponentType, COL ) typedef vector TypeName; -#define FFX_MIN16_MATRIX( TypeName, BaseComponentType, ROW, COL ) typedef matrix TypeName; - -#define FFX_16BIT_SCALAR( TypeName, BaseComponentType ) typedef BaseComponentType TypeName; -#define FFX_16BIT_VECTOR( TypeName, BaseComponentType, COL ) typedef vector TypeName; -#define FFX_16BIT_MATRIX( TypeName, BaseComponentType, ROW, COL ) typedef matrix TypeName; - -#endif //FFX_HALF - -#if defined(FFX_GPU) -// Common typedefs: -#if defined(FFX_HLSL) && !defined(SHADER_API_PSSL) -FFX_MIN16_SCALAR( FFX_MIN16_F , float ); -FFX_MIN16_VECTOR( FFX_MIN16_F2, float, 2 ); -FFX_MIN16_VECTOR( FFX_MIN16_F3, float, 3 ); -FFX_MIN16_VECTOR( FFX_MIN16_F4, float, 4 ); - -FFX_MIN16_SCALAR( FFX_MIN16_I, int ); -FFX_MIN16_VECTOR( FFX_MIN16_I2, int, 2 ); -FFX_MIN16_VECTOR( FFX_MIN16_I3, int, 3 ); -FFX_MIN16_VECTOR( FFX_MIN16_I4, int, 4 ); - -FFX_MIN16_SCALAR( FFX_MIN16_U, uint ); -FFX_MIN16_VECTOR( FFX_MIN16_U2, uint, 2 ); -FFX_MIN16_VECTOR( FFX_MIN16_U3, uint, 3 ); -FFX_MIN16_VECTOR( FFX_MIN16_U4, uint, 4 ); - -FFX_16BIT_SCALAR( FFX_F16_t , float ); -FFX_16BIT_VECTOR( FFX_F16_t2, float, 2 ); -FFX_16BIT_VECTOR( FFX_F16_t3, float, 3 ); -FFX_16BIT_VECTOR( FFX_F16_t4, float, 4 ); - -FFX_16BIT_SCALAR( FFX_I16_t, int ); -FFX_16BIT_VECTOR( FFX_I16_t2, int, 2 ); -FFX_16BIT_VECTOR( FFX_I16_t3, int, 3 ); -FFX_16BIT_VECTOR( FFX_I16_t4, int, 4 ); - -FFX_16BIT_SCALAR( FFX_U16_t, uint ); -FFX_16BIT_VECTOR( FFX_U16_t2, uint, 2 ); -FFX_16BIT_VECTOR( FFX_U16_t3, uint, 3 ); -FFX_16BIT_VECTOR( FFX_U16_t4, uint, 4 ); - -#define TYPEDEF_MIN16_TYPES(Prefix) \ -typedef FFX_MIN16_F Prefix##_F; \ -typedef FFX_MIN16_F2 Prefix##_F2; \ -typedef FFX_MIN16_F3 Prefix##_F3; \ -typedef FFX_MIN16_F4 Prefix##_F4; \ -typedef FFX_MIN16_I Prefix##_I; \ -typedef FFX_MIN16_I2 Prefix##_I2; \ -typedef FFX_MIN16_I3 Prefix##_I3; \ -typedef FFX_MIN16_I4 Prefix##_I4; \ -typedef FFX_MIN16_U Prefix##_U; \ -typedef FFX_MIN16_U2 Prefix##_U2; \ -typedef FFX_MIN16_U3 Prefix##_U3; \ -typedef FFX_MIN16_U4 Prefix##_U4; - -#define TYPEDEF_16BIT_TYPES(Prefix) \ -typedef FFX_16BIT_F Prefix##_F; \ -typedef FFX_16BIT_F2 Prefix##_F2; \ -typedef FFX_16BIT_F3 Prefix##_F3; \ -typedef FFX_16BIT_F4 Prefix##_F4; \ -typedef FFX_16BIT_I Prefix##_I; \ -typedef FFX_16BIT_I2 Prefix##_I2; \ -typedef FFX_16BIT_I3 Prefix##_I3; \ -typedef FFX_16BIT_I4 Prefix##_I4; \ -typedef FFX_16BIT_U Prefix##_U; \ -typedef FFX_16BIT_U2 Prefix##_U2; \ -typedef FFX_16BIT_U3 Prefix##_U3; \ -typedef FFX_16BIT_U4 Prefix##_U4; - -#define TYPEDEF_FULL_PRECISION_TYPES(Prefix) \ -typedef FfxFloat32 Prefix##_F; \ -typedef FfxFloat32x2 Prefix##_F2; \ -typedef FfxFloat32x3 Prefix##_F3; \ -typedef FfxFloat32x4 Prefix##_F4; \ -typedef FfxInt32 Prefix##_I; \ -typedef FfxInt32x2 Prefix##_I2; \ -typedef FfxInt32x3 Prefix##_I3; \ -typedef FfxInt32x4 Prefix##_I4; \ -typedef FfxUInt32 Prefix##_U; \ -typedef FfxUInt32x2 Prefix##_U2; \ -typedef FfxUInt32x3 Prefix##_U3; \ -typedef FfxUInt32x4 Prefix##_U4; -#endif // #if defined(FFX_HLSL) - -#if defined(SHADER_API_PSSL) - -#define unorm -#define globallycoherent - -#if FFX_HALF - -#define FFX_MIN16_F half -#define FFX_MIN16_F2 half2 -#define FFX_MIN16_F3 half3 -#define FFX_MIN16_F4 half4 - -#define FFX_MIN16_I short -#define FFX_MIN16_I2 short2 -#define FFX_MIN16_I3 short3 -#define FFX_MIN16_I4 short4 - -#define FFX_MIN16_U ushort -#define FFX_MIN16_U2 ushort2 -#define FFX_MIN16_U3 ushort3 -#define FFX_MIN16_U4 ushort4 - -#define FFX_16BIT_F half -#define FFX_16BIT_F2 half2 -#define FFX_16BIT_F3 half3 -#define FFX_16BIT_F4 half4 - -#define FFX_16BIT_I short -#define FFX_16BIT_I2 short2 -#define FFX_16BIT_I3 short3 -#define FFX_16BIT_I4 short4 - -#define FFX_16BIT_U ushort -#define FFX_16BIT_U2 ushort2 -#define FFX_16BIT_U3 ushort3 -#define FFX_16BIT_U4 ushort4 - -#else // FFX_HALF - -#define FFX_MIN16_F float -#define FFX_MIN16_F2 float2 -#define FFX_MIN16_F3 float3 -#define FFX_MIN16_F4 float4 - -#define FFX_MIN16_I int -#define FFX_MIN16_I2 int2 -#define FFX_MIN16_I3 int3 -#define FFX_MIN16_I4 int4 - -#define FFX_MIN16_U uint -#define FFX_MIN16_U2 uint2 -#define FFX_MIN16_U3 uint3 -#define FFX_MIN16_U4 uint4 - -#define FFX_16BIT_F float -#define FFX_16BIT_F2 float2 -#define FFX_16BIT_F3 float3 -#define FFX_16BIT_F4 float4 - -#define FFX_16BIT_I int -#define FFX_16BIT_I2 int2 -#define FFX_16BIT_I3 int3 -#define FFX_16BIT_I4 int4 - -#define FFX_16BIT_U uint -#define FFX_16BIT_U2 uint2 -#define FFX_16BIT_U3 uint3 -#define FFX_16BIT_U4 uint4 - -#endif // FFX_HALF - -#endif // #if defined(SHADER_API_PSSL) - -#if defined(FFX_GLSL) - -#if FFX_HALF - -#define FFX_MIN16_F float16_t -#define FFX_MIN16_F2 f16vec2 -#define FFX_MIN16_F3 f16vec3 -#define FFX_MIN16_F4 f16vec4 - -#define FFX_MIN16_I int16_t -#define FFX_MIN16_I2 i16vec2 -#define FFX_MIN16_I3 i16vec3 -#define FFX_MIN16_I4 i16vec4 - -#define FFX_MIN16_U uint16_t -#define FFX_MIN16_U2 u16vec2 -#define FFX_MIN16_U3 u16vec3 -#define FFX_MIN16_U4 u16vec4 - -#define FFX_16BIT_F float16_t -#define FFX_16BIT_F2 f16vec2 -#define FFX_16BIT_F3 f16vec3 -#define FFX_16BIT_F4 f16vec4 - -#define FFX_16BIT_I int16_t -#define FFX_16BIT_I2 i16vec2 -#define FFX_16BIT_I3 i16vec3 -#define FFX_16BIT_I4 i16vec4 - -#define FFX_16BIT_U uint16_t -#define FFX_16BIT_U2 u16vec2 -#define FFX_16BIT_U3 u16vec3 -#define FFX_16BIT_U4 u16vec4 - -#else // FFX_HALF - -#define FFX_MIN16_F float -#define FFX_MIN16_F2 vec2 -#define FFX_MIN16_F3 vec3 -#define FFX_MIN16_F4 vec4 - -#define FFX_MIN16_I int -#define FFX_MIN16_I2 ivec2 -#define FFX_MIN16_I3 ivec3 -#define FFX_MIN16_I4 ivec4 - -#define FFX_MIN16_U uint -#define FFX_MIN16_U2 uvec2 -#define FFX_MIN16_U3 uvec3 -#define FFX_MIN16_U4 uvec4 - -#define FFX_16BIT_F float -#define FFX_16BIT_F2 vec2 -#define FFX_16BIT_F3 vec3 -#define FFX_16BIT_F4 vec4 - -#define FFX_16BIT_I int -#define FFX_16BIT_I2 ivec2 -#define FFX_16BIT_I3 ivec3 -#define FFX_16BIT_I4 ivec4 - -#define FFX_16BIT_U uint -#define FFX_16BIT_U2 uvec2 -#define FFX_16BIT_U3 uvec3 -#define FFX_16BIT_U4 uvec4 - -#endif // FFX_HALF - -#endif // #if defined(FFX_GLSL) - -#endif // #if defined(FFX_GPU) -#endif // #ifndef FFX_COMMON_TYPES_H diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_common_types.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_common_types.h.meta deleted file mode 100644 index 0570595..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_common_types.h.meta +++ /dev/null @@ -1,60 +0,0 @@ -fileFormatVersion: 2 -guid: 7974b728d5c1b6d4a8a8e3965d03f96d -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude WebGL: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core.h deleted file mode 100644 index 02f6b3f..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core.h +++ /dev/null @@ -1,80 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -/// @defgroup FfxGPU GPU -/// The FidelityFX SDK GPU References -/// -/// @ingroup ffxSDK - -/// @defgroup FfxHLSL HLSL References -/// FidelityFX SDK HLSL GPU References -/// -/// @ingroup FfxGPU - -/// @defgroup FfxGLSL GLSL References -/// FidelityFX SDK GLSL GPU References -/// -/// @ingroup FfxGPU - -/// @defgroup FfxGPUEffects FidelityFX GPU References -/// FidelityFX Effect GPU Reference Documentation -/// -/// @ingroup FfxGPU - -/// @defgroup GPUCore GPU Core -/// GPU defines and functions -/// -/// @ingroup FfxGPU - -#if !defined(FFX_CORE_H) -#define FFX_CORE_H - -#ifdef __hlsl_dx_compiler -#pragma dxc diagnostic push -#pragma dxc diagnostic ignored "-Wambig-lit-shift" -#endif //__hlsl_dx_compiler - -#include "ffx_common_types.h" - -#if defined(FFX_CPU) - #include "ffx_core_cpu.h" -#endif // #if defined(FFX_CPU) - -#if defined(FFX_GLSL) && defined(FFX_GPU) - #include "ffx_core_glsl.h" -#endif // #if defined(FFX_GLSL) && defined(FFX_GPU) - -#if defined(FFX_HLSL) && defined(FFX_GPU) - #include "ffx_core_hlsl.h" -#endif // #if defined(FFX_HLSL) && defined(FFX_GPU) - -#if defined(FFX_GPU) - #include "ffx_core_gpu_common.h" - #include "ffx_core_gpu_common_half.h" - #include "ffx_core_portability.h" -#endif // #if defined(FFX_GPU) - -#ifdef __hlsl_dx_compiler -#pragma dxc diagnostic pop -#endif //__hlsl_dx_compiler - -#endif // #if !defined(FFX_CORE_H) diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core.h.meta deleted file mode 100644 index 8533462..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: b91c5f52b89ff554dacb51045a802ed8 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_cpu.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_cpu.h deleted file mode 100644 index 865258d..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_cpu.h +++ /dev/null @@ -1,338 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -/// A define for a true value in a boolean expression. -/// -/// @ingroup CPUTypes -#define FFX_TRUE (1) - -/// A define for a false value in a boolean expression. -/// -/// @ingroup CPUTypes -#define FFX_FALSE (0) - -#if !defined(FFX_STATIC) -/// A define to abstract declaration of static variables and functions. -/// -/// @ingroup CPUTypes -#define FFX_STATIC static -#endif // #if !defined(FFX_STATIC) - -/// @defgroup CPUCore CPU Core -/// Core CPU-side defines and functions -/// -/// @ingroup ffxHost - -#ifdef __clang__ -#pragma clang diagnostic ignored "-Wunused-variable" -#endif - -/// Interpret the bit layout of an IEEE-754 floating point value as an unsigned integer. -/// -/// @param [in] x A 32bit floating value. -/// -/// @returns -/// An unsigned 32bit integer value containing the bit pattern of x. -/// -/// @ingroup CPUCore -FFX_STATIC FfxUInt32 ffxAsUInt32(FfxFloat32 x) -{ - union - { - FfxFloat32 f; - FfxUInt32 u; - } bits; - - bits.f = x; - return bits.u; -} - -FFX_STATIC FfxFloat32 ffxDot2(FfxFloat32x2 a, FfxFloat32x2 b) -{ - return a[0] * b[0] + a[1] * b[1]; -} - -FFX_STATIC FfxFloat32 ffxDot3(FfxFloat32x3 a, FfxFloat32x3 b) -{ - return a[0] * b[0] + a[1] * b[1] + a[2] * b[2]; -} - -FFX_STATIC FfxFloat32 ffxDot4(FfxFloat32x4 a, FfxFloat32x4 b) -{ - return a[0] * b[0] + a[1] * b[1] + a[2] * b[2] + a[3] * b[3]; -} - -/// Compute the linear interopation between two values. -/// -/// Implemented by calling the GLSL mix instrinsic function. Implements the -/// following math: -/// -/// (1 - t) * x + t * y -/// -/// @param [in] x The first value to lerp between. -/// @param [in] y The second value to lerp between. -/// @param [in] t The value to determine how much of x and how much of y. -/// -/// @returns -/// A linearly interpolated value between x and y according to t. -/// -/// @ingroup CPUCore -FFX_STATIC FfxFloat32 ffxLerp(FfxFloat32 x, FfxFloat32 y, FfxFloat32 t) -{ - return y * t + (-x * t + x); -} - -/// Compute the reciprocal of a value. -/// -/// @param [in] x The value to compute the reciprocal for. -/// -/// @returns -/// The reciprocal value of x. -/// -/// @ingroup CPUCore -FFX_STATIC FfxFloat32 ffxReciprocal(FfxFloat32 x) -{ - return 1.0f / x; -} - -/// Compute the square root of a value. -/// -/// @param [in] x The first value to compute the min of. -/// -/// @returns -/// The the square root of x. -/// -/// @ingroup CPUCore -FFX_STATIC FfxFloat32 ffxSqrt(FfxFloat32 x) -{ - return sqrt(x); -} - -FFX_STATIC FfxUInt32 AShrSU1(FfxUInt32 a, FfxUInt32 b) -{ - return FfxUInt32(FfxInt32(a) >> FfxInt32(b)); -} - -/// Compute the factional part of a decimal value. -/// -/// This function calculates x - floor(x). -/// -/// @param [in] x The value to compute the fractional part from. -/// -/// @returns -/// The fractional part of x. -/// -/// @ingroup CPUCore -FFX_STATIC FfxFloat32 ffxFract(FfxFloat32 x) -{ - return x - floor(x); -} - -/// Compute the reciprocal square root of a value. -/// -/// @param [in] x The value to compute the reciprocal for. -/// -/// @returns -/// The reciprocal square root value of x. -/// -/// @ingroup CPUCore -FFX_STATIC FfxFloat32 rsqrt(FfxFloat32 x) -{ - return ffxReciprocal(ffxSqrt(x)); -} - -FFX_STATIC FfxFloat32 ffxMin(FfxFloat32 x, FfxFloat32 y) -{ - return x < y ? x : y; -} - -FFX_STATIC FfxUInt32 ffxMin(FfxUInt32 x, FfxUInt32 y) -{ - return x < y ? x : y; -} - -FFX_STATIC FfxFloat32 ffxMax(FfxFloat32 x, FfxFloat32 y) -{ - return x > y ? x : y; -} - -FFX_STATIC FfxUInt32 ffxMax(FfxUInt32 x, FfxUInt32 y) -{ - return x > y ? x : y; -} - -/// Clamp a value to a [0..1] range. -/// -/// @param [in] x The value to clamp to [0..1] range. -/// -/// @returns -/// The clamped version of x. -/// -/// @ingroup CPUCore -FFX_STATIC FfxFloat32 ffxSaturate(FfxFloat32 x) -{ - return ffxMin(1.0f, ffxMax(0.0f, x)); -} - -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// - -FFX_STATIC void opAAddOneF3(FfxFloat32x3 d, FfxFloat32x3 a, FfxFloat32 b) -{ - d[0] = a[0] + b; - d[1] = a[1] + b; - d[2] = a[2] + b; - return; -} - -FFX_STATIC void opACpyF3(FfxFloat32x3 d, FfxFloat32x3 a) -{ - d[0] = a[0]; - d[1] = a[1]; - d[2] = a[2]; - return; -} - -FFX_STATIC void opAMulF3(FfxFloat32x3 d, FfxFloat32x3 a, FfxFloat32x3 b) -{ - d[0] = a[0] * b[0]; - d[1] = a[1] * b[1]; - d[2] = a[2] * b[2]; - return; -} - -FFX_STATIC void opAMulOneF3(FfxFloat32x3 d, FfxFloat32x3 a, FfxFloat32 b) -{ - d[0] = a[0] * b; - d[1] = a[1] * b; - d[2] = a[2] * b; - return; -} - -FFX_STATIC void opARcpF3(FfxFloat32x3 d, FfxFloat32x3 a) -{ - d[0] = ffxReciprocal(a[0]); - d[1] = ffxReciprocal(a[1]); - d[2] = ffxReciprocal(a[2]); - return; -} - -/// Convert FfxFloat32 to half (in lower 16-bits of output). -/// -/// This function implements the same fast technique that is documented here: ftp://ftp.fox-toolkit.org/pub/fasthalffloatconversion.pdf -/// -/// The function supports denormals. -/// -/// Some conversion rules are to make computations possibly "safer" on the GPU, -/// -INF & -NaN -> -65504 -/// +INF & +NaN -> +65504 -/// -/// @param [in] f The 32bit floating point value to convert. -/// -/// @returns -/// The closest 16bit floating point value to f. -/// -/// @ingroup CPUCore -FFX_STATIC FfxUInt32 f32tof16(FfxFloat32 f) -{ - static FfxUInt16 base[512] = { - 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, - 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, - 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, - 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, - 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, - 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0001, 0x0002, 0x0004, 0x0008, 0x0010, 0x0020, 0x0040, 0x0080, 0x0100, 0x0200, 0x0400, - 0x0800, 0x0c00, 0x1000, 0x1400, 0x1800, 0x1c00, 0x2000, 0x2400, 0x2800, 0x2c00, 0x3000, 0x3400, 0x3800, 0x3c00, 0x4000, 0x4400, 0x4800, 0x4c00, 0x5000, - 0x5400, 0x5800, 0x5c00, 0x6000, 0x6400, 0x6800, 0x6c00, 0x7000, 0x7400, 0x7800, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, - 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, - 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, - 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, - 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, - 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, - 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x7bff, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, - 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, - 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, - 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, - 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, - 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8000, 0x8001, 0x8002, - 0x8004, 0x8008, 0x8010, 0x8020, 0x8040, 0x8080, 0x8100, 0x8200, 0x8400, 0x8800, 0x8c00, 0x9000, 0x9400, 0x9800, 0x9c00, 0xa000, 0xa400, 0xa800, 0xac00, - 0xb000, 0xb400, 0xb800, 0xbc00, 0xc000, 0xc400, 0xc800, 0xcc00, 0xd000, 0xd400, 0xd800, 0xdc00, 0xe000, 0xe400, 0xe800, 0xec00, 0xf000, 0xf400, 0xf800, - 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, - 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, - 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, - 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, - 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, - 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff, 0xfbff - }; - - static FfxUInt8 shift[512] = { - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x17, 0x16, 0x15, 0x14, 0x13, 0x12, 0x11, 0x10, 0x0f, 0x0e, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, - 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x17, 0x16, 0x15, 0x14, 0x13, 0x12, 0x11, 0x10, 0x0f, 0x0e, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, - 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x0d, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, - 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18 - }; - - union - { - FfxFloat32 f; - FfxUInt32 u; - } bits; - - bits.f = f; - FfxUInt32 u = bits.u; - FfxUInt32 i = u >> 23; - return (FfxUInt32)(base[i]) + ((u & 0x7fffff) >> shift[i]); -} - -/// Pack 2x32-bit floating point values in a single 32bit value. -/// -/// This function first converts each component of value into their nearest 16-bit floating -/// point representation, and then stores the X and Y components in the lower and upper 16 bits of the -/// 32bit unsigned integer respectively. -/// -/// @param [in] x A 2-dimensional floating point value to convert and pack. -/// -/// @returns -/// A packed 32bit value containing 2 16bit floating point values. -/// -/// @ingroup CPUCore -FFX_STATIC FfxUInt32 packHalf2x16(FfxFloat32x2 x) -{ - return f32tof16(x[0]) + (f32tof16(x[1]) << 16); -} diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_cpu.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_cpu.h.meta deleted file mode 100644 index f6508bc..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_cpu.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 4c88c0b7a4dec1e479272449c19ca981 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common.h deleted file mode 100644 index 2f687df..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common.h +++ /dev/null @@ -1,2784 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -/// A define for a true value in a boolean expression. -/// -/// @ingroup GPUCore -#define FFX_TRUE (true) - -/// A define for a false value in a boolean expression. -/// -/// @ingroup GPUCore -#define FFX_FALSE (false) - -/// A define value for positive infinity. -/// -/// @ingroup GPUCore -#define FFX_POSITIVE_INFINITY_FLOAT ffxAsFloat(0x7f800000u) - -/// A define value for negative infinity. -/// -/// @ingroup GPUCore -#define FFX_NEGATIVE_INFINITY_FLOAT ffxAsFloat(0xff800000u) - -/// A define value for PI. -/// -/// @ingroup GPUCore -#define FFX_PI (3.14159) - -FFX_STATIC const FfxFloat32 FFX_FP16_MIN = 6.10e-05f; -FFX_STATIC const FfxFloat32 FFX_FP16_MAX = 65504.0f; -FFX_STATIC const FfxFloat32 FFX_TONEMAP_EPSILON = 1.0f / FFX_FP16_MAX; - -/// Compute the reciprocal of value. -/// -/// @param [in] value The value to compute the reciprocal of. -/// -/// @returns -/// The 1 / value. -/// -/// @ingroup GPUCore -FfxFloat32 ffxReciprocal(FfxFloat32 value) -{ - return rcp(value); -} - -/// Compute the reciprocal of value. -/// -/// @param [in] value The value to compute the reciprocal of. -/// -/// @returns -/// The 1 / value. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxReciprocal(FfxFloat32x2 value) -{ - return rcp(value); -} - -/// Compute the reciprocal of value. -/// -/// @param [in] value The value to compute the reciprocal of. -/// -/// @returns -/// The 1 / value. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxReciprocal(FfxFloat32x3 value) -{ - return rcp(value); -} - -/// Compute the reciprocal of value. -/// -/// @param [in] value The value to compute the reciprocal of. -/// -/// @returns -/// The 1 / value. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxReciprocal(FfxFloat32x4 value) -{ - return rcp(value); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat32 ffxMin(FfxFloat32 x, FfxFloat32 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxMin(FfxFloat32x2 x, FfxFloat32x2 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxMin(FfxFloat32x3 x, FfxFloat32x3 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxMin(FfxFloat32x4 x, FfxFloat32x4 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt32 ffxMin(FfxInt32 x, FfxInt32 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt32x2 ffxMin(FfxInt32x2 x, FfxInt32x2 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt32x3 ffxMin(FfxInt32x3 x, FfxInt32x3 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt32x4 ffxMin(FfxInt32x4 x, FfxInt32x4 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt32 ffxMin(FfxUInt32 x, FfxUInt32 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxMin(FfxUInt32x2 x, FfxUInt32x2 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt32x3 ffxMin(FfxUInt32x3 x, FfxUInt32x3 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt32x4 ffxMin(FfxUInt32x4 x, FfxUInt32x4 y) -{ - return min(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat32 ffxMax(FfxFloat32 x, FfxFloat32 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxMax(FfxFloat32x2 x, FfxFloat32x2 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxMax(FfxFloat32x3 x, FfxFloat32x3 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxMax(FfxFloat32x4 x, FfxFloat32x4 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt32 ffxMax(FfxInt32 x, FfxInt32 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt32x2 ffxMax(FfxInt32x2 x, FfxInt32x2 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt32x3 ffxMax(FfxInt32x3 x, FfxInt32x3 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt32x4 ffxMax(FfxInt32x4 x, FfxInt32x4 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt32 ffxMax(FfxUInt32 x, FfxUInt32 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxMax(FfxUInt32x2 x, FfxUInt32x2 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt32x3 ffxMax(FfxUInt32x3 x, FfxUInt32x3 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt32x4 ffxMax(FfxUInt32x4 x, FfxUInt32x4 y) -{ - return max(x, y); -} - -/// Compute the value of the first parameter raised to the power of the second. -/// -/// @param [in] x The value to raise to the power y. -/// @param [in] y The power to which to raise x. -/// -/// @returns -/// The value of the first parameter raised to the power of the second. -/// -/// @ingroup GPUCore -FfxFloat32 ffxPow(FfxFloat32 x, FfxFloat32 y) -{ - return pow(x, y); -} - -/// Compute the value of the first parameter raised to the power of the second. -/// -/// @param [in] x The value to raise to the power y. -/// @param [in] y The power to which to raise x. -/// -/// @returns -/// The value of the first parameter raised to the power of the second. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxPow(FfxFloat32x2 x, FfxFloat32x2 y) -{ - return pow(x, y); -} - -/// Compute the value of the first parameter raised to the power of the second. -/// -/// @param [in] x The value to raise to the power y. -/// @param [in] y The power to which to raise x. -/// -/// @returns -/// The value of the first parameter raised to the power of the second. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxPow(FfxFloat32x3 x, FfxFloat32x3 y) -{ - return pow(x, y); -} - -/// Compute the value of the first parameter raised to the power of the second. -/// -/// @param [in] x The value to raise to the power y. -/// @param [in] y The power to which to raise x. -/// -/// @returns -/// The value of the first parameter raised to the power of the second. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxPow(FfxFloat32x4 x, FfxFloat32x4 y) -{ - return pow(x, y); -} - -/// Compute the square root of a value. -/// -/// @param [in] x The first value to compute the min of. -/// -/// @returns -/// The the square root of x. -/// -/// @ingroup GPUCore -FfxFloat32 ffxSqrt(FfxFloat32 x) -{ - return sqrt(x); -} - -/// Compute the square root of a value. -/// -/// @param [in] x The first value to compute the min of. -/// -/// @returns -/// The the square root of x. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxSqrt(FfxFloat32x2 x) -{ - return sqrt(x); -} - -/// Compute the square root of a value. -/// -/// @param [in] x The first value to compute the min of. -/// -/// @returns -/// The the square root of x. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxSqrt(FfxFloat32x3 x) -{ - return sqrt(x); -} - -/// Compute the square root of a value. -/// -/// @param [in] x The first value to compute the min of. -/// -/// @returns -/// The the square root of x. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxSqrt(FfxFloat32x4 x) -{ - return sqrt(x); -} - -/// Copy the sign bit from 's' to positive 'd'. -/// -/// @param [in] d The value to copy the sign bit into. -/// @param [in] s The value to copy the sign bit from. -/// -/// @returns -/// The value of d with the sign bit from s. -/// -/// @ingroup GPUCore -FfxFloat32 ffxCopySignBit(FfxFloat32 d, FfxFloat32 s) -{ - return ffxAsFloat(ffxAsUInt32(d) | (ffxAsUInt32(s) & FfxUInt32(0x80000000u))); -} - -/// Copy the sign bit from 's' to positive 'd'. -/// -/// @param [in] d The value to copy the sign bit into. -/// @param [in] s The value to copy the sign bit from. -/// -/// @returns -/// The value of d with the sign bit from s. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxCopySignBit(FfxFloat32x2 d, FfxFloat32x2 s) -{ - return ffxAsFloat(ffxAsUInt32(d) | (ffxAsUInt32(s) & ffxBroadcast2(0x80000000u))); -} - -/// Copy the sign bit from 's' to positive 'd'. -/// -/// @param [in] d The value to copy the sign bit into. -/// @param [in] s The value to copy the sign bit from. -/// -/// @returns -/// The value of d with the sign bit from s. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxCopySignBit(FfxFloat32x3 d, FfxFloat32x3 s) -{ - return ffxAsFloat(ffxAsUInt32(d) | (ffxAsUInt32(s) & ffxBroadcast3(0x80000000u))); -} - -/// Copy the sign bit from 's' to positive 'd'. -/// -/// @param [in] d The value to copy the sign bit into. -/// @param [in] s The value to copy the sign bit from. -/// -/// @returns -/// The value of d with the sign bit from s. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxCopySignBit(FfxFloat32x4 d, FfxFloat32x4 s) -{ - return ffxAsFloat(ffxAsUInt32(d) | (ffxAsUInt32(s) & ffxBroadcast4(0x80000000u))); -} - -/// A single operation to return the following: -/// m = NaN := 0 -/// m >= 0 := 0 -/// m < 0 := 1 -/// -/// Uses the following useful floating point logic, -/// saturate(+a*(-INF)==-INF) := 0 -/// saturate( 0*(-INF)== NaN) := 0 -/// saturate(-a*(-INF)==+INF) := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against 0. -/// -/// @returns -/// 1.0 when the value is negative, or 0.0 when the value is 0 or position. -/// -/// @ingroup GPUCore -FfxFloat32 ffxIsSigned(FfxFloat32 m) -{ - return ffxSaturate(m * FfxFloat32(FFX_NEGATIVE_INFINITY_FLOAT)); -} - -/// A single operation to return the following: -/// m = NaN := 0 -/// m >= 0 := 0 -/// m < 0 := 1 -/// -/// Uses the following useful floating point logic, -/// saturate(+a*(-INF)==-INF) := 0 -/// saturate( 0*(-INF)== NaN) := 0 -/// saturate(-a*(-INF)==+INF) := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against 0. -/// -/// @returns -/// 1.0 when the value is negative, or 0.0 when the value is 0 or position. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxIsSigned(FfxFloat32x2 m) -{ - return ffxSaturate(m * ffxBroadcast2(FFX_NEGATIVE_INFINITY_FLOAT)); -} - -/// A single operation to return the following: -/// m = NaN := 0 -/// m >= 0 := 0 -/// m < 0 := 1 -/// -/// Uses the following useful floating point logic, -/// saturate(+a*(-INF)==-INF) := 0 -/// saturate( 0*(-INF)== NaN) := 0 -/// saturate(-a*(-INF)==+INF) := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against 0. -/// -/// @returns -/// 1.0 when the value is negative, or 0.0 when the value is 0 or position. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxIsSigned(FfxFloat32x3 m) -{ - return ffxSaturate(m * ffxBroadcast3(FFX_NEGATIVE_INFINITY_FLOAT)); -} - -/// A single operation to return the following: -/// m = NaN := 0 -/// m >= 0 := 0 -/// m < 0 := 1 -/// -/// Uses the following useful floating point logic, -/// saturate(+a*(-INF)==-INF) := 0 -/// saturate( 0*(-INF)== NaN) := 0 -/// saturate(-a*(-INF)==+INF) := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against for have the sign set. -/// -/// @returns -/// 1.0 when the value is negative, or 0.0 when the value is 0 or positive. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxIsSigned(FfxFloat32x4 m) -{ - return ffxSaturate(m * ffxBroadcast4(FFX_NEGATIVE_INFINITY_FLOAT)); -} - -/// A single operation to return the following: -/// m = NaN := 1 -/// m > 0 := 0 -/// m <= 0 := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against zero. -/// -/// @returns -/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. -/// -/// @ingroup GPUCore -FfxFloat32 ffxIsGreaterThanZero(FfxFloat32 m) -{ - return ffxSaturate(m * FfxFloat32(FFX_POSITIVE_INFINITY_FLOAT)); -} - -/// A single operation to return the following: -/// m = NaN := 1 -/// m > 0 := 0 -/// m <= 0 := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against zero. -/// -/// @returns -/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxIsGreaterThanZero(FfxFloat32x2 m) -{ - return ffxSaturate(m * ffxBroadcast2(FFX_POSITIVE_INFINITY_FLOAT)); -} - -/// A single operation to return the following: -/// m = NaN := 1 -/// m > 0 := 0 -/// m <= 0 := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against zero. -/// -/// @returns -/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxIsGreaterThanZero(FfxFloat32x3 m) -{ - return ffxSaturate(m * ffxBroadcast3(FFX_POSITIVE_INFINITY_FLOAT)); -} - -/// A single operation to return the following: -/// m = NaN := 1 -/// m > 0 := 0 -/// m <= 0 := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against zero. -/// -/// @returns -/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxIsGreaterThanZero(FfxFloat32x4 m) -{ - return ffxSaturate(m * ffxBroadcast4(FFX_POSITIVE_INFINITY_FLOAT)); -} - -/// Convert a 32bit floating point value to sortable integer. -/// -/// - If sign bit=0, flip the sign bit (positives). -/// - If sign bit=1, flip all bits (negatives). -/// -/// The function has the side effects that: -/// - Larger integers are more positive values. -/// - Float zero is mapped to center of integers (so clear to integer zero is a nice default for atomic max usage). -/// -/// @param [in] value The floating point value to make sortable. -/// -/// @returns -/// The sortable integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxFloatToSortableInteger(FfxUInt32 value) -{ - return value ^ ((AShrSU1(value, FfxUInt32(31))) | FfxUInt32(0x80000000)); -} - -/// Convert a sortable integer to a 32bit floating point value. -/// -/// The function has the side effects that: -/// - If sign bit=1, flip the sign bit (positives). -/// - If sign bit=0, flip all bits (negatives). -/// -/// @param [in] value The floating point value to make sortable. -/// -/// @returns -/// The sortable integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxSortableIntegerToFloat(FfxUInt32 value) -{ - return value ^ ((~AShrSU1(value, FfxUInt32(31))) | FfxUInt32(0x80000000)); -} - -/// Calculate a low-quality approximation for the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the square root for. -/// -/// @returns -/// An approximation of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximateSqrt(FfxFloat32 value) -{ - return ffxAsFloat((ffxAsUInt32(value) >> FfxUInt32(1)) + FfxUInt32(0x1fbc4639)); -} - -/// Calculate a low-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximateReciprocal(FfxFloat32 value) -{ - return ffxAsFloat(FfxUInt32(0x7ef07ebb) - ffxAsUInt32(value)); -} - -/// Calculate a medium-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to medium quality. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximateReciprocalMedium(FfxFloat32 value) -{ - FfxFloat32 b = ffxAsFloat(FfxUInt32(0x7ef19fff) - ffxAsUInt32(value)); - return b * (-b * value + FfxFloat32(2.0)); -} - -/// Calculate a low-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the reciprocal square root for. -/// -/// @returns -/// An approximation of the reciprocal square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximateReciprocalSquareRoot(FfxFloat32 value) -{ - return ffxAsFloat(FfxUInt32(0x5f347d74) - (ffxAsUInt32(value) >> FfxUInt32(1))); -} - -/// Calculate a low-quality approximation for the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the square root for. -/// -/// @returns -/// An approximation of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximateSqrt(FfxFloat32x2 value) -{ - return ffxAsFloat((ffxAsUInt32(value) >> ffxBroadcast2(1u)) + ffxBroadcast2(0x1fbc4639u)); -} - -/// Calculate a low-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximateReciprocal(FfxFloat32x2 value) -{ - return ffxAsFloat(ffxBroadcast2(0x7ef07ebbu) - ffxAsUInt32(value)); -} - -/// Calculate a medium-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to medium quality. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximateReciprocalMedium(FfxFloat32x2 value) -{ - FfxFloat32x2 b = ffxAsFloat(ffxBroadcast2(0x7ef19fffu) - ffxAsUInt32(value)); - return b * (-b * value + ffxBroadcast2(2.0f)); -} - -/// Calculate a low-quality approximation for the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the square root for. -/// -/// @returns -/// An approximation of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximateReciprocalSquareRoot(FfxFloat32x2 value) -{ - return ffxAsFloat(ffxBroadcast2(0x5f347d74u) - (ffxAsUInt32(value) >> ffxBroadcast2(1u))); -} - -/// Calculate a low-quality approximation for the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the square root for. -/// -/// @returns -/// An approximation of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximateSqrt(FfxFloat32x3 value) -{ - return ffxAsFloat((ffxAsUInt32(value) >> ffxBroadcast3(1u)) + ffxBroadcast3(0x1fbc4639u)); -} - -/// Calculate a low-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximateReciprocal(FfxFloat32x3 value) -{ - return ffxAsFloat(ffxBroadcast3(0x7ef07ebbu) - ffxAsUInt32(value)); -} - -/// Calculate a medium-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to medium quality. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximateReciprocalMedium(FfxFloat32x3 value) -{ - FfxFloat32x3 b = ffxAsFloat(ffxBroadcast3(0x7ef19fffu) - ffxAsUInt32(value)); - return b * (-b * value + ffxBroadcast3(2.0f)); -} - -/// Calculate a low-quality approximation for the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the square root for. -/// -/// @returns -/// An approximation of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximateReciprocalSquareRoot(FfxFloat32x3 value) -{ - return ffxAsFloat(ffxBroadcast3(0x5f347d74u) - (ffxAsUInt32(value) >> ffxBroadcast3(1u))); -} - -/// Calculate a low-quality approximation for the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the square root for. -/// -/// @returns -/// An approximation of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximateSqrt(FfxFloat32x4 value) -{ - return ffxAsFloat((ffxAsUInt32(value) >> ffxBroadcast4(1u)) + ffxBroadcast4(0x1fbc4639u)); -} - -/// Calculate a low-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximateReciprocal(FfxFloat32x4 value) -{ - return ffxAsFloat(ffxBroadcast4(0x7ef07ebbu) - ffxAsUInt32(value)); -} - -/// Calculate a medium-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to medium quality. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximateReciprocalMedium(FfxFloat32x4 value) -{ - FfxFloat32x4 b = ffxAsFloat(ffxBroadcast4(0x7ef19fffu) - ffxAsUInt32(value)); - return b * (-b * value + ffxBroadcast4(2.0f)); -} - -/// Calculate a low-quality approximation for the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] value The value to calculate an approximate to the square root for. -/// -/// @returns -/// An approximation of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximateReciprocalSquareRoot(FfxFloat32x4 value) -{ - return ffxAsFloat(ffxBroadcast4(0x5f347d74u) - (ffxAsUInt32(value) >> ffxBroadcast4(1u))); -} - -/// Calculate dot product of 'a' and 'b'. -/// -/// @param [in] a First vector input. -/// @param [in] b Second vector input. -/// -/// @returns -/// The value of a dot b. -/// -/// @ingroup GPUCore -FfxFloat32 ffxDot2(FfxFloat32x2 a, FfxFloat32x2 b) -{ - return dot(a, b); -} - -/// Calculate dot product of 'a' and 'b'. -/// -/// @param [in] a First vector input. -/// @param [in] b Second vector input. -/// -/// @returns -/// The value of a dot b. -/// -/// @ingroup GPUCore -FfxFloat32 ffxDot3(FfxFloat32x3 a, FfxFloat32x3 b) -{ - return dot(a, b); -} - -/// Calculate dot product of 'a' and 'b'. -/// -/// @param [in] a First vector input. -/// @param [in] b Second vector input. -/// -/// @returns -/// The value of a dot b. -/// -/// @ingroup GPUCore -FfxFloat32 ffxDot4(FfxFloat32x4 a, FfxFloat32x4 b) -{ - return dot(a, b); -} - - -/// Compute an approximate conversion from PQ to Gamma2 space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between PQ and Gamma2. -/// -/// @returns -/// The value a converted into Gamma2. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximatePQToGamma2Medium(FfxFloat32 a) -{ - return a * a * a * a; -} - -/// Compute an approximate conversion from PQ to linear space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between PQ and linear. -/// -/// @returns -/// The value a converted into linear. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximatePQToLinear(FfxFloat32 a) -{ - return a * a * a * a * a * a * a * a; -} - -/// Compute an approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximateGamma2ToPQ(FfxFloat32 a) -{ - return ffxAsFloat((ffxAsUInt32(a) >> FfxUInt32(2)) + FfxUInt32(0x2F9A4E46)); -} - -/// Compute a more accurate approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximateGamma2ToPQMedium(FfxFloat32 a) -{ - FfxFloat32 b = ffxAsFloat((ffxAsUInt32(a) >> FfxUInt32(2)) + FfxUInt32(0x2F9A4E46)); - FfxFloat32 b4 = b * b * b * b; - return b - b * (b4 - a) / (FfxFloat32(4.0) * b4); -} - -/// Compute a high accuracy approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximateGamma2ToPQHigh(FfxFloat32 a) -{ - return ffxSqrt(ffxSqrt(a)); -} - -/// Compute an approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximateLinearToPQ(FfxFloat32 a) -{ - return ffxAsFloat((ffxAsUInt32(a) >> FfxUInt32(3)) + FfxUInt32(0x378D8723)); -} - -/// Compute a more accurate approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximateLinearToPQMedium(FfxFloat32 a) -{ - FfxFloat32 b = ffxAsFloat((ffxAsUInt32(a) >> FfxUInt32(3)) + FfxUInt32(0x378D8723)); - FfxFloat32 b8 = b * b * b * b * b * b * b * b; - return b - b * (b8 - a) / (FfxFloat32(8.0) * b8); -} - -/// Compute a very accurate approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32 ffxApproximateLinearToPQHigh(FfxFloat32 a) -{ - return ffxSqrt(ffxSqrt(ffxSqrt(a))); -} - -/// Compute an approximate conversion from PQ to Gamma2 space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between PQ and Gamma2. -/// -/// @returns -/// The value a converted into Gamma2. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximatePQToGamma2Medium(FfxFloat32x2 a) -{ - return a * a * a * a; -} - -/// Compute an approximate conversion from PQ to linear space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between PQ and linear. -/// -/// @returns -/// The value a converted into linear. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximatePQToLinear(FfxFloat32x2 a) -{ - return a * a * a * a * a * a * a * a; -} - -/// Compute an approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximateGamma2ToPQ(FfxFloat32x2 a) -{ - return ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast2(2u)) + ffxBroadcast2(0x2F9A4E46u)); -} - -/// Compute a more accurate approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximateGamma2ToPQMedium(FfxFloat32x2 a) -{ - FfxFloat32x2 b = ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast2(2u)) + ffxBroadcast2(0x2F9A4E46u)); - FfxFloat32x2 b4 = b * b * b * b; - return b - b * (b4 - a) / (FfxFloat32(4.0) * b4); -} - -/// Compute a high accuracy approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximateGamma2ToPQHigh(FfxFloat32x2 a) -{ - return ffxSqrt(ffxSqrt(a)); -} - -/// Compute an approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximateLinearToPQ(FfxFloat32x2 a) -{ - return ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast2(3u)) + ffxBroadcast2(0x378D8723u)); -} - -/// Compute a more accurate approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximateLinearToPQMedium(FfxFloat32x2 a) -{ - FfxFloat32x2 b = ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast2(3u)) + ffxBroadcast2(0x378D8723u)); - FfxFloat32x2 b8 = b * b * b * b * b * b * b * b; - return b - b * (b8 - a) / (FfxFloat32(8.0) * b8); -} - -/// Compute a very accurate approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxApproximateLinearToPQHigh(FfxFloat32x2 a) -{ - return ffxSqrt(ffxSqrt(ffxSqrt(a))); -} - -/// Compute an approximate conversion from PQ to Gamma2 space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between PQ and Gamma2. -/// -/// @returns -/// The value a converted into Gamma2. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximatePQToGamma2Medium(FfxFloat32x3 a) -{ - return a * a * a * a; -} - -/// Compute an approximate conversion from PQ to linear space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between PQ and linear. -/// -/// @returns -/// The value a converted into linear. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximatePQToLinear(FfxFloat32x3 a) -{ - return a * a * a * a * a * a * a * a; -} - -/// Compute an approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximateGamma2ToPQ(FfxFloat32x3 a) -{ - return ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast3(2u)) + ffxBroadcast3(0x2F9A4E46u)); -} - -/// Compute a more accurate approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximateGamma2ToPQMedium(FfxFloat32x3 a) -{ - FfxFloat32x3 b = ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast3(2u)) + ffxBroadcast3(0x2F9A4E46u)); - FfxFloat32x3 b4 = b * b * b * b; - return b - b * (b4 - a) / (FfxFloat32(4.0) * b4); -} - -/// Compute a high accuracy approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximateGamma2ToPQHigh(FfxFloat32x3 a) -{ - return ffxSqrt(ffxSqrt(a)); -} - -/// Compute an approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximateLinearToPQ(FfxFloat32x3 a) -{ - return ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast3(3u)) + ffxBroadcast3(0x378D8723u)); -} - -/// Compute a more accurate approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximateLinearToPQMedium(FfxFloat32x3 a) -{ - FfxFloat32x3 b = ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast3(3u)) + ffxBroadcast3(0x378D8723u)); - FfxFloat32x3 b8 = b * b * b * b * b * b * b * b; - return b - b * (b8 - a) / (FfxFloat32(8.0) * b8); -} - -/// Compute a very accurate approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxApproximateLinearToPQHigh(FfxFloat32x3 a) -{ - return ffxSqrt(ffxSqrt(ffxSqrt(a))); -} - -/// Compute an approximate conversion from PQ to Gamma2 space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between PQ and Gamma2. -/// -/// @returns -/// The value a converted into Gamma2. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximatePQToGamma2Medium(FfxFloat32x4 a) -{ - return a * a * a * a; -} - -/// Compute an approximate conversion from PQ to linear space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between PQ and linear. -/// -/// @returns -/// The value a converted into linear. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximatePQToLinear(FfxFloat32x4 a) -{ - return a * a * a * a * a * a * a * a; -} - -/// Compute an approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximateGamma2ToPQ(FfxFloat32x4 a) -{ - return ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast4(2u)) + ffxBroadcast4(0x2F9A4E46u)); -} - -/// Compute a more accurate approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximateGamma2ToPQMedium(FfxFloat32x4 a) -{ - FfxFloat32x4 b = ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast4(2u)) + ffxBroadcast4(0x2F9A4E46u)); - FfxFloat32x4 b4 = b * b * b * b * b * b * b * b; - return b - b * (b4 - a) / (FfxFloat32(4.0) * b4); -} - -/// Compute a high accuracy approximate conversion from gamma2 to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between gamma2 and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximateGamma2ToPQHigh(FfxFloat32x4 a) -{ - return ffxSqrt(ffxSqrt(a)); -} - -/// Compute an approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximateLinearToPQ(FfxFloat32x4 a) -{ - return ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast4(3u)) + ffxBroadcast4(0x378D8723u)); -} - -/// Compute a more accurate approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximateLinearToPQMedium(FfxFloat32x4 a) -{ - FfxFloat32x4 b = ffxAsFloat((ffxAsUInt32(a) >> ffxBroadcast4(3u)) + ffxBroadcast4(0x378D8723u)); - FfxFloat32x4 b8 = b * b * b * b * b * b * b * b; - return b - b * (b8 - a) / (FfxFloat32(8.0) * b8); -} - -/// Compute a very accurate approximate conversion from linear to PQ space. -/// -/// PQ is very close to x^(1/8). The functions below Use the fast FfxFloat32 approximation method to do -/// PQ conversions to and from Gamma2 (4th power and fast 4th root), and PQ to and from Linear -/// (8th power and fast 8th root). The maximum error is approximately 0.2%. -/// -/// @param a The value to convert between linear and PQ. -/// -/// @returns -/// The value a converted into PQ. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxApproximateLinearToPQHigh(FfxFloat32x4 a) -{ - return ffxSqrt(ffxSqrt(ffxSqrt(a))); -} - -// An approximation of sine. -// -// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range -// is {-1/4 to 1/4} representing {-1 to 1}. -// -// @param [in] value The value to calculate approximate sine for. -// -// @returns -// The approximate sine of value. -FfxFloat32 ffxParabolicSin(FfxFloat32 value) -{ - return value * abs(value) - value; -} - -// An approximation of sine. -// -// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range -// is {-1/4 to 1/4} representing {-1 to 1}. -// -// @param [in] value The value to calculate approximate sine for. -// -// @returns -// The approximate sine of value. -FfxFloat32x2 ffxParabolicSin(FfxFloat32x2 x) -{ - return x * abs(x) - x; -} - -// An approximation of cosine. -// -// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range -// is {-1/4 to 1/4} representing {-1 to 1}. -// -// @param [in] value The value to calculate approximate cosine for. -// -// @returns -// The approximate cosine of value. -FfxFloat32 ffxParabolicCos(FfxFloat32 x) -{ - x = ffxFract(x * FfxFloat32(0.5) + FfxFloat32(0.75)); - x = x * FfxFloat32(2.0) - FfxFloat32(1.0); - return ffxParabolicSin(x); -} - -// An approximation of cosine. -// -// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range -// is {-1/4 to 1/4} representing {-1 to 1}. -// -// @param [in] value The value to calculate approximate cosine for. -// -// @returns -// The approximate cosine of value. -FfxFloat32x2 ffxParabolicCos(FfxFloat32x2 x) -{ - x = ffxFract(x * ffxBroadcast2(0.5f) + ffxBroadcast2(0.75f)); - x = x * ffxBroadcast2(2.0f) - ffxBroadcast2(1.0f); - return ffxParabolicSin(x); -} - -// An approximation of both sine and cosine. -// -// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range -// is {-1/4 to 1/4} representing {-1 to 1}. -// -// @param [in] value The value to calculate approximate cosine for. -// -// @returns -// A FfxFloat32x2 containing approximations of both sine and cosine of value. -FfxFloat32x2 ffxParabolicSinCos(FfxFloat32 x) -{ - FfxFloat32 y = ffxFract(x * FfxFloat32(0.5) + FfxFloat32(0.75)); - y = y * FfxFloat32(2.0) - FfxFloat32(1.0); - return ffxParabolicSin(FfxFloat32x2(x, y)); -} - -/// Conditional free logic AND operation using values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxUInt32 ffxZeroOneAnd(FfxUInt32 x, FfxUInt32 y) -{ - return min(x, y); -} - -/// Conditional free logic AND operation using two values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxZeroOneAnd(FfxUInt32x2 x, FfxUInt32x2 y) -{ - return min(x, y); -} - -/// Conditional free logic AND operation using two values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxUInt32x3 ffxZeroOneAnd(FfxUInt32x3 x, FfxUInt32x3 y) -{ - return min(x, y); -} - -/// Conditional free logic AND operation using two values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxUInt32x4 ffxZeroOneAnd(FfxUInt32x4 x, FfxUInt32x4 y) -{ - return min(x, y); -} - -/// Conditional free logic NOT operation using two values. -/// -/// @param [in] x The first value to be fed into the NOT operator. -/// -/// @returns -/// Result of the NOT operation. -/// -/// @ingroup GPUCore -FfxUInt32 ffxZeroOneAnd(FfxUInt32 x) -{ - return x ^ FfxUInt32(1); -} - -/// Conditional free logic NOT operation using two values. -/// -/// @param [in] x The first value to be fed into the NOT operator. -/// -/// @returns -/// Result of the NOT operation. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxZeroOneAnd(FfxUInt32x2 x) -{ - return x ^ ffxBroadcast2(1u); -} - -/// Conditional free logic NOT operation using two values. -/// -/// @param [in] x The first value to be fed into the NOT operator. -/// -/// @returns -/// Result of the NOT operation. -/// -/// @ingroup GPUCore -FfxUInt32x3 ffxZeroOneAnd(FfxUInt32x3 x) -{ - return x ^ ffxBroadcast3(1u); -} - -/// Conditional free logic NOT operation using two values. -/// -/// @param [in] x The first value to be fed into the NOT operator. -/// -/// @returns -/// Result of the NOT operation. -/// -/// @ingroup GPUCore -FfxUInt32x4 ffxZeroOneAnd(FfxUInt32x4 x) -{ - return x ^ ffxBroadcast4(1u); -} - -/// Conditional free logic OR operation using two values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxUInt32 ffxZeroOneOr(FfxUInt32 x, FfxUInt32 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxZeroOneOr(FfxUInt32x2 x, FfxUInt32x2 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxUInt32x3 ffxZeroOneOr(FfxUInt32x3 x, FfxUInt32x3 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxUInt32x4 ffxZeroOneOr(FfxUInt32x4 x, FfxUInt32x4 y) -{ - return max(x, y); -} - -/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxUInt32 ffxZeroOneAndToU1(FfxFloat32 x) -{ - return FfxUInt32(FfxFloat32(1.0) - x); -} - -/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxZeroOneAndToU2(FfxFloat32x2 x) -{ - return FfxUInt32x2(ffxBroadcast2(1.0) - x); -} - -/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxUInt32x3 ffxZeroOneAndToU3(FfxFloat32x3 x) -{ - return FfxUInt32x3(ffxBroadcast3(1.0) - x); -} - -/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxUInt32x4 ffxZeroOneAndToU4(FfxFloat32x4 x) -{ - return FfxUInt32x4(ffxBroadcast4(1.0) - x); -} - -/// Conditional free logic AND operation using two values followed by a NOT operation -/// using the resulting value and a third value. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// @param [in] z The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat32 ffxZeroOneAndOr(FfxFloat32 x, FfxFloat32 y, FfxFloat32 z) -{ - return ffxSaturate(x * y + z); -} - -/// Conditional free logic AND operation using two values followed by a NOT operation -/// using the resulting value and a third value. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// @param [in] z The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxZeroOneAndOr(FfxFloat32x2 x, FfxFloat32x2 y, FfxFloat32x2 z) -{ - return ffxSaturate(x * y + z); -} - -/// Conditional free logic AND operation using two values followed by a NOT operation -/// using the resulting value and a third value. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// @param [in] z The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxZeroOneAndOr(FfxFloat32x3 x, FfxFloat32x3 y, FfxFloat32x3 z) -{ - return ffxSaturate(x * y + z); -} - -/// Conditional free logic AND operation using two values followed by a NOT operation -/// using the resulting value and a third value. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// @param [in] z The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxZeroOneAndOr(FfxFloat32x4 x, FfxFloat32x4 y, FfxFloat32x4 z) -{ - return ffxSaturate(x * y + z); -} - -/// Given a value, returns 1.0 if greater than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the greater than zero comparison. -/// -/// @ingroup GPUCore -FfxFloat32 ffxZeroOneIsGreaterThanZero(FfxFloat32 x) -{ - return ffxSaturate(x * FfxFloat32(FFX_POSITIVE_INFINITY_FLOAT)); -} - -/// Given a value, returns 1.0 if greater than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the greater than zero comparison. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxZeroOneIsGreaterThanZero(FfxFloat32x2 x) -{ - return ffxSaturate(x * ffxBroadcast2(FFX_POSITIVE_INFINITY_FLOAT)); -} - -/// Given a value, returns 1.0 if greater than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the greater than zero comparison. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxZeroOneIsGreaterThanZero(FfxFloat32x3 x) -{ - return ffxSaturate(x * ffxBroadcast3(FFX_POSITIVE_INFINITY_FLOAT)); -} - -/// Given a value, returns 1.0 if greater than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the greater than zero comparison. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxZeroOneIsGreaterThanZero(FfxFloat32x4 x) -{ - return ffxSaturate(x * ffxBroadcast4(FFX_POSITIVE_INFINITY_FLOAT)); -} - -/// Conditional free logic signed NOT operation using two FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat32 ffxZeroOneAnd(FfxFloat32 x) -{ - return FfxFloat32(1.0) - x; -} - -/// Conditional free logic signed NOT operation using two FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxZeroOneAnd(FfxFloat32x2 x) -{ - return ffxBroadcast2(1.0) - x; -} - -/// Conditional free logic signed NOT operation using two FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxZeroOneAnd(FfxFloat32x3 x) -{ - return ffxBroadcast3(1.0) - x; -} - -/// Conditional free logic signed NOT operation using two FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxZeroOneAnd(FfxFloat32x4 x) -{ - return ffxBroadcast4(1.0) - x; -} - -/// Conditional free logic OR operation using two FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxFloat32 ffxZeroOneOr(FfxFloat32 x, FfxFloat32 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxZeroOneOr(FfxFloat32x2 x, FfxFloat32x2 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxZeroOneOr(FfxFloat32x3 x, FfxFloat32x3 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxZeroOneOr(FfxFloat32x4 x, FfxFloat32x4 y) -{ - return max(x, y); -} - -/// Choose between two FfxFloat32 values if the first paramter is greater than zero. -/// -/// @param [in] x The value to compare against zero. -/// @param [in] y The value to return if the comparision is greater than zero. -/// @param [in] z The value to return if the comparision is less than or equal to zero. -/// -/// @returns -/// The selected value. -/// -/// @ingroup GPUCore -FfxFloat32 ffxZeroOneSelect(FfxFloat32 x, FfxFloat32 y, FfxFloat32 z) -{ - FfxFloat32 r = (-x) * z + z; - return x * y + r; -} - -/// Choose between two FfxFloat32 values if the first paramter is greater than zero. -/// -/// @param [in] x The value to compare against zero. -/// @param [in] y The value to return if the comparision is greater than zero. -/// @param [in] z The value to return if the comparision is less than or equal to zero. -/// -/// @returns -/// The selected value. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxZeroOneSelect(FfxFloat32x2 x, FfxFloat32x2 y, FfxFloat32x2 z) -{ - FfxFloat32x2 r = (-x) * z + z; - return x * y + r; -} - -/// Choose between two FfxFloat32 values if the first paramter is greater than zero. -/// -/// @param [in] x The value to compare against zero. -/// @param [in] y The value to return if the comparision is greater than zero. -/// @param [in] z The value to return if the comparision is less than or equal to zero. -/// -/// @returns -/// The selected value. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxZeroOneSelect(FfxFloat32x3 x, FfxFloat32x3 y, FfxFloat32x3 z) -{ - FfxFloat32x3 r = (-x) * z + z; - return x * y + r; -} - -/// Choose between two FfxFloat32 values if the first paramter is greater than zero. -/// -/// @param [in] x The value to compare against zero. -/// @param [in] y The value to return if the comparision is greater than zero. -/// @param [in] z The value to return if the comparision is less than or equal to zero. -/// -/// @returns -/// The selected value. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxZeroOneSelect(FfxFloat32x4 x, FfxFloat32x4 y, FfxFloat32x4 z) -{ - FfxFloat32x4 r = (-x) * z + z; - return x * y + r; -} - -/// Given a value, returns 1.0 if less than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the sign value. -/// -/// @ingroup GPUCore -FfxFloat32 ffxZeroOneIsSigned(FfxFloat32 x) -{ - return ffxSaturate(x * FfxFloat32(FFX_NEGATIVE_INFINITY_FLOAT)); -} - -/// Given a value, returns 1.0 if less than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the sign value. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxZeroOneIsSigned(FfxFloat32x2 x) -{ - return ffxSaturate(x * ffxBroadcast2(FFX_NEGATIVE_INFINITY_FLOAT)); -} - -/// Given a value, returns 1.0 if less than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the sign value. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxZeroOneIsSigned(FfxFloat32x3 x) -{ - return ffxSaturate(x * ffxBroadcast3(FFX_NEGATIVE_INFINITY_FLOAT)); -} - -/// Given a value, returns 1.0 if less than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the sign value. -/// -/// @ingroup GPUCore -FfxFloat32x4 ffxZeroOneIsSigned(FfxFloat32x4 x) -{ - return ffxSaturate(x * ffxBroadcast4(FFX_NEGATIVE_INFINITY_FLOAT)); -} - -/// Compute a Rec.709 color space. -/// -/// Rec.709 is used for some HDTVs. -/// -/// Both Rec.709 and sRGB have a linear segment which as spec'ed would intersect the curved segment 2 times. -/// (a.) For 8-bit sRGB, steps {0 to 10.3} are in the linear region (4% of the encoding range). -/// (b.) For 8-bit 709, steps {0 to 20.7} are in the linear region (8% of the encoding range). -/// -/// @param [in] color The color to convert to Rec. 709. -/// -/// @returns -/// The color in linear space. -/// -/// @ingroup GPUCore -FfxFloat32 ffxRec709FromLinear(FfxFloat32 color) -{ - FfxFloat32x3 j = FfxFloat32x3(0.018 * 4.5, 4.5, 0.45); - FfxFloat32x2 k = FfxFloat32x2(1.099, -0.099); - return clamp(j.x, color * j.y, pow(color, j.z) * k.x + k.y); -} - -/// Compute a Rec.709 color space. -/// -/// Rec.709 is used for some HDTVs. -/// -/// Both Rec.709 and sRGB have a linear segment which as spec'ed would intersect the curved segment 2 times. -/// (a.) For 8-bit sRGB, steps {0 to 10.3} are in the linear region (4% of the encoding range). -/// (b.) For 8-bit 709, steps {0 to 20.7} are in the linear region (8% of the encoding range). -/// -/// @param [in] color The color to convert to Rec. 709. -/// -/// @returns -/// The color in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxRec709FromLinear(FfxFloat32x2 color) -{ - FfxFloat32x3 j = FfxFloat32x3(0.018 * 4.5, 4.5, 0.45); - FfxFloat32x2 k = FfxFloat32x2(1.099, -0.099); - return clamp(j.xx, color * j.yy, pow(color, j.zz) * k.xx + k.yy); -} - -/// Compute a Rec.709 color space. -/// -/// Rec.709 is used for some HDTVs. -/// -/// Both Rec.709 and sRGB have a linear segment which as spec'ed would intersect the curved segment 2 times. -/// (a.) For 8-bit sRGB, steps {0 to 10.3} are in the linear region (4% of the encoding range). -/// (b.) For 8-bit 709, steps {0 to 20.7} are in the linear region (8% of the encoding range). -/// -/// @param [in] color The color to convert to Rec. 709. -/// -/// @returns -/// The color in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxRec709FromLinear(FfxFloat32x3 color) -{ - FfxFloat32x3 j = FfxFloat32x3(0.018 * 4.5, 4.5, 0.45); - FfxFloat32x2 k = FfxFloat32x2(1.099, -0.099); - return clamp(j.xxx, color * j.yyy, pow(color, j.zzz) * k.xxx + k.yyy); -} - -/// Compute a gamma value from a linear value. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// Note: 'rcpX' is '1/x', where the 'x' is what would be used in ffxLinearFromGamma. -/// -/// @param [in] value The value to convert to gamma space from linear. -/// @param [in] power The reciprocal of power value used for the gamma curve. -/// -/// @returns -/// A value in gamma space. -/// -/// @ingroup GPUCore -FfxFloat32 ffxGammaFromLinear(FfxFloat32 value, FfxFloat32 power) -{ - return pow(value, FfxFloat32(power)); -} - -/// Compute a gamma value from a linear value. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// Note: 'rcpX' is '1/x', where the 'x' is what would be used in ffxLinearFromGamma. -/// -/// @param [in] value The value to convert to gamma space from linear. -/// @param [in] power The reciprocal of power value used for the gamma curve. -/// -/// @returns -/// A value in gamma space. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxGammaFromLinear(FfxFloat32x2 value, FfxFloat32 power) -{ - return pow(value, ffxBroadcast2(power)); -} - -/// Compute a gamma value from a linear value. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// Note: 'rcpX' is '1/x', where the 'x' is what would be used in ffxLinearFromGamma. -/// -/// @param [in] value The value to convert to gamma space from linear. -/// @param [in] power The reciprocal of power value used for the gamma curve. -/// -/// @returns -/// A value in gamma space. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxGammaFromLinear(FfxFloat32x3 value, FfxFloat32 power) -{ - return pow(value, ffxBroadcast3(power)); -} - -/// Compute a PQ value from a linear value. -/// -/// @param [in] value The value to convert to PQ from linear. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32 ffxPQToLinear(FfxFloat32 value) -{ - FfxFloat32 p = pow(value, FfxFloat32(0.159302)); - return pow((FfxFloat32(0.835938) + FfxFloat32(18.8516) * p) / (FfxFloat32(1.0) + FfxFloat32(18.6875) * p), FfxFloat32(78.8438)); -} - -/// Compute a PQ value from a linear value. -/// -/// @param [in] value The value to convert to PQ from linear. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxPQToLinear(FfxFloat32x2 value) -{ - FfxFloat32x2 p = pow(value, ffxBroadcast2(0.159302)); - return pow((ffxBroadcast2(0.835938) + ffxBroadcast2(18.8516) * p) / (ffxBroadcast2(1.0) + ffxBroadcast2(18.6875) * p), ffxBroadcast2(78.8438)); -} - -/// Compute a PQ value from a linear value. -/// -/// @param [in] value The value to convert to PQ from linear. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxPQToLinear(FfxFloat32x3 value) -{ - FfxFloat32x3 p = pow(value, ffxBroadcast3(0.159302)); - return pow((ffxBroadcast3(0.835938) + ffxBroadcast3(18.8516) * p) / (ffxBroadcast3(1.0) + ffxBroadcast3(18.6875) * p), ffxBroadcast3(78.8438)); -} - -/// Compute a linear value from a SRGB value. -/// -/// @param [in] value The value to convert to linear from SRGB. -/// -/// @returns -/// A value in SRGB space. -/// -/// @ingroup GPUCore -FfxFloat32 ffxSrgbToLinear(FfxFloat32 value) -{ - FfxFloat32x3 j = FfxFloat32x3(0.0031308 * 12.92, 12.92, 1.0 / 2.4); - FfxFloat32x2 k = FfxFloat32x2(1.055, -0.055); - return clamp(j.x, value * j.y, pow(value, j.z) * k.x + k.y); -} - -/// Compute a linear value from a SRGB value. -/// -/// @param [in] value The value to convert to linear from SRGB. -/// -/// @returns -/// A value in SRGB space. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxSrgbToLinear(FfxFloat32x2 value) -{ - FfxFloat32x3 j = FfxFloat32x3(0.0031308 * 12.92, 12.92, 1.0 / 2.4); - FfxFloat32x2 k = FfxFloat32x2(1.055, -0.055); - return clamp(j.xx, value * j.yy, pow(value, j.zz) * k.xx + k.yy); -} - -/// Compute a linear value from a SRGB value. -/// -/// @param [in] value The value to convert to linear from SRGB. -/// -/// @returns -/// A value in SRGB space. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxSrgbToLinear(FfxFloat32x3 value) -{ - FfxFloat32x3 j = FfxFloat32x3(0.0031308 * 12.92, 12.92, 1.0 / 2.4); - FfxFloat32x2 k = FfxFloat32x2(1.055, -0.055); - return clamp(j.xxx, value * j.yyy, pow(value, j.zzz) * k.xxx + k.yyy); -} - -/// Compute a linear value from a REC.709 value. -/// -/// @param [in] color The value to convert to linear from REC.709. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32 ffxLinearFromRec709(FfxFloat32 color) -{ - FfxFloat32x3 j = FfxFloat32x3(0.081 / 4.5, 1.0 / 4.5, 1.0 / 0.45); - FfxFloat32x2 k = FfxFloat32x2(1.0 / 1.099, 0.099 / 1.099); - return ffxZeroOneSelect(ffxZeroOneIsSigned(color - j.x), color * j.y, pow(color * k.x + k.y, j.z)); -} - -/// Compute a linear value from a REC.709 value. -/// -/// @param [in] color The value to convert to linear from REC.709. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxLinearFromRec709(FfxFloat32x2 color) -{ - FfxFloat32x3 j = FfxFloat32x3(0.081 / 4.5, 1.0 / 4.5, 1.0 / 0.45); - FfxFloat32x2 k = FfxFloat32x2(1.0 / 1.099, 0.099 / 1.099); - return ffxZeroOneSelect(ffxZeroOneIsSigned(color - j.xx), color * j.yy, pow(color * k.xx + k.yy, j.zz)); -} - -/// Compute a linear value from a REC.709 value. -/// -/// @param [in] color The value to convert to linear from REC.709. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxLinearFromRec709(FfxFloat32x3 color) -{ - FfxFloat32x3 j = FfxFloat32x3(0.081 / 4.5, 1.0 / 4.5, 1.0 / 0.45); - FfxFloat32x2 k = FfxFloat32x2(1.0 / 1.099, 0.099 / 1.099); - return ffxZeroOneSelect(ffxZeroOneIsSigned(color - j.xxx), color * j.yyy, pow(color * k.xxx + k.yyy, j.zzz)); -} - -/// Compute a linear value from a value in a gamma space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] color The value to convert to linear in gamma space. -/// @param [in] power The power value used for the gamma curve. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32 ffxLinearFromGamma(FfxFloat32 color, FfxFloat32 power) -{ - return pow(color, FfxFloat32(power)); -} - -/// Compute a linear value from a value in a gamma space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] color The value to convert to linear in gamma space. -/// @param [in] power The power value used for the gamma curve. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxLinearFromGamma(FfxFloat32x2 color, FfxFloat32 power) -{ - return pow(color, ffxBroadcast2(power)); -} - -/// Compute a linear value from a value in a gamma space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] color The value to convert to linear in gamma space. -/// @param [in] power The power value used for the gamma curve. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxLinearFromGamma(FfxFloat32x3 color, FfxFloat32 power) -{ - return pow(color, ffxBroadcast3(power)); -} - -/// Compute a linear value from a value in a PQ space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] value The value to convert to linear in PQ space. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32 ffxLinearFromPQ(FfxFloat32 value) -{ - FfxFloat32 p = pow(value, FfxFloat32(0.0126833)); - return pow(ffxSaturate(p - FfxFloat32(0.835938)) / (FfxFloat32(18.8516) - FfxFloat32(18.6875) * p), FfxFloat32(6.27739)); -} - -/// Compute a linear value from a value in a PQ space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] value The value to convert to linear in PQ space. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxLinearFromPQ(FfxFloat32x2 value) -{ - FfxFloat32x2 p = pow(value, ffxBroadcast2(0.0126833)); - return pow(ffxSaturate(p - ffxBroadcast2(0.835938)) / (ffxBroadcast2(18.8516) - ffxBroadcast2(18.6875) * p), ffxBroadcast2(6.27739)); -} - -/// Compute a linear value from a value in a PQ space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] value The value to convert to linear in PQ space. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxLinearFromPQ(FfxFloat32x3 value) -{ - FfxFloat32x3 p = pow(value, ffxBroadcast3(0.0126833)); - return pow(ffxSaturate(p - ffxBroadcast3(0.835938)) / (ffxBroadcast3(18.8516) - ffxBroadcast3(18.6875) * p), ffxBroadcast3(6.27739)); -} - -/// Compute a linear value from a value in a SRGB space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] value The value to convert to linear in SRGB space. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32 ffxLinearFromSrgb(FfxFloat32 value) -{ - FfxFloat32x3 j = FfxFloat32x3(0.04045 / 12.92, 1.0 / 12.92, 2.4); - FfxFloat32x2 k = FfxFloat32x2(1.0 / 1.055, 0.055 / 1.055); - return ffxZeroOneSelect(ffxZeroOneIsSigned(value - j.x), value * j.y, pow(value * k.x + k.y, j.z)); -} - -/// Compute a linear value from a value in a SRGB space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] value The value to convert to linear in SRGB space. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x2 ffxLinearFromSrgb(FfxFloat32x2 value) -{ - FfxFloat32x3 j = FfxFloat32x3(0.04045 / 12.92, 1.0 / 12.92, 2.4); - FfxFloat32x2 k = FfxFloat32x2(1.0 / 1.055, 0.055 / 1.055); - return ffxZeroOneSelect(ffxZeroOneIsSigned(value - j.xx), value * j.yy, pow(value * k.xx + k.yy, j.zz)); -} - -/// Compute a linear value from a value in a SRGB space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] value The value to convert to linear in SRGB space. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat32x3 ffxLinearFromSrgb(FfxFloat32x3 value) -{ - FfxFloat32x3 j = FfxFloat32x3(0.04045 / 12.92, 1.0 / 12.92, 2.4); - FfxFloat32x2 k = FfxFloat32x2(1.0 / 1.055, 0.055 / 1.055); - return ffxZeroOneSelect(ffxZeroOneIsSigned(value - j.xxx), value * j.yyy, pow(value * k.xxx + k.yyy, j.zzz)); -} - -/// A remapping of 64x1 to 8x8 imposing rotated 2x2 pixel quads in quad linear. -/// -/// 543210 -/// ====== -/// ..xxx. -/// yy...y -/// -/// @param [in] a The input 1D coordinates to remap. -/// -/// @returns -/// The remapped 2D coordinates. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxRemapForQuad(FfxUInt32 a) -{ - return FfxUInt32x2(bitfieldExtract(a, 1u, 3u), bitfieldInsertMask(bitfieldExtract(a, 3u, 3u), a, 1u)); -} - -/// A helper function performing a remap 64x1 to 8x8 remapping which is necessary for 2D wave reductions. -/// -/// The 64-wide lane indices to 8x8 remapping is performed as follows: -/// -/// 00 01 08 09 10 11 18 19 -/// 02 03 0a 0b 12 13 1a 1b -/// 04 05 0c 0d 14 15 1c 1d -/// 06 07 0e 0f 16 17 1e 1f -/// 20 21 28 29 30 31 38 39 -/// 22 23 2a 2b 32 33 3a 3b -/// 24 25 2c 2d 34 35 3c 3d -/// 26 27 2e 2f 36 37 3e 3f -/// -/// @param [in] a The input 1D coordinate to remap. -/// -/// @returns -/// The remapped 2D coordinates. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxRemapForWaveReduction(FfxUInt32 a) -{ - return FfxUInt32x2(bitfieldInsertMask(bitfieldExtract(a, 2u, 3u), a, 1u), bitfieldInsertMask(bitfieldExtract(a, 3u, 3u), bitfieldExtract(a, 1u, 2u), 2u)); -} diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common.h.meta deleted file mode 100644 index 14292a9..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 05b921699d1374a429e32afca13137e2 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common_half.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common_half.h deleted file mode 100644 index 4c73daf..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common_half.h +++ /dev/null @@ -1,2979 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#if FFX_HALF -#if FFX_HLSL_SM >= 62 -/// A define value for 16bit positive infinity. -/// -/// @ingroup GPUCore -#define FFX_POSITIVE_INFINITY_HALF FFX_TO_FLOAT16((uint16_t)0x7c00u) - -/// A define value for 16bit negative infinity. -/// -/// @ingroup GPUCore -#define FFX_NEGATIVE_INFINITY_HALF FFX_TO_FLOAT16((uint16_t)0xfc00u) -#else -/// A define value for 16bit positive infinity. -/// -/// @ingroup GPUCore -#define FFX_POSITIVE_INFINITY_HALF FFX_TO_FLOAT16(0x7c00u) - -/// A define value for 16bit negative infinity. -/// -/// @ingroup GPUCore -#define FFX_NEGATIVE_INFINITY_HALF FFX_TO_FLOAT16(0xfc00u) -#endif // #if FFX_HLSL_SM>=62 - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat16 ffxMin(FfxFloat16 x, FfxFloat16 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxMin(FfxFloat16x2 x, FfxFloat16x2 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxMin(FfxFloat16x3 x, FfxFloat16x3 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxMin(FfxFloat16x4 x, FfxFloat16x4 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt16 ffxMin(FfxInt16 x, FfxInt16 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt16x2 ffxMin(FfxInt16x2 x, FfxInt16x2 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt16x3 ffxMin(FfxInt16x3 x, FfxInt16x3 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt16x4 ffxMin(FfxInt16x4 x, FfxInt16x4 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt16 ffxMin(FfxUInt16 x, FfxUInt16 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxMin(FfxUInt16x2 x, FfxUInt16x2 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt16x3 ffxMin(FfxUInt16x3 x, FfxUInt16x3 y) -{ - return min(x, y); -} - -/// Compute the min of two values. -/// -/// @param [in] x The first value to compute the min of. -/// @param [in] y The second value to compute the min of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt16x4 ffxMin(FfxUInt16x4 x, FfxUInt16x4 y) -{ - return min(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat16 ffxMax(FfxFloat16 x, FfxFloat16 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxMax(FfxFloat16x2 x, FfxFloat16x2 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxMax(FfxFloat16x3 x, FfxFloat16x3 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxMax(FfxFloat16x4 x, FfxFloat16x4 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt16 ffxMax(FfxInt16 x, FfxInt16 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt16x2 ffxMax(FfxInt16x2 x, FfxInt16x2 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt16x3 ffxMax(FfxInt16x3 x, FfxInt16x3 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxInt16x4 ffxMax(FfxInt16x4 x, FfxInt16x4 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt16 ffxMax(FfxUInt16 x, FfxUInt16 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxMax(FfxUInt16x2 x, FfxUInt16x2 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt16x3 ffxMax(FfxUInt16x3 x, FfxUInt16x3 y) -{ - return max(x, y); -} - -/// Compute the max of two values. -/// -/// @param [in] x The first value to compute the max of. -/// @param [in] y The second value to compute the max of. -/// -/// @returns -/// The the lowest of two values. -/// -/// @ingroup GPUCore -FfxUInt16x4 ffxMax(FfxUInt16x4 x, FfxUInt16x4 y) -{ - return max(x, y); -} - -/// Compute the value of the first parameter raised to the power of the second. -/// -/// @param [in] x The value to raise to the power y. -/// @param [in] y The power to which to raise x. -/// -/// @returns -/// The value of the first parameter raised to the power of the second. -/// -/// @ingroup GPUCore -FfxFloat16 ffxPow(FfxFloat16 x, FfxFloat16 y) -{ - return pow(x, y); -} - -/// Compute the value of the first parameter raised to the power of the second. -/// -/// @param [in] x The value to raise to the power y. -/// @param [in] y The power to which to raise x. -/// -/// @returns -/// The value of the first parameter raised to the power of the second. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPow(FfxFloat16x2 x, FfxFloat16x2 y) -{ - return pow(x, y); -} - -/// Compute the value of the first parameter raised to the power of the second. -/// -/// @param [in] x The value to raise to the power y. -/// @param [in] y The power to which to raise x. -/// -/// @returns -/// The value of the first parameter raised to the power of the second. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxPow(FfxFloat16x3 x, FfxFloat16x3 y) -{ - return pow(x, y); -} - -/// Compute the value of the first parameter raised to the power of the second. -/// -/// @param [in] x The value to raise to the power y. -/// @param [in] y The power to which to raise x. -/// -/// @returns -/// The value of the first parameter raised to the power of the second. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxPow(FfxFloat16x4 x, FfxFloat16x4 y) -{ - return pow(x, y); -} - -/// Compute the square root of a value. -/// -/// @param [in] x The first value to compute the min of. -/// -/// @returns -/// The the square root of x. -/// -/// @ingroup GPUCore -FfxFloat16 ffxSqrt(FfxFloat16 x) -{ - return sqrt(x); -} - -/// Compute the square root of a value. -/// -/// @param [in] x The first value to compute the min of. -/// -/// @returns -/// The the square root of x. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxSqrt(FfxFloat16x2 x) -{ - return sqrt(x); -} - -/// Compute the square root of a value. -/// -/// @param [in] x The first value to compute the min of. -/// -/// @returns -/// The the square root of x. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxSqrt(FfxFloat16x3 x) -{ - return sqrt(x); -} - -/// Compute the square root of a value. -/// -/// @param [in] x The first value to compute the min of. -/// -/// @returns -/// The the square root of x. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxSqrt(FfxFloat16x4 x) -{ - return sqrt(x); -} - -/// Copy the sign bit from 's' to positive 'd'. -/// -/// @param [in] d The value to copy the sign bit into. -/// @param [in] s The value to copy the sign bit from. -/// -/// @returns -/// The value of d with the sign bit from s. -/// -/// @ingroup GPUCore -FfxFloat16 ffxCopySignBitHalf(FfxFloat16 d, FfxFloat16 s) -{ - return FFX_TO_FLOAT16(FFX_TO_UINT16(d) | (FFX_TO_UINT16(s) & FFX_BROADCAST_UINT16(0x8000u))); -} - -/// Copy the sign bit from 's' to positive 'd'. -/// -/// @param [in] d The value to copy the sign bit into. -/// @param [in] s The value to copy the sign bit from. -/// -/// @returns -/// The value of d with the sign bit from s. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxCopySignBitHalf(FfxFloat16x2 d, FfxFloat16x2 s) -{ - return FFX_TO_FLOAT16X2(FFX_TO_UINT16X2(d) | (FFX_TO_UINT16X2(s) & FFX_BROADCAST_UINT16X2(0x8000u))); -} - -/// Copy the sign bit from 's' to positive 'd'. -/// -/// @param [in] d The value to copy the sign bit into. -/// @param [in] s The value to copy the sign bit from. -/// -/// @returns -/// The value of d with the sign bit from s. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxCopySignBitHalf(FfxFloat16x3 d, FfxFloat16x3 s) -{ - return FFX_TO_FLOAT16X3(FFX_TO_UINT16X3(d) | (FFX_TO_UINT16X3(s) & FFX_BROADCAST_UINT16X3(0x8000u))); -} - -/// Copy the sign bit from 's' to positive 'd'. -/// -/// @param [in] d The value to copy the sign bit into. -/// @param [in] s The value to copy the sign bit from. -/// -/// @returns -/// The value of d with the sign bit from s. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxCopySignBitHalf(FfxFloat16x4 d, FfxFloat16x4 s) -{ - return FFX_TO_FLOAT16X4(FFX_TO_UINT16X4(d) | (FFX_TO_UINT16X4(s) & FFX_BROADCAST_UINT16X4(0x8000u))); -} - -/// A single operation to return the following: -/// m = NaN := 0 -/// m >= 0 := 0 -/// m < 0 := 1 -/// -/// Uses the following useful floating point logic, -/// saturate(+a*(-INF)==-INF) := 0 -/// saturate( 0*(-INF)== NaN) := 0 -/// saturate(-a*(-INF)==+INF) := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against 0. -/// -/// @returns -/// 1.0 when the value is negative, or 0.0 when the value is 0 or position. -/// -/// @ingroup GPUCore -FfxFloat16 ffxIsSignedHalf(FfxFloat16 m) -{ - return FfxFloat16(ffxSaturate(m * FFX_BROADCAST_FLOAT16(FFX_NEGATIVE_INFINITY_HALF))); -} - -/// A single operation to return the following: -/// m = NaN := 0 -/// m >= 0 := 0 -/// m < 0 := 1 -/// -/// Uses the following useful floating point logic, -/// saturate(+a*(-INF)==-INF) := 0 -/// saturate( 0*(-INF)== NaN) := 0 -/// saturate(-a*(-INF)==+INF) := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against 0. -/// -/// @returns -/// 1.0 when the value is negative, or 0.0 when the value is 0 or position. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxIsSignedHalf(FfxFloat16x2 m) -{ - return FfxFloat16x2(ffxSaturate(m * FFX_BROADCAST_FLOAT16X2(FFX_NEGATIVE_INFINITY_HALF))); -} - -/// A single operation to return the following: -/// m = NaN := 0 -/// m >= 0 := 0 -/// m < 0 := 1 -/// -/// Uses the following useful floating point logic, -/// saturate(+a*(-INF)==-INF) := 0 -/// saturate( 0*(-INF)== NaN) := 0 -/// saturate(-a*(-INF)==+INF) := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against 0. -/// -/// @returns -/// 1.0 when the value is negative, or 0.0 when the value is 0 or position. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxIsSignedHalf(FfxFloat16x3 m) -{ - return FfxFloat16x3(ffxSaturate(m * FFX_BROADCAST_FLOAT16X3(FFX_NEGATIVE_INFINITY_HALF))); -} - -/// A single operation to return the following: -/// m = NaN := 0 -/// m >= 0 := 0 -/// m < 0 := 1 -/// -/// Uses the following useful floating point logic, -/// saturate(+a*(-INF)==-INF) := 0 -/// saturate( 0*(-INF)== NaN) := 0 -/// saturate(-a*(-INF)==+INF) := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against 0. -/// -/// @returns -/// 1.0 when the value is negative, or 0.0 when the value is 0 or position. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxIsSignedHalf(FfxFloat16x4 m) -{ - return FfxFloat16x4(ffxSaturate(m * FFX_BROADCAST_FLOAT16X4(FFX_NEGATIVE_INFINITY_HALF))); -} - -/// A single operation to return the following: -/// m = NaN := 1 -/// m > 0 := 0 -/// m <= 0 := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against zero. -/// -/// @returns -/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. -/// -/// @ingroup GPUCore -FfxFloat16 ffxIsGreaterThanZeroHalf(FfxFloat16 m) -{ - return FfxFloat16(ffxSaturate(m * FFX_BROADCAST_FLOAT16(FFX_POSITIVE_INFINITY_HALF))); -} - -/// A single operation to return the following: -/// m = NaN := 1 -/// m > 0 := 0 -/// m <= 0 := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against zero. -/// -/// @returns -/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxIsGreaterThanZeroHalf(FfxFloat16x2 m) -{ - return FfxFloat16x2(ffxSaturate(m * FFX_BROADCAST_FLOAT16X2(FFX_POSITIVE_INFINITY_HALF))); -} - -/// A single operation to return the following: -/// m = NaN := 1 -/// m > 0 := 0 -/// m <= 0 := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against zero. -/// -/// @returns -/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxIsGreaterThanZeroHalf(FfxFloat16x3 m) -{ - return FfxFloat16x3(ffxSaturate(m * FFX_BROADCAST_FLOAT16X3(FFX_POSITIVE_INFINITY_HALF))); -} - -/// A single operation to return the following: -/// m = NaN := 1 -/// m > 0 := 0 -/// m <= 0 := 1 -/// -/// This function is useful when creating masks for branch-free logic. -/// -/// @param [in] m The value to test against zero. -/// -/// @returns -/// 1.0 when the value is position, or 0.0 when the value is 0 or negative. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxIsGreaterThanZeroHalf(FfxFloat16x4 m) -{ - return FfxFloat16x4(ffxSaturate(m * FFX_BROADCAST_FLOAT16X4(FFX_POSITIVE_INFINITY_HALF))); -} - -/// Convert a 16bit floating point value to sortable integer. -/// -/// - If sign bit=0, flip the sign bit (positives). -/// - If sign bit=1, flip all bits (negatives). -/// -/// The function has the side effects that: -/// - Larger integers are more positive values. -/// - Float zero is mapped to center of integers (so clear to integer zero is a nice default for atomic max usage). -/// -/// @param [in] x The floating point value to make sortable. -/// -/// @returns -/// The sortable integer value. -/// -/// @ingroup GPUCore -FfxUInt16 ffxFloatToSortableIntegerHalf(FfxUInt16 x) -{ - return x ^ ((ffxBitShiftRightHalf(x, FFX_BROADCAST_UINT16(15))) | FFX_BROADCAST_UINT16(0x8000)); -} - -/// Convert a sortable integer to a 16bit floating point value. -/// -/// The function has the side effects that: -/// - If sign bit=1, flip the sign bit (positives). -/// - If sign bit=0, flip all bits (negatives). -/// -/// @param [in] x The sortable integer value to make floating point. -/// -/// @returns -/// The floating point value. -/// -/// @ingroup GPUCore -FfxUInt16 ffxSortableIntegerToFloatHalf(FfxUInt16 x) -{ - return x ^ ((~ffxBitShiftRightHalf(x, FFX_BROADCAST_UINT16(15))) | FFX_BROADCAST_UINT16(0x8000)); -} - -/// Convert a pair of 16bit floating point values to a pair of sortable integers. -/// -/// - If sign bit=0, flip the sign bit (positives). -/// - If sign bit=1, flip all bits (negatives). -/// -/// The function has the side effects that: -/// - Larger integers are more positive values. -/// - Float zero is mapped to center of integers (so clear to integer zero is a nice default for atomic max usage). -/// -/// @param [in] x The floating point values to make sortable. -/// -/// @returns -/// The sortable integer values. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxFloatToSortableIntegerHalf(FfxUInt16x2 x) -{ - return x ^ ((ffxBitShiftRightHalf(x, FFX_BROADCAST_UINT16X2(15))) | FFX_BROADCAST_UINT16X2(0x8000)); -} - -/// Convert a pair of sortable integers to a pair of 16bit floating point values. -/// -/// The function has the side effects that: -/// - If sign bit=1, flip the sign bit (positives). -/// - If sign bit=0, flip all bits (negatives). -/// -/// @param [in] x The sortable integer values to make floating point. -/// -/// @returns -/// The floating point values. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxSortableIntegerToFloatHalf(FfxUInt16x2 x) -{ - return x ^ ((~ffxBitShiftRightHalf(x, FFX_BROADCAST_UINT16X2(15))) | FFX_BROADCAST_UINT16X2(0x8000)); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// [Zero] Y0 [Zero] X0 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesZeroY0ZeroX0(FfxUInt32x2 i) -{ - return ((i.x) & 0xffu) | ((i.y << 16) & 0xff0000u); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// [Zero] Y1 [Zero] X1 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesZeroY1ZeroX1(FfxUInt32x2 i) -{ - return ((i.x >> 8) & 0xffu) | ((i.y << 8) & 0xff0000u); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// [Zero] Y2 [Zero] X2 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesZeroY2ZeroX2(FfxUInt32x2 i) -{ - return ((i.x >> 16) & 0xffu) | ((i.y) & 0xff0000u); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// [Zero] Y3 [Zero] X3 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesZeroY3ZeroX3(FfxUInt32x2 i) -{ - return ((i.x >> 24) & 0xffu) | ((i.y >> 8) & 0xff0000u); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// Y3 Y2 Y1 X0 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesY3Y2Y1X0(FfxUInt32x2 i) -{ - return ((i.x) & 0x000000ffu) | (i.y & 0xffffff00u); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// Y3 Y2 Y1 X2 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesY3Y2Y1X2(FfxUInt32x2 i) -{ - return ((i.x >> 16) & 0x000000ffu) | (i.y & 0xffffff00u); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// Y3 Y2 X0 Y0 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesY3Y2X0Y0(FfxUInt32x2 i) -{ - return ((i.x << 8) & 0x0000ff00u) | (i.y & 0xffff00ffu); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// Y3 Y2 X2 Y0 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesY3Y2X2Y0(FfxUInt32x2 i) -{ - return ((i.x >> 8) & 0x0000ff00u) | (i.y & 0xffff00ffu); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// Y3 X0 Y1 Y0 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesY3X0Y1Y0(FfxUInt32x2 i) -{ - return ((i.x << 16) & 0x00ff0000u) | (i.y & 0xff00ffffu); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// Y3 X2 Y1 Y0 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesY3X2Y1Y0(FfxUInt32x2 i) -{ - return ((i.x) & 0x00ff0000u) | (i.y & 0xff00ffffu); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// X0 Y2 Y1 Y0 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesX0Y2Y1Y0(FfxUInt32x2 i) -{ - return ((i.x << 24) & 0xff000000u) | (i.y & 0x00ffffffu); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// X2 Y2 Y1 Y0 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesX2Y2Y1Y0(FfxUInt32x2 i) -{ - return ((i.x << 8) & 0xff000000u) | (i.y & 0x00ffffffu); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// Y2 X2 Y0 X0 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesY2X2Y0X0(FfxUInt32x2 i) -{ - return ((i.x) & 0x00ff00ffu) | ((i.y << 8) & 0xff00ff00u); -} - -/// Packs the bytes from the X and Y components of a FfxUInt32x2 into a single 32-bit integer. -/// -/// The resulting integer will contain bytes in the following order, from most to least significant: -/// Y2 Y0 X2 X0 -/// -/// @param [in] i The integer pair to pack. -/// -/// @returns -/// The packed integer value. -/// -/// @ingroup GPUCore -FfxUInt32 ffxPackBytesY2Y0X2X0(FfxUInt32x2 i) -{ - return (((i.x) & 0xffu) | ((i.x >> 8) & 0xff00u) | ((i.y << 16) & 0xff0000u) | ((i.y << 8) & 0xff000000u)); -} - -/// Takes two Float16x2 values x and y, normalizes them and builds a single Uint16x2 value in the format {{x0,y0},{x1,y1}}. -/// -/// @param [in] x The first float16x2 value to pack. -/// @param [in] y The second float16x2 value to pack. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxPackX0Y0X1Y1UnsignedToUint16x2(FfxFloat16x2 x, FfxFloat16x2 y) -{ - x *= FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0); - y *= FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0); - return FFX_UINT32_TO_UINT16X2(ffxPackBytesY2X2Y0X0(FfxUInt32x2(FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(x)), FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(y))))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[0:7], -/// d.y[0:7] into r.y[0:7], i.x[8:15] into r.x[8:15], r.y[8:15] and i.y[0:15] into r.x[16:31], r.y[16:31] using 3 ops. -/// -/// r=ffxPermuteUByte0Float16x2ToUint2(d,i) -/// Where 'k0' is an SGPR with {1.0/32768.0} packed into the lower 16-bits -/// Where 'k1' is an SGPR with 0x???? -/// Where 'k2' is an SGPR with 0x???? -/// V_PK_FMA_F16 i,i,k0.x,0 -/// V_PERM_B32 r.x,i,i,k1 -/// V_PERM_B32 r.y,i,i,k2 -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteUByte0Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0))); - return FfxUInt32x2(ffxPackBytesY3Y2Y1X0(FfxUInt32x2(d.x, b)), ffxPackBytesY3Y2Y1X2(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[8:15], -/// d.y[0:7] into r.y[8:15], i.x[0:7] into r.x[0:7], r.y[0:7] and i.y[0:15] into r.x[16:31], r.y[16:31] using 3 ops. -/// -/// r=ffxPermuteUByte1Float16x2ToUint2(d,i) -/// Where 'k0' is an SGPR with {1.0/32768.0} packed into the lower 16-bits -/// Where 'k1' is an SGPR with 0x???? -/// Where 'k2' is an SGPR with 0x???? -/// V_PK_FMA_F16 i,i,k0.x,0 -/// V_PERM_B32 r.x,i,i,k1 -/// V_PERM_B32 r.y,i,i,k2 -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteUByte1Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0))); - return FfxUInt32x2(ffxPackBytesY3Y2X0Y0(FfxUInt32x2(d.x, b)), ffxPackBytesY3Y2X2Y0(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[16:23], -/// d.y[0:7] into r.y[16:23], i.x[0:15] into r.x[0:15], r.y[0:15] and i.y[8:15] into r.x[24:31], r.y[24:31] using 3 ops. -/// -/// r=ffxPermuteUByte2Float16x2ToUint2(d,i) -/// Where 'k0' is an SGPR with {1.0/32768.0} packed into the lower 16-bits -/// Where 'k1' is an SGPR with 0x???? -/// Where 'k2' is an SGPR with 0x???? -/// V_PK_FMA_F16 i,i,k0.x,0 -/// V_PERM_B32 r.x,i,i,k1 -/// V_PERM_B32 r.y,i,i,k2 -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteUByte2Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0))); - return FfxUInt32x2(ffxPackBytesY3X0Y1Y0(FfxUInt32x2(d.x, b)), ffxPackBytesY3X2Y1Y0(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[24:31], -/// d.y[0:7] into r.y[24:31], i.x[0:15] into r.x[0:15], r.y[0:15] and i.y[0:7] into r.x[16:23], r.y[16:23] using 3 ops. -/// -/// r=ffxPermuteUByte3Float16x2ToUint2(d,i) -/// Where 'k0' is an SGPR with {1.0/32768.0} packed into the lower 16-bits -/// Where 'k1' is an SGPR with 0x???? -/// Where 'k2' is an SGPR with 0x???? -/// V_PK_FMA_F16 i,i,k0.x,0 -/// V_PERM_B32 r.x,i,i,k1 -/// V_PERM_B32 r.y,i,i,k2 -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteUByte3Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0))); - return FfxUInt32x2(ffxPackBytesX0Y2Y1Y0(FfxUInt32x2(d.x, b)), ffxPackBytesX2Y2Y1Y0(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[0:7] into r.x[0:7] and i.y[0:7] into r.y[0:7] using 2 ops. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteUByte0Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY0ZeroX0(i))) * FFX_BROADCAST_FLOAT16X2(32768.0); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[8:15] into r.x[0:7] and i.y[8:15] into r.y[0:7] using 2 ops. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteUByte1Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY1ZeroX1(i))) * FFX_BROADCAST_FLOAT16X2(32768.0); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[16:23] into r.x[0:7] and i.y[16:23] into r.y[0:7] using 2 ops. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteUByte2Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY2ZeroX2(i))) * FFX_BROADCAST_FLOAT16X2(32768.0); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[24:31] into r.x[0:7] and i.y[24:31] into r.y[0:7] using 2 ops. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteUByte3Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY3ZeroX3(i))) * FFX_BROADCAST_FLOAT16X2(32768.0); -} - -/// Takes two Float16x2 values x and y, normalizes them and builds a single Uint16x2 value in the format {{x0,y0},{x1,y1}}. -/// -/// @param [in] x The first float16x2 value to pack. -/// @param [in] y The second float16x2 value to pack. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxPackX0Y0X1Y1SignedToUint16x2(FfxFloat16x2 x, FfxFloat16x2 y) -{ - x = x * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0); - y = y * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0); - return FFX_UINT32_TO_UINT16X2(ffxPackBytesY2X2Y0X0(FfxUInt32x2(FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(x)), FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(y))))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[0:7], -/// d.y[0:7] into r.y[0:7], i.x[8:15] into r.x[8:15], r.y[8:15] and i.y[0:15] into r.x[16:31], r.y[16:31] using 3 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteSByte0Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))); - return FfxUInt32x2(ffxPackBytesY3Y2Y1X0(FfxUInt32x2(d.x, b)), ffxPackBytesY3Y2Y1X2(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[8:15], -/// d.y[0:7] into r.y[8:15], i.x[0:7] into r.x[0:7], r.y[0:7] and i.y[0:15] into r.x[16:31], r.y[16:31] using 3 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteSByte1Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))); - return FfxUInt32x2(ffxPackBytesY3Y2X0Y0(FfxUInt32x2(d.x, b)), ffxPackBytesY3Y2X2Y0(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[16:23], -/// d.y[0:7] into r.y[16:23], i.x[0:15] into r.x[0:15], r.y[0:15] and i.y[8:15] into r.x[24:31], r.y[24:31] using 3 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteSByte2Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))); - return FfxUInt32x2(ffxPackBytesY3X0Y1Y0(FfxUInt32x2(d.x, b)), ffxPackBytesY3X2Y1Y0(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[24:31], -/// d.y[0:7] into r.y[24:31], i.x[0:15] into r.x[0:15], r.y[0:15] and i.y[0:7] into r.x[16:23], r.y[16:23] using 3 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteSByte3Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))); - return FfxUInt32x2(ffxPackBytesX0Y2Y1Y0(FfxUInt32x2(d.x, b)), ffxPackBytesX2Y2Y1Y0(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[0:7], -/// d.y[0:7] into r.y[0:7], i.x[8:15] into r.x[8:15], r.y[8:15] and i.y[0:15] into r.x[16:31], r.y[16:31] using 3 ops. -/// -/// Zero-based flips the MSB bit of the byte (making 128 "exact zero" actually zero). -/// This is useful if there is a desire for cleared values to decode as zero. -/// -/// Handles signed byte values. -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteZeroBasedSByte0Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))) ^ 0x00800080u; - return FfxUInt32x2(ffxPackBytesY3Y2Y1X0(FfxUInt32x2(d.x, b)), ffxPackBytesY3Y2Y1X2(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[8:15], -/// d.y[0:7] into r.y[8:15], i.x[0:7] into r.x[0:7], r.y[0:7] and i.y[0:15] into r.x[16:31], r.y[16:31] using 3 ops. -/// -/// Zero-based flips the MSB bit of the byte (making 128 "exact zero" actually zero). -/// This is useful if there is a desire for cleared values to decode as zero. -/// -/// Handles signed byte values. -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteZeroBasedSByte1Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))) ^ 0x00800080u; - return FfxUInt32x2(ffxPackBytesY3Y2X0Y0(FfxUInt32x2(d.x, b)), ffxPackBytesY3Y2X2Y0(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[16:23], -/// d.y[0:7] into r.y[16:23], i.x[0:15] into r.x[0:15], r.y[0:15] and i.y[8:15] into r.x[24:31], r.y[24:31] using 3 ops. -/// -/// Zero-based flips the MSB bit of the byte (making 128 "exact zero" actually zero). -/// This is useful if there is a desire for cleared values to decode as zero. -/// -/// Handles signed byte values. -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteZeroBasedSByte2Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))) ^ 0x00800080u; - return FfxUInt32x2(ffxPackBytesY3X0Y1Y0(FfxUInt32x2(d.x, b)), ffxPackBytesY3X2Y1Y0(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value d, Float16x2 value i and a resulting FfxUInt32x2 value r, this function packs d.x[0:7] into r.x[24:31], -/// d.y[0:7] into r.y[24:31], i.x[0:15] into r.x[0:15], r.y[0:15] and i.y[0:7] into r.x[16:23], r.y[16:23] using 3 ops. -/// -/// Zero-based flips the MSB bit of the byte (making 128 "exact zero" actually zero). -/// This is useful if there is a desire for cleared values to decode as zero. -/// -/// Handles signed byte values. -/// -/// @param [in] d The FfxUInt32x2 value to be packed. -/// @param [in] i The FfxFloat16x2 value to be packed. -/// -/// @returns -/// The packed FfxUInt32x2 value. -/// -/// @ingroup GPUCore -FfxUInt32x2 ffxPermuteZeroBasedSByte3Float16x2ToUint2(FfxUInt32x2 d, FfxFloat16x2 i) -{ - FfxUInt32 b = FFX_UINT16X2_TO_UINT32(FFX_TO_UINT16X2(i * FFX_BROADCAST_FLOAT16X2(1.0 / 32768.0) + FFX_BROADCAST_FLOAT16X2(0.25 / 32768.0))) ^ 0x00800080u; - return FfxUInt32x2(ffxPackBytesX0Y2Y1Y0(FfxUInt32x2(d.x, b)), ffxPackBytesX2Y2Y1Y0(FfxUInt32x2(d.y, b))); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[0:7] into r.x[0:7] and i.y[0:7] into r.y[0:7] using 2 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteSByte0Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY0ZeroX0(i))) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[8:15] into r.x[0:7] and i.y[8:15] into r.y[0:7] using 2 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteSByte1Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY1ZeroX1(i))) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[16:23] into r.x[0:7] and i.y[16:23] into r.y[0:7] using 2 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteSByte2Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY2ZeroX2(i))) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[24:31] into r.x[0:7] and i.y[24:31] into r.y[0:7] using 2 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteSByte3Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY3ZeroX3(i))) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[0:7] into r.x[0:7] and i.y[0:7] into r.y[0:7] using 2 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteZeroBasedSByte0Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY0ZeroX0(i) ^ 0x00800080u)) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[8:15] into r.x[0:7] and i.y[8:15] into r.y[0:7] using 2 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteZeroBasedSByte1Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY1ZeroX1(i) ^ 0x00800080u)) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[16:23] into r.x[0:7] and i.y[16:23] into r.y[0:7] using 2 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteZeroBasedSByte2Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY2ZeroX2(i) ^ 0x00800080u)) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); -} - -/// Given a FfxUInt32x2 value i and a resulting Float16x2 value r, this function packs i.x[24:31] into r.x[0:7] and i.y[24:31] into r.y[0:7] using 2 ops. -/// -/// Handles signed byte values. -/// -/// @param [in] i The FfxUInt32x2 value to be unpacked. -/// -/// @returns -/// The unpacked FfxFloat16x2. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxPermuteZeroBasedSByte3Uint2ToFloat16x2(FfxUInt32x2 i) -{ - return FFX_TO_FLOAT16X2(FFX_UINT32_TO_UINT16X2(ffxPackBytesZeroY3ZeroX3(i) ^ 0x00800080u)) * FFX_BROADCAST_FLOAT16X2(32768.0) - FFX_BROADCAST_FLOAT16X2(0.25); -} - -/// Calculate a half-precision low-quality approximation for the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the square root for. -/// -/// @returns -/// An approximation of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat16 ffxApproximateSqrtHalf(FfxFloat16 a) -{ - return FFX_TO_FLOAT16((FFX_TO_UINT16(a) >> FFX_BROADCAST_UINT16(1)) + FFX_BROADCAST_UINT16(0x1de2)); -} - -/// Calculate a half-precision low-quality approximation for the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the square root for. -/// -/// @returns -/// An approximation of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxApproximateSqrtHalf(FfxFloat16x2 a) -{ - return FFX_TO_FLOAT16X2((FFX_TO_UINT16X2(a) >> FFX_BROADCAST_UINT16X2(1)) + FFX_BROADCAST_UINT16X2(0x1de2)); -} - -/// Calculate a half-precision low-quality approximation for the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the square root for. -/// -/// @returns -/// An approximation of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxApproximateSqrtHalf(FfxFloat16x3 a) -{ - return FFX_TO_FLOAT16X3((FFX_TO_UINT16X3(a) >> FFX_BROADCAST_UINT16X3(1)) + FFX_BROADCAST_UINT16X3(0x1de2)); -} - -/// Calculate a half-precision low-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat16 ffxApproximateReciprocalHalf(FfxFloat16 a) -{ - return FFX_TO_FLOAT16(FFX_BROADCAST_UINT16(0x7784) - FFX_TO_UINT16(a)); -} - -/// Calculate a half-precision low-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxApproximateReciprocalHalf(FfxFloat16x2 a) -{ - return FFX_TO_FLOAT16X2(FFX_BROADCAST_UINT16X2(0x7784) - FFX_TO_UINT16X2(a)); -} - -/// Calculate a half-precision low-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxApproximateReciprocalHalf(FfxFloat16x3 a) -{ - return FFX_TO_FLOAT16X3(FFX_BROADCAST_UINT16X3(0x7784) - FFX_TO_UINT16X3(a)); -} - -/// Calculate a half-precision low-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxApproximateReciprocalHalf(FfxFloat16x4 a) -{ - return FFX_TO_FLOAT16X4(FFX_BROADCAST_UINT16X4(0x7784) - FFX_TO_UINT16X4(a)); -} - -/// Calculate a half-precision medium-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to medium quality. -/// -/// @ingroup GPUCore -FfxFloat16 ffxApproximateReciprocalMediumHalf(FfxFloat16 a) -{ - FfxFloat16 b = FFX_TO_FLOAT16(FFX_BROADCAST_UINT16(0x778d) - FFX_TO_UINT16(a)); - return b * (-b * a + FFX_BROADCAST_FLOAT16(2.0)); -} - -/// Calculate a half-precision medium-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to medium quality. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxApproximateReciprocalMediumHalf(FfxFloat16x2 a) -{ - FfxFloat16x2 b = FFX_TO_FLOAT16X2(FFX_BROADCAST_UINT16X2(0x778d) - FFX_TO_UINT16X2(a)); - return b * (-b * a + FFX_BROADCAST_FLOAT16X2(2.0)); -} - -/// Calculate a half-precision medium-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to medium quality. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxApproximateReciprocalMediumHalf(FfxFloat16x3 a) -{ - FfxFloat16x3 b = FFX_TO_FLOAT16X3(FFX_BROADCAST_UINT16X3(0x778d) - FFX_TO_UINT16X3(a)); - return b * (-b * a + FFX_BROADCAST_FLOAT16X3(2.0)); -} - -/// Calculate a half-precision medium-quality approximation for the reciprocal of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal for. -/// -/// @returns -/// An approximation of the reciprocal, estimated to medium quality. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxApproximateReciprocalMediumHalf(FfxFloat16x4 a) -{ - FfxFloat16x4 b = FFX_TO_FLOAT16X4(FFX_BROADCAST_UINT16X4(0x778d) - FFX_TO_UINT16X4(a)); - return b * (-b * a + FFX_BROADCAST_FLOAT16X4(2.0)); -} - -/// Calculate a half-precision low-quality approximation for the reciprocal of the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal of the square root for. -/// -/// @returns -/// An approximation of the reciprocal of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat16 ffxApproximateReciprocalSquareRootHalf(FfxFloat16 a) -{ - return FFX_TO_FLOAT16(FFX_BROADCAST_UINT16(0x59a3) - (FFX_TO_UINT16(a) >> FFX_BROADCAST_UINT16(1))); -} - -/// Calculate a half-precision low-quality approximation for the reciprocal of the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal of the square root for. -/// -/// @returns -/// An approximation of the reciprocal of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxApproximateReciprocalSquareRootHalf(FfxFloat16x2 a) -{ - return FFX_TO_FLOAT16X2(FFX_BROADCAST_UINT16X2(0x59a3) - (FFX_TO_UINT16X2(a) >> FFX_BROADCAST_UINT16X2(1))); -} - -/// Calculate a half-precision low-quality approximation for the reciprocal of the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal of the square root for. -/// -/// @returns -/// An approximation of the reciprocal of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxApproximateReciprocalSquareRootHalf(FfxFloat16x3 a) -{ - return FFX_TO_FLOAT16X3(FFX_BROADCAST_UINT16X3(0x59a3) - (FFX_TO_UINT16X3(a) >> FFX_BROADCAST_UINT16X3(1))); -} - -/// Calculate a half-precision low-quality approximation for the reciprocal of the square root of a value. -/// -/// For additional information on the approximation family of functions, you can refer to Michal Drobot's excellent -/// presentation materials: -/// -/// - https://michaldrobot.files.wordpress.com/2014/05/gcn_alu_opt_digitaldragons2014.pdf -/// - https://github.com/michaldrobot/ShaderFastLibs/blob/master/ShaderFastMathLib.h -/// -/// @param [in] a The value to calculate an approximate to the reciprocal of the square root for. -/// -/// @returns -/// An approximation of the reciprocal of the square root, estimated to low quality. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxApproximateReciprocalSquareRootHalf(FfxFloat16x4 a) -{ - return FFX_TO_FLOAT16X4(FFX_BROADCAST_UINT16X4(0x59a3) - (FFX_TO_UINT16X4(a) >> FFX_BROADCAST_UINT16X4(1))); -} - -/// An approximation of sine. -/// -/// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range -/// is {-1/4 to 1/4} representing {-1 to 1}. -/// -/// @param [in] x The value to calculate approximate sine for. -/// -/// @returns -/// The approximate sine of value. -FfxFloat16 ffxParabolicSinHalf(FfxFloat16 x) -{ - return x * abs(x) - x; -} - -/// An approximation of sine. -/// -/// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range -/// is {-1/4 to 1/4} representing {-1 to 1}. -/// -/// @param [in] x The value to calculate approximate sine for. -/// -/// @returns -/// The approximate sine of value. -FfxFloat16x2 ffxParabolicSinHalf(FfxFloat16x2 x) -{ - return x * abs(x) - x; -} - -/// An approximation of cosine. -/// -/// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range -/// is {-1/4 to 1/4} representing {-1 to 1}. -/// -/// @param [in] x The value to calculate approximate cosine for. -/// -/// @returns -/// The approximate cosine of value. -FfxFloat16 ffxParabolicCosHalf(FfxFloat16 x) -{ - x = ffxFract(x * FFX_BROADCAST_FLOAT16(0.5) + FFX_BROADCAST_FLOAT16(0.75)); - x = x * FFX_BROADCAST_FLOAT16(2.0) - FFX_BROADCAST_FLOAT16(1.0); - return ffxParabolicSinHalf(x); -} - -/// An approximation of cosine. -/// -/// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range -/// is {-1/4 to 1/4} representing {-1 to 1}. -/// -/// @param [in] x The value to calculate approximate cosine for. -/// -/// @returns -/// The approximate cosine of value. -FfxFloat16x2 ffxParabolicCosHalf(FfxFloat16x2 x) -{ - x = ffxFract(x * FFX_BROADCAST_FLOAT16X2(0.5) + FFX_BROADCAST_FLOAT16X2(0.75)); - x = x * FFX_BROADCAST_FLOAT16X2(2.0) - FFX_BROADCAST_FLOAT16X2(1.0); - return ffxParabolicSinHalf(x); -} - -/// An approximation of both sine and cosine. -/// -/// Valid input range is {-1 to 1} representing {0 to 2 pi}, and the output range -/// is {-1/4 to 1/4} representing {-1 to 1}. -/// -/// @param [in] x The value to calculate approximate cosine for. -/// -/// @returns -/// A FfxFloat32x2 containing approximations of both sine and cosine of value. -FfxFloat16x2 ffxParabolicSinCosHalf(FfxFloat16 x) -{ - FfxFloat16 y = ffxFract(x * FFX_BROADCAST_FLOAT16(0.5) + FFX_BROADCAST_FLOAT16(0.75)); - y = y * FFX_BROADCAST_FLOAT16(2.0) - FFX_BROADCAST_FLOAT16(1.0); - return ffxParabolicSinHalf(FfxFloat16x2(x, y)); -} - -/// Conditional free logic AND operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxUInt16 ffxZeroOneAndHalf(FfxUInt16 x, FfxUInt16 y) -{ - return min(x, y); -} - -/// Conditional free logic AND operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxZeroOneAndHalf(FfxUInt16x2 x, FfxUInt16x2 y) -{ - return min(x, y); -} - -/// Conditional free logic AND operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxUInt16x3 ffxZeroOneAndHalf(FfxUInt16x3 x, FfxUInt16x3 y) -{ - return min(x, y); -} - -/// Conditional free logic AND operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxUInt16x4 ffxZeroOneAndHalf(FfxUInt16x4 x, FfxUInt16x4 y) -{ - return min(x, y); -} - -/// Conditional free logic NOT operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the NOT operator. -/// @param [in] y The second value to be fed into the NOT operator. -/// -/// @returns -/// Result of the NOT operation. -/// -/// @ingroup GPUCore -FfxUInt16 ffxZeroOneNotHalf(FfxUInt16 x) -{ - return x ^ FFX_BROADCAST_UINT16(1); -} - -/// Conditional free logic NOT operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the NOT operator. -/// @param [in] y The second value to be fed into the NOT operator. -/// -/// @returns -/// Result of the NOT operation. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxZeroOneNotHalf(FfxUInt16x2 x) -{ - return x ^ FFX_BROADCAST_UINT16X2(1); -} - -/// Conditional free logic NOT operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the NOT operator. -/// @param [in] y The second value to be fed into the NOT operator. -/// -/// @returns -/// Result of the NOT operation. -/// -/// @ingroup GPUCore -FfxUInt16x3 ffxZeroOneNotHalf(FfxUInt16x3 x) -{ - return x ^ FFX_BROADCAST_UINT16X3(1); -} - -/// Conditional free logic NOT operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the NOT operator. -/// @param [in] y The second value to be fed into the NOT operator. -/// -/// @returns -/// Result of the NOT operation. -/// -/// @ingroup GPUCore -FfxUInt16x4 ffxZeroOneNotHalf(FfxUInt16x4 x) -{ - return x ^ FFX_BROADCAST_UINT16X4(1); -} - -/// Conditional free logic OR operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxUInt16 ffxZeroOneOrHalf(FfxUInt16 x, FfxUInt16 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxZeroOneOrHalf(FfxUInt16x2 x, FfxUInt16x2 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxUInt16x3 ffxZeroOneOrHalf(FfxUInt16x3 x, FfxUInt16x3 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxUInt16x4 ffxZeroOneOrHalf(FfxUInt16x4 x, FfxUInt16x4 y) -{ - return max(x, y); -} - -/// Convert a half-precision FfxFloat32 value between 0.0f and 1.0f to a half-precision Uint. -/// -/// @param [in] x The value to converted to a Uint. -/// -/// @returns -/// The converted Uint value. -/// -/// @ingroup GPUCore -FfxUInt16 ffxZeroOneFloat16ToUint16(FfxFloat16 x) -{ - return FFX_TO_UINT16(x * FFX_TO_FLOAT16(FFX_TO_UINT16(1))); -} - -/// Convert a half-precision FfxFloat32 value between 0.0f and 1.0f to a half-precision Uint. -/// -/// @param [in] x The value to converted to a Uint. -/// -/// @returns -/// The converted Uint value. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxZeroOneFloat16x2ToUint16x2(FfxFloat16x2 x) -{ - return FFX_TO_UINT16X2(x * FFX_TO_FLOAT16X2(FfxUInt16x2(1, 1))); -} - -/// Convert a half-precision FfxFloat32 value between 0.0f and 1.0f to a half-precision Uint. -/// -/// @param [in] x The value to converted to a Uint. -/// -/// @returns -/// The converted Uint value. -/// -/// @ingroup GPUCore -FfxUInt16x3 ffxZeroOneFloat16x3ToUint16x3(FfxFloat16x3 x) -{ - return FFX_TO_UINT16X3(x * FFX_TO_FLOAT16X3(FfxUInt16x3(1, 1, 1))); -} - -/// Convert a half-precision FfxFloat32 value between 0.0f and 1.0f to a half-precision Uint. -/// -/// @param [in] x The value to converted to a Uint. -/// -/// @returns -/// The converted Uint value. -/// -/// @ingroup GPUCore -FfxUInt16x4 ffxZeroOneFloat16x4ToUint16x4(FfxFloat16x4 x) -{ - return FFX_TO_UINT16X4(x * FFX_TO_FLOAT16X4(FfxUInt16x4(1, 1, 1, 1))); -} - -/// Convert a half-precision FfxUInt32 value between 0 and 1 to a half-precision FfxFloat32. -/// -/// @param [in] x The value to converted to a half-precision FfxFloat32. -/// -/// @returns -/// The converted half-precision FfxFloat32 value. -/// -/// @ingroup GPUCore -FfxFloat16 ffxZeroOneUint16ToFloat16(FfxUInt16 x) -{ - return FFX_TO_FLOAT16(x * FFX_TO_UINT16(FFX_TO_FLOAT16(1.0))); -} - -/// Convert a half-precision FfxUInt32 value between 0 and 1 to a half-precision FfxFloat32. -/// -/// @param [in] x The value to converted to a half-precision FfxFloat32. -/// -/// @returns -/// The converted half-precision FfxFloat32 value. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxZeroOneUint16x2ToFloat16x2(FfxUInt16x2 x) -{ - return FFX_TO_FLOAT16X2(x * FFX_TO_UINT16X2(FfxUInt16x2(FFX_TO_FLOAT16(1.0), FFX_TO_FLOAT16(1.0)))); -} - -/// Convert a half-precision FfxUInt32 value between 0 and 1 to a half-precision FfxFloat32. -/// -/// @param [in] x The value to converted to a half-precision FfxFloat32. -/// -/// @returns -/// The converted half-precision FfxFloat32 value. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxZeroOneUint16x3ToFloat16x3(FfxUInt16x3 x) -{ - return FFX_TO_FLOAT16X3(x * FFX_TO_UINT16X3(FfxUInt16x3(FFX_TO_FLOAT16(1.0), FFX_TO_FLOAT16(1.0), FFX_TO_FLOAT16(1.0)))); -} - -/// Convert a half-precision FfxUInt32 value between 0 and 1 to a half-precision FfxFloat32. -/// -/// @param [in] x The value to converted to a half-precision FfxFloat32. -/// -/// @returns -/// The converted half-precision FfxFloat32 value. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxZeroOneUint16x4ToFloat16x4(FfxUInt16x4 x) -{ - return FFX_TO_FLOAT16X4(x * FFX_TO_UINT16X4(FfxUInt16x4(FFX_TO_FLOAT16(1.0), FFX_TO_FLOAT16(1.0), FFX_TO_FLOAT16(1.0), FFX_TO_FLOAT16(1.0)))); -} - -/// Conditional free logic AND operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxFloat16 ffxZeroOneAndHalf(FfxFloat16 x, FfxFloat16 y) -{ - return min(x, y); -} - -/// Conditional free logic AND operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxZeroOneAndHalf(FfxFloat16x2 x, FfxFloat16x2 y) -{ - return min(x, y); -} - -/// Conditional free logic AND operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxZeroOneAndHalf(FfxFloat16x3 x, FfxFloat16x3 y) -{ - return min(x, y); -} - -/// Conditional free logic AND operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// -/// @returns -/// Result of the AND operation. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxZeroOneAndHalf(FfxFloat16x4 x, FfxFloat16x4 y) -{ - return min(x, y); -} - -/// Conditional free logic AND NOT operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND NOT operator. -/// @param [in] y The second value to be fed into the AND NOT operator. -/// -/// @returns -/// Result of the AND NOT operation. -/// -/// @ingroup GPUCore -FfxFloat16 ffxSignedZeroOneAndOrHalf(FfxFloat16 x, FfxFloat16 y) -{ - return (-x) * y + FFX_BROADCAST_FLOAT16(1.0); -} - -/// Conditional free logic AND NOT operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND NOT operator. -/// @param [in] y The second value to be fed into the AND NOT operator. -/// -/// @returns -/// Result of the AND NOT operation. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxSignedZeroOneAndOrHalf(FfxFloat16x2 x, FfxFloat16x2 y) -{ - return (-x) * y + FFX_BROADCAST_FLOAT16X2(1.0); -} - -/// Conditional free logic AND NOT operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND NOT operator. -/// @param [in] y The second value to be fed into the AND NOT operator. -/// -/// @returns -/// Result of the AND NOT operation. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxSignedZeroOneAndOrHalf(FfxFloat16x3 x, FfxFloat16x3 y) -{ - return (-x) * y + FFX_BROADCAST_FLOAT16X3(1.0); -} - -/// Conditional free logic AND NOT operation using two half-precision values. -/// -/// @param [in] x The first value to be fed into the AND NOT operator. -/// @param [in] y The second value to be fed into the AND NOT operator. -/// -/// @returns -/// Result of the AND NOT operation. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxSignedZeroOneAndOrHalf(FfxFloat16x4 x, FfxFloat16x4 y) -{ - return (-x) * y + FFX_BROADCAST_FLOAT16X4(1.0); -} - -/// Conditional free logic AND operation using two half-precision values followed by -/// a NOT operation using the resulting value and a third half-precision value. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// @param [in] z The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat16 ffxZeroOneAndOrHalf(FfxFloat16 x, FfxFloat16 y, FfxFloat16 z) -{ - return FfxFloat16(ffxSaturate(x * y + z)); -} - -/// Conditional free logic AND operation using two half-precision values followed by -/// a NOT operation using the resulting value and a third half-precision value. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// @param [in] z The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxZeroOneAndOrHalf(FfxFloat16x2 x, FfxFloat16x2 y, FfxFloat16x2 z) -{ - return FfxFloat16x2(ffxSaturate(x * y + z)); -} - -/// Conditional free logic AND operation using two half-precision values followed by -/// a NOT operation using the resulting value and a third half-precision value. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// @param [in] z The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxZeroOneAndOrHalf(FfxFloat16x3 x, FfxFloat16x3 y, FfxFloat16x3 z) -{ - return FfxFloat16x3(ffxSaturate(x * y + z)); -} - -/// Conditional free logic AND operation using two half-precision values followed by -/// a NOT operation using the resulting value and a third half-precision value. -/// -/// @param [in] x The first value to be fed into the AND operator. -/// @param [in] y The second value to be fed into the AND operator. -/// @param [in] z The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxZeroOneAndOrHalf(FfxFloat16x4 x, FfxFloat16x4 y, FfxFloat16x4 z) -{ - return FfxFloat16x4(ffxSaturate(x * y + z)); -} - -/// Given a half-precision value, returns 1.0 if greater than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the greater than zero comparison. -/// -/// @ingroup GPUCore -FfxFloat16 ffxZeroOneIsGreaterThanZeroHalf(FfxFloat16 x) -{ - return FfxFloat16(ffxSaturate(x * FFX_BROADCAST_FLOAT16(FFX_POSITIVE_INFINITY_HALF))); -} - -/// Given a half-precision value, returns 1.0 if greater than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the greater than zero comparison. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxZeroOneIsGreaterThanZeroHalf(FfxFloat16x2 x) -{ - return FfxFloat16x2(ffxSaturate(x * FFX_BROADCAST_FLOAT16X2(FFX_POSITIVE_INFINITY_HALF))); -} - -/// Given a half-precision value, returns 1.0 if greater than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the greater than zero comparison. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxZeroOneIsGreaterThanZeroHalf(FfxFloat16x3 x) -{ - return FfxFloat16x3(ffxSaturate(x * FFX_BROADCAST_FLOAT16X3(FFX_POSITIVE_INFINITY_HALF))); -} - -/// Given a half-precision value, returns 1.0 if greater than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the greater than zero comparison. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxZeroOneIsGreaterThanZeroHalf(FfxFloat16x4 x) -{ - return FfxFloat16x4(ffxSaturate(x * FFX_BROADCAST_FLOAT16X4(FFX_POSITIVE_INFINITY_HALF))); -} - -/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat16 ffxZeroOneNotHalf(FfxFloat16 x) -{ - return FFX_BROADCAST_FLOAT16(1.0) - x; -} - -/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxZeroOneNotHalf(FfxFloat16x2 x) -{ - return FFX_BROADCAST_FLOAT16X2(1.0) - x; -} - -/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxZeroOneNotHalf(FfxFloat16x3 x) -{ - return FFX_BROADCAST_FLOAT16X3(1.0) - x; -} - -/// Conditional free logic signed NOT operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the AND OR operator. -/// -/// @returns -/// Result of the AND OR operation. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxZeroOneNotHalf(FfxFloat16x4 x) -{ - return FFX_BROADCAST_FLOAT16X4(1.0) - x; -} - -/// Conditional free logic OR operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxFloat16 ffxZeroOneOrHalf(FfxFloat16 x, FfxFloat16 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxZeroOneOrHalf(FfxFloat16x2 x, FfxFloat16x2 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxZeroOneOrHalf(FfxFloat16x3 x, FfxFloat16x3 y) -{ - return max(x, y); -} - -/// Conditional free logic OR operation using two half-precision FfxFloat32 values. -/// -/// @param [in] x The first value to be fed into the OR operator. -/// @param [in] y The second value to be fed into the OR operator. -/// -/// @returns -/// Result of the OR operation. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxZeroOneOrHalf(FfxFloat16x4 x, FfxFloat16x4 y) -{ - return max(x, y); -} - -/// Choose between two half-precision FfxFloat32 values if the first paramter is greater than zero. -/// -/// @param [in] x The value to compare against zero. -/// @param [in] y The value to return if the comparision is greater than zero. -/// @param [in] z The value to return if the comparision is less than or equal to zero. -/// -/// @returns -/// The selected value. -/// -/// @ingroup GPUCore -FfxFloat16 ffxZeroOneSelectHalf(FfxFloat16 x, FfxFloat16 y, FfxFloat16 z) -{ - FfxFloat16 r = (-x) * z + z; - return x * y + r; -} - -/// Choose between two half-precision FfxFloat32 values if the first paramter is greater than zero. -/// -/// @param [in] x The value to compare against zero. -/// @param [in] y The value to return if the comparision is greater than zero. -/// @param [in] z The value to return if the comparision is less than or equal to zero. -/// -/// @returns -/// The selected value. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxZeroOneSelectHalf(FfxFloat16x2 x, FfxFloat16x2 y, FfxFloat16x2 z) -{ - FfxFloat16x2 r = (-x) * z + z; - return x * y + r; -} - -/// Choose between two half-precision FfxFloat32 values if the first paramter is greater than zero. -/// -/// @param [in] x The value to compare against zero. -/// @param [in] y The value to return if the comparision is greater than zero. -/// @param [in] z The value to return if the comparision is less than or equal to zero. -/// -/// @returns -/// The selected value. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxZeroOneSelectHalf(FfxFloat16x3 x, FfxFloat16x3 y, FfxFloat16x3 z) -{ - FfxFloat16x3 r = (-x) * z + z; - return x * y + r; -} - -/// Choose between two half-precision FfxFloat32 values if the first paramter is greater than zero. -/// -/// @param [in] x The value to compare against zero. -/// @param [in] y The value to return if the comparision is greater than zero. -/// @param [in] z The value to return if the comparision is less than or equal to zero. -/// -/// @returns -/// The selected value. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxZeroOneSelectHalf(FfxFloat16x4 x, FfxFloat16x4 y, FfxFloat16x4 z) -{ - FfxFloat16x4 r = (-x) * z + z; - return x * y + r; -} - -/// Given a half-precision value, returns 1.0 if less than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the sign value. -/// -/// @ingroup GPUCore -FfxFloat16 ffxZeroOneIsSignedHalf(FfxFloat16 x) -{ - return FfxFloat16(ffxSaturate(x * FFX_BROADCAST_FLOAT16(FFX_NEGATIVE_INFINITY_HALF))); -} - -/// Given a half-precision value, returns 1.0 if less than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the sign value. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxZeroOneIsSignedHalf(FfxFloat16x2 x) -{ - return FfxFloat16x2(ffxSaturate(x * FFX_BROADCAST_FLOAT16X2(FFX_NEGATIVE_INFINITY_HALF))); -} - -/// Given a half-precision value, returns 1.0 if less than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the sign value. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxZeroOneIsSignedHalf(FfxFloat16x3 x) -{ - return FfxFloat16x3(ffxSaturate(x * FFX_BROADCAST_FLOAT16X3(FFX_NEGATIVE_INFINITY_HALF))); -} - -/// Given a half-precision value, returns 1.0 if less than zero and 0.0 if not. -/// -/// @param [in] x The value to be compared. -/// -/// @returns -/// Result of the sign value. -/// -/// @ingroup GPUCore -FfxFloat16x4 ffxZeroOneIsSignedHalf(FfxFloat16x4 x) -{ - return FfxFloat16x4(ffxSaturate(x * FFX_BROADCAST_FLOAT16X4(FFX_NEGATIVE_INFINITY_HALF))); -} - -/// Compute a Rec.709 color space. -/// -/// Rec.709 is used for some HDTVs. -/// -/// Both Rec.709 and sRGB have a linear segment which as spec'ed would intersect the curved segment 2 times. -/// (a.) For 8-bit sRGB, steps {0 to 10.3} are in the linear region (4% of the encoding range). -/// (b.) For 8-bit 709, steps {0 to 20.7} are in the linear region (8% of the encoding range). -/// -/// @param [in] c The color to convert to Rec. 709. -/// -/// @returns -/// The color in Rec.709 space. -/// -/// @ingroup GPUCore -FfxFloat16 ffxRec709FromLinearHalf(FfxFloat16 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.018 * 4.5, 4.5, 0.45); - FfxFloat16x2 k = FfxFloat16x2(1.099, -0.099); - return clamp(j.x, c * j.y, pow(c, j.z) * k.x + k.y); -} - -/// Compute a Rec.709 color space. -/// -/// Rec.709 is used for some HDTVs. -/// -/// Both Rec.709 and sRGB have a linear segment which as spec'ed would intersect the curved segment 2 times. -/// (a.) For 8-bit sRGB, steps {0 to 10.3} are in the linear region (4% of the encoding range). -/// (b.) For 8-bit 709, steps {0 to 20.7} are in the linear region (8% of the encoding range). -/// -/// @param [in] c The color to convert to Rec. 709. -/// -/// @returns -/// The color in Rec.709 space. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxRec709FromLinearHalf(FfxFloat16x2 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.018 * 4.5, 4.5, 0.45); - FfxFloat16x2 k = FfxFloat16x2(1.099, -0.099); - return clamp(j.xx, c * j.yy, pow(c, j.zz) * k.xx + k.yy); -} - -/// Compute a Rec.709 color space. -/// -/// Rec.709 is used for some HDTVs. -/// -/// Both Rec.709 and sRGB have a linear segment which as spec'ed would intersect the curved segment 2 times. -/// (a.) For 8-bit sRGB, steps {0 to 10.3} are in the linear region (4% of the encoding range). -/// (b.) For 8-bit 709, steps {0 to 20.7} are in the linear region (8% of the encoding range). -/// -/// @param [in] c The color to convert to Rec. 709. -/// -/// @returns -/// The color in Rec.709 space. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxRec709FromLinearHalf(FfxFloat16x3 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.018 * 4.5, 4.5, 0.45); - FfxFloat16x2 k = FfxFloat16x2(1.099, -0.099); - return clamp(j.xxx, c * j.yyy, pow(c, j.zzz) * k.xxx + k.yyy); -} - -/// Compute a gamma value from a linear value. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// Note: 'rcpX' is '1/x', where the 'x' is what would be used in ffxLinearFromGammaHalf. -/// -/// @param [in] c The value to convert to gamma space from linear. -/// @param [in] rcpX The reciprocal of power value used for the gamma curve. -/// -/// @returns -/// A value in gamma space. -/// -/// @ingroup GPUCore -FfxFloat16 ffxGammaFromLinearHalf(FfxFloat16 c, FfxFloat16 rcpX) -{ - return pow(c, FFX_BROADCAST_FLOAT16(rcpX)); -} - -/// Compute a gamma value from a linear value. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// Note: 'rcpX' is '1/x', where the 'x' is what would be used in ffxLinearFromGammaHalf. -/// -/// @param [in] c The value to convert to gamma space from linear. -/// @param [in] rcpX The reciprocal of power value used for the gamma curve. -/// -/// @returns -/// A value in gamma space. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxGammaFromLinearHalf(FfxFloat16x2 c, FfxFloat16 rcpX) -{ - return pow(c, FFX_BROADCAST_FLOAT16X2(rcpX)); -} - -/// Compute a gamma value from a linear value. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// Note: 'rcpX' is '1/x', where the 'x' is what would be used in ffxLinearFromGammaHalf. -/// -/// @param [in] c The value to convert to gamma space from linear. -/// @param [in] rcpX The reciprocal of power value used for the gamma curve. -/// -/// @returns -/// A value in gamma space. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxGammaFromLinearHalf(FfxFloat16x3 c, FfxFloat16 rcpX) -{ - return pow(c, FFX_BROADCAST_FLOAT16X3(rcpX)); -} - -/// Compute an SRGB value from a linear value. -/// -/// @param [in] c The value to convert to SRGB from linear. -/// -/// @returns -/// A value in SRGB space. -/// -/// @ingroup GPUCore -FfxFloat16 ffxSrgbFromLinearHalf(FfxFloat16 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.0031308 * 12.92, 12.92, 1.0 / 2.4); - FfxFloat16x2 k = FfxFloat16x2(1.055, -0.055); - return clamp(j.x, c * j.y, pow(c, j.z) * k.x + k.y); -} - -/// Compute an SRGB value from a linear value. -/// -/// @param [in] c The value to convert to SRGB from linear. -/// -/// @returns -/// A value in SRGB space. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxSrgbFromLinearHalf(FfxFloat16x2 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.0031308 * 12.92, 12.92, 1.0 / 2.4); - FfxFloat16x2 k = FfxFloat16x2(1.055, -0.055); - return clamp(j.xx, c * j.yy, pow(c, j.zz) * k.xx + k.yy); -} - -/// Compute an SRGB value from a linear value. -/// -/// @param [in] c The value to convert to SRGB from linear. -/// -/// @returns -/// A value in SRGB space. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxSrgbFromLinearHalf(FfxFloat16x3 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.0031308 * 12.92, 12.92, 1.0 / 2.4); - FfxFloat16x2 k = FfxFloat16x2(1.055, -0.055); - return clamp(j.xxx, c * j.yyy, pow(c, j.zzz) * k.xxx + k.yyy); -} - -/// Compute the square root of a value. -/// -/// @param [in] c The value to compute the square root for. -/// -/// @returns -/// A square root of the input value. -/// -/// @ingroup GPUCore -FfxFloat16 ffxSquareRootHalf(FfxFloat16 c) -{ - return sqrt(c); -} - -/// Compute the square root of a value. -/// -/// @param [in] c The value to compute the square root for. -/// -/// @returns -/// A square root of the input value. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxSquareRootHalf(FfxFloat16x2 c) -{ - return sqrt(c); -} - -/// Compute the square root of a value. -/// -/// @param [in] c The value to compute the square root for. -/// -/// @returns -/// A square root of the input value. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxSquareRootHalf(FfxFloat16x3 c) -{ - return sqrt(c); -} - -/// Compute the cube root of a value. -/// -/// @param [in] c The value to compute the cube root for. -/// -/// @returns -/// A cube root of the input value. -/// -/// @ingroup GPUCore -FfxFloat16 ffxCubeRootHalf(FfxFloat16 c) -{ - return pow(c, FFX_BROADCAST_FLOAT16(1.0 / 3.0)); -} - -/// Compute the cube root of a value. -/// -/// @param [in] c The value to compute the cube root for. -/// -/// @returns -/// A cube root of the input value. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxCubeRootHalf(FfxFloat16x2 c) -{ - return pow(c, FFX_BROADCAST_FLOAT16X2(1.0 / 3.0)); -} - -/// Compute the cube root of a value. -/// -/// @param [in] c The value to compute the cube root for. -/// -/// @returns -/// A cube root of the input value. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxCubeRootHalf(FfxFloat16x3 c) -{ - return pow(c, FFX_BROADCAST_FLOAT16X3(1.0 / 3.0)); -} - -/// Compute a linear value from a REC.709 value. -/// -/// @param [in] c The value to convert to linear from REC.709. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat16 ffxLinearFromRec709Half(FfxFloat16 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.081 / 4.5, 1.0 / 4.5, 1.0 / 0.45); - FfxFloat16x2 k = FfxFloat16x2(1.0 / 1.099, 0.099 / 1.099); - return ffxZeroOneSelectHalf(ffxZeroOneIsSignedHalf(c - j.x), c * j.y, pow(c * k.x + k.y, j.z)); -} - -/// Compute a linear value from a REC.709 value. -/// -/// @param [in] c The value to convert to linear from REC.709. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxLinearFromRec709Half(FfxFloat16x2 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.081 / 4.5, 1.0 / 4.5, 1.0 / 0.45); - FfxFloat16x2 k = FfxFloat16x2(1.0 / 1.099, 0.099 / 1.099); - return ffxZeroOneSelectHalf(ffxZeroOneIsSignedHalf(c - j.xx), c * j.yy, pow(c * k.xx + k.yy, j.zz)); -} - -/// Compute a linear value from a REC.709 value. -/// -/// @param [in] c The value to convert to linear from REC.709. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxLinearFromRec709Half(FfxFloat16x3 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.081 / 4.5, 1.0 / 4.5, 1.0 / 0.45); - FfxFloat16x2 k = FfxFloat16x2(1.0 / 1.099, 0.099 / 1.099); - return ffxZeroOneSelectHalf(ffxZeroOneIsSignedHalf(c - j.xxx), c * j.yyy, pow(c * k.xxx + k.yyy, j.zzz)); -} - -/// Compute a linear value from a value in a gamma space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] c The value to convert to linear in gamma space. -/// @param [in] x The power value used for the gamma curve. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat16 ffxLinearFromGammaHalf(FfxFloat16 c, FfxFloat16 x) -{ - return pow(c, FFX_BROADCAST_FLOAT16(x)); -} - -/// Compute a linear value from a value in a gamma space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] c The value to convert to linear in gamma space. -/// @param [in] x The power value used for the gamma curve. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxLinearFromGammaHalf(FfxFloat16x2 c, FfxFloat16 x) -{ - return pow(c, FFX_BROADCAST_FLOAT16X2(x)); -} - -/// Compute a linear value from a value in a gamma space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] c The value to convert to linear in gamma space. -/// @param [in] x The power value used for the gamma curve. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxLinearFromGammaHalf(FfxFloat16x3 c, FfxFloat16 x) -{ - return pow(c, FFX_BROADCAST_FLOAT16X3(x)); -} - -/// Compute a linear value from a value in a SRGB space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] c The value to convert to linear in SRGB space. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat16 ffxLinearFromSrgbHalf(FfxFloat16 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.04045 / 12.92, 1.0 / 12.92, 2.4); - FfxFloat16x2 k = FfxFloat16x2(1.0 / 1.055, 0.055 / 1.055); - return ffxZeroOneSelectHalf(ffxZeroOneIsSignedHalf(c - j.x), c * j.y, pow(c * k.x + k.y, j.z)); -} - -/// Compute a linear value from a value in a SRGB space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] c The value to convert to linear in SRGB space. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat16x2 ffxLinearFromSrgbHalf(FfxFloat16x2 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.04045 / 12.92, 1.0 / 12.92, 2.4); - FfxFloat16x2 k = FfxFloat16x2(1.0 / 1.055, 0.055 / 1.055); - return ffxZeroOneSelectHalf(ffxZeroOneIsSignedHalf(c - j.xx), c * j.yy, pow(c * k.xx + k.yy, j.zz)); -} - -/// Compute a linear value from a value in a SRGB space. -/// -/// Typically 2.2 for some PC displays, or 2.4-2.5 for CRTs, or 2.2 FreeSync2 native. -/// -/// @param [in] c The value to convert to linear in SRGB space. -/// -/// @returns -/// A value in linear space. -/// -/// @ingroup GPUCore -FfxFloat16x3 ffxLinearFromSrgbHalf(FfxFloat16x3 c) -{ - FfxFloat16x3 j = FfxFloat16x3(0.04045 / 12.92, 1.0 / 12.92, 2.4); - FfxFloat16x2 k = FfxFloat16x2(1.0 / 1.055, 0.055 / 1.055); - return ffxZeroOneSelectHalf(ffxZeroOneIsSignedHalf(c - j.xxx), c * j.yyy, pow(c * k.xxx + k.yyy, j.zzz)); -} - -/// A remapping of 64x1 to 8x8 imposing rotated 2x2 pixel quads in quad linear. -/// -/// 543210 -/// ====== -/// ..xxx. -/// yy...y -/// -/// @param [in] a The input 1D coordinates to remap. -/// -/// @returns -/// The remapped 2D coordinates. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxRemapForQuadHalf(FfxUInt32 a) -{ - return FfxUInt16x2(bitfieldExtract(a, 1u, 3u), bitfieldInsertMask(bitfieldExtract(a, 3u, 3u), a, 1u)); -} - -/// A helper function performing a remap 64x1 to 8x8 remapping which is necessary for 2D wave reductions. -/// -/// The 64-wide lane indices to 8x8 remapping is performed as follows: -/// -/// 00 01 08 09 10 11 18 19 -/// 02 03 0a 0b 12 13 1a 1b -/// 04 05 0c 0d 14 15 1c 1d -/// 06 07 0e 0f 16 17 1e 1f -/// 20 21 28 29 30 31 38 39 -/// 22 23 2a 2b 32 33 3a 3b -/// 24 25 2c 2d 34 35 3c 3d -/// 26 27 2e 2f 36 37 3e 3f -/// -/// @param [in] a The input 1D coordinate to remap. -/// -/// @returns -/// The remapped 2D coordinates. -/// -/// @ingroup GPUCore -FfxUInt16x2 ffxRemapForWaveReductionHalf(FfxUInt32 a) -{ - return FfxUInt16x2(bitfieldInsertMask(bitfieldExtract(a, 2u, 3u), a, 1u), bitfieldInsertMask(bitfieldExtract(a, 3u, 3u), bitfieldExtract(a, 1u, 2u), 2u)); -} - -#endif // FFX_HALF diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common_half.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common_half.h.meta deleted file mode 100644 index 234aa90..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_gpu_common_half.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 1bdb323791a91a5438ee8e1e63187840 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_hlsl.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_hlsl.h deleted file mode 100644 index 337eb06..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_hlsl.h +++ /dev/null @@ -1,1651 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -/// @defgroup HLSLCore HLSL Core -/// HLSL core defines and functions -/// -/// @ingroup FfxHLSL - -#define DECLARE_SRV_REGISTER(regIndex) t##regIndex -#define DECLARE_UAV_REGISTER(regIndex) u##regIndex -#define DECLARE_CB_REGISTER(regIndex) b##regIndex -#define FFX_DECLARE_SRV(regIndex) register(DECLARE_SRV_REGISTER(regIndex)) -#define FFX_DECLARE_UAV(regIndex) register(DECLARE_UAV_REGISTER(regIndex)) -#define FFX_DECLARE_CB(regIndex) register(DECLARE_CB_REGISTER(regIndex)) - -/// A define for abstracting shared memory between shading languages. -/// -/// @ingroup HLSLCore -#define FFX_GROUPSHARED groupshared - -/// A define for abstracting compute memory barriers between shading languages. -/// -/// @ingroup HLSLCore -#define FFX_GROUP_MEMORY_BARRIER GroupMemoryBarrierWithGroupSync - -/// A define for abstracting compute atomic additions between shading languages. -/// -/// @ingroup HLSLCore -#define FFX_ATOMIC_ADD(x, y) InterlockedAdd(x, y) - -/// A define added to accept static markup on functions to aid CPU/GPU portability of code. -/// -/// @ingroup HLSLCore -#define FFX_STATIC static - -/// A define for abstracting loop unrolling between shading languages. -/// -/// @ingroup HLSLCore -#define FFX_UNROLL [unroll] - -/// A define for abstracting a 'greater than' comparison operator between two types. -/// -/// @ingroup HLSLCore -#define FFX_GREATER_THAN(x, y) x > y - -/// A define for abstracting a 'greater than or equal' comparison operator between two types. -/// -/// @ingroup HLSLCore -#define FFX_GREATER_THAN_EQUAL(x, y) x >= y - -/// A define for abstracting a 'less than' comparison operator between two types. -/// -/// @ingroup HLSLCore -#define FFX_LESS_THAN(x, y) x < y - -/// A define for abstracting a 'less than or equal' comparison operator between two types. -/// -/// @ingroup HLSLCore -#define FFX_LESS_THAN_EQUAL(x, y) x <= y - -/// A define for abstracting an 'equal' comparison operator between two types. -/// -/// @ingroup HLSLCore -#define FFX_EQUAL(x, y) x == y - -/// A define for abstracting a 'not equal' comparison operator between two types. -/// -/// @ingroup HLSLCore -#define FFX_NOT_EQUAL(x, y) x != y - -/// A define for abstracting matrix multiply operations between shading languages. -/// -/// @ingroup HLSLCore -#define FFX_MATRIX_MULTIPLY(a, b) mul(a, b) - -/// A define for abstracting vector transformations between shading languages. -/// -/// @ingroup HLSLCore -#define FFX_TRANSFORM_VECTOR(a, b) mul(a, b) - -/// A define for abstracting modulo operations between shading languages. -/// -/// @ingroup HLSLCore -#define FFX_MODULO(a, b) (fmod(a, b)) - -/// Broadcast a scalar value to a 1-dimensional floating point vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_FLOAT32(x) FfxFloat32(x) - -/// Broadcast a scalar value to a 2-dimensional floating point vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_FLOAT32X2(x) FfxFloat32(x) - -/// Broadcast a scalar value to a 3-dimensional floating point vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_FLOAT32X3(x) FfxFloat32(x) - -/// Broadcast a scalar value to a 4-dimensional floating point vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_FLOAT32X4(x) FfxFloat32(x) - -/// Broadcast a scalar value to a 1-dimensional unsigned integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_UINT32(x) FfxUInt32(x) - -/// Broadcast a scalar value to a 2-dimensional unsigned integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_UINT32X2(x) FfxUInt32(x) - -/// Broadcast a scalar value to a 4-dimensional unsigned integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_UINT32X3(x) FfxUInt32(x) - -/// Broadcast a scalar value to a 4-dimensional unsigned integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_UINT32X4(x) FfxUInt32(x) - -/// Broadcast a scalar value to a 1-dimensional signed integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_INT32(x) FfxInt32(x) - -/// Broadcast a scalar value to a 2-dimensional signed integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_INT32X2(x) FfxInt32(x) - -/// Broadcast a scalar value to a 3-dimensional signed integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_INT32X3(x) FfxInt32(x) - -/// Broadcast a scalar value to a 4-dimensional signed integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_INT32X4(x) FfxInt32(x) - -/// Broadcast a scalar value to a 1-dimensional half-precision floating point vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_FLOAT16(a) FFX_MIN16_F(a) - -/// Broadcast a scalar value to a 2-dimensional half-precision floating point vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_FLOAT16X2(a) FFX_MIN16_F(a) - -/// Broadcast a scalar value to a 3-dimensional half-precision floating point vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_FLOAT16X3(a) FFX_MIN16_F(a) - -/// Broadcast a scalar value to a 4-dimensional half-precision floating point vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_FLOAT16X4(a) FFX_MIN16_F(a) - -/// Broadcast a scalar value to a 1-dimensional half-precision unsigned integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_UINT16(a) FFX_MIN16_U(a) - -/// Broadcast a scalar value to a 2-dimensional half-precision unsigned integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_UINT16X2(a) FFX_MIN16_U(a) - -/// Broadcast a scalar value to a 3-dimensional half-precision unsigned integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_UINT16X3(a) FFX_MIN16_U(a) - -/// Broadcast a scalar value to a 4-dimensional half-precision unsigned integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_UINT16X4(a) FFX_MIN16_U(a) - -/// Broadcast a scalar value to a 1-dimensional half-precision signed integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_INT16(a) FFX_MIN16_I(a) - -/// Broadcast a scalar value to a 2-dimensional half-precision signed integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_INT16X2(a) FFX_MIN16_I(a) - -/// Broadcast a scalar value to a 3-dimensional half-precision signed integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_INT16X3(a) FFX_MIN16_I(a) - -/// Broadcast a scalar value to a 4-dimensional half-precision signed integer vector. -/// -/// @ingroup HLSLCore -#define FFX_BROADCAST_MIN_INT16X4(a) FFX_MIN16_I(a) - -/// Pack 2x32-bit floating point values in a single 32bit value. -/// -/// This function first converts each component of value into their nearest 16-bit floating -/// point representation, and then stores the X and Y components in the lower and upper 16 bits of the -/// 32bit unsigned integer respectively. -/// -/// @param [in] value A 2-dimensional floating point value to convert and pack. -/// -/// @returns -/// A packed 32bit value containing 2 16bit floating point values. -/// -/// @ingroup HLSLCore -FfxUInt32 packHalf2x16(FfxFloat32x2 value) -{ - return f32tof16(value.x) | (f32tof16(value.y) << 16); -} - -/// Broadcast a scalar value to a 2-dimensional floating point vector. -/// -/// @param [in] value The value to to broadcast. -/// -/// @returns -/// A 2-dimensional floating point vector with value in each component. -/// -/// @ingroup HLSLCore -FfxFloat32x2 ffxBroadcast2(FfxFloat32 value) -{ - return FfxFloat32x2(value, value); -} - -/// Broadcast a scalar value to a 3-dimensional floating point vector. -/// -/// @param [in] value The value to to broadcast. -/// -/// @returns -/// A 3-dimensional floating point vector with value in each component. -/// -/// @ingroup HLSLCore -FfxFloat32x3 ffxBroadcast3(FfxFloat32 value) -{ - return FfxFloat32x3(value, value, value); -} - -/// Broadcast a scalar value to a 4-dimensional floating point vector. -/// -/// @param [in] value The value to to broadcast. -/// -/// @returns -/// A 4-dimensional floating point vector with value in each component. -/// -/// @ingroup HLSLCore -FfxFloat32x4 ffxBroadcast4(FfxFloat32 value) -{ - return FfxFloat32x4(value, value, value, value); -} - -/// Broadcast a scalar value to a 2-dimensional signed integer vector. -/// -/// @param [in] value The value to to broadcast. -/// -/// @returns -/// A 2-dimensional signed integer vector with value in each component. -/// -/// @ingroup HLSLCore -FfxInt32x2 ffxBroadcast2(FfxInt32 value) -{ - return FfxInt32x2(value, value); -} - -/// Broadcast a scalar value to a 3-dimensional signed integer vector. -/// -/// @param [in] value The value to to broadcast. -/// -/// @returns -/// A 3-dimensional signed integer vector with value in each component. -/// -/// @ingroup HLSLCore -FfxUInt32x3 ffxBroadcast3(FfxInt32 value) -{ - return FfxUInt32x3(value, value, value); -} - -/// Broadcast a scalar value to a 4-dimensional signed integer vector. -/// -/// @param [in] value The value to to broadcast. -/// -/// @returns -/// A 4-dimensional signed integer vector with value in each component. -/// -/// @ingroup HLSLCore -FfxInt32x4 ffxBroadcast4(FfxInt32 value) -{ - return FfxInt32x4(value, value, value, value); -} - -/// Broadcast a scalar value to a 2-dimensional unsigned integer vector. -/// -/// @param [in] value The value to to broadcast. -/// -/// @returns -/// A 2-dimensional unsigned integer vector with value in each component. -/// -/// @ingroup HLSLCore -FfxUInt32x2 ffxBroadcast2(FfxUInt32 value) -{ - return FfxUInt32x2(value, value); -} - -/// Broadcast a scalar value to a 3-dimensional unsigned integer vector. -/// -/// @param [in] value The value to to broadcast. -/// -/// @returns -/// A 3-dimensional unsigned integer vector with value in each component. -/// -/// @ingroup HLSLCore -FfxUInt32x3 ffxBroadcast3(FfxUInt32 value) -{ - return FfxUInt32x3(value, value, value); -} - -/// Broadcast a scalar value to a 4-dimensional unsigned integer vector. -/// -/// @param [in] value The value to to broadcast. -/// -/// @returns -/// A 4-dimensional unsigned integer vector with value in each component. -/// -/// @ingroup HLSLCore -FfxUInt32x4 ffxBroadcast4(FfxUInt32 value) -{ - return FfxUInt32x4(value, value, value, value); -} - -FfxUInt32 bitfieldExtract(FfxUInt32 src, FfxUInt32 off, FfxUInt32 bits) -{ - FfxUInt32 mask = (1u << bits) - 1; - return (src >> off) & mask; -} - -FfxUInt32 bitfieldInsert(FfxUInt32 src, FfxUInt32 ins, FfxUInt32 mask) -{ - return (ins & mask) | (src & (~mask)); -} - -FfxUInt32 bitfieldInsertMask(FfxUInt32 src, FfxUInt32 ins, FfxUInt32 bits) -{ - FfxUInt32 mask = (1u << bits) - 1; - return (ins & mask) | (src & (~mask)); -} - -/// Interprets the bit pattern of x as an unsigned integer. -/// -/// @param [in] x The input value. -/// -/// @returns -/// The input interpreted as an unsigned integer. -/// -/// @ingroup HLSLCore -FfxUInt32 ffxAsUInt32(FfxFloat32 x) -{ - return asuint(x); -} - -/// Interprets the bit pattern of x as an unsigned integer. -/// -/// @param [in] x The input value. -/// -/// @returns -/// The input interpreted as an unsigned integer. -/// -/// @ingroup HLSLCore -FfxUInt32x2 ffxAsUInt32(FfxFloat32x2 x) -{ - return asuint(x); -} - -/// Interprets the bit pattern of x as an unsigned integer. -/// -/// @param [in] x The input value. -/// -/// @returns -/// The input interpreted as an unsigned integer. -/// -/// @ingroup HLSLCore -FfxUInt32x3 ffxAsUInt32(FfxFloat32x3 x) -{ - return asuint(x); -} - -/// Interprets the bit pattern of x as an unsigned integer. -/// -/// @param [in] x The input value. -/// -/// @returns -/// The input interpreted as an unsigned integer. -/// -/// @ingroup HLSLCore -FfxUInt32x4 ffxAsUInt32(FfxFloat32x4 x) -{ - return asuint(x); -} - -/// Interprets the bit pattern of x as a floating-point number. -/// -/// @param [in] x The input value. -/// -/// @returns -/// The input interpreted as a floating-point number. -/// -/// @ingroup HLSLCore -FfxFloat32 ffxAsFloat(FfxUInt32 x) -{ - return asfloat(x); -} - -/// Interprets the bit pattern of x as a floating-point number. -/// -/// @param [in] x The input value. -/// -/// @returns -/// The input interpreted as a floating-point number. -/// -/// @ingroup HLSLCore -FfxFloat32x2 ffxAsFloat(FfxUInt32x2 x) -{ - return asfloat(x); -} - -/// Interprets the bit pattern of x as a floating-point number. -/// -/// @param [in] x The input value. -/// -/// @returns -/// The input interpreted as a floating-point number. -/// -/// @ingroup HLSLCore -FfxFloat32x3 ffxAsFloat(FfxUInt32x3 x) -{ - return asfloat(x); -} - -/// Interprets the bit pattern of x as a floating-point number. -/// -/// @param [in] x The input value. -/// -/// @returns -/// The input interpreted as a floating-point number. -/// -/// @ingroup HLSLCore -FfxFloat32x4 ffxAsFloat(FfxUInt32x4 x) -{ - return asfloat(x); -} - -/// Compute the linear interopation between two values. -/// -/// Implemented by calling the HLSL mix instrinsic function. Implements the -/// following math: -/// -/// (1 - t) * x + t * y -/// -/// @param [in] x The first value to lerp between. -/// @param [in] y The second value to lerp between. -/// @param [in] t The value to determine how much of x and how much of y. -/// -/// @returns -/// A linearly interpolated value between x and y according to t. -/// -/// @ingroup HLSLCore -FfxFloat32 ffxLerp(FfxFloat32 x, FfxFloat32 y, FfxFloat32 t) -{ - return lerp(x, y, t); -} - -/// Compute the linear interopation between two values. -/// -/// Implemented by calling the HLSL mix instrinsic function. Implements the -/// following math: -/// -/// (1 - t) * x + t * y -/// -/// @param [in] x The first value to lerp between. -/// @param [in] y The second value to lerp between. -/// @param [in] t The value to determine how much of x and how much of y. -/// -/// @returns -/// A linearly interpolated value between x and y according to t. -/// -/// @ingroup HLSLCore -FfxFloat32x2 ffxLerp(FfxFloat32x2 x, FfxFloat32x2 y, FfxFloat32 t) -{ - return lerp(x, y, t); -} - -/// Compute the linear interopation between two values. -/// -/// Implemented by calling the HLSL mix instrinsic function. Implements the -/// following math: -/// -/// (1 - t) * x + t * y -/// -/// @param [in] x The first value to lerp between. -/// @param [in] y The second value to lerp between. -/// @param [in] t The value to determine how much of x and how much of y. -/// -/// @returns -/// A linearly interpolated value between x and y according to t. -/// -/// @ingroup HLSLCore -FfxFloat32x2 ffxLerp(FfxFloat32x2 x, FfxFloat32x2 y, FfxFloat32x2 t) -{ - return lerp(x, y, t); -} - -/// Compute the linear interopation between two values. -/// -/// Implemented by calling the HLSL mix instrinsic function. Implements the -/// following math: -/// -/// (1 - t) * x + t * y -/// -/// @param [in] x The first value to lerp between. -/// @param [in] y The second value to lerp between. -/// @param [in] t The value to determine how much of x and how much of y. -/// -/// @returns -/// A linearly interpolated value between x and y according to t. -/// -/// @ingroup HLSLCore -FfxFloat32x3 ffxLerp(FfxFloat32x3 x, FfxFloat32x3 y, FfxFloat32 t) -{ - return lerp(x, y, t); -} - -/// Compute the linear interopation between two values. -/// -/// Implemented by calling the HLSL mix instrinsic function. Implements the -/// following math: -/// -/// (1 - t) * x + t * y -/// -/// @param [in] x The first value to lerp between. -/// @param [in] y The second value to lerp between. -/// @param [in] t The value to determine how much of x and how much of y. -/// -/// @returns -/// A linearly interpolated value between x and y according to t. -/// -/// @ingroup HLSLCore -FfxFloat32x3 ffxLerp(FfxFloat32x3 x, FfxFloat32x3 y, FfxFloat32x3 t) -{ - return lerp(x, y, t); -} - -/// Compute the linear interopation between two values. -/// -/// Implemented by calling the HLSL mix instrinsic function. Implements the -/// following math: -/// -/// (1 - t) * x + t * y -/// -/// @param [in] x The first value to lerp between. -/// @param [in] y The second value to lerp between. -/// @param [in] t The value to determine how much of x and how much of y. -/// -/// @returns -/// A linearly interpolated value between x and y according to t. -/// -/// @ingroup HLSLCore -FfxFloat32x4 ffxLerp(FfxFloat32x4 x, FfxFloat32x4 y, FfxFloat32 t) -{ - return lerp(x, y, t); -} - -/// Compute the linear interopation between two values. -/// -/// Implemented by calling the HLSL mix instrinsic function. Implements the -/// following math: -/// -/// (1 - t) * x + t * y -/// -/// @param [in] x The first value to lerp between. -/// @param [in] y The second value to lerp between. -/// @param [in] t The value to determine how much of x and how much of y. -/// -/// @returns -/// A linearly interpolated value between x and y according to t. -/// -/// @ingroup HLSLCore -FfxFloat32x4 ffxLerp(FfxFloat32x4 x, FfxFloat32x4 y, FfxFloat32x4 t) -{ - return lerp(x, y, t); -} - -/// Clamp a value to a [0..1] range. -/// -/// @param [in] x The value to clamp to [0..1] range. -/// -/// @returns -/// The clamped version of x. -/// -/// @ingroup HLSLCore -FfxFloat32 ffxSaturate(FfxFloat32 x) -{ - return saturate(x); -} - -/// Clamp a value to a [0..1] range. -/// -/// @param [in] x The value to clamp to [0..1] range. -/// -/// @returns -/// The clamped version of x. -/// -/// @ingroup HLSLCore -FfxFloat32x2 ffxSaturate(FfxFloat32x2 x) -{ - return saturate(x); -} - -/// Clamp a value to a [0..1] range. -/// -/// @param [in] x The value to clamp to [0..1] range. -/// -/// @returns -/// The clamped version of x. -/// -/// @ingroup HLSLCore -FfxFloat32x3 ffxSaturate(FfxFloat32x3 x) -{ - return saturate(x); -} - -/// Clamp a value to a [0..1] range. -/// -/// @param [in] x The value to clamp to [0..1] range. -/// -/// @returns -/// The clamped version of x. -/// -/// @ingroup HLSLCore -FfxFloat32x4 ffxSaturate(FfxFloat32x4 x) -{ - return saturate(x); -} - -/// Compute the factional part of a decimal value. -/// -/// This function calculates x - floor(x). Where floor is the intrinsic HLSL function. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. It is -/// worth further noting that this function is intentionally distinct from the HLSL frac intrinsic -/// function. -/// -/// @param [in] x The value to compute the fractional part from. -/// -/// @returns -/// The fractional part of x. -/// -/// @ingroup HLSLCore -FfxFloat32 ffxFract(FfxFloat32 x) -{ - return x - floor(x); -} - -/// Compute the factional part of a decimal value. -/// -/// This function calculates x - floor(x). Where floor is the intrinsic HLSL function. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. It is -/// worth further noting that this function is intentionally distinct from the HLSL frac intrinsic -/// function. -/// -/// @param [in] x The value to compute the fractional part from. -/// -/// @returns -/// The fractional part of x. -/// -/// @ingroup HLSLCore -FfxFloat32x2 ffxFract(FfxFloat32x2 x) -{ - return x - floor(x); -} - -/// Compute the factional part of a decimal value. -/// -/// This function calculates x - floor(x). Where floor is the intrinsic HLSL function. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. It is -/// worth further noting that this function is intentionally distinct from the HLSL frac intrinsic -/// function. -/// -/// @param [in] x The value to compute the fractional part from. -/// -/// @returns -/// The fractional part of x. -/// -/// @ingroup HLSLCore -FfxFloat32x3 ffxFract(FfxFloat32x3 x) -{ - return x - floor(x); -} - -/// Compute the factional part of a decimal value. -/// -/// This function calculates x - floor(x). Where floor is the intrinsic HLSL function. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. It is -/// worth further noting that this function is intentionally distinct from the HLSL frac intrinsic -/// function. -/// -/// @param [in] x The value to compute the fractional part from. -/// -/// @returns -/// The fractional part of x. -/// -/// @ingroup HLSLCore -FfxFloat32x4 ffxFract(FfxFloat32x4 x) -{ - return x - floor(x); -} - -/// Compute the maximum of three values. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the max calculation. -/// @param [in] y The second value to include in the max calcuation. -/// @param [in] z The third value to include in the max calcuation. -/// -/// @returns -/// The maximum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32 ffxMax3(FfxFloat32 x, FfxFloat32 y, FfxFloat32 z) -{ - return max(x, max(y, z)); -} - -/// Compute the maximum of three values. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the max calculation. -/// @param [in] y The second value to include in the max calcuation. -/// @param [in] z The third value to include in the max calcuation. -/// -/// @returns -/// The maximum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32x2 ffxMax3(FfxFloat32x2 x, FfxFloat32x2 y, FfxFloat32x2 z) -{ - return max(x, max(y, z)); -} - -/// Compute the maximum of three values. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the max calculation. -/// @param [in] y The second value to include in the max calcuation. -/// @param [in] z The third value to include in the max calcuation. -/// -/// @returns -/// The maximum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32x3 ffxMax3(FfxFloat32x3 x, FfxFloat32x3 y, FfxFloat32x3 z) -{ - return max(x, max(y, z)); -} - -/// Compute the maximum of three values. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the max calculation. -/// @param [in] y The second value to include in the max calcuation. -/// @param [in] z The third value to include in the max calcuation. -/// -/// @returns -/// The maximum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32x4 ffxMax3(FfxFloat32x4 x, FfxFloat32x4 y, FfxFloat32x4 z) -{ - return max(x, max(y, z)); -} - -/// Compute the maximum of three values. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the max calculation. -/// @param [in] y The second value to include in the max calcuation. -/// @param [in] z The third value to include in the max calcuation. -/// -/// @returns -/// The maximum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxUInt32 ffxMax3(FfxUInt32 x, FfxUInt32 y, FfxUInt32 z) -{ - return max(x, max(y, z)); -} - -/// Compute the maximum of three values. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the max calculation. -/// @param [in] y The second value to include in the max calcuation. -/// @param [in] z The third value to include in the max calcuation. -/// -/// @returns -/// The maximum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxUInt32x2 ffxMax3(FfxUInt32x2 x, FfxUInt32x2 y, FfxUInt32x2 z) -{ - return max(x, max(y, z)); -} - -/// Compute the maximum of three values. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the max calculation. -/// @param [in] y The second value to include in the max calcuation. -/// @param [in] z The third value to include in the max calcuation. -/// -/// @returns -/// The maximum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxUInt32x3 ffxMax3(FfxUInt32x3 x, FfxUInt32x3 y, FfxUInt32x3 z) -{ - return max(x, max(y, z)); -} - -/// Compute the maximum of three values. -/// -/// NOTE: This function should compile down to a single V_MAX3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the max calculation. -/// @param [in] y The second value to include in the max calcuation. -/// @param [in] z The third value to include in the max calcuation. -/// -/// @returns -/// The maximum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxUInt32x4 ffxMax3(FfxUInt32x4 x, FfxUInt32x4 y, FfxUInt32x4 z) -{ - return max(x, max(y, z)); -} - -/// Compute the median of three values. -/// -/// NOTE: This function should compile down to a single V_MED3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the median calculation. -/// @param [in] y The second value to include in the median calcuation. -/// @param [in] z The third value to include in the median calcuation. -/// -/// @returns -/// The median value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32 ffxMed3(FfxFloat32 x, FfxFloat32 y, FfxFloat32 z) -{ - return max(min(x, y), min(max(x, y), z)); -} - -/// Compute the median of three values. -/// -/// NOTE: This function should compile down to a single V_MED3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the median calculation. -/// @param [in] y The second value to include in the median calcuation. -/// @param [in] z The third value to include in the median calcuation. -/// -/// @returns -/// The median value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32x2 ffxMed3(FfxFloat32x2 x, FfxFloat32x2 y, FfxFloat32x2 z) -{ - return max(min(x, y), min(max(x, y), z)); -} - -/// Compute the median of three values. -/// -/// NOTE: This function should compile down to a single V_MED3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the median calculation. -/// @param [in] y The second value to include in the median calcuation. -/// @param [in] z The third value to include in the median calcuation. -/// -/// @returns -/// The median value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32x3 ffxMed3(FfxFloat32x3 x, FfxFloat32x3 y, FfxFloat32x3 z) -{ - return max(min(x, y), min(max(x, y), z)); -} - -/// Compute the median of three values. -/// -/// NOTE: This function should compile down to a single V_MED3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the median calculation. -/// @param [in] y The second value to include in the median calcuation. -/// @param [in] z The third value to include in the median calcuation. -/// -/// @returns -/// The median value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32x4 ffxMed3(FfxFloat32x4 x, FfxFloat32x4 y, FfxFloat32x4 z) -{ - return max(min(x, y), min(max(x, y), z)); -} - -/// Compute the median of three values. -/// -/// NOTE: This function should compile down to a single V_MED3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the median calculation. -/// @param [in] y The second value to include in the median calcuation. -/// @param [in] z The third value to include in the median calcuation. -/// -/// @returns -/// The median value of x, y, and z. -/// -/// @ingroup HLSL -FfxInt32 ffxMed3(FfxInt32 x, FfxInt32 y, FfxInt32 z) -{ - return max(min(x, y), min(max(x, y), z)); - // return min(max(min(y, z), x), max(y, z)); - // return max(max(x, y), z) == x ? max(y, z) : (max(max(x, y), z) == y ? max(x, z) : max(x, y)); -} - -/// Compute the median of three values. -/// -/// NOTE: This function should compile down to a single V_MED3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the median calculation. -/// @param [in] y The second value to include in the median calcuation. -/// @param [in] z The third value to include in the median calcuation. -/// -/// @returns -/// The median value of x, y, and z. -/// -/// @ingroup HLSL -FfxInt32x2 ffxMed3(FfxInt32x2 x, FfxInt32x2 y, FfxInt32x2 z) -{ - return max(min(x, y), min(max(x, y), z)); - // return min(max(min(y, z), x), max(y, z)); - // return max(max(x, y), z) == x ? max(y, z) : (max(max(x, y), z) == y ? max(x, z) : max(x, y)); -} - -/// Compute the median of three values. -/// -/// NOTE: This function should compile down to a single V_MED3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the median calculation. -/// @param [in] y The second value to include in the median calcuation. -/// @param [in] z The third value to include in the median calcuation. -/// -/// @returns -/// The median value of x, y, and z. -/// -/// @ingroup HLSL -FfxInt32x3 ffxMed3(FfxInt32x3 x, FfxInt32x3 y, FfxInt32x3 z) -{ - return max(min(x, y), min(max(x, y), z)); -} - -/// Compute the median of three values. -/// -/// NOTE: This function should compile down to a single V_MED3_I32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the median calculation. -/// @param [in] y The second value to include in the median calcuation. -/// @param [in] z The third value to include in the median calcuation. -/// -/// @returns -/// The median value of x, y, and z. -/// -/// @ingroup HLSL -FfxInt32x4 ffxMed3(FfxInt32x4 x, FfxInt32x4 y, FfxInt32x4 z) -{ - return max(min(x, y), min(max(x, y), z)); -} - -/// Compute the minimum of three values. -/// -/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the min calculation. -/// @param [in] y The second value to include in the min calcuation. -/// @param [in] z The third value to include in the min calcuation. -/// -/// @returns -/// The minimum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32 ffxMin3(FfxFloat32 x, FfxFloat32 y, FfxFloat32 z) -{ - return min(x, min(y, z)); -} - -/// Compute the minimum of three values. -/// -/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the min calculation. -/// @param [in] y The second value to include in the min calcuation. -/// @param [in] z The third value to include in the min calcuation. -/// -/// @returns -/// The minimum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32x2 ffxMin3(FfxFloat32x2 x, FfxFloat32x2 y, FfxFloat32x2 z) -{ - return min(x, min(y, z)); -} - -/// Compute the minimum of three values. -/// -/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the min calculation. -/// @param [in] y The second value to include in the min calcuation. -/// @param [in] z The third value to include in the min calcuation. -/// -/// @returns -/// The minimum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32x3 ffxMin3(FfxFloat32x3 x, FfxFloat32x3 y, FfxFloat32x3 z) -{ - return min(x, min(y, z)); -} - -/// Compute the minimum of three values. -/// -/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the min calculation. -/// @param [in] y The second value to include in the min calcuation. -/// @param [in] z The third value to include in the min calcuation. -/// -/// @returns -/// The minimum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxFloat32x4 ffxMin3(FfxFloat32x4 x, FfxFloat32x4 y, FfxFloat32x4 z) -{ - return min(x, min(y, z)); -} - -/// Compute the minimum of three values. -/// -/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the min calculation. -/// @param [in] y The second value to include in the min calcuation. -/// @param [in] z The third value to include in the min calcuation. -/// -/// @returns -/// The minimum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxUInt32 ffxMin3(FfxUInt32 x, FfxUInt32 y, FfxUInt32 z) -{ - return min(x, min(y, z)); -} - -/// Compute the minimum of three values. -/// -/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the min calculation. -/// @param [in] y The second value to include in the min calcuation. -/// @param [in] z The third value to include in the min calcuation. -/// -/// @returns -/// The minimum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxUInt32x2 ffxMin3(FfxUInt32x2 x, FfxUInt32x2 y, FfxUInt32x2 z) -{ - return min(x, min(y, z)); -} - -/// Compute the minimum of three values. -/// -/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the min calculation. -/// @param [in] y The second value to include in the min calculation. -/// @param [in] z The third value to include in the min calculation. -/// -/// @returns -/// The minimum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxUInt32x3 ffxMin3(FfxUInt32x3 x, FfxUInt32x3 y, FfxUInt32x3 z) -{ - return min(x, min(y, z)); -} - -/// Compute the minimum of three values. -/// -/// NOTE: This function should compile down to a single V_MIN3_F32 operation on GCN/RDNA hardware. -/// -/// @param [in] x The first value to include in the min calculation. -/// @param [in] y The second value to include in the min calcuation. -/// @param [in] z The third value to include in the min calcuation. -/// -/// @returns -/// The minimum value of x, y, and z. -/// -/// @ingroup HLSLCore -FfxUInt32x4 ffxMin3(FfxUInt32x4 x, FfxUInt32x4 y, FfxUInt32x4 z) -{ - return min(x, min(y, z)); -} - - -FfxUInt32 AShrSU1(FfxUInt32 a, FfxUInt32 b) -{ - return FfxUInt32(FfxInt32(a) >> FfxInt32(b)); -} - -FfxUInt32 ffxPackF32(FfxFloat32x2 v){ - FfxUInt32x2 p = FfxUInt32x2(f32tof16(FfxFloat32x2(v).x), f32tof16(FfxFloat32x2(v).y)); - return p.x | (p.y << 16); -} - -FfxFloat32x2 ffxUnpackF32(FfxUInt32 a){ - return f16tof32(FfxUInt32x2(a & 0xFFFF, a >> 16)); -} - -//============================================================================================================================== -// HLSL HALF -//============================================================================================================================== -//============================================================================================================================== -// Need to use manual unpack to get optimal execution (don't use packed types in buffers directly). -// Unpack requires this pattern: https://gpuopen.com/first-steps-implementing-fp16/ -FFX_MIN16_F2 ffxUint32ToFloat16x2(FfxUInt32 x) -{ - FfxFloat32x2 t = f16tof32(FfxUInt32x2(x & 0xFFFF, x >> 16)); - return FFX_MIN16_F2(t); -} -FFX_MIN16_F4 ffxUint32x2ToFloat16x4(FfxUInt32x2 x) -{ - return FFX_MIN16_F4(ffxUint32ToFloat16x2(x.x), ffxUint32ToFloat16x2(x.y)); -} -FFX_MIN16_U2 ffxUint32ToUint16x2(FfxUInt32 x) -{ - FfxUInt32x2 t = FfxUInt32x2(x & 0xFFFF, x >> 16); - return FFX_MIN16_U2(t); -} -FFX_MIN16_U4 ffxUint32x2ToUint16x4(FfxUInt32x2 x) -{ - return FFX_MIN16_U4(ffxUint32ToUint16x2(x.x), ffxUint32ToUint16x2(x.y)); -} - -/// @brief Inverts the value while avoiding division by zero. If the value is zero, zero is returned. -/// @param v Value to invert. -/// @return If v = 0 returns 0. If v != 0 returns 1/v. -FfxFloat32 ffxInvertSafe(FfxFloat32 v){ - FfxFloat32 s = sign(v); - FfxFloat32 s2 = s*s; - return s2/(v + s2 - 1.0); -} - -/// @brief Inverts the value while avoiding division by zero. If the value is zero, zero is returned. -/// @param v Value to invert. -/// @return If v = 0 returns 0. If v != 0 returns 1/v. -FfxFloat32x2 ffxInvertSafe(FfxFloat32x2 v){ - FfxFloat32x2 s = sign(v); - FfxFloat32x2 s2 = s*s; - return s2/(v + s2 - FfxFloat32x2(1.0, 1.0)); -} - -/// @brief Inverts the value while avoiding division by zero. If the value is zero, zero is returned. -/// @param v Value to invert. -/// @return If v = 0 returns 0. If v != 0 returns 1/v. -FfxFloat32x3 ffxInvertSafe(FfxFloat32x3 v){ - FfxFloat32x3 s = sign(v); - FfxFloat32x3 s2 = s*s; - return s2/(v + s2 - FfxFloat32x3(1.0, 1.0, 1.0)); -} - -/// @brief Inverts the value while avoiding division by zero. If the value is zero, zero is returned. -/// @param v Value to invert. -/// @return If v = 0 returns 0. If v != 0 returns 1/v. -FfxFloat32x4 ffxInvertSafe(FfxFloat32x4 v){ - FfxFloat32x4 s = sign(v); - FfxFloat32x4 s2 = s*s; - return s2/(v + s2 - FfxFloat32x4(1.0, 1.0, 1.0, 1.0)); -} - -#define FFX_UINT32_TO_FLOAT16X2(x) ffxUint32ToFloat16x2(FfxUInt32(x)) -#if FFX_HALF - -#define FFX_UINT32X2_TO_FLOAT16X4(x) ffxUint32x2ToFloat16x4(FfxUInt32x2(x)) -#define FFX_UINT32_TO_UINT16X2(x) ffxUint32ToUint16x2(FfxUInt32(x)) -#define FFX_UINT32X2_TO_UINT16X4(x) ffxUint32x2ToUint16x4(FfxUInt32x2(x)) - -FfxUInt32 ffxPackF16(FfxFloat16x2 v){ - FfxUInt32x2 p = FfxUInt32x2(f32tof16(FfxFloat32x2(v).x), f32tof16(FfxFloat32x2(v).y)); - return p.x | (p.y << 16); -} - -FfxFloat16x2 ffxUnpackF16(FfxUInt32 a){ - return FfxFloat16x2(f16tof32(FfxUInt32x2(a & 0xFFFF, a >> 16))); -} - -//------------------------------------------------------------------------------------------------------------------------------ -FfxUInt32 FFX_MIN16_F2ToUint32(FFX_MIN16_F2 x) -{ - return f32tof16(x.x) + (f32tof16(x.y) << 16); -} -FfxUInt32x2 FFX_MIN16_F4ToUint32x2(FFX_MIN16_F4 x) -{ - return FfxUInt32x2(FFX_MIN16_F2ToUint32(x.xy), FFX_MIN16_F2ToUint32(x.zw)); -} -FfxUInt32 FFX_MIN16_U2ToUint32(FFX_MIN16_U2 x) -{ - return FfxUInt32(x.x) + (FfxUInt32(x.y) << 16); -} -FfxUInt32x2 FFX_MIN16_U4ToUint32x2(FFX_MIN16_U4 x) -{ - return FfxUInt32x2(FFX_MIN16_U2ToUint32(x.xy), FFX_MIN16_U2ToUint32(x.zw)); -} -#define FFX_FLOAT16X2_TO_UINT32(x) FFX_MIN16_F2ToUint32(FFX_MIN16_F2(x)) -#define FFX_FLOAT16X4_TO_UINT32X2(x) FFX_MIN16_F4ToUint32x2(FFX_MIN16_F4(x)) -#define FFX_UINT16X2_TO_UINT32(x) FFX_MIN16_U2ToUint32(FFX_MIN16_U2(x)) -#define FFX_UINT16X4_TO_UINT32X2(x) FFX_MIN16_U4ToUint32x2(FFX_MIN16_U4(x)) - -#if (FFX_HLSL_SM >= 62) && !defined(FFX_NO_16_BIT_CAST) -#define FFX_TO_UINT16(x) asuint16(x) -#define FFX_TO_UINT16X2(x) asuint16(x) -#define FFX_TO_UINT16X3(x) asuint16(x) -#define FFX_TO_UINT16X4(x) asuint16(x) -#else -#define FFX_TO_UINT16(a) FFX_MIN16_U(f32tof16(FfxFloat32(a))) -#define FFX_TO_UINT16X2(a) FFX_MIN16_U2(FFX_TO_UINT16((a).x), FFX_TO_UINT16((a).y)) -#define FFX_TO_UINT16X3(a) FFX_MIN16_U3(FFX_TO_UINT16((a).x), FFX_TO_UINT16((a).y), FFX_TO_UINT16((a).z)) -#define FFX_TO_UINT16X4(a) FFX_MIN16_U4(FFX_TO_UINT16((a).x), FFX_TO_UINT16((a).y), FFX_TO_UINT16((a).z), FFX_TO_UINT16((a).w)) -#endif // #if (FFX_HLSL_SM>=62) && !defined(FFX_NO_16_BIT_CAST) - -#if (FFX_HLSL_SM >= 62) && !defined(FFX_NO_16_BIT_CAST) -#define FFX_TO_FLOAT16(x) asfloat16(x) -#define FFX_TO_FLOAT16X2(x) asfloat16(x) -#define FFX_TO_FLOAT16X3(x) asfloat16(x) -#define FFX_TO_FLOAT16X4(x) asfloat16(x) -#else -#define FFX_TO_FLOAT16(a) FFX_MIN16_F(f16tof32(FfxUInt32(a))) -#define FFX_TO_FLOAT16X2(a) FFX_MIN16_F2(FFX_TO_FLOAT16((a).x), FFX_TO_FLOAT16((a).y)) -#define FFX_TO_FLOAT16X3(a) FFX_MIN16_F3(FFX_TO_FLOAT16((a).x), FFX_TO_FLOAT16((a).y), FFX_TO_FLOAT16((a).z)) -#define FFX_TO_FLOAT16X4(a) FFX_MIN16_F4(FFX_TO_FLOAT16((a).x), FFX_TO_FLOAT16((a).y), FFX_TO_FLOAT16((a).z), FFX_TO_FLOAT16((a).w)) -#endif // #if (FFX_HLSL_SM>=62) && !defined(FFX_NO_16_BIT_CAST) - -//============================================================================================================================== -#define FFX_BROADCAST_FLOAT16(a) FFX_MIN16_F(a) -#define FFX_BROADCAST_FLOAT16X2(a) FFX_MIN16_F(a) -#define FFX_BROADCAST_FLOAT16X3(a) FFX_MIN16_F(a) -#define FFX_BROADCAST_FLOAT16X4(a) FFX_MIN16_F(a) - -//------------------------------------------------------------------------------------------------------------------------------ -#define FFX_BROADCAST_INT16(a) FFX_MIN16_I(a) -#define FFX_BROADCAST_INT16X2(a) FFX_MIN16_I(a) -#define FFX_BROADCAST_INT16X3(a) FFX_MIN16_I(a) -#define FFX_BROADCAST_INT16X4(a) FFX_MIN16_I(a) - -//------------------------------------------------------------------------------------------------------------------------------ -#define FFX_BROADCAST_UINT16(a) FFX_MIN16_U(a) -#define FFX_BROADCAST_UINT16X2(a) FFX_MIN16_U(a) -#define FFX_BROADCAST_UINT16X3(a) FFX_MIN16_U(a) -#define FFX_BROADCAST_UINT16X4(a) FFX_MIN16_U(a) - -//============================================================================================================================== -FFX_MIN16_U ffxAbsHalf(FFX_MIN16_U a) -{ - return FFX_MIN16_U(abs(FFX_MIN16_I(a))); -} -FFX_MIN16_U2 ffxAbsHalf(FFX_MIN16_U2 a) -{ - return FFX_MIN16_U2(abs(FFX_MIN16_I2(a))); -} -FFX_MIN16_U3 ffxAbsHalf(FFX_MIN16_U3 a) -{ - return FFX_MIN16_U3(abs(FFX_MIN16_I3(a))); -} -FFX_MIN16_U4 ffxAbsHalf(FFX_MIN16_U4 a) -{ - return FFX_MIN16_U4(abs(FFX_MIN16_I4(a))); -} -//------------------------------------------------------------------------------------------------------------------------------ -FFX_MIN16_F ffxClampHalf(FFX_MIN16_F x, FFX_MIN16_F n, FFX_MIN16_F m) -{ - return max(n, min(x, m)); -} -FFX_MIN16_F2 ffxClampHalf(FFX_MIN16_F2 x, FFX_MIN16_F2 n, FFX_MIN16_F2 m) -{ - return max(n, min(x, m)); -} -FFX_MIN16_F3 ffxClampHalf(FFX_MIN16_F3 x, FFX_MIN16_F3 n, FFX_MIN16_F3 m) -{ - return max(n, min(x, m)); -} -FFX_MIN16_F4 ffxClampHalf(FFX_MIN16_F4 x, FFX_MIN16_F4 n, FFX_MIN16_F4 m) -{ - return max(n, min(x, m)); -} -//------------------------------------------------------------------------------------------------------------------------------ -// V_FRACT_F16 (note DX frac() is different). -FFX_MIN16_F ffxFract(FFX_MIN16_F x) -{ - return x - floor(x); -} -FFX_MIN16_F2 ffxFract(FFX_MIN16_F2 x) -{ - return x - floor(x); -} -FFX_MIN16_F3 ffxFract(FFX_MIN16_F3 x) -{ - return x - floor(x); -} -FFX_MIN16_F4 ffxFract(FFX_MIN16_F4 x) -{ - return x - floor(x); -} -//------------------------------------------------------------------------------------------------------------------------------ -FFX_MIN16_F ffxLerp(FFX_MIN16_F x, FFX_MIN16_F y, FFX_MIN16_F a) -{ - return lerp(x, y, a); -} -FFX_MIN16_F2 ffxLerp(FFX_MIN16_F2 x, FFX_MIN16_F2 y, FFX_MIN16_F a) -{ - return lerp(x, y, a); -} -FFX_MIN16_F2 ffxLerp(FFX_MIN16_F2 x, FFX_MIN16_F2 y, FFX_MIN16_F2 a) -{ - return lerp(x, y, a); -} -FFX_MIN16_F3 ffxLerp(FFX_MIN16_F3 x, FFX_MIN16_F3 y, FFX_MIN16_F a) -{ - return lerp(x, y, a); -} -FFX_MIN16_F3 ffxLerp(FFX_MIN16_F3 x, FFX_MIN16_F3 y, FFX_MIN16_F3 a) -{ - return lerp(x, y, a); -} -FFX_MIN16_F4 ffxLerp(FFX_MIN16_F4 x, FFX_MIN16_F4 y, FFX_MIN16_F a) -{ - return lerp(x, y, a); -} -FFX_MIN16_F4 ffxLerp(FFX_MIN16_F4 x, FFX_MIN16_F4 y, FFX_MIN16_F4 a) -{ - return lerp(x, y, a); -} -//------------------------------------------------------------------------------------------------------------------------------ -FFX_MIN16_F ffxMax3Half(FFX_MIN16_F x, FFX_MIN16_F y, FFX_MIN16_F z) -{ - return max(x, max(y, z)); -} -FFX_MIN16_F2 ffxMax3Half(FFX_MIN16_F2 x, FFX_MIN16_F2 y, FFX_MIN16_F2 z) -{ - return max(x, max(y, z)); -} -FFX_MIN16_F3 ffxMax3Half(FFX_MIN16_F3 x, FFX_MIN16_F3 y, FFX_MIN16_F3 z) -{ - return max(x, max(y, z)); -} -FFX_MIN16_F4 ffxMax3Half(FFX_MIN16_F4 x, FFX_MIN16_F4 y, FFX_MIN16_F4 z) -{ - return max(x, max(y, z)); -} -//------------------------------------------------------------------------------------------------------------------------------ -FFX_MIN16_F ffxMin3Half(FFX_MIN16_F x, FFX_MIN16_F y, FFX_MIN16_F z) -{ - return min(x, min(y, z)); -} -FFX_MIN16_F2 ffxMin3Half(FFX_MIN16_F2 x, FFX_MIN16_F2 y, FFX_MIN16_F2 z) -{ - return min(x, min(y, z)); -} -FFX_MIN16_F3 ffxMin3Half(FFX_MIN16_F3 x, FFX_MIN16_F3 y, FFX_MIN16_F3 z) -{ - return min(x, min(y, z)); -} -FFX_MIN16_F4 ffxMin3Half(FFX_MIN16_F4 x, FFX_MIN16_F4 y, FFX_MIN16_F4 z) -{ - return min(x, min(y, z)); -} -//------------------------------------------------------------------------------------------------------------------------------ -FFX_MIN16_F ffxMed3Half(FFX_MIN16_F x, FFX_MIN16_F y, FFX_MIN16_F z) -{ - return max(min(x, y), min(max(x, y), z)); -} -FFX_MIN16_F2 ffxMed3Half(FFX_MIN16_F2 x, FFX_MIN16_F2 y, FFX_MIN16_F2 z) -{ - return max(min(x, y), min(max(x, y), z)); -} -FFX_MIN16_F3 ffxMed3Half(FFX_MIN16_F3 x, FFX_MIN16_F3 y, FFX_MIN16_F3 z) -{ - return max(min(x, y), min(max(x, y), z)); -} -FFX_MIN16_F4 ffxMed3Half(FFX_MIN16_F4 x, FFX_MIN16_F4 y, FFX_MIN16_F4 z) -{ - return max(min(x, y), min(max(x, y), z)); -} -//------------------------------------------------------------------------------------------------------------------------------ -FFX_MIN16_I ffxMed3Half(FFX_MIN16_I x, FFX_MIN16_I y, FFX_MIN16_I z) -{ - return max(min(x, y), min(max(x, y), z)); -} -FFX_MIN16_I2 ffxMed3Half(FFX_MIN16_I2 x, FFX_MIN16_I2 y, FFX_MIN16_I2 z) -{ - return max(min(x, y), min(max(x, y), z)); -} -FFX_MIN16_I3 ffxMed3Half(FFX_MIN16_I3 x, FFX_MIN16_I3 y, FFX_MIN16_I3 z) -{ - return max(min(x, y), min(max(x, y), z)); -} -FFX_MIN16_I4 ffxMed3Half(FFX_MIN16_I4 x, FFX_MIN16_I4 y, FFX_MIN16_I4 z) -{ - return max(min(x, y), min(max(x, y), z)); -} -//------------------------------------------------------------------------------------------------------------------------------ -FFX_MIN16_F ffxReciprocalHalf(FFX_MIN16_F x) -{ - return rcp(x); -} -FFX_MIN16_F2 ffxReciprocalHalf(FFX_MIN16_F2 x) -{ - return rcp(x); -} -FFX_MIN16_F3 ffxReciprocalHalf(FFX_MIN16_F3 x) -{ - return rcp(x); -} -FFX_MIN16_F4 ffxReciprocalHalf(FFX_MIN16_F4 x) -{ - return rcp(x); -} -//------------------------------------------------------------------------------------------------------------------------------ -FFX_MIN16_F ffxReciprocalSquareRootHalf(FFX_MIN16_F x) -{ - return rsqrt(x); -} -FFX_MIN16_F2 ffxReciprocalSquareRootHalf(FFX_MIN16_F2 x) -{ - return rsqrt(x); -} -FFX_MIN16_F3 ffxReciprocalSquareRootHalf(FFX_MIN16_F3 x) -{ - return rsqrt(x); -} -FFX_MIN16_F4 ffxReciprocalSquareRootHalf(FFX_MIN16_F4 x) -{ - return rsqrt(x); -} -//------------------------------------------------------------------------------------------------------------------------------ -FFX_MIN16_F ffxSaturate(FFX_MIN16_F x) -{ - return saturate(x); -} -FFX_MIN16_F2 ffxSaturate(FFX_MIN16_F2 x) -{ - return saturate(x); -} -FFX_MIN16_F3 ffxSaturate(FFX_MIN16_F3 x) -{ - return saturate(x); -} -FFX_MIN16_F4 ffxSaturate(FFX_MIN16_F4 x) -{ - return saturate(x); -} -//------------------------------------------------------------------------------------------------------------------------------ -FFX_MIN16_U ffxBitShiftRightHalf(FFX_MIN16_U a, FFX_MIN16_U b) -{ - return FFX_MIN16_U(FFX_MIN16_I(a) >> FFX_MIN16_I(b)); -} -FFX_MIN16_U2 ffxBitShiftRightHalf(FFX_MIN16_U2 a, FFX_MIN16_U2 b) -{ - return FFX_MIN16_U2(FFX_MIN16_I2(a) >> FFX_MIN16_I2(b)); -} -FFX_MIN16_U3 ffxBitShiftRightHalf(FFX_MIN16_U3 a, FFX_MIN16_U3 b) -{ - return FFX_MIN16_U3(FFX_MIN16_I3(a) >> FFX_MIN16_I3(b)); -} -FFX_MIN16_U4 ffxBitShiftRightHalf(FFX_MIN16_U4 a, FFX_MIN16_U4 b) -{ - return FFX_MIN16_U4(FFX_MIN16_I4(a) >> FFX_MIN16_I4(b)); -} -#endif // FFX_HALF - -//============================================================================================================================== -// HLSL WAVE -//============================================================================================================================== -#if defined(FFX_WAVE) -// Where 'x' must be a compile time literal. -FfxFloat32 AWaveXorF1(FfxFloat32 v, FfxUInt32 x) -{ - return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); -} -FfxFloat32x2 AWaveXorF2(FfxFloat32x2 v, FfxUInt32 x) -{ - return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); -} -FfxFloat32x3 AWaveXorF3(FfxFloat32x3 v, FfxUInt32 x) -{ - return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); -} -FfxFloat32x4 AWaveXorF4(FfxFloat32x4 v, FfxUInt32 x) -{ - return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); -} -FfxUInt32 AWaveXorU1(FfxUInt32 v, FfxUInt32 x) -{ - return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); -} -FfxUInt32x2 AWaveXorU1(FfxUInt32x2 v, FfxUInt32 x) -{ - return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); -} -FfxUInt32x3 AWaveXorU1(FfxUInt32x3 v, FfxUInt32 x) -{ - return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); -} -FfxUInt32x4 AWaveXorU1(FfxUInt32x4 v, FfxUInt32 x) -{ - return WaveReadLaneAt(v, WaveGetLaneIndex() ^ x); -} -FfxBoolean AWaveIsFirstLane() -{ - return WaveIsFirstLane(); -} -FfxUInt32 AWaveLaneIndex() -{ - return WaveGetLaneIndex(); -} -FfxBoolean AWaveReadAtLaneIndexB1(FfxBoolean v, FfxUInt32 x) -{ - return WaveReadLaneAt(v, x); -} -FfxUInt32 AWavePrefixCountBits(FfxBoolean v) -{ - return WavePrefixCountBits(v); -} -FfxUInt32 AWaveActiveCountBits(FfxBoolean v) -{ - return WaveActiveCountBits(v); -} -FfxUInt32 AWaveReadLaneFirstU1(FfxUInt32 v) -{ - return WaveReadLaneFirst(v); -} -FfxUInt32 WaveOr(FfxUInt32 a) -{ - return WaveActiveBitOr(a); -} -FfxFloat32 WaveMin(FfxFloat32 a) -{ - return WaveActiveMin(a); -} -FfxFloat32 WaveMax(FfxFloat32 a) -{ - return WaveActiveMax(a); -} -FfxUInt32 WaveLaneCount() -{ - return WaveGetLaneCount(); -} -FfxBoolean WaveAllTrue(FfxBoolean v) -{ - return WaveActiveAllTrue(v); -} -FfxFloat32 QuadReadX(FfxFloat32 v) -{ - return QuadReadAcrossX(v); -} -FfxFloat32x2 QuadReadX(FfxFloat32x2 v) -{ - return QuadReadAcrossX(v); -} -FfxFloat32 QuadReadY(FfxFloat32 v) -{ - return QuadReadAcrossY(v); -} -FfxFloat32x2 QuadReadY(FfxFloat32x2 v) -{ - return QuadReadAcrossY(v); -} - -#if FFX_HALF -FfxFloat16x2 ffxWaveXorFloat16x2(FfxFloat16x2 v, FfxUInt32 x) -{ - return FFX_UINT32_TO_FLOAT16X2(WaveReadLaneAt(FFX_FLOAT16X2_TO_UINT32(v), WaveGetLaneIndex() ^ x)); -} -FfxFloat16x4 ffxWaveXorFloat16x4(FfxFloat16x4 v, FfxUInt32 x) -{ - return FFX_UINT32X2_TO_FLOAT16X4(WaveReadLaneAt(FFX_FLOAT16X4_TO_UINT32X2(v), WaveGetLaneIndex() ^ x)); -} -FfxUInt16x2 ffxWaveXorUint16x2(FfxUInt16x2 v, FfxUInt32 x) -{ - return FFX_UINT32_TO_UINT16X2(WaveReadLaneAt(FFX_UINT16X2_TO_UINT32(v), WaveGetLaneIndex() ^ x)); -} -FfxUInt16x4 ffxWaveXorUint16x4(FfxUInt16x4 v, FfxUInt32 x) -{ - return FFX_UINT32X2_TO_UINT16X4(WaveReadLaneAt(FFX_UINT16X4_TO_UINT32X2(v), WaveGetLaneIndex() ^ x)); -} -#endif // FFX_HALF -#endif // #if defined(FFX_WAVE) diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_hlsl.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_hlsl.h.meta deleted file mode 100644 index 2aa0691..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_hlsl.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 299f67e8b7e1d1a48a577bf8b328ac92 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_portability.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_portability.h deleted file mode 100644 index 84a62d6..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_portability.h +++ /dev/null @@ -1,51 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -FfxFloat32x3 opAAddOneF3(FfxFloat32x3 d, FfxFloat32x3 a, FfxFloat32 b) -{ - d = a + ffxBroadcast3(b); - return d; -} - -FfxFloat32x3 opACpyF3(FfxFloat32x3 d, FfxFloat32x3 a) -{ - d = a; - return d; -} - -FfxFloat32x3 opAMulF3(FfxFloat32x3 d, FfxFloat32x3 a, FfxFloat32x3 b) -{ - d = a * b; - return d; -} - -FfxFloat32x3 opAMulOneF3(FfxFloat32x3 d, FfxFloat32x3 a, FfxFloat32 b) -{ - d = a * ffxBroadcast3(b); - return d; -} - -FfxFloat32x3 opARcpF3(FfxFloat32x3 d, FfxFloat32x3 a) -{ - d = rcp(a); - return d; -} diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_portability.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_portability.h.meta deleted file mode 100644 index b333bf0..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_core_portability.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 8d2ace0bd52e0e1438e08ddaccd3ba24 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_accumulate.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_accumulate.h deleted file mode 100644 index c425de7..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_accumulate.h +++ /dev/null @@ -1,288 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#ifndef FFX_FSR3UPSCALER_ACCUMULATE_H -#define FFX_FSR3UPSCALER_ACCUMULATE_H - -FfxFloat32 GetPxHrVelocity(FfxFloat32x2 fMotionVector) -{ - return length(fMotionVector * DisplaySize()); -} -#if FFX_HALF -FFX_MIN16_F GetPxHrVelocity(FFX_MIN16_F2 fMotionVector) -{ - return length(fMotionVector * FFX_MIN16_F2(DisplaySize())); -} -#endif - -void Accumulate(const AccumulationPassCommonParams params, FFX_PARAMETER_INOUT FfxFloat32x3 fHistoryColor, FfxFloat32x3 fAccumulation, FFX_PARAMETER_IN FfxFloat32x4 fUpsampledColorAndWeight) -{ - // Aviod invalid values when accumulation and upsampled weight is 0 - fAccumulation = ffxMax(FSR3UPSCALER_EPSILON.xxx, fAccumulation + fUpsampledColorAndWeight.www); - -#if FFX_FSR3UPSCALER_OPTION_HDR_COLOR_INPUT - //YCoCg -> RGB -> Tonemap -> YCoCg (Use RGB tonemapper to avoid color desaturation) - fUpsampledColorAndWeight.xyz = RGBToYCoCg(Tonemap(YCoCgToRGB(fUpsampledColorAndWeight.xyz))); - fHistoryColor = RGBToYCoCg(Tonemap(YCoCgToRGB(fHistoryColor))); -#endif - - const FfxFloat32x3 fAlpha = fUpsampledColorAndWeight.www / fAccumulation; - fHistoryColor = ffxLerp(fHistoryColor, fUpsampledColorAndWeight.xyz, fAlpha); - - fHistoryColor = YCoCgToRGB(fHistoryColor); - -#if FFX_FSR3UPSCALER_OPTION_HDR_COLOR_INPUT - fHistoryColor = InverseTonemap(fHistoryColor); -#endif -} - -void RectifyHistory( - const AccumulationPassCommonParams params, - RectificationBox clippingBox, - FFX_PARAMETER_INOUT FfxFloat32x3 fHistoryColor, - FFX_PARAMETER_INOUT FfxFloat32x3 fAccumulation, - FfxFloat32 fLockContributionThisFrame, - FfxFloat32 fTemporalReactiveFactor, - FfxFloat32 fLumaInstabilityFactor) -{ - const FfxFloat32 fVecolityFactor = ffxSaturate(params.fHrVelocity / 20.0f); - const FfxFloat32 fBoxScaleT = ffxMax(params.fDepthClipFactor, ffxMax(params.fAccumulationMask, fVecolityFactor)); - const FfxFloat32 fBoxScale = ffxLerp(3.0f, 1.0f, fBoxScaleT); - - const FfxFloat32x3 fScaledBoxVec = clippingBox.boxVec * fBoxScale; - const FfxFloat32x3 boxMin = clippingBox.boxCenter - fScaledBoxVec; - const FfxFloat32x3 boxMax = clippingBox.boxCenter + fScaledBoxVec; - - if (any(FFX_GREATER_THAN(boxMin, fHistoryColor)) || any(FFX_GREATER_THAN(fHistoryColor, boxMax))) { - - const FfxFloat32x3 fClampedHistoryColor = clamp(fHistoryColor, boxMin, boxMax); - - FfxFloat32x3 fHistoryContribution = ffxMax(fLumaInstabilityFactor, fLockContributionThisFrame).xxx; - - const FfxFloat32 fReactiveFactor = params.fDilatedReactiveFactor; - const FfxFloat32 fReactiveContribution = 1.0f - ffxPow(fReactiveFactor, 1.0f / 2.0f); - fHistoryContribution *= fReactiveContribution; - - // Scale history color using rectification info, also using accumulation mask to avoid potential invalid color protection - fHistoryColor = ffxLerp(fClampedHistoryColor, fHistoryColor, ffxSaturate(fHistoryContribution)); - - // Scale accumulation using rectification info - const FfxFloat32x3 fAccumulationMin = ffxMin(fAccumulation, FFX_BROADCAST_FLOAT32X3(0.1f)); - fAccumulation = ffxLerp(fAccumulationMin, fAccumulation, ffxSaturate(fHistoryContribution)); - } -} - -void WriteUpscaledOutput(FfxInt32x2 iPxHrPos, FfxFloat32x3 fUpscaledColor) -{ - StoreUpscaledOutput(iPxHrPos, fUpscaledColor); -} - -void FinalizeLockStatus(const AccumulationPassCommonParams params, FfxFloat32x2 fLockStatus, FfxFloat32 fUpsampledWeight) -{ - // we expect similar motion for next frame - // kill lock if that location is outside screen, avoid locks to be clamped to screen borders - FfxFloat32x2 fEstimatedUvNextFrame = params.fHrUv - params.fMotionVector; - if (IsUvInside(fEstimatedUvNextFrame) == false) { - KillLock(fLockStatus); - } - else { - // Decrease lock lifetime - const FfxFloat32 fLifetimeDecreaseLanczosMax = FfxFloat32(JitterSequenceLength()) * FfxFloat32(fAverageLanczosWeightPerFrame); - const FfxFloat32 fLifetimeDecrease = FfxFloat32(fUpsampledWeight / fLifetimeDecreaseLanczosMax); - fLockStatus[LOCK_LIFETIME_REMAINING] = ffxMax(FfxFloat32(0), fLockStatus[LOCK_LIFETIME_REMAINING] - fLifetimeDecrease); - } - - StoreLockStatus(params.iPxHrPos, fLockStatus); -} - - -FfxFloat32x3 ComputeBaseAccumulationWeight(const AccumulationPassCommonParams params, FfxFloat32 fThisFrameReactiveFactor, FfxBoolean bInMotionLastFrame, FfxFloat32 fUpsampledWeight, LockState lockState) -{ - // Always assume max accumulation was reached - FfxFloat32 fBaseAccumulation = fMaxAccumulationLanczosWeight * FfxFloat32(params.bIsExistingSample) * (1.0f - fThisFrameReactiveFactor) * (1.0f - params.fDepthClipFactor); - - fBaseAccumulation = ffxMin(fBaseAccumulation, ffxLerp(fBaseAccumulation, fUpsampledWeight * 10.0f, ffxMax(FfxFloat32(bInMotionLastFrame), ffxSaturate(params.fHrVelocity * FfxFloat32(10))))); - - fBaseAccumulation = ffxMin(fBaseAccumulation, ffxLerp(fBaseAccumulation, fUpsampledWeight, ffxSaturate(params.fHrVelocity / FfxFloat32(20)))); - - return fBaseAccumulation.xxx; -} - -FfxFloat32 ComputeLumaInstabilityFactor(const AccumulationPassCommonParams params, RectificationBox clippingBox, FfxFloat32 fThisFrameReactiveFactor, FfxFloat32 fLuminanceDiff) -{ - const FfxFloat32 fUnormThreshold = 1.0f / 255.0f; - const FfxInt32 N_MINUS_1 = 0; - const FfxInt32 N_MINUS_2 = 1; - const FfxInt32 N_MINUS_3 = 2; - const FfxInt32 N_MINUS_4 = 3; - - FfxFloat32 fCurrentFrameLuma = clippingBox.boxCenter.x; - -#if FFX_FSR3UPSCALER_OPTION_HDR_COLOR_INPUT - fCurrentFrameLuma = fCurrentFrameLuma / (1.0f + ffxMax(0.0f, fCurrentFrameLuma)); -#endif - - fCurrentFrameLuma = round(fCurrentFrameLuma * 255.0f) / 255.0f; - - const FfxBoolean bSampleLumaHistory = (ffxMax(ffxMax(params.fDepthClipFactor, params.fAccumulationMask), fLuminanceDiff) < 0.1f) && (params.bIsNewSample == false); - FfxFloat32x4 fCurrentFrameLumaHistory = bSampleLumaHistory ? SampleLumaHistory(params.fReprojectedHrUv) : FFX_BROADCAST_FLOAT32X4(0.0f); - - FfxFloat32 fLumaInstability = 0.0f; - FfxFloat32 fDiffs0 = (fCurrentFrameLuma - fCurrentFrameLumaHistory[N_MINUS_1]); - - FfxFloat32 fMin = abs(fDiffs0); - - if (fMin >= fUnormThreshold) { - for (int i = N_MINUS_2; i <= N_MINUS_4; i++) { - FfxFloat32 fDiffs1 = (fCurrentFrameLuma - fCurrentFrameLumaHistory[i]); - - if (sign(fDiffs0) == sign(fDiffs1)) { - - // Scale difference to protect historically similar values - const FfxFloat32 fMinBias = 1.0f; - fMin = ffxMin(fMin, abs(fDiffs1) * fMinBias); - } - } - - const FfxFloat32 fBoxSize = clippingBox.boxVec.x; - const FfxFloat32 fBoxSizeFactor = ffxPow(ffxSaturate(fBoxSize / 0.1f), 6.0f); - - fLumaInstability = FfxFloat32(fMin != abs(fDiffs0)) * fBoxSizeFactor; - fLumaInstability = FfxFloat32(fLumaInstability > fUnormThreshold); - - fLumaInstability *= 1.0f - ffxMax(params.fAccumulationMask, ffxPow(fThisFrameReactiveFactor, 1.0f / 6.0f)); - } - - //shift history - fCurrentFrameLumaHistory[N_MINUS_4] = fCurrentFrameLumaHistory[N_MINUS_3]; - fCurrentFrameLumaHistory[N_MINUS_3] = fCurrentFrameLumaHistory[N_MINUS_2]; - fCurrentFrameLumaHistory[N_MINUS_2] = fCurrentFrameLumaHistory[N_MINUS_1]; - fCurrentFrameLumaHistory[N_MINUS_1] = fCurrentFrameLuma; - - StoreLumaHistory(params.iPxHrPos, fCurrentFrameLumaHistory); - - return fLumaInstability * FfxFloat32(fCurrentFrameLumaHistory[N_MINUS_4] != 0); -} - -FfxFloat32 ComputeTemporalReactiveFactor(const AccumulationPassCommonParams params, FfxFloat32 fTemporalReactiveFactor) -{ - FfxFloat32 fNewFactor = ffxMin(0.99f, fTemporalReactiveFactor); - - fNewFactor = ffxMax(fNewFactor, ffxLerp(fNewFactor, 0.4f, ffxSaturate(params.fHrVelocity))); - - fNewFactor = ffxMax(fNewFactor * fNewFactor, ffxMax(params.fDepthClipFactor * 0.1f, params.fDilatedReactiveFactor)); - - // Force reactive factor for new samples - fNewFactor = params.bIsNewSample ? 1.0f : fNewFactor; - - if (ffxSaturate(params.fHrVelocity * 10.0f) >= 1.0f) { - fNewFactor = ffxMax(FSR3UPSCALER_EPSILON, fNewFactor) * -1.0f; - } - - return fNewFactor; -} - -AccumulationPassCommonParams InitParams(FfxInt32x2 iPxHrPos) -{ - AccumulationPassCommonParams params; - - params.iPxHrPos = iPxHrPos; - const FfxFloat32x2 fHrUv = (iPxHrPos + 0.5f) / DisplaySize(); - params.fHrUv = fHrUv; - - const FfxFloat32x2 fLrUvJittered = fHrUv + Jitter() / RenderSize(); - params.fLrUv_HwSampler = ClampUv(fLrUvJittered, RenderSize(), MaxRenderSize()); - - params.fMotionVector = GetMotionVector(iPxHrPos, fHrUv); - params.fHrVelocity = GetPxHrVelocity(params.fMotionVector); - - ComputeReprojectedUVs(params, params.fReprojectedHrUv, params.bIsExistingSample); - - params.fDepthClipFactor = ffxSaturate(SampleDepthClip(params.fLrUv_HwSampler)); - - const FfxFloat32x2 fDilatedReactiveMasks = SampleDilatedReactiveMasks(params.fLrUv_HwSampler); - params.fDilatedReactiveFactor = fDilatedReactiveMasks.x; - params.fAccumulationMask = fDilatedReactiveMasks.y; - params.bIsResetFrame = (0 == FrameIndex()); - - params.bIsNewSample = (params.bIsExistingSample == false || params.bIsResetFrame); - - return params; -} - -void Accumulate(FfxInt32x2 iPxHrPos) -{ - const AccumulationPassCommonParams params = InitParams(iPxHrPos); - - FfxFloat32x3 fHistoryColor = FfxFloat32x3(0, 0, 0); - FfxFloat32x2 fLockStatus; - InitializeNewLockSample(fLockStatus); - - FfxFloat32 fTemporalReactiveFactor = 0.0f; - FfxBoolean bInMotionLastFrame = FFX_FALSE; - LockState lockState = { FFX_FALSE , FFX_FALSE }; - if (params.bIsExistingSample && !params.bIsResetFrame) { - ReprojectHistoryColor(params, fHistoryColor, fTemporalReactiveFactor, bInMotionLastFrame); - lockState = ReprojectHistoryLockStatus(params, fLockStatus); - } - - FfxFloat32 fThisFrameReactiveFactor = ffxMax(params.fDilatedReactiveFactor, fTemporalReactiveFactor); - - FfxFloat32 fLuminanceDiff = 0.0f; - FfxFloat32 fLockContributionThisFrame = 0.0f; - UpdateLockStatus(params, fThisFrameReactiveFactor, lockState, fLockStatus, fLockContributionThisFrame, fLuminanceDiff); - - // Load upsampled input color - RectificationBox clippingBox; - FfxFloat32x4 fUpsampledColorAndWeight = ComputeUpsampledColorAndWeight(params, clippingBox, fThisFrameReactiveFactor); - - const FfxFloat32 fLumaInstabilityFactor = ComputeLumaInstabilityFactor(params, clippingBox, fThisFrameReactiveFactor, fLuminanceDiff); - - - FfxFloat32x3 fAccumulation = ComputeBaseAccumulationWeight(params, fThisFrameReactiveFactor, bInMotionLastFrame, fUpsampledColorAndWeight.w, lockState); - - if (params.bIsNewSample) { - fHistoryColor = YCoCgToRGB(fUpsampledColorAndWeight.xyz); - } - else { - RectifyHistory(params, clippingBox, fHistoryColor, fAccumulation, fLockContributionThisFrame, fThisFrameReactiveFactor, fLumaInstabilityFactor); - - Accumulate(params, fHistoryColor, fAccumulation, fUpsampledColorAndWeight); - } - - fHistoryColor = UnprepareRgb(fHistoryColor, Exposure()); - - FinalizeLockStatus(params, fLockStatus, fUpsampledColorAndWeight.w); - - // Get new temporal reactive factor - fTemporalReactiveFactor = ComputeTemporalReactiveFactor(params, fThisFrameReactiveFactor); - - StoreInternalColorAndWeight(iPxHrPos, FfxFloat32x4(fHistoryColor, fTemporalReactiveFactor)); - - // Output final color when RCAS is disabled -#if FFX_FSR3UPSCALER_OPTION_APPLY_SHARPENING == 0 - WriteUpscaledOutput(iPxHrPos, fHistoryColor); -#endif - StoreNewLocks(iPxHrPos, 0); -} - -#endif // FFX_FSR3UPSCALER_ACCUMULATE_H diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_accumulate.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_accumulate.h.meta deleted file mode 100644 index 79fa7a3..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_accumulate.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 3fc2f7a2c8c31324a949e1761bf599cc -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h deleted file mode 100644 index 13b317a..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h +++ /dev/null @@ -1,928 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#include "ffx_fsr3upscaler_resources.h" - -#if defined(FFX_GPU) -#ifdef __hlsl_dx_compiler -#pragma dxc diagnostic push -#pragma dxc diagnostic ignored "-Wambig-lit-shift" -#endif //__hlsl_dx_compiler -#include "ffx_core.h" -#ifdef __hlsl_dx_compiler -#pragma dxc diagnostic pop -#endif //__hlsl_dx_compiler -#endif // #if defined(FFX_GPU) - -#if defined(FFX_GPU) -#ifndef FFX_PREFER_WAVE64 -#define FFX_PREFER_WAVE64 -#endif // FFX_PREFER_WAVE64 - -#if defined(FFX_GPU) -#pragma warning(disable: 3205) // conversion from larger type to smaller -#endif // #if defined(FFX_GPU) - -#define DECLARE_SRV_REGISTER(regIndex) t##regIndex -#define DECLARE_UAV_REGISTER(regIndex) u##regIndex -#define DECLARE_CB_REGISTER(regIndex) b##regIndex -#define FFX_FSR3UPSCALER_DECLARE_SRV(regIndex) register(DECLARE_SRV_REGISTER(regIndex)) -#define FFX_FSR3UPSCALER_DECLARE_UAV(regIndex) register(DECLARE_UAV_REGISTER(regIndex)) -#define FFX_FSR3UPSCALER_DECLARE_CB(regIndex) register(DECLARE_CB_REGISTER(regIndex)) - -#if defined(FSR3UPSCALER_BIND_CB_FSR3UPSCALER) - cbuffer cbFSR3Upscaler : FFX_FSR3UPSCALER_DECLARE_CB(FSR3UPSCALER_BIND_CB_FSR3UPSCALER) - { - FfxInt32x2 iRenderSize; - FfxInt32x2 iMaxRenderSize; - FfxInt32x2 iDisplaySize; - FfxInt32x2 iInputColorResourceDimensions; - FfxInt32x2 iLumaMipDimensions; - FfxInt32 iLumaMipLevelToUse; - FfxInt32 iFrameIndex; - - FfxFloat32x4 fDeviceToViewDepth; - FfxFloat32x2 fJitter; - FfxFloat32x2 fMotionVectorScale; - FfxFloat32x2 fDownscaleFactor; - FfxFloat32x2 fMotionVectorJitterCancellation; - FfxFloat32 fPreExposure; - FfxFloat32 fPreviousFramePreExposure; - FfxFloat32 fTanHalfFOV; - FfxFloat32 fJitterSequenceLength; - FfxFloat32 fDeltaTime; - FfxFloat32 fDynamicResChangeFactor; - FfxFloat32 fViewSpaceToMetersFactor; - - FfxInt32 iDummy; - }; - -#define FFX_FSR3UPSCALER_CONSTANT_BUFFER_1_SIZE (sizeof(cbFSR3Upscaler) / 4) // Number of 32-bit values. This must be kept in sync with the cbFSR3Upscaler size. - -/* Define getter functions in the order they are defined in the CB! */ -FfxInt32x2 RenderSize() -{ - return iRenderSize; -} - -FfxInt32x2 MaxRenderSize() -{ - return iMaxRenderSize; -} - -FfxInt32x2 DisplaySize() -{ - return iDisplaySize; -} - -FfxInt32x2 InputColorResourceDimensions() -{ - return iInputColorResourceDimensions; -} - -FfxInt32x2 LumaMipDimensions() -{ - return iLumaMipDimensions; -} - -FfxInt32 LumaMipLevelToUse() -{ - return iLumaMipLevelToUse; -} - -FfxInt32 FrameIndex() -{ - return iFrameIndex; -} - -FfxFloat32x2 Jitter() -{ - return fJitter; -} - -FfxFloat32x4 DeviceToViewSpaceTransformFactors() -{ - return fDeviceToViewDepth; -} - -FfxFloat32x2 MotionVectorScale() -{ - return fMotionVectorScale; -} - -FfxFloat32x2 DownscaleFactor() -{ - return fDownscaleFactor; -} - -FfxFloat32x2 MotionVectorJitterCancellation() -{ - return fMotionVectorJitterCancellation; -} - -FfxFloat32 PreExposure() -{ - return fPreExposure; -} - -FfxFloat32 PreviousFramePreExposure() -{ - return fPreviousFramePreExposure; -} - -FfxFloat32 TanHalfFoV() -{ - return fTanHalfFOV; -} - -FfxFloat32 JitterSequenceLength() -{ - return fJitterSequenceLength; -} - -FfxFloat32 DeltaTime() -{ - return fDeltaTime; -} - -FfxFloat32 DynamicResChangeFactor() -{ - return fDynamicResChangeFactor; -} - -FfxFloat32 ViewSpaceToMetersFactor() -{ - return fViewSpaceToMetersFactor; -} -#endif // #if defined(FSR3UPSCALER_BIND_CB_FSR3UPSCALER) - -#define FFX_FSR3UPSCALER_ROOTSIG_STRINGIFY(p) FFX_FSR3UPSCALER_ROOTSIG_STR(p) -#define FFX_FSR3UPSCALER_ROOTSIG_STR(p) #p -#define FFX_FSR3UPSCALER_ROOTSIG [RootSignature( "DescriptorTable(UAV(u0, numDescriptors = " FFX_FSR3UPSCALER_ROOTSIG_STRINGIFY(FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_COUNT) ")), " \ - "DescriptorTable(SRV(t0, numDescriptors = " FFX_FSR3UPSCALER_ROOTSIG_STRINGIFY(FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_COUNT) ")), " \ - "RootConstants(num32BitConstants=" FFX_FSR3UPSCALER_ROOTSIG_STRINGIFY(FFX_FSR3UPSCALER_CONSTANT_BUFFER_1_SIZE) ", b0), " \ - "StaticSampler(s0, filter = FILTER_MIN_MAG_MIP_POINT, " \ - "addressU = TEXTURE_ADDRESS_CLAMP, " \ - "addressV = TEXTURE_ADDRESS_CLAMP, " \ - "addressW = TEXTURE_ADDRESS_CLAMP, " \ - "comparisonFunc = COMPARISON_NEVER, " \ - "borderColor = STATIC_BORDER_COLOR_TRANSPARENT_BLACK), " \ - "StaticSampler(s1, filter = FILTER_MIN_MAG_MIP_LINEAR, " \ - "addressU = TEXTURE_ADDRESS_CLAMP, " \ - "addressV = TEXTURE_ADDRESS_CLAMP, " \ - "addressW = TEXTURE_ADDRESS_CLAMP, " \ - "comparisonFunc = COMPARISON_NEVER, " \ - "borderColor = STATIC_BORDER_COLOR_TRANSPARENT_BLACK)" )] - -#define FFX_FSR3UPSCALER_CONSTANT_BUFFER_2_SIZE 6 // Number of 32-bit values. This must be kept in sync with max( cbRCAS , cbSPD) size. - -#define FFX_FSR3UPSCALER_CB2_ROOTSIG [RootSignature( "DescriptorTable(UAV(u0, numDescriptors = " FFX_FSR3UPSCALER_ROOTSIG_STRINGIFY(FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_COUNT) ")), " \ - "DescriptorTable(SRV(t0, numDescriptors = " FFX_FSR3UPSCALER_ROOTSIG_STRINGIFY(FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_COUNT) ")), " \ - "RootConstants(num32BitConstants=" FFX_FSR3UPSCALER_ROOTSIG_STRINGIFY(FFX_FSR3UPSCALER_CONSTANT_BUFFER_1_SIZE) ", b0), " \ - "RootConstants(num32BitConstants=" FFX_FSR3UPSCALER_ROOTSIG_STRINGIFY(FFX_FSR3UPSCALER_CONSTANT_BUFFER_2_SIZE) ", b1), " \ - "StaticSampler(s0, filter = FILTER_MIN_MAG_MIP_POINT, " \ - "addressU = TEXTURE_ADDRESS_CLAMP, " \ - "addressV = TEXTURE_ADDRESS_CLAMP, " \ - "addressW = TEXTURE_ADDRESS_CLAMP, " \ - "comparisonFunc = COMPARISON_NEVER, " \ - "borderColor = STATIC_BORDER_COLOR_TRANSPARENT_BLACK), " \ - "StaticSampler(s1, filter = FILTER_MIN_MAG_MIP_LINEAR, " \ - "addressU = TEXTURE_ADDRESS_CLAMP, " \ - "addressV = TEXTURE_ADDRESS_CLAMP, " \ - "addressW = TEXTURE_ADDRESS_CLAMP, " \ - "comparisonFunc = COMPARISON_NEVER, " \ - "borderColor = STATIC_BORDER_COLOR_TRANSPARENT_BLACK)" )] -#if defined(FFX_FSR3UPSCALER_EMBED_ROOTSIG) -#define FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT FFX_FSR3UPSCALER_ROOTSIG -#define FFX_FSR3UPSCALER_EMBED_CB2_ROOTSIG_CONTENT FFX_FSR3UPSCALER_CB2_ROOTSIG -#else -#define FFX_FSR3UPSCALER_EMBED_ROOTSIG_CONTENT -#define FFX_FSR3UPSCALER_EMBED_CB2_ROOTSIG_CONTENT -#endif // #if FFX_FSR3UPSCALER_EMBED_ROOTSIG - -#if defined(FSR3UPSCALER_BIND_CB_AUTOREACTIVE) -cbuffer cbGenerateReactive : FFX_FSR3UPSCALER_DECLARE_CB(FSR3UPSCALER_BIND_CB_AUTOREACTIVE) -{ - FfxFloat32 fTcThreshold; // 0.1 is a good starting value, lower will result in more TC pixels - FfxFloat32 fTcScale; - FfxFloat32 fReactiveScale; - FfxFloat32 fReactiveMax; -}; - -FfxFloat32 TcThreshold() -{ - return fTcThreshold; -} - -FfxFloat32 TcScale() -{ - return fTcScale; -} - -FfxFloat32 ReactiveScale() -{ - return fReactiveScale; -} - -FfxFloat32 ReactiveMax() -{ - return fReactiveMax; -} -#endif // #if defined(FSR3UPSCALER_BIND_CB_AUTOREACTIVE) - -#if defined(FSR3UPSCALER_BIND_CB_RCAS) -cbuffer cbRCAS : FFX_FSR3UPSCALER_DECLARE_CB(FSR3UPSCALER_BIND_CB_RCAS) -{ - FfxUInt32x4 rcasConfig; -}; - -FfxUInt32x4 RCASConfig() -{ - return rcasConfig; -} -#endif // #if defined(FSR3UPSCALER_BIND_CB_RCAS) - - -#if defined(FSR3UPSCALER_BIND_CB_REACTIVE) -cbuffer cbGenerateReactive : FFX_FSR3UPSCALER_DECLARE_CB(FSR3UPSCALER_BIND_CB_REACTIVE) -{ - FfxFloat32 gen_reactive_scale; - FfxFloat32 gen_reactive_threshold; - FfxFloat32 gen_reactive_binaryValue; - FfxUInt32 gen_reactive_flags; -}; - -FfxFloat32 GenReactiveScale() -{ - return gen_reactive_scale; -} - -FfxFloat32 GenReactiveThreshold() -{ - return gen_reactive_threshold; -} - -FfxFloat32 GenReactiveBinaryValue() -{ - return gen_reactive_binaryValue; -} - -FfxUInt32 GenReactiveFlags() -{ - return gen_reactive_flags; -} -#endif // #if defined(FSR3UPSCALER_BIND_CB_REACTIVE) - -#if defined(FSR3UPSCALER_BIND_CB_SPD) -cbuffer cbSPD : FFX_FSR3UPSCALER_DECLARE_CB(FSR3UPSCALER_BIND_CB_SPD) { - - FfxUInt32 mips; - FfxUInt32 numWorkGroups; - FfxUInt32x2 workGroupOffset; - FfxUInt32x2 renderSize; -}; - -FfxUInt32 MipCount() -{ - return mips; -} - -FfxUInt32 NumWorkGroups() -{ - return numWorkGroups; -} - -FfxUInt32x2 WorkGroupOffset() -{ - return workGroupOffset; -} - -FfxUInt32x2 SPD_RenderSize() -{ - return renderSize; -} -#endif // #if defined(FSR3UPSCALER_BIND_CB_SPD) - -// Declare and sample camera buffers as regular textures, unless overridden -#if !defined(UNITY_FSR3_TEX2D) -#define UNITY_FSR3_TEX2D(type) Texture2D -#endif -#if !defined(UNITY_FSR3_RWTEX2D) -#define UNITY_FSR3_RWTEX2D(type) RWTexture2D -#endif -#if !defined(UNITY_FSR3_POS) -#define UNITY_FSR3_POS(pxPos) (pxPos) -#endif -#if !defined(UNITY_FSR3_UV) -#define UNITY_FSR3_UV(uv) (uv) -#endif - -SamplerState s_PointClamp : register(s0); -SamplerState s_LinearClamp : register(s1); - - // SRVs - #if defined FSR3UPSCALER_BIND_SRV_INPUT_COLOR - UNITY_FSR3_TEX2D(FfxFloat32x4) r_input_color_jittered : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_INPUT_COLOR); - #endif - #if defined FSR3UPSCALER_BIND_SRV_INPUT_OPAQUE_ONLY - UNITY_FSR3_TEX2D(FfxFloat32x4) r_input_opaque_only : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_INPUT_OPAQUE_ONLY); - #endif - #if defined FSR3UPSCALER_BIND_SRV_INPUT_MOTION_VECTORS - UNITY_FSR3_TEX2D(FfxFloat32x4) r_input_motion_vectors : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_INPUT_MOTION_VECTORS); - #endif - #if defined FSR3UPSCALER_BIND_SRV_INPUT_DEPTH - UNITY_FSR3_TEX2D(FfxFloat32) r_input_depth : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_INPUT_DEPTH); - #endif - #if defined FSR3UPSCALER_BIND_SRV_INPUT_EXPOSURE - Texture2D r_input_exposure : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_INPUT_EXPOSURE); - #endif - #if defined FSR3UPSCALER_BIND_SRV_AUTO_EXPOSURE - Texture2D r_auto_exposure : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_AUTO_EXPOSURE); - #endif - #if defined FSR3UPSCALER_BIND_SRV_REACTIVE_MASK - UNITY_FSR3_TEX2D(FfxFloat32) r_reactive_mask : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_REACTIVE_MASK); - #endif - #if defined FSR3UPSCALER_BIND_SRV_TRANSPARENCY_AND_COMPOSITION_MASK - UNITY_FSR3_TEX2D(FfxFloat32) r_transparency_and_composition_mask : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_TRANSPARENCY_AND_COMPOSITION_MASK); - #endif - #if defined FSR3UPSCALER_BIND_SRV_RECONSTRUCTED_PREV_NEAREST_DEPTH - Texture2D r_reconstructed_previous_nearest_depth : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_RECONSTRUCTED_PREV_NEAREST_DEPTH); - #endif - #if defined FSR3UPSCALER_BIND_SRV_DILATED_MOTION_VECTORS - Texture2D r_dilated_motion_vectors : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_DILATED_MOTION_VECTORS); - #endif - #if defined FSR3UPSCALER_BIND_SRV_PREVIOUS_DILATED_MOTION_VECTORS - Texture2D r_previous_dilated_motion_vectors : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_PREVIOUS_DILATED_MOTION_VECTORS); - #endif - #if defined FSR3UPSCALER_BIND_SRV_DILATED_DEPTH - Texture2D r_dilated_depth : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_DILATED_DEPTH); - #endif - #if defined FSR3UPSCALER_BIND_SRV_INTERNAL_UPSCALED - Texture2D r_internal_upscaled_color : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_INTERNAL_UPSCALED); - #endif - #if defined FSR3UPSCALER_BIND_SRV_LOCK_STATUS - Texture2D r_lock_status : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_LOCK_STATUS); - #endif - #if defined FSR3UPSCALER_BIND_SRV_LOCK_INPUT_LUMA - Texture2D r_lock_input_luma : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_LOCK_INPUT_LUMA); - #endif - #if defined FSR3UPSCALER_BIND_SRV_NEW_LOCKS - Texture2D r_new_locks : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_NEW_LOCKS); - #endif - #if defined FSR3UPSCALER_BIND_SRV_PREPARED_INPUT_COLOR - Texture2D r_prepared_input_color : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_PREPARED_INPUT_COLOR); - #endif - #if defined FSR3UPSCALER_BIND_SRV_LUMA_HISTORY - Texture2D r_luma_history : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_LUMA_HISTORY); - #endif - #if defined FSR3UPSCALER_BIND_SRV_RCAS_INPUT - Texture2D r_rcas_input : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_RCAS_INPUT); - #endif - #if defined FSR3UPSCALER_BIND_SRV_LANCZOS_LUT - Texture2D r_lanczos_lut : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_LANCZOS_LUT); - #endif - #if defined FSR3UPSCALER_BIND_SRV_SCENE_LUMINANCE_MIPS - Texture2D r_imgMips : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_SCENE_LUMINANCE_MIPS); - #endif - #if defined FSR3UPSCALER_BIND_SRV_UPSCALE_MAXIMUM_BIAS_LUT - Texture2D r_upsample_maximum_bias_lut : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_UPSCALE_MAXIMUM_BIAS_LUT); - #endif - #if defined FSR3UPSCALER_BIND_SRV_DILATED_REACTIVE_MASKS - Texture2D r_dilated_reactive_masks : FFX_FSR3UPSCALER_DECLARE_SRV(FSR3UPSCALER_BIND_SRV_DILATED_REACTIVE_MASKS); - #endif - - #if defined FSR3UPSCALER_BIND_SRV_PREV_PRE_ALPHA_COLOR - Texture2D r_input_prev_color_pre_alpha : FFX_FSR3UPSCALER_DECLARE_SRV(FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_PREV_PRE_ALPHA_COLOR); - #endif - #if defined FSR3UPSCALER_BIND_SRV_PREV_POST_ALPHA_COLOR - Texture2D r_input_prev_color_post_alpha : FFX_FSR3UPSCALER_DECLARE_SRV(FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_PREV_POST_ALPHA_COLOR); - #endif - - // UAV declarations - #if defined FSR3UPSCALER_BIND_UAV_RECONSTRUCTED_PREV_NEAREST_DEPTH - RWTexture2D rw_reconstructed_previous_nearest_depth : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_RECONSTRUCTED_PREV_NEAREST_DEPTH); - #endif - #if defined FSR3UPSCALER_BIND_UAV_DILATED_MOTION_VECTORS - RWTexture2D rw_dilated_motion_vectors : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_DILATED_MOTION_VECTORS); - #endif - #if defined FSR3UPSCALER_BIND_UAV_DILATED_DEPTH - RWTexture2D rw_dilated_depth : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_DILATED_DEPTH); - #endif - #if defined FSR3UPSCALER_BIND_UAV_INTERNAL_UPSCALED - RWTexture2D rw_internal_upscaled_color : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_INTERNAL_UPSCALED); - #endif - #if defined FSR3UPSCALER_BIND_UAV_LOCK_STATUS - RWTexture2D rw_lock_status : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_LOCK_STATUS); - #endif - #if defined FSR3UPSCALER_BIND_UAV_LOCK_INPUT_LUMA - RWTexture2D rw_lock_input_luma : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_LOCK_INPUT_LUMA); - #endif - #if defined FSR3UPSCALER_BIND_UAV_NEW_LOCKS - RWTexture2D rw_new_locks : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_NEW_LOCKS); - #endif - #if defined FSR3UPSCALER_BIND_UAV_PREPARED_INPUT_COLOR - RWTexture2D rw_prepared_input_color : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_PREPARED_INPUT_COLOR); - #endif - #if defined FSR3UPSCALER_BIND_UAV_LUMA_HISTORY - RWTexture2D rw_luma_history : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_LUMA_HISTORY); - #endif - #if defined FSR3UPSCALER_BIND_UAV_UPSCALED_OUTPUT - UNITY_FSR3_RWTEX2D(FfxFloat32x4) rw_upscaled_output : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_UPSCALED_OUTPUT); - #endif - #if defined FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_LUMA_CHANGE - globallycoherent RWTexture2D rw_img_mip_shading_change : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_LUMA_CHANGE); - #endif - #if defined FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_5 - globallycoherent RWTexture2D rw_img_mip_5 : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_5); - #endif - #if defined FSR3UPSCALER_BIND_UAV_DILATED_REACTIVE_MASKS - RWTexture2D rw_dilated_reactive_masks : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_DILATED_REACTIVE_MASKS); - #endif - #if defined FSR3UPSCALER_BIND_UAV_EXPOSURE - RWTexture2D rw_exposure : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_EXPOSURE); - #endif - #if defined FSR3UPSCALER_BIND_UAV_AUTO_EXPOSURE - RWTexture2D rw_auto_exposure : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_AUTO_EXPOSURE); - #endif - #if defined FSR3UPSCALER_BIND_UAV_SPD_GLOBAL_ATOMIC - globallycoherent RWTexture2D rw_spd_global_atomic : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_SPD_GLOBAL_ATOMIC); - #endif - - #if defined FSR3UPSCALER_BIND_UAV_AUTOREACTIVE - RWTexture2D rw_output_autoreactive : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_AUTOREACTIVE); - #endif - #if defined FSR3UPSCALER_BIND_UAV_AUTOCOMPOSITION - RWTexture2D rw_output_autocomposition : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_AUTOCOMPOSITION); - #endif - #if defined FSR3UPSCALER_BIND_UAV_PREV_PRE_ALPHA_COLOR - RWTexture2D rw_output_prev_color_pre_alpha : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_PREV_PRE_ALPHA_COLOR); - #endif - #if defined FSR3UPSCALER_BIND_UAV_PREV_POST_ALPHA_COLOR - RWTexture2D rw_output_prev_color_post_alpha : FFX_FSR3UPSCALER_DECLARE_UAV(FSR3UPSCALER_BIND_UAV_PREV_POST_ALPHA_COLOR); - #endif - -#if defined(FSR3UPSCALER_BIND_SRV_SCENE_LUMINANCE_MIPS) -FfxFloat32 LoadMipLuma(FfxUInt32x2 iPxPos, FfxUInt32 mipLevel) -{ - return r_imgMips.mips[mipLevel][iPxPos]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_SCENE_LUMINANCE_MIPS) -FfxFloat32 SampleMipLuma(FfxFloat32x2 fUV, FfxUInt32 mipLevel) -{ - return r_imgMips.SampleLevel(s_LinearClamp, fUV, mipLevel); -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_INPUT_DEPTH) -FfxFloat32 LoadInputDepth(FfxUInt32x2 iPxPos) -{ - return r_input_depth[UNITY_FSR3_POS(iPxPos)]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_INPUT_DEPTH) -FfxFloat32 SampleInputDepth(FfxFloat32x2 fUV) -{ - return r_input_depth.SampleLevel(s_LinearClamp, UNITY_FSR3_UV(fUV), 0).x; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_REACTIVE_MASK) -FfxFloat32 LoadReactiveMask(FfxUInt32x2 iPxPos) -{ - return r_reactive_mask[UNITY_FSR3_POS(iPxPos)]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_TRANSPARENCY_AND_COMPOSITION_MASK) -FfxFloat32 LoadTransparencyAndCompositionMask(FfxUInt32x2 iPxPos) -{ - return r_transparency_and_composition_mask[UNITY_FSR3_POS(iPxPos)]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_INPUT_COLOR) -FfxFloat32x3 LoadInputColor(FfxUInt32x2 iPxPos) -{ - return r_input_color_jittered[UNITY_FSR3_POS(iPxPos)].rgb; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_INPUT_COLOR) -FfxFloat32x3 SampleInputColor(FfxFloat32x2 fUV) -{ - return r_input_color_jittered.SampleLevel(s_LinearClamp, UNITY_FSR3_UV(fUV), 0).rgb; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_PREPARED_INPUT_COLOR) -FfxFloat32x3 LoadPreparedInputColor(FfxUInt32x2 iPxPos) -{ - return r_prepared_input_color[iPxPos].xyz; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_INPUT_MOTION_VECTORS) -FfxFloat32x2 LoadInputMotionVector(FfxUInt32x2 iPxDilatedMotionVectorPos) -{ - FfxFloat32x2 fSrcMotionVector = r_input_motion_vectors[UNITY_FSR3_POS(iPxDilatedMotionVectorPos)].xy; - - FfxFloat32x2 fUvMotionVector = fSrcMotionVector * MotionVectorScale(); - -#if FFX_FSR3UPSCALER_OPTION_JITTERED_MOTION_VECTORS - fUvMotionVector -= MotionVectorJitterCancellation(); -#endif - - return fUvMotionVector; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_INTERNAL_UPSCALED) -FfxFloat32x4 LoadHistory(FfxUInt32x2 iPxHistory) -{ - return r_internal_upscaled_color[iPxHistory]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_LUMA_HISTORY) -void StoreLumaHistory(FfxUInt32x2 iPxPos, FfxFloat32x4 fLumaHistory) -{ - rw_luma_history[iPxPos] = fLumaHistory; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_LUMA_HISTORY) -FfxFloat32x4 SampleLumaHistory(FfxFloat32x2 fUV) -{ - return r_luma_history.SampleLevel(s_LinearClamp, fUV, 0); -} -#endif - -FfxFloat32x4 LoadRCAS_Input(FfxInt32x2 iPxPos) -{ -#if defined(FSR3UPSCALER_BIND_SRV_RCAS_INPUT) - return r_rcas_input[iPxPos]; -#else - return 0.0; -#endif -} - -#if defined(FSR3UPSCALER_BIND_UAV_INTERNAL_UPSCALED) -void StoreReprojectedHistory(FfxUInt32x2 iPxHistory, FfxFloat32x4 fHistory) -{ - rw_internal_upscaled_color[iPxHistory] = fHistory; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_INTERNAL_UPSCALED) -void StoreInternalColorAndWeight(FfxUInt32x2 iPxPos, FfxFloat32x4 fColorAndWeight) -{ - rw_internal_upscaled_color[iPxPos] = fColorAndWeight; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_UPSCALED_OUTPUT) -void StoreUpscaledOutput(FfxUInt32x2 iPxPos, FfxFloat32x3 fColor) -{ - rw_upscaled_output[UNITY_FSR3_POS(iPxPos)] = FfxFloat32x4(fColor, 1.f); -} -#endif - -//LOCK_LIFETIME_REMAINING == 0 -//Should make LockInitialLifetime() return a const 1.0f later -#if defined(FSR3UPSCALER_BIND_SRV_LOCK_STATUS) -FfxFloat32x2 LoadLockStatus(FfxUInt32x2 iPxPos) -{ - return r_lock_status[iPxPos]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_LOCK_STATUS) -void StoreLockStatus(FfxUInt32x2 iPxPos, FfxFloat32x2 fLockStatus) -{ - rw_lock_status[iPxPos] = fLockStatus; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_LOCK_INPUT_LUMA) -FfxFloat32 LoadLockInputLuma(FfxUInt32x2 iPxPos) -{ - return r_lock_input_luma[iPxPos]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_LOCK_INPUT_LUMA) -void StoreLockInputLuma(FfxUInt32x2 iPxPos, FfxFloat32 fLuma) -{ - rw_lock_input_luma[iPxPos] = fLuma; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_NEW_LOCKS) -FfxFloat32 LoadNewLocks(FfxUInt32x2 iPxPos) -{ - return r_new_locks[iPxPos]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_NEW_LOCKS) -FfxFloat32 LoadRwNewLocks(FfxUInt32x2 iPxPos) -{ - return rw_new_locks[iPxPos]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_NEW_LOCKS) -void StoreNewLocks(FfxUInt32x2 iPxPos, FfxFloat32 newLock) -{ - rw_new_locks[iPxPos] = newLock; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_PREPARED_INPUT_COLOR) -void StorePreparedInputColor(FFX_PARAMETER_IN FfxUInt32x2 iPxPos, FFX_PARAMETER_IN FfxFloat32x4 fTonemapped) -{ - rw_prepared_input_color[iPxPos] = fTonemapped; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_PREPARED_INPUT_COLOR) -FfxFloat32 SampleDepthClip(FfxFloat32x2 fUV) -{ - return r_prepared_input_color.SampleLevel(s_LinearClamp, fUV, 0).w; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_LOCK_STATUS) -FfxFloat32x2 SampleLockStatus(FfxFloat32x2 fUV) -{ - FfxFloat32x2 fLockStatus = r_lock_status.SampleLevel(s_LinearClamp, fUV, 0); - return fLockStatus; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_RECONSTRUCTED_PREV_NEAREST_DEPTH) -FfxFloat32 LoadReconstructedPrevDepth(FfxUInt32x2 iPxPos) -{ - return asfloat(r_reconstructed_previous_nearest_depth[iPxPos]); -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_RECONSTRUCTED_PREV_NEAREST_DEPTH) -void StoreReconstructedDepth(FfxUInt32x2 iPxSample, FfxFloat32 fDepth) -{ - FfxUInt32 uDepth = asuint(fDepth); - - #if FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - InterlockedMax(rw_reconstructed_previous_nearest_depth[iPxSample], uDepth); - #else - InterlockedMin(rw_reconstructed_previous_nearest_depth[iPxSample], uDepth); // min for standard, max for inverted depth - #endif -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_RECONSTRUCTED_PREV_NEAREST_DEPTH) -void SetReconstructedDepth(FfxUInt32x2 iPxSample, const FfxUInt32 uValue) -{ - rw_reconstructed_previous_nearest_depth[iPxSample] = uValue; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_DILATED_DEPTH) -void StoreDilatedDepth(FFX_PARAMETER_IN FfxUInt32x2 iPxPos, FFX_PARAMETER_IN FfxFloat32 fDepth) -{ - rw_dilated_depth[iPxPos] = fDepth; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_DILATED_MOTION_VECTORS) -void StoreDilatedMotionVector(FFX_PARAMETER_IN FfxUInt32x2 iPxPos, FFX_PARAMETER_IN FfxFloat32x2 fMotionVector) -{ - rw_dilated_motion_vectors[iPxPos] = fMotionVector; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_DILATED_MOTION_VECTORS) -FfxFloat32x2 LoadDilatedMotionVector(FfxUInt32x2 iPxInput) -{ - return r_dilated_motion_vectors[iPxInput].xy; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_PREVIOUS_DILATED_MOTION_VECTORS) -FfxFloat32x2 LoadPreviousDilatedMotionVector(FfxUInt32x2 iPxInput) -{ - return r_previous_dilated_motion_vectors[iPxInput].xy; -} - -FfxFloat32x2 SamplePreviousDilatedMotionVector(FfxFloat32x2 uv) -{ - return r_previous_dilated_motion_vectors.SampleLevel(s_LinearClamp, uv, 0).xy; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_DILATED_DEPTH) -FfxFloat32 LoadDilatedDepth(FfxUInt32x2 iPxInput) -{ - return r_dilated_depth[iPxInput]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_INPUT_EXPOSURE) -FfxFloat32 Exposure() -{ - FfxFloat32 exposure = r_input_exposure[FfxUInt32x2(0, 0)].x; - - if (exposure == 0.0f) { - exposure = 1.0f; - } - - return exposure; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_AUTO_EXPOSURE) -FfxFloat32 AutoExposure() -{ - FfxFloat32 exposure = r_auto_exposure[FfxUInt32x2(0, 0)].x; - - if (exposure == 0.0f) { - exposure = 1.0f; - } - - return exposure; -} -#endif - -FfxFloat32 SampleLanczos2Weight(FfxFloat32 x) -{ -#if defined(FSR3UPSCALER_BIND_SRV_LANCZOS_LUT) - return r_lanczos_lut.SampleLevel(s_LinearClamp, FfxFloat32x2(x / 2, 0.5f), 0); -#else - return 0.f; -#endif -} - -#if defined(FSR3UPSCALER_BIND_SRV_UPSCALE_MAXIMUM_BIAS_LUT) -FfxFloat32 SampleUpsampleMaximumBias(FfxFloat32x2 uv) -{ - // Stored as a SNORM, so make sure to multiply by 2 to retrieve the actual expected range. - return FfxFloat32(2.0) * r_upsample_maximum_bias_lut.SampleLevel(s_LinearClamp, abs(uv) * 2.0, 0); -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_DILATED_REACTIVE_MASKS) -FfxFloat32x2 SampleDilatedReactiveMasks(FfxFloat32x2 fUV) -{ - return r_dilated_reactive_masks.SampleLevel(s_LinearClamp, fUV, 0); -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_DILATED_REACTIVE_MASKS) -FfxFloat32x2 LoadDilatedReactiveMasks(FFX_PARAMETER_IN FfxUInt32x2 iPxPos) -{ - return r_dilated_reactive_masks[iPxPos]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_DILATED_REACTIVE_MASKS) -void StoreDilatedReactiveMasks(FFX_PARAMETER_IN FfxUInt32x2 iPxPos, FFX_PARAMETER_IN FfxFloat32x2 fDilatedReactiveMasks) -{ - rw_dilated_reactive_masks[iPxPos] = fDilatedReactiveMasks; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_INPUT_OPAQUE_ONLY) -FfxFloat32x3 LoadOpaqueOnly(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos) -{ - return r_input_opaque_only[UNITY_FSR3_POS(iPxPos)].xyz; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_PREV_PRE_ALPHA_COLOR) -FfxFloat32x3 LoadPrevPreAlpha(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos) -{ - return r_input_prev_color_pre_alpha[iPxPos]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_SRV_PREV_POST_ALPHA_COLOR) -FfxFloat32x3 LoadPrevPostAlpha(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos) -{ - return r_input_prev_color_post_alpha[iPxPos]; -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_AUTOREACTIVE) -#if defined(FSR3UPSCALER_BIND_UAV_AUTOCOMPOSITION) -void StoreAutoReactive(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos, FFX_PARAMETER_IN FFX_MIN16_F2 fReactive) -{ - rw_output_autoreactive[iPxPos] = fReactive.x; - - rw_output_autocomposition[iPxPos] = fReactive.y; -} -#endif -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_PREV_PRE_ALPHA_COLOR) -void StorePrevPreAlpha(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos, FFX_PARAMETER_IN FFX_MIN16_F3 color) -{ - rw_output_prev_color_pre_alpha[iPxPos] = color; - -} -#endif - -#if defined(FSR3UPSCALER_BIND_UAV_PREV_POST_ALPHA_COLOR) -void StorePrevPostAlpha(FFX_PARAMETER_IN FFX_MIN16_I2 iPxPos, FFX_PARAMETER_IN FFX_MIN16_F3 color) -{ - rw_output_prev_color_post_alpha[iPxPos] = color; -} -#endif - -FfxFloat32x2 SPD_LoadExposureBuffer() -{ -#if defined FSR3UPSCALER_BIND_UAV_AUTO_EXPOSURE - return rw_auto_exposure[FfxInt32x2(0, 0)]; -#else - return FfxFloat32x2(0.f, 0.f); -#endif // #if defined FSR3UPSCALER_BIND_UAV_AUTO_EXPOSURE -} - -void SPD_SetExposureBuffer(FfxFloat32x2 value) -{ -#if defined FSR3UPSCALER_BIND_UAV_AUTO_EXPOSURE - rw_auto_exposure[FfxInt32x2(0, 0)] = value; -#endif // #if defined FSR3UPSCALER_BIND_UAV_AUTO_EXPOSURE -} - -FfxFloat32x4 SPD_LoadMipmap5(FfxInt32x2 iPxPos) -{ -#if defined FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_5 - return FfxFloat32x4(rw_img_mip_5[iPxPos], 0, 0, 0); -#else - return FfxFloat32x4(0.f, 0.f, 0.f, 0.f); -#endif // #if defined FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_5 -} - -void SPD_SetMipmap(FfxInt32x2 iPxPos, FfxUInt32 slice, FfxFloat32 value) -{ - switch (slice) - { - case FFX_FSR3UPSCALER_SHADING_CHANGE_MIP_LEVEL: -#if defined FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_LUMA_CHANGE - rw_img_mip_shading_change[iPxPos] = value; -#endif // #if defined FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_LUMA_CHANGE - break; - case 5: -#if defined FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_5 - rw_img_mip_5[iPxPos] = value; -#endif // #if defined FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_5 - break; - default: - - // avoid flattened side effect -#if defined(FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_LUMA_CHANGE) - rw_img_mip_shading_change[iPxPos] = rw_img_mip_shading_change[iPxPos]; -#elif defined(FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_5) - rw_img_mip_5[iPxPos] = rw_img_mip_5[iPxPos]; -#endif // #if defined FSR3UPSCALER_BIND_UAV_EXPOSURE_MIP_5 - break; - } -} - -void SPD_IncreaseAtomicCounter(inout FfxUInt32 spdCounter) -{ -#if defined FSR3UPSCALER_BIND_UAV_SPD_GLOBAL_ATOMIC - InterlockedAdd(rw_spd_global_atomic[FfxInt32x2(0, 0)], 1, spdCounter); -#endif // #if defined FSR3UPSCALER_BIND_UAV_SPD_GLOBAL_ATOMIC -} - -void SPD_ResetAtomicCounter() -{ -#if defined FSR3UPSCALER_BIND_UAV_SPD_GLOBAL_ATOMIC - rw_spd_global_atomic[FfxInt32x2(0, 0)] = 0; -#endif // #if defined FSR3UPSCALER_BIND_UAV_SPD_GLOBAL_ATOMIC -} - -#endif // #if defined(FFX_GPU) diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h.meta deleted file mode 100644 index e78e6a1..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_callbacks_hlsl.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: ba849fdeb042e7f458c81408414db834 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_common.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_common.h deleted file mode 100644 index 1f78a29..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_common.h +++ /dev/null @@ -1,566 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#if !defined(FFX_FSR3UPSCALER_COMMON_H) -#define FFX_FSR3UPSCALER_COMMON_H - -#if defined(FFX_CPU) || defined(FFX_GPU) -//Locks -#define LOCK_LIFETIME_REMAINING 0 -#define LOCK_TEMPORAL_LUMA 1 -#endif // #if defined(FFX_CPU) || defined(FFX_GPU) - -#if defined(FFX_GPU) -FFX_STATIC const FfxFloat32 FSR3UPSCALER_FP16_MIN = 6.10e-05f; -FFX_STATIC const FfxFloat32 FSR3UPSCALER_FP16_MAX = 65504.0f; -FFX_STATIC const FfxFloat32 FSR3UPSCALER_EPSILON = 1e-03f; -FFX_STATIC const FfxFloat32 FSR3UPSCALER_TONEMAP_EPSILON = 1.0f / FSR3UPSCALER_FP16_MAX; -FFX_STATIC const FfxFloat32 FSR3UPSCALER_FLT_MAX = 3.402823466e+38f; -FFX_STATIC const FfxFloat32 FSR3UPSCALER_FLT_MIN = 1.175494351e-38f; - -// treat vector truncation warnings as errors -#pragma warning(error: 3206) - -// suppress warnings -#pragma warning(disable: 3205) // conversion from larger type to smaller -#pragma warning(disable: 3571) // in ffxPow(f, e), f could be negative - -// Reconstructed depth usage -FFX_STATIC const FfxFloat32 fReconstructedDepthBilinearWeightThreshold = 0.01f; - -// Accumulation -FFX_STATIC const FfxFloat32 fUpsampleLanczosWeightScale = 1.0f / 12.0f; -FFX_STATIC const FfxFloat32 fMaxAccumulationLanczosWeight = 1.0f; -FFX_STATIC const FfxFloat32 fAverageLanczosWeightPerFrame = 0.74f * fUpsampleLanczosWeightScale; // Average lanczos weight for jitter accumulated samples -FFX_STATIC const FfxFloat32 fAccumulationMaxOnMotion = 3.0f * fUpsampleLanczosWeightScale; - -// Auto exposure -FFX_STATIC const FfxFloat32 resetAutoExposureAverageSmoothing = 1e8f; - -struct AccumulationPassCommonParams -{ - FfxInt32x2 iPxHrPos; - FfxFloat32x2 fHrUv; - FfxFloat32x2 fLrUv_HwSampler; - FfxFloat32x2 fMotionVector; - FfxFloat32x2 fReprojectedHrUv; - FfxFloat32 fHrVelocity; - FfxFloat32 fDepthClipFactor; - FfxFloat32 fDilatedReactiveFactor; - FfxFloat32 fAccumulationMask; - - FfxBoolean bIsResetFrame; - FfxBoolean bIsExistingSample; - FfxBoolean bIsNewSample; -}; - -struct LockState -{ - FfxBoolean NewLock; //Set for both unique new and re-locked new - FfxBoolean WasLockedPrevFrame; //Set to identify if the pixel was already locked (relock) -}; - -void InitializeNewLockSample(FFX_PARAMETER_OUT FfxFloat32x2 fLockStatus) -{ - fLockStatus = FfxFloat32x2(0, 0); -} - -#if FFX_HALF -void InitializeNewLockSample(FFX_PARAMETER_OUT FFX_MIN16_F2 fLockStatus) -{ - fLockStatus = FFX_MIN16_F2(0, 0); -} -#endif - - -void KillLock(FFX_PARAMETER_INOUT FfxFloat32x2 fLockStatus) -{ - fLockStatus[LOCK_LIFETIME_REMAINING] = 0; -} - -#if FFX_HALF -void KillLock(FFX_PARAMETER_INOUT FFX_MIN16_F2 fLockStatus) -{ - fLockStatus[LOCK_LIFETIME_REMAINING] = FFX_MIN16_F(0); -} -#endif - -struct RectificationBox -{ - FfxFloat32x3 boxCenter; - FfxFloat32x3 boxVec; - FfxFloat32x3 aabbMin; - FfxFloat32x3 aabbMax; - FfxFloat32 fBoxCenterWeight; -}; -#if FFX_HALF -struct RectificationBoxMin16 -{ - FFX_MIN16_F3 boxCenter; - FFX_MIN16_F3 boxVec; - FFX_MIN16_F3 aabbMin; - FFX_MIN16_F3 aabbMax; - FFX_MIN16_F fBoxCenterWeight; -}; -#endif - -void RectificationBoxReset(FFX_PARAMETER_INOUT RectificationBox rectificationBox) -{ - rectificationBox.fBoxCenterWeight = FfxFloat32(0); - - rectificationBox.boxCenter = FfxFloat32x3(0, 0, 0); - rectificationBox.boxVec = FfxFloat32x3(0, 0, 0); - rectificationBox.aabbMin = FfxFloat32x3(FSR3UPSCALER_FLT_MAX, FSR3UPSCALER_FLT_MAX, FSR3UPSCALER_FLT_MAX); - rectificationBox.aabbMax = -FfxFloat32x3(FSR3UPSCALER_FLT_MAX, FSR3UPSCALER_FLT_MAX, FSR3UPSCALER_FLT_MAX); -} -#if FFX_HALF -void RectificationBoxReset(FFX_PARAMETER_INOUT RectificationBoxMin16 rectificationBox) -{ - rectificationBox.fBoxCenterWeight = FFX_MIN16_F(0); - - rectificationBox.boxCenter = FFX_MIN16_F3(0, 0, 0); - rectificationBox.boxVec = FFX_MIN16_F3(0, 0, 0); - rectificationBox.aabbMin = FFX_MIN16_F3(FSR3UPSCALER_FP16_MAX, FSR3UPSCALER_FP16_MAX, FSR3UPSCALER_FP16_MAX); - rectificationBox.aabbMax = -FFX_MIN16_F3(FSR3UPSCALER_FP16_MAX, FSR3UPSCALER_FP16_MAX, FSR3UPSCALER_FP16_MAX); -} -#endif - -void RectificationBoxAddInitialSample(FFX_PARAMETER_INOUT RectificationBox rectificationBox, const FfxFloat32x3 colorSample, const FfxFloat32 fSampleWeight) -{ - rectificationBox.aabbMin = colorSample; - rectificationBox.aabbMax = colorSample; - - FfxFloat32x3 weightedSample = colorSample * fSampleWeight; - rectificationBox.boxCenter = weightedSample; - rectificationBox.boxVec = colorSample * weightedSample; - rectificationBox.fBoxCenterWeight = fSampleWeight; -} - -void RectificationBoxAddSample(FfxBoolean bInitialSample, FFX_PARAMETER_INOUT RectificationBox rectificationBox, const FfxFloat32x3 colorSample, const FfxFloat32 fSampleWeight) -{ - if (bInitialSample) { - RectificationBoxAddInitialSample(rectificationBox, colorSample, fSampleWeight); - } else { - rectificationBox.aabbMin = ffxMin(rectificationBox.aabbMin, colorSample); - rectificationBox.aabbMax = ffxMax(rectificationBox.aabbMax, colorSample); - - FfxFloat32x3 weightedSample = colorSample * fSampleWeight; - rectificationBox.boxCenter += weightedSample; - rectificationBox.boxVec += colorSample * weightedSample; - rectificationBox.fBoxCenterWeight += fSampleWeight; - } -} -#if FFX_HALF -void RectificationBoxAddInitialSample(FFX_PARAMETER_INOUT RectificationBoxMin16 rectificationBox, const FFX_MIN16_F3 colorSample, const FFX_MIN16_F fSampleWeight) -{ - rectificationBox.aabbMin = colorSample; - rectificationBox.aabbMax = colorSample; - - FFX_MIN16_F3 weightedSample = colorSample * fSampleWeight; - rectificationBox.boxCenter = weightedSample; - rectificationBox.boxVec = colorSample * weightedSample; - rectificationBox.fBoxCenterWeight = fSampleWeight; -} - -void RectificationBoxAddSample(FfxBoolean bInitialSample, FFX_PARAMETER_INOUT RectificationBoxMin16 rectificationBox, const FFX_MIN16_F3 colorSample, const FFX_MIN16_F fSampleWeight) -{ - if (bInitialSample) { - RectificationBoxAddInitialSample(rectificationBox, colorSample, fSampleWeight); - } else { - rectificationBox.aabbMin = ffxMin(rectificationBox.aabbMin, colorSample); - rectificationBox.aabbMax = ffxMax(rectificationBox.aabbMax, colorSample); - - FFX_MIN16_F3 weightedSample = colorSample * fSampleWeight; - rectificationBox.boxCenter += weightedSample; - rectificationBox.boxVec += colorSample * weightedSample; - rectificationBox.fBoxCenterWeight += fSampleWeight; - } -} -#endif - -void RectificationBoxComputeVarianceBoxData(FFX_PARAMETER_INOUT RectificationBox rectificationBox) -{ - rectificationBox.fBoxCenterWeight = (abs(rectificationBox.fBoxCenterWeight) > FfxFloat32(FSR3UPSCALER_EPSILON) ? rectificationBox.fBoxCenterWeight : FfxFloat32(1.f)); - rectificationBox.boxCenter /= rectificationBox.fBoxCenterWeight; - rectificationBox.boxVec /= rectificationBox.fBoxCenterWeight; - FfxFloat32x3 stdDev = sqrt(abs(rectificationBox.boxVec - rectificationBox.boxCenter * rectificationBox.boxCenter)); - rectificationBox.boxVec = stdDev; -} -#if FFX_HALF -void RectificationBoxComputeVarianceBoxData(FFX_PARAMETER_INOUT RectificationBoxMin16 rectificationBox) -{ - rectificationBox.fBoxCenterWeight = (abs(rectificationBox.fBoxCenterWeight) > FFX_MIN16_F(FSR3UPSCALER_EPSILON) ? rectificationBox.fBoxCenterWeight : FFX_MIN16_F(1.f)); - rectificationBox.boxCenter /= rectificationBox.fBoxCenterWeight; - rectificationBox.boxVec /= rectificationBox.fBoxCenterWeight; - FFX_MIN16_F3 stdDev = sqrt(abs(rectificationBox.boxVec - rectificationBox.boxCenter * rectificationBox.boxCenter)); - rectificationBox.boxVec = stdDev; -} -#endif - -FfxFloat32x3 SafeRcp3(FfxFloat32x3 v) -{ - return (all(FFX_NOT_EQUAL(v, FfxFloat32x3(0, 0, 0)))) ? (FfxFloat32x3(1, 1, 1) / v) : FfxFloat32x3(0, 0, 0); -} -#if FFX_HALF -FFX_MIN16_F3 SafeRcp3(FFX_MIN16_F3 v) -{ - return (all(FFX_NOT_EQUAL(v, FFX_MIN16_F3(0, 0, 0)))) ? (FFX_MIN16_F3(1, 1, 1) / v) : FFX_MIN16_F3(0, 0, 0); -} -#endif - -FfxFloat32 MinDividedByMax(const FfxFloat32 v0, const FfxFloat32 v1) -{ - const FfxFloat32 m = ffxMax(v0, v1); - return m != 0 ? ffxMin(v0, v1) / m : 0; -} - -#if FFX_HALF -FFX_MIN16_F MinDividedByMax(const FFX_MIN16_F v0, const FFX_MIN16_F v1) -{ - const FFX_MIN16_F m = ffxMax(v0, v1); - return m != FFX_MIN16_F(0) ? ffxMin(v0, v1) / m : FFX_MIN16_F(0); -} -#endif - -FfxFloat32x3 YCoCgToRGB(FfxFloat32x3 fYCoCg) -{ - FfxFloat32x3 fRgb; - - fRgb = FfxFloat32x3( - fYCoCg.x + fYCoCg.y - fYCoCg.z, - fYCoCg.x + fYCoCg.z, - fYCoCg.x - fYCoCg.y - fYCoCg.z); - - return fRgb; -} -#if FFX_HALF -FFX_MIN16_F3 YCoCgToRGB(FFX_MIN16_F3 fYCoCg) -{ - FFX_MIN16_F3 fRgb; - - fRgb = FFX_MIN16_F3( - fYCoCg.x + fYCoCg.y - fYCoCg.z, - fYCoCg.x + fYCoCg.z, - fYCoCg.x - fYCoCg.y - fYCoCg.z); - - return fRgb; -} -#endif - -FfxFloat32x3 RGBToYCoCg(FfxFloat32x3 fRgb) -{ - FfxFloat32x3 fYCoCg; - - fYCoCg = FfxFloat32x3( - 0.25f * fRgb.r + 0.5f * fRgb.g + 0.25f * fRgb.b, - 0.5f * fRgb.r - 0.5f * fRgb.b, - -0.25f * fRgb.r + 0.5f * fRgb.g - 0.25f * fRgb.b); - - return fYCoCg; -} -#if FFX_HALF -FFX_MIN16_F3 RGBToYCoCg(FFX_MIN16_F3 fRgb) -{ - FFX_MIN16_F3 fYCoCg; - - fYCoCg = FFX_MIN16_F3( - 0.25 * fRgb.r + 0.5 * fRgb.g + 0.25 * fRgb.b, - 0.5 * fRgb.r - 0.5 * fRgb.b, - -0.25 * fRgb.r + 0.5 * fRgb.g - 0.25 * fRgb.b); - - return fYCoCg; -} -#endif - -FfxFloat32 RGBToLuma(FfxFloat32x3 fLinearRgb) -{ - return dot(fLinearRgb, FfxFloat32x3(0.2126f, 0.7152f, 0.0722f)); -} -#if FFX_HALF -FFX_MIN16_F RGBToLuma(FFX_MIN16_F3 fLinearRgb) -{ - return dot(fLinearRgb, FFX_MIN16_F3(0.2126f, 0.7152f, 0.0722f)); -} -#endif - -FfxFloat32 RGBToPerceivedLuma(FfxFloat32x3 fLinearRgb) -{ - FfxFloat32 fLuminance = RGBToLuma(fLinearRgb); - - FfxFloat32 fPercievedLuminance = 0; - if (fLuminance <= 216.0f / 24389.0f) { - fPercievedLuminance = fLuminance * (24389.0f / 27.0f); - } - else { - fPercievedLuminance = ffxPow(fLuminance, 1.0f / 3.0f) * 116.0f - 16.0f; - } - - return fPercievedLuminance * 0.01f; -} -#if FFX_HALF -FFX_MIN16_F RGBToPerceivedLuma(FFX_MIN16_F3 fLinearRgb) -{ - FFX_MIN16_F fLuminance = RGBToLuma(fLinearRgb); - - FFX_MIN16_F fPercievedLuminance = FFX_MIN16_F(0); - if (fLuminance <= FFX_MIN16_F(216.0f / 24389.0f)) { - fPercievedLuminance = fLuminance * FFX_MIN16_F(24389.0f / 27.0f); - } - else { - fPercievedLuminance = ffxPow(fLuminance, FFX_MIN16_F(1.0f / 3.0f)) * FFX_MIN16_F(116.0f) - FFX_MIN16_F(16.0f); - } - - return fPercievedLuminance * FFX_MIN16_F(0.01f); -} -#endif - -FfxFloat32x3 Tonemap(FfxFloat32x3 fRgb) -{ - return fRgb / (ffxMax(ffxMax(0.f, fRgb.r), ffxMax(fRgb.g, fRgb.b)) + 1.f).xxx; -} - -FfxFloat32x3 InverseTonemap(FfxFloat32x3 fRgb) -{ - return fRgb / ffxMax(FSR3UPSCALER_TONEMAP_EPSILON, 1.f - ffxMax(fRgb.r, ffxMax(fRgb.g, fRgb.b))).xxx; -} - -#if FFX_HALF -FFX_MIN16_F3 Tonemap(FFX_MIN16_F3 fRgb) -{ - return fRgb / (ffxMax(ffxMax(FFX_MIN16_F(0.f), fRgb.r), ffxMax(fRgb.g, fRgb.b)) + FFX_MIN16_F(1.f)).xxx; -} - -FFX_MIN16_F3 InverseTonemap(FFX_MIN16_F3 fRgb) -{ - return fRgb / ffxMax(FFX_MIN16_F(FSR3UPSCALER_TONEMAP_EPSILON), FFX_MIN16_F(1.f) - ffxMax(fRgb.r, ffxMax(fRgb.g, fRgb.b))).xxx; -} -#endif - -FfxInt32x2 ClampLoad(FfxInt32x2 iPxSample, FfxInt32x2 iPxOffset, FfxInt32x2 iTextureSize) -{ - FfxInt32x2 result = iPxSample + iPxOffset; - result.x = (iPxOffset.x < 0) ? ffxMax(result.x, 0) : result.x; - result.x = (iPxOffset.x > 0) ? ffxMin(result.x, iTextureSize.x - 1) : result.x; - result.y = (iPxOffset.y < 0) ? ffxMax(result.y, 0) : result.y; - result.y = (iPxOffset.y > 0) ? ffxMin(result.y, iTextureSize.y - 1) : result.y; - return result; - - // return ffxMed3(iPxSample + iPxOffset, FfxInt32x2(0, 0), iTextureSize - FfxInt32x2(1, 1)); -} -#if FFX_HALF -FFX_MIN16_I2 ClampLoad(FFX_MIN16_I2 iPxSample, FFX_MIN16_I2 iPxOffset, FFX_MIN16_I2 iTextureSize) -{ - FFX_MIN16_I2 result = iPxSample + iPxOffset; - result.x = (iPxOffset.x < 0) ? ffxMax(result.x, FFX_MIN16_I(0)) : result.x; - result.x = (iPxOffset.x > 0) ? ffxMin(result.x, iTextureSize.x - FFX_MIN16_I(1)) : result.x; - result.y = (iPxOffset.y < 0) ? ffxMax(result.y, FFX_MIN16_I(0)) : result.y; - result.y = (iPxOffset.y > 0) ? ffxMin(result.y, iTextureSize.y - FFX_MIN16_I(1)) : result.y; - return result; - - // return ffxMed3Half(iPxSample + iPxOffset, FFX_MIN16_I2(0, 0), iTextureSize - FFX_MIN16_I2(1, 1)); -} -#endif - -FfxFloat32x2 ClampUv(FfxFloat32x2 fUv, FfxInt32x2 iTextureSize, FfxInt32x2 iResourceSize) -{ - const FfxFloat32x2 fSampleLocation = fUv * iTextureSize; - const FfxFloat32x2 fClampedLocation = ffxMax(FfxFloat32x2(0.5f, 0.5f), ffxMin(fSampleLocation, FfxFloat32x2(iTextureSize) - FfxFloat32x2(0.5f, 0.5f))); - const FfxFloat32x2 fClampedUv = fClampedLocation / FfxFloat32x2(iResourceSize); - - return fClampedUv; -} - -FfxBoolean IsOnScreen(FfxInt32x2 pos, FfxInt32x2 size) -{ - return all(FFX_LESS_THAN(FfxUInt32x2(pos), FfxUInt32x2(size))); -} -#if FFX_HALF -FfxBoolean IsOnScreen(FFX_MIN16_I2 pos, FFX_MIN16_I2 size) -{ - return all(FFX_LESS_THAN(FFX_MIN16_U2(pos), FFX_MIN16_U2(size))); -} -#endif - -FfxFloat32 ComputeAutoExposureFromLavg(FfxFloat32 Lavg) -{ - Lavg = exp(Lavg); - - const FfxFloat32 S = 100.0f; //ISO arithmetic speed - const FfxFloat32 K = 12.5f; - FfxFloat32 ExposureISO100 = log2((Lavg * S) / K); - - const FfxFloat32 q = 0.65f; - FfxFloat32 Lmax = (78.0f / (q * S)) * ffxPow(2.0f, ExposureISO100); - - return 1 / Lmax; -} -#if FFX_HALF -FFX_MIN16_F ComputeAutoExposureFromLavg(FFX_MIN16_F Lavg) -{ - Lavg = exp(Lavg); - - const FFX_MIN16_F S = FFX_MIN16_F(100.0f); //ISO arithmetic speed - const FFX_MIN16_F K = FFX_MIN16_F(12.5f); - const FFX_MIN16_F ExposureISO100 = log2((Lavg * S) / K); - - const FFX_MIN16_F q = FFX_MIN16_F(0.65f); - const FFX_MIN16_F Lmax = (FFX_MIN16_F(78.0f) / (q * S)) * ffxPow(FFX_MIN16_F(2.0f), ExposureISO100); - - return FFX_MIN16_F(1) / Lmax; -} -#endif - -FfxInt32x2 ComputeHrPosFromLrPos(FfxInt32x2 iPxLrPos) -{ - FfxFloat32x2 fSrcJitteredPos = FfxFloat32x2(iPxLrPos) + 0.5f - Jitter(); - FfxFloat32x2 fLrPosInHr = (fSrcJitteredPos / RenderSize()) * DisplaySize(); - FfxInt32x2 iPxHrPos = FfxInt32x2(floor(fLrPosInHr)); - return iPxHrPos; -} -#if FFX_HALF -FFX_MIN16_I2 ComputeHrPosFromLrPos(FFX_MIN16_I2 iPxLrPos) -{ - FFX_MIN16_F2 fSrcJitteredPos = FFX_MIN16_F2(iPxLrPos) + FFX_MIN16_F(0.5f) - FFX_MIN16_F2(Jitter()); - FFX_MIN16_F2 fLrPosInHr = (fSrcJitteredPos / FFX_MIN16_F2(RenderSize())) * FFX_MIN16_F2(DisplaySize()); - FFX_MIN16_I2 iPxHrPos = FFX_MIN16_I2(floor(fLrPosInHr)); - return iPxHrPos; -} -#endif - -FfxFloat32x2 ComputeNdc(FfxFloat32x2 fPxPos, FfxInt32x2 iSize) -{ - return fPxPos / FfxFloat32x2(iSize) * FfxFloat32x2(2.0f, -2.0f) + FfxFloat32x2(-1.0f, 1.0f); -} - -FfxFloat32 GetViewSpaceDepth(FfxFloat32 fDeviceDepth) -{ - const FfxFloat32x4 fDeviceToViewDepth = DeviceToViewSpaceTransformFactors(); - - // fDeviceToViewDepth details found in ffx_fsr3upscaler.cpp - return (fDeviceToViewDepth[1] / (fDeviceDepth - fDeviceToViewDepth[0])); -} - -FfxFloat32 GetViewSpaceDepthInMeters(FfxFloat32 fDeviceDepth) -{ - return GetViewSpaceDepth(fDeviceDepth) * ViewSpaceToMetersFactor(); -} - -FfxFloat32x3 GetViewSpacePosition(FfxInt32x2 iViewportPos, FfxInt32x2 iViewportSize, FfxFloat32 fDeviceDepth) -{ - const FfxFloat32x4 fDeviceToViewDepth = DeviceToViewSpaceTransformFactors(); - - const FfxFloat32 Z = GetViewSpaceDepth(fDeviceDepth); - - const FfxFloat32x2 fNdcPos = ComputeNdc(iViewportPos, iViewportSize); - const FfxFloat32 X = fDeviceToViewDepth[2] * fNdcPos.x * Z; - const FfxFloat32 Y = fDeviceToViewDepth[3] * fNdcPos.y * Z; - - return FfxFloat32x3(X, Y, Z); -} - -FfxFloat32x3 GetViewSpacePositionInMeters(FfxInt32x2 iViewportPos, FfxInt32x2 iViewportSize, FfxFloat32 fDeviceDepth) -{ - return GetViewSpacePosition(iViewportPos, iViewportSize, fDeviceDepth) * ViewSpaceToMetersFactor(); -} - -FfxFloat32 GetMaxDistanceInMeters() -{ -#if FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - return GetViewSpaceDepth(0.0f) * ViewSpaceToMetersFactor(); -#else - return GetViewSpaceDepth(1.0f) * ViewSpaceToMetersFactor(); -#endif -} - -FfxFloat32x3 PrepareRgb(FfxFloat32x3 fRgb, FfxFloat32 fExposure, FfxFloat32 fPreExposure) -{ - fRgb /= fPreExposure; - fRgb *= fExposure; - - fRgb = clamp(fRgb, 0.0f, FSR3UPSCALER_FP16_MAX); - - return fRgb; -} - -FfxFloat32x3 UnprepareRgb(FfxFloat32x3 fRgb, FfxFloat32 fExposure) -{ - fRgb /= fExposure; - fRgb *= PreExposure(); - - return fRgb; -} - - -struct BilinearSamplingData -{ - FfxInt32x2 iOffsets[4]; - FfxFloat32 fWeights[4]; - FfxInt32x2 iBasePos; -}; - -BilinearSamplingData GetBilinearSamplingData(FfxFloat32x2 fUv, FfxInt32x2 iSize) -{ - BilinearSamplingData data; - - FfxFloat32x2 fPxSample = (fUv * iSize) - FfxFloat32x2(0.5f, 0.5f); - data.iBasePos = FfxInt32x2(floor(fPxSample)); - FfxFloat32x2 fPxFrac = ffxFract(fPxSample); - - data.iOffsets[0] = FfxInt32x2(0, 0); - data.iOffsets[1] = FfxInt32x2(1, 0); - data.iOffsets[2] = FfxInt32x2(0, 1); - data.iOffsets[3] = FfxInt32x2(1, 1); - - data.fWeights[0] = (1 - fPxFrac.x) * (1 - fPxFrac.y); - data.fWeights[1] = (fPxFrac.x) * (1 - fPxFrac.y); - data.fWeights[2] = (1 - fPxFrac.x) * (fPxFrac.y); - data.fWeights[3] = (fPxFrac.x) * (fPxFrac.y); - - return data; -} - -struct PlaneData -{ - FfxFloat32x3 fNormal; - FfxFloat32 fDistanceFromOrigin; -}; - -PlaneData GetPlaneFromPoints(FfxFloat32x3 fP0, FfxFloat32x3 fP1, FfxFloat32x3 fP2) -{ - PlaneData plane; - - FfxFloat32x3 v0 = fP0 - fP1; - FfxFloat32x3 v1 = fP0 - fP2; - plane.fNormal = normalize(cross(v0, v1)); - plane.fDistanceFromOrigin = -dot(fP0, plane.fNormal); - - return plane; -} - -FfxFloat32 PointToPlaneDistance(PlaneData plane, FfxFloat32x3 fPoint) -{ - return abs(dot(plane.fNormal, fPoint) + plane.fDistanceFromOrigin); -} - -#endif // #if defined(FFX_GPU) - -#endif //!defined(FFX_FSR3UPSCALER_COMMON_H) diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_common.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_common.h.meta deleted file mode 100644 index 08b2046..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_common.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 6a638bec681caac4fa8e2ca198726694 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_compute_luminance_pyramid.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_compute_luminance_pyramid.h deleted file mode 100644 index d26cf23..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_compute_luminance_pyramid.h +++ /dev/null @@ -1,176 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -FFX_GROUPSHARED FfxUInt32 spdCounter; - -void SpdIncreaseAtomicCounter(FfxUInt32 slice) -{ - SPD_IncreaseAtomicCounter(spdCounter); -} - -FfxUInt32 SpdGetAtomicCounter() -{ - return spdCounter; -} - -void SpdResetAtomicCounter(FfxUInt32 slice) -{ - SPD_ResetAtomicCounter(); -} - -#ifndef SPD_PACKED_ONLY -FFX_GROUPSHARED FfxFloat32 spdIntermediateR[16][16]; -FFX_GROUPSHARED FfxFloat32 spdIntermediateG[16][16]; -FFX_GROUPSHARED FfxFloat32 spdIntermediateB[16][16]; -FFX_GROUPSHARED FfxFloat32 spdIntermediateA[16][16]; - -FfxFloat32x4 SpdLoadSourceImage(FfxFloat32x2 tex, FfxUInt32 slice) -{ - FfxFloat32x2 fUv = (tex + 0.5f + Jitter()) / RenderSize(); - fUv = ClampUv(fUv, RenderSize(), InputColorResourceDimensions()); - FfxFloat32x3 fRgb = SampleInputColor(fUv); - - fRgb /= PreExposure(); - - //compute log luma - const FfxFloat32 fLogLuma = log(ffxMax(FSR3UPSCALER_EPSILON, RGBToLuma(fRgb))); - - // Make sure out of screen pixels contribute no value to the end result - const FfxFloat32 result = all(FFX_LESS_THAN(tex, RenderSize())) ? fLogLuma : 0.0f; - - return FfxFloat32x4(result, 0, 0, 0); -} - -FfxFloat32x4 SpdLoad(FfxInt32x2 tex, FfxUInt32 slice) -{ - return SPD_LoadMipmap5(tex); -} - -void SpdStore(FfxInt32x2 pix, FfxFloat32x4 outValue, FfxUInt32 index, FfxUInt32 slice) -{ - if (index == LumaMipLevelToUse() || index == 5) - { - SPD_SetMipmap(pix, index, outValue.r); - } - - if (index == MipCount() - 1) { //accumulate on 1x1 level - - if (all(FFX_EQUAL(pix, FfxInt32x2(0, 0)))) - { - FfxFloat32 prev = SPD_LoadExposureBuffer().y; - FfxFloat32 result = outValue.r; - - if (prev < resetAutoExposureAverageSmoothing) // Compare Lavg, so small or negative values - { - FfxFloat32 rate = 1.0f; - result = prev + (result - prev) * (1 - exp(-DeltaTime() * rate)); - } - FfxFloat32x2 spdOutput = FfxFloat32x2(ComputeAutoExposureFromLavg(result), result); - SPD_SetExposureBuffer(spdOutput); - } - } -} - -FfxFloat32x4 SpdLoadIntermediate(FfxUInt32 x, FfxUInt32 y) -{ - return FfxFloat32x4( - spdIntermediateR[x][y], - spdIntermediateG[x][y], - spdIntermediateB[x][y], - spdIntermediateA[x][y]); -} -void SpdStoreIntermediate(FfxUInt32 x, FfxUInt32 y, FfxFloat32x4 value) -{ - spdIntermediateR[x][y] = value.x; - spdIntermediateG[x][y] = value.y; - spdIntermediateB[x][y] = value.z; - spdIntermediateA[x][y] = value.w; -} -FfxFloat32x4 SpdReduce4(FfxFloat32x4 v0, FfxFloat32x4 v1, FfxFloat32x4 v2, FfxFloat32x4 v3) -{ - return (v0 + v1 + v2 + v3) * 0.25f; -} -#endif - -// define fetch and store functions Packed -#if FFX_HALF - -FFX_GROUPSHARED FfxFloat16x2 spdIntermediateRG[16][16]; -FFX_GROUPSHARED FfxFloat16x2 spdIntermediateBA[16][16]; - -FfxFloat16x4 SpdLoadSourceImageH(FfxFloat32x2 tex, FfxUInt32 slice) -{ - return FfxFloat16x4(0, 0, 0, 0); -} - -FfxFloat16x4 SpdLoadH(FfxInt32x2 p, FfxUInt32 slice) -{ - return FfxFloat16x4(0, 0, 0, 0); -} - -void SpdStoreH(FfxInt32x2 p, FfxFloat16x4 value, FfxUInt32 mip, FfxUInt32 slice) -{ -} - -FfxFloat16x4 SpdLoadIntermediateH(FfxUInt32 x, FfxUInt32 y) -{ - return FfxFloat16x4( - spdIntermediateRG[x][y].x, - spdIntermediateRG[x][y].y, - spdIntermediateBA[x][y].x, - spdIntermediateBA[x][y].y); -} - -void SpdStoreIntermediateH(FfxUInt32 x, FfxUInt32 y, FfxFloat16x4 value) -{ - spdIntermediateRG[x][y] = value.xy; - spdIntermediateBA[x][y] = value.zw; -} - -FfxFloat16x4 SpdReduce4H(FfxFloat16x4 v0, FfxFloat16x4 v1, FfxFloat16x4 v2, FfxFloat16x4 v3) -{ - return (v0 + v1 + v2 + v3) * FfxFloat16(0.25); -} -#endif - -#include "spd/ffx_spd.h" - -void ComputeAutoExposure(FfxUInt32x3 WorkGroupId, FfxUInt32 LocalThreadIndex) -{ -#if FFX_HALF - SpdDownsampleH( - FfxUInt32x2(WorkGroupId.xy), - FfxUInt32(LocalThreadIndex), - FfxUInt32(MipCount()), - FfxUInt32(NumWorkGroups()), - FfxUInt32(WorkGroupId.z), - FfxUInt32x2(WorkGroupOffset())); -#else - SpdDownsample( - FfxUInt32x2(WorkGroupId.xy), - FfxUInt32(LocalThreadIndex), - FfxUInt32(MipCount()), - FfxUInt32(NumWorkGroups()), - FfxUInt32(WorkGroupId.z), - FfxUInt32x2(WorkGroupOffset())); -#endif -} diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_compute_luminance_pyramid.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_compute_luminance_pyramid.h.meta deleted file mode 100644 index 9fb7653..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_compute_luminance_pyramid.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 61bd10363d44ee2478461c9e9efbcb67 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_depth_clip.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_depth_clip.h deleted file mode 100644 index 53763c8..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_depth_clip.h +++ /dev/null @@ -1,259 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#ifndef FFX_FSR3UPSCALER_DEPTH_CLIP_H -#define FFX_FSR3UPSCALER_DEPTH_CLIP_H - -FFX_STATIC const FfxFloat32 DepthClipBaseScale = 4.0f; - -FfxFloat32 ComputeDepthClip(FfxFloat32x2 fUvSample, FfxFloat32 fCurrentDepthSample) -{ - FfxFloat32 fCurrentDepthViewSpace = GetViewSpaceDepth(fCurrentDepthSample); - BilinearSamplingData bilinearInfo = GetBilinearSamplingData(fUvSample, RenderSize()); - - FfxFloat32 fDilatedSum = 0.0f; - FfxFloat32 fDepth = 0.0f; - FfxFloat32 fWeightSum = 0.0f; - for (FfxInt32 iSampleIndex = 0; iSampleIndex < 4; iSampleIndex++) { - - const FfxInt32x2 iOffset = bilinearInfo.iOffsets[iSampleIndex]; - const FfxInt32x2 iSamplePos = bilinearInfo.iBasePos + iOffset; - - if (IsOnScreen(iSamplePos, RenderSize())) { - const FfxFloat32 fWeight = bilinearInfo.fWeights[iSampleIndex]; - if (fWeight > fReconstructedDepthBilinearWeightThreshold) { - - const FfxFloat32 fPrevDepthSample = LoadReconstructedPrevDepth(iSamplePos); - const FfxFloat32 fPrevNearestDepthViewSpace = GetViewSpaceDepth(fPrevDepthSample); - - const FfxFloat32 fDepthDiff = fCurrentDepthViewSpace - fPrevNearestDepthViewSpace; - - if (fDepthDiff > 0.0f) { - -#if FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - const FfxFloat32 fPlaneDepth = ffxMin(fPrevDepthSample, fCurrentDepthSample); -#else - const FfxFloat32 fPlaneDepth = ffxMax(fPrevDepthSample, fCurrentDepthSample); -#endif - - const FfxFloat32x3 fCenter = GetViewSpacePosition(FfxInt32x2(RenderSize() * 0.5f), RenderSize(), fPlaneDepth); - const FfxFloat32x3 fCorner = GetViewSpacePosition(FfxInt32x2(0, 0), RenderSize(), fPlaneDepth); - - const FfxFloat32 fHalfViewportWidth = length(FfxFloat32x2(RenderSize())); - const FfxFloat32 fDepthThreshold = ffxMax(fCurrentDepthViewSpace, fPrevNearestDepthViewSpace); - - const FfxFloat32 Ksep = 1.37e-05f; - const FfxFloat32 Kfov = length(fCorner) / length(fCenter); - const FfxFloat32 fRequiredDepthSeparation = Ksep * Kfov * fHalfViewportWidth * fDepthThreshold; - - const FfxFloat32 fResolutionFactor = ffxSaturate(length(FfxFloat32x2(RenderSize())) / length(FfxFloat32x2(1920.0f, 1080.0f))); - const FfxFloat32 fPower = ffxLerp(1.0f, 3.0f, fResolutionFactor); - fDepth += ffxPow(ffxSaturate(FfxFloat32(fRequiredDepthSeparation / fDepthDiff)), fPower) * fWeight; - fWeightSum += fWeight; - } - } - } - } - - return (fWeightSum > 0) ? ffxSaturate(1.0f - fDepth / fWeightSum) : 0.0f; -} - -FfxFloat32 ComputeMotionDivergence(FfxInt32x2 iPxPos, FfxInt32x2 iPxInputMotionVectorSize) -{ - FfxFloat32 minconvergence = 1.0f; - - FfxFloat32x2 fMotionVectorNucleus = LoadInputMotionVector(iPxPos); - FfxFloat32 fNucleusVelocityLr = length(fMotionVectorNucleus * RenderSize()); - FfxFloat32 fMaxVelocityUv = length(fMotionVectorNucleus); - - const FfxFloat32 MotionVectorVelocityEpsilon = 1e-02f; - - if (fNucleusVelocityLr > MotionVectorVelocityEpsilon) { - for (FfxInt32 y = -1; y <= 1; ++y) { - for (FfxInt32 x = -1; x <= 1; ++x) { - - FfxInt32x2 sp = ClampLoad(iPxPos, FfxInt32x2(x, y), iPxInputMotionVectorSize); - - FfxFloat32x2 fMotionVector = LoadInputMotionVector(sp); - FfxFloat32 fVelocityUv = length(fMotionVector); - - fMaxVelocityUv = ffxMax(fVelocityUv, fMaxVelocityUv); - fVelocityUv = ffxMax(fVelocityUv, fMaxVelocityUv); - minconvergence = ffxMin(minconvergence, dot(fMotionVector / fVelocityUv, fMotionVectorNucleus / fVelocityUv)); - } - } - } - - return ffxSaturate(1.0f - minconvergence) * ffxSaturate(fMaxVelocityUv / 0.01f); -} - -FfxFloat32 ComputeDepthDivergence(FfxInt32x2 iPxPos) -{ - const FfxFloat32 fMaxDistInMeters = GetMaxDistanceInMeters(); - FfxFloat32 fDepthMax = 0.0f; - FfxFloat32 fDepthMin = fMaxDistInMeters; - - FfxInt32 iMaxDistFound = 0; - - for (FfxInt32 y = -1; y < 2; y++) { - for (FfxInt32 x = -1; x < 2; x++) { - - const FfxInt32x2 iOffset = FfxInt32x2(x, y); - const FfxInt32x2 iSamplePos = iPxPos + iOffset; - - const FfxFloat32 fOnScreenFactor = IsOnScreen(iSamplePos, RenderSize()) ? 1.0f : 0.0f; - FfxFloat32 fDepth = GetViewSpaceDepthInMeters(LoadDilatedDepth(iSamplePos)) * fOnScreenFactor; - - iMaxDistFound |= FfxInt32(fMaxDistInMeters == fDepth); - - fDepthMin = ffxMin(fDepthMin, fDepth); - fDepthMax = ffxMax(fDepthMax, fDepth); - } - } - - return (1.0f - fDepthMin / fDepthMax) * (FfxBoolean(iMaxDistFound) ? 0.0f : 1.0f); -} - -FfxFloat32 ComputeTemporalMotionDivergence(FfxInt32x2 iPxPos) -{ - const FfxFloat32x2 fUv = FfxFloat32x2(iPxPos + 0.5f) / RenderSize(); - - FfxFloat32x2 fMotionVector = LoadDilatedMotionVector(iPxPos); - FfxFloat32x2 fReprojectedUv = fUv + fMotionVector; - fReprojectedUv = ClampUv(fReprojectedUv, RenderSize(), MaxRenderSize()); - FfxFloat32x2 fPrevMotionVector = SamplePreviousDilatedMotionVector(fReprojectedUv); - - float fPxDistance = length(fMotionVector * DisplaySize()); - return fPxDistance > 1.0f ? ffxLerp(0.0f, 1.0f - ffxSaturate(length(fPrevMotionVector) / length(fMotionVector)), ffxSaturate(ffxPow(fPxDistance / 20.0f, 3.0f))) : 0; -} - -void PreProcessReactiveMasks(FfxInt32x2 iPxLrPos, FfxFloat32 fMotionDivergence) -{ - // Compensate for bilinear sampling in accumulation pass - - FfxFloat32x3 fReferenceColor = LoadInputColor(iPxLrPos).xyz; - FfxFloat32x2 fReactiveFactor = FfxFloat32x2(0.0f, fMotionDivergence); - - float fMasksSum = 0.0f; - - FfxFloat32x3 fColorSamples[9]; - FfxFloat32 fReactiveSamples[9]; - FfxFloat32 fTransparencyAndCompositionSamples[9]; - - FFX_UNROLL - for (FfxInt32 y = -1; y < 2; y++) { - FFX_UNROLL - for (FfxInt32 x = -1; x < 2; x++) { - - const FfxInt32x2 sampleCoord = ClampLoad(iPxLrPos, FfxInt32x2(x, y), FfxInt32x2(RenderSize())); - - FfxInt32 sampleIdx = (y + 1) * 3 + x + 1; - - FfxFloat32x3 fColorSample = LoadInputColor(sampleCoord).xyz; - FfxFloat32 fReactiveSample = LoadReactiveMask(sampleCoord); - FfxFloat32 fTransparencyAndCompositionSample = LoadTransparencyAndCompositionMask(sampleCoord); - - fColorSamples[sampleIdx] = fColorSample; - fReactiveSamples[sampleIdx] = fReactiveSample; - fTransparencyAndCompositionSamples[sampleIdx] = fTransparencyAndCompositionSample; - - fMasksSum += (fReactiveSample + fTransparencyAndCompositionSample); - } - } - - if (fMasksSum > 0) - { - for (FfxInt32 sampleIdx = 0; sampleIdx < 9; sampleIdx++) - { - FfxFloat32x3 fColorSample = fColorSamples[sampleIdx]; - FfxFloat32 fReactiveSample = fReactiveSamples[sampleIdx]; - FfxFloat32 fTransparencyAndCompositionSample = fTransparencyAndCompositionSamples[sampleIdx]; - - const FfxFloat32 fMaxLenSq = ffxMax(dot(fReferenceColor, fReferenceColor), dot(fColorSample, fColorSample)); - const FfxFloat32 fSimilarity = dot(fReferenceColor, fColorSample) / fMaxLenSq; - - // Increase power for non-similar samples - const FfxFloat32 fPowerBiasMax = 6.0f; - const FfxFloat32 fSimilarityPower = 1.0f + (fPowerBiasMax - fSimilarity * fPowerBiasMax); - const FfxFloat32 fWeightedReactiveSample = ffxPow(fReactiveSample, fSimilarityPower); - const FfxFloat32 fWeightedTransparencyAndCompositionSample = ffxPow(fTransparencyAndCompositionSample, fSimilarityPower); - - fReactiveFactor = ffxMax(fReactiveFactor, FfxFloat32x2(fWeightedReactiveSample, fWeightedTransparencyAndCompositionSample)); - } - } - - StoreDilatedReactiveMasks(iPxLrPos, fReactiveFactor); -} - -FfxFloat32x3 ComputePreparedInputColor(FfxInt32x2 iPxLrPos) -{ - //We assume linear data. if non-linear input (sRGB, ...), - //then we should convert to linear first and back to sRGB on output. - FfxFloat32x3 fRgb = ffxMax(FfxFloat32x3(0, 0, 0), LoadInputColor(iPxLrPos)); - - fRgb = PrepareRgb(fRgb, Exposure(), PreExposure()); - - const FfxFloat32x3 fPreparedYCoCg = RGBToYCoCg(fRgb); - - return fPreparedYCoCg; -} - -FfxFloat32 EvaluateSurface(FfxInt32x2 iPxPos, FfxFloat32x2 fMotionVector) -{ - FfxFloat32 d0 = GetViewSpaceDepth(LoadReconstructedPrevDepth(iPxPos + FfxInt32x2(0, -1))); - FfxFloat32 d1 = GetViewSpaceDepth(LoadReconstructedPrevDepth(iPxPos + FfxInt32x2(0, 0))); - FfxFloat32 d2 = GetViewSpaceDepth(LoadReconstructedPrevDepth(iPxPos + FfxInt32x2(0, 1))); - - return 1.0f - FfxFloat32(((d0 - d1) > (d1 * 0.01f)) && ((d1 - d2) > (d2 * 0.01f))); -} - -void DepthClip(FfxInt32x2 iPxPos) -{ - FfxFloat32x2 fDepthUv = (iPxPos + 0.5f) / RenderSize(); - FfxFloat32x2 fMotionVector = LoadDilatedMotionVector(iPxPos); - - // Discard tiny mvs - fMotionVector *= FfxFloat32(length(fMotionVector * DisplaySize()) > 0.01f); - - const FfxFloat32x2 fDilatedUv = fDepthUv + fMotionVector; - const FfxFloat32 fDilatedDepth = LoadDilatedDepth(iPxPos); - const FfxFloat32 fCurrentDepthViewSpace = GetViewSpaceDepth(LoadInputDepth(iPxPos)); - - // Compute prepared input color and depth clip - FfxFloat32 fDepthClip = ComputeDepthClip(fDilatedUv, fDilatedDepth) * EvaluateSurface(iPxPos, fMotionVector); - FfxFloat32x3 fPreparedYCoCg = ComputePreparedInputColor(iPxPos); - StorePreparedInputColor(iPxPos, FfxFloat32x4(fPreparedYCoCg, fDepthClip)); - - // Compute dilated reactive mask -#if FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS - FfxInt32x2 iSamplePos = iPxPos; -#else - FfxInt32x2 iSamplePos = ComputeHrPosFromLrPos(iPxPos); -#endif - - FfxFloat32 fMotionDivergence = ComputeMotionDivergence(iSamplePos, RenderSize()); - FfxFloat32 fTemporalMotionDifference = ffxSaturate(ComputeTemporalMotionDivergence(iPxPos) - ComputeDepthDivergence(iPxPos)); - - PreProcessReactiveMasks(iPxPos, ffxMax(fTemporalMotionDifference, fMotionDivergence)); -} - -#endif //!defined( FFX_FSR3UPSCALER_DEPTH_CLIPH ) diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_depth_clip.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_depth_clip.h.meta deleted file mode 100644 index 891d3d1..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_depth_clip.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 7c662249d70c4434da4f2da00e432c38 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_lock.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_lock.h deleted file mode 100644 index e1a0d06..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_lock.h +++ /dev/null @@ -1,116 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#ifndef FFX_FSR3UPSCALER_LOCK_H -#define FFX_FSR3UPSCALER_LOCK_H - -void ClearResourcesForNextFrame(in FfxInt32x2 iPxHrPos) -{ - if (all(FFX_LESS_THAN(iPxHrPos, FfxInt32x2(RenderSize())))) - { -#if FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - const FfxUInt32 farZ = 0x0; -#else - const FfxUInt32 farZ = 0x3f800000; -#endif - SetReconstructedDepth(iPxHrPos, farZ); - } -} - -FfxBoolean ComputeThinFeatureConfidence(FfxInt32x2 pos) -{ - const FfxInt32 RADIUS = 1; - - FfxFloat32 fNucleus = LoadLockInputLuma(pos); - - FfxFloat32 similar_threshold = 1.05f; - FfxFloat32 dissimilarLumaMin = FSR3UPSCALER_FLT_MAX; - FfxFloat32 dissimilarLumaMax = 0; - - /* - 0 1 2 - 3 4 5 - 6 7 8 - */ - - #define SETBIT(x) (1U << x) - - FfxUInt32 mask = SETBIT(4); //flag fNucleus as similar - - const FfxUInt32 uNumRejectionMasks = 4; - const FfxUInt32 uRejectionMasks[uNumRejectionMasks] = { - SETBIT(0) | SETBIT(1) | SETBIT(3) | SETBIT(4), //Upper left - SETBIT(1) | SETBIT(2) | SETBIT(4) | SETBIT(5), //Upper right - SETBIT(3) | SETBIT(4) | SETBIT(6) | SETBIT(7), //Lower left - SETBIT(4) | SETBIT(5) | SETBIT(7) | SETBIT(8), //Lower right - }; - - FfxInt32 idx = 0; - FFX_UNROLL - for (FfxInt32 y = -RADIUS; y <= RADIUS; y++) { - FFX_UNROLL - for (FfxInt32 x = -RADIUS; x <= RADIUS; x++, idx++) { - if (x == 0 && y == 0) continue; - - FfxInt32x2 samplePos = ClampLoad(pos, FfxInt32x2(x, y), FfxInt32x2(RenderSize())); - - FfxFloat32 sampleLuma = LoadLockInputLuma(samplePos); - FfxFloat32 difference = ffxMax(sampleLuma, fNucleus) / ffxMin(sampleLuma, fNucleus); - - if (difference > 0 && (difference < similar_threshold)) { - mask |= SETBIT(idx); - } else { - dissimilarLumaMin = ffxMin(dissimilarLumaMin, sampleLuma); - dissimilarLumaMax = ffxMax(dissimilarLumaMax, sampleLuma); - } - } - } - - FfxBoolean isRidge = fNucleus > dissimilarLumaMax || fNucleus < dissimilarLumaMin; - - if (FFX_FALSE == isRidge) { - - return false; - } - - FFX_UNROLL - for (FfxInt32 i = 0; i < 4; i++) { - - if ((mask & uRejectionMasks[i]) == uRejectionMasks[i]) { - return false; - } - } - - return true; -} - -void ComputeLock(FfxInt32x2 iPxLrPos) -{ - if (ComputeThinFeatureConfidence(iPxLrPos)) - { - StoreNewLocks(ComputeHrPosFromLrPos(iPxLrPos), 1.f); - } - - // ClearResourcesForNextFrame(iPxLrPos); -} - -#endif // FFX_FSR3UPSCALER_LOCK_H diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_lock.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_lock.h.meta deleted file mode 100644 index 4013169..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_lock.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: c7e9f53dd040b2645af5ccd936a94b0e -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_postprocess_lock_status.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_postprocess_lock_status.h deleted file mode 100644 index 3709113..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_postprocess_lock_status.h +++ /dev/null @@ -1,107 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#ifndef FFX_FSR3UPSCALER_POSTPROCESS_LOCK_STATUS_H -#define FFX_FSR3UPSCALER_POSTPROCESS_LOCK_STATUS_H - -FfxFloat32x4 WrapShadingChangeLuma(FfxInt32x2 iPxSample) -{ - return FfxFloat32x4(LoadMipLuma(iPxSample, LumaMipLevelToUse()), 0, 0, 0); -} - -#if FFX_HALF -FFX_MIN16_F4 WrapShadingChangeLuma(FFX_MIN16_I2 iPxSample) -{ - return FFX_MIN16_F4(LoadMipLuma(iPxSample, LumaMipLevelToUse()), 0, 0, 0); -} -#endif - -#if FFX_FSR3UPSCALER_OPTION_POSTPROCESSLOCKSTATUS_SAMPLERS_USE_DATA_HALF && FFX_HALF -DeclareCustomFetchBilinearSamplesMin16(FetchShadingChangeLumaSamples, WrapShadingChangeLuma) -#else -DeclareCustomFetchBicubicSamples(FetchShadingChangeLumaSamples, WrapShadingChangeLuma) -#endif -DeclareCustomTextureSample(ShadingChangeLumaSample, Lanczos2, FetchShadingChangeLumaSamples) - -FfxFloat32 GetShadingChangeLuma(FfxInt32x2 iPxHrPos, FfxFloat32x2 fUvCoord) -{ - FfxFloat32 fShadingChangeLuma = 0; - -#if 0 - fShadingChangeLuma = Exposure() * exp(ShadingChangeLumaSample(fUvCoord, LumaMipDimensions()).x); -#else - - const FfxFloat32 fDiv = FfxFloat32(2u << LumaMipLevelToUse()); - FfxInt32x2 iMipRenderSize = FfxInt32x2(RenderSize() / fDiv); - - fUvCoord = ClampUv(fUvCoord, iMipRenderSize, LumaMipDimensions()); - fShadingChangeLuma = Exposure() * exp(FfxFloat32(SampleMipLuma(fUvCoord, LumaMipLevelToUse()))); -#endif - - fShadingChangeLuma = ffxPow(fShadingChangeLuma, 1.0f / 6.0f); - - return fShadingChangeLuma; -} - -void UpdateLockStatus(AccumulationPassCommonParams params, - FFX_PARAMETER_INOUT FfxFloat32 fReactiveFactor, LockState state, - FFX_PARAMETER_INOUT FfxFloat32x2 fLockStatus, - FFX_PARAMETER_OUT FfxFloat32 fLockContributionThisFrame, - FFX_PARAMETER_OUT FfxFloat32 fLuminanceDiff) { - - const FfxFloat32 fShadingChangeLuma = GetShadingChangeLuma(params.iPxHrPos, params.fHrUv); - - //init temporal shading change factor, init to -1 or so in reproject to know if "true new"? - fLockStatus[LOCK_TEMPORAL_LUMA] = (fLockStatus[LOCK_TEMPORAL_LUMA] == FfxFloat32(0.0f)) ? fShadingChangeLuma : fLockStatus[LOCK_TEMPORAL_LUMA]; - - FfxFloat32 fPreviousShadingChangeLuma = fLockStatus[LOCK_TEMPORAL_LUMA]; - - fLuminanceDiff = 1.0f - MinDividedByMax(fPreviousShadingChangeLuma, fShadingChangeLuma); - - if (state.NewLock) { - fLockStatus[LOCK_TEMPORAL_LUMA] = fShadingChangeLuma; - - fLockStatus[LOCK_LIFETIME_REMAINING] = (fLockStatus[LOCK_LIFETIME_REMAINING] != 0.0f) ? 2.0f : 1.0f; - } - else if(fLockStatus[LOCK_LIFETIME_REMAINING] <= 1.0f) { - fLockStatus[LOCK_TEMPORAL_LUMA] = ffxLerp(fLockStatus[LOCK_TEMPORAL_LUMA], FfxFloat32(fShadingChangeLuma), 0.5f); - } - else { - if (fLuminanceDiff > 0.1f) { - KillLock(fLockStatus); - } - } - - fReactiveFactor = ffxMax(fReactiveFactor, ffxSaturate((fLuminanceDiff - 0.1f) * 10.0f)); - fLockStatus[LOCK_LIFETIME_REMAINING] *= (1.0f - fReactiveFactor); - - fLockStatus[LOCK_LIFETIME_REMAINING] *= ffxSaturate(1.0f - params.fAccumulationMask); - fLockStatus[LOCK_LIFETIME_REMAINING] *= FfxFloat32(params.fDepthClipFactor < 0.1f); - - // Compute this frame lock contribution - const FfxFloat32 fLifetimeContribution = ffxSaturate(fLockStatus[LOCK_LIFETIME_REMAINING] - 1.0f); - const FfxFloat32 fShadingChangeContribution = ffxSaturate(MinDividedByMax(fLockStatus[LOCK_TEMPORAL_LUMA], fShadingChangeLuma)); - - fLockContributionThisFrame = ffxSaturate(ffxSaturate(fLifetimeContribution * 4.0f) * fShadingChangeContribution); -} - -#endif //!defined( FFX_FSR3UPSCALER_POSTPROCESS_LOCK_STATUS_H ) diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_postprocess_lock_status.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_postprocess_lock_status.h.meta deleted file mode 100644 index 8c8bf49..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_postprocess_lock_status.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 67a8b72ceb93d634f883b086fdccb348 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_rcas.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_rcas.h deleted file mode 100644 index 77619a5..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_rcas.h +++ /dev/null @@ -1,67 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#define GROUP_SIZE 8 -#define FSR_RCAS_DENOISE 1 - -#include "ffx_core.h" - -void WriteUpscaledOutput(FFX_MIN16_U2 iPxHrPos, FfxFloat32x3 fUpscaledColor) -{ - StoreUpscaledOutput(FFX_MIN16_I2(iPxHrPos), fUpscaledColor); -} - -#define FSR_RCAS_F 1 -FfxFloat32x4 FsrRcasLoadF(FfxInt32x2 p) -{ - FfxFloat32x4 fColor = LoadRCAS_Input(p); - - fColor.rgb = PrepareRgb(fColor.rgb, Exposure(), PreExposure()); - - return fColor; -} -void FsrRcasInputF(inout FfxFloat32 r, inout FfxFloat32 g, inout FfxFloat32 b) {} - -#include "fsr1/ffx_fsr1.h" - -void CurrFilter(FFX_MIN16_U2 pos) -{ - FfxFloat32x3 c; - FsrRcasF(c.r, c.g, c.b, pos, RCASConfig()); - - c = UnprepareRgb(c, Exposure()); - - WriteUpscaledOutput(pos, c); -} - -void RCAS(FfxUInt32x3 LocalThreadId, FfxUInt32x3 WorkGroupId, FfxUInt32x3 Dtid) -{ - // Do remapping of local xy in workgroup for a more PS-like swizzle pattern. - FfxUInt32x2 gxy = ffxRemapForQuad(LocalThreadId.x) + FfxUInt32x2(WorkGroupId.x << 4u, WorkGroupId.y << 4u); - CurrFilter(FFX_MIN16_U2(gxy)); - gxy.x += 8u; - CurrFilter(FFX_MIN16_U2(gxy)); - gxy.y += 8u; - CurrFilter(FFX_MIN16_U2(gxy)); - gxy.x -= 8u; - CurrFilter(FFX_MIN16_U2(gxy)); -} diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_rcas.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_rcas.h.meta deleted file mode 100644 index a315002..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_rcas.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 692efb7cec0df67408a583a7ff34146a -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reconstruct_dilated_velocity_and_previous_depth.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reconstruct_dilated_velocity_and_previous_depth.h deleted file mode 100644 index a822dfc..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reconstruct_dilated_velocity_and_previous_depth.h +++ /dev/null @@ -1,146 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#ifndef FFX_FSR3UPSCALER_RECONSTRUCT_DILATED_VELOCITY_AND_PREVIOUS_DEPTH_H -#define FFX_FSR3UPSCALER_RECONSTRUCT_DILATED_VELOCITY_AND_PREVIOUS_DEPTH_H - -void ReconstructPrevDepth(FfxInt32x2 iPxPos, FfxFloat32 fDepth, FfxFloat32x2 fMotionVector, FfxInt32x2 iPxDepthSize) -{ - fMotionVector *= FfxFloat32(length(fMotionVector * DisplaySize()) > 0.1f); - - FfxFloat32x2 fUv = (iPxPos + FfxFloat32(0.5)) / iPxDepthSize; - FfxFloat32x2 fReprojectedUv = fUv + fMotionVector; - - BilinearSamplingData bilinearInfo = GetBilinearSamplingData(fReprojectedUv, RenderSize()); - - // Project current depth into previous frame locations. - // Push to all pixels having some contribution if reprojection is using bilinear logic. - for (FfxInt32 iSampleIndex = 0; iSampleIndex < 4; iSampleIndex++) { - - const FfxInt32x2 iOffset = bilinearInfo.iOffsets[iSampleIndex]; - FfxFloat32 fWeight = bilinearInfo.fWeights[iSampleIndex]; - - if (fWeight > fReconstructedDepthBilinearWeightThreshold) { - - FfxInt32x2 iStorePos = bilinearInfo.iBasePos + iOffset; - if (IsOnScreen(iStorePos, iPxDepthSize)) { - StoreReconstructedDepth(iStorePos, fDepth); - } - } - } -} - -void FindNearestDepth(FFX_PARAMETER_IN FfxInt32x2 iPxPos, FFX_PARAMETER_IN FfxInt32x2 iPxSize, FFX_PARAMETER_OUT FfxFloat32 fNearestDepth, FFX_PARAMETER_OUT FfxInt32x2 fNearestDepthCoord) -{ - const FfxInt32 iSampleCount = 9; - const FfxInt32x2 iSampleOffsets[iSampleCount] = { - FfxInt32x2(+0, +0), - FfxInt32x2(+1, +0), - FfxInt32x2(+0, +1), - FfxInt32x2(+0, -1), - FfxInt32x2(-1, +0), - FfxInt32x2(-1, +1), - FfxInt32x2(+1, +1), - FfxInt32x2(-1, -1), - FfxInt32x2(+1, -1), - }; - - // pull out the depth loads to allow SC to batch them - FfxFloat32 depth[9]; - FfxInt32 iSampleIndex = 0; - FFX_UNROLL - for (iSampleIndex = 0; iSampleIndex < iSampleCount; ++iSampleIndex) { - - FfxInt32x2 iPos = iPxPos + iSampleOffsets[iSampleIndex]; - depth[iSampleIndex] = LoadInputDepth(iPos); - } - - // find closest depth - fNearestDepthCoord = iPxPos; - fNearestDepth = depth[0]; - FFX_UNROLL - for (iSampleIndex = 1; iSampleIndex < iSampleCount; ++iSampleIndex) { - - FfxInt32x2 iPos = iPxPos + iSampleOffsets[iSampleIndex]; - if (IsOnScreen(iPos, iPxSize)) { - - FfxFloat32 fNdDepth = depth[iSampleIndex]; -#if FFX_FSR3UPSCALER_OPTION_INVERTED_DEPTH - if (fNdDepth > fNearestDepth) { -#else - if (fNdDepth < fNearestDepth) { -#endif - fNearestDepthCoord = iPos; - fNearestDepth = fNdDepth; - } - } - } -} - -FfxFloat32 ComputeLockInputLuma(FfxInt32x2 iPxLrPos) -{ - //We assume linear data. if non-linear input (sRGB, ...), - //then we should convert to linear first and back to sRGB on output. - FfxFloat32x3 fRgb = ffxMax(FfxFloat32x3(0, 0, 0), LoadInputColor(iPxLrPos)); - - // Use internal auto exposure for locking logic - fRgb /= PreExposure(); - fRgb *= Exposure(); - -#if FFX_FSR3UPSCALER_OPTION_HDR_COLOR_INPUT - fRgb = Tonemap(fRgb); -#endif - - //compute luma used to lock pixels, if used elsewhere the ffxPow must be moved! - const FfxFloat32 fLockInputLuma = ffxPow(RGBToPerceivedLuma(fRgb), FfxFloat32(1.0 / 6.0)); - - return fLockInputLuma; -} - -void ReconstructAndDilate(FfxInt32x2 iPxLrPos) -{ - FfxFloat32 fDilatedDepth; - FfxInt32x2 iNearestDepthCoord; - - FindNearestDepth(iPxLrPos, RenderSize(), fDilatedDepth, iNearestDepthCoord); - -#if FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS - FfxInt32x2 iSamplePos = iPxLrPos; - FfxInt32x2 iMotionVectorPos = iNearestDepthCoord; -#else - FfxInt32x2 iSamplePos = ComputeHrPosFromLrPos(iPxLrPos); - FfxInt32x2 iMotionVectorPos = ComputeHrPosFromLrPos(iNearestDepthCoord); -#endif - - FfxFloat32x2 fDilatedMotionVector = LoadInputMotionVector(iMotionVectorPos); - - StoreDilatedDepth(iPxLrPos, fDilatedDepth); - StoreDilatedMotionVector(iPxLrPos, fDilatedMotionVector); - - ReconstructPrevDepth(iPxLrPos, fDilatedDepth, fDilatedMotionVector, RenderSize()); - - FfxFloat32 fLockInputLuma = ComputeLockInputLuma(iPxLrPos); - StoreLockInputLuma(iPxLrPos, fLockInputLuma); -} - - -#endif //!defined( FFX_FSR3UPSCALER_RECONSTRUCT_DILATED_VELOCITY_AND_PREVIOUS_DEPTH_H ) diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reconstruct_dilated_velocity_and_previous_depth.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reconstruct_dilated_velocity_and_previous_depth.h.meta deleted file mode 100644 index a1fd018..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reconstruct_dilated_velocity_and_previous_depth.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: c8b3854bad30a8b40babc5a9805f294e -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reproject.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reproject.h deleted file mode 100644 index 29b7584..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reproject.h +++ /dev/null @@ -1,137 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#ifndef FFX_FSR3UPSCALER_REPROJECT_H -#define FFX_FSR3UPSCALER_REPROJECT_H - -#ifndef FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE -#define FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE 0 // Reference -#endif - -FfxFloat32x4 WrapHistory(FfxInt32x2 iPxSample) -{ - return LoadHistory(iPxSample); -} - -#if FFX_HALF -FFX_MIN16_F4 WrapHistory(FFX_MIN16_I2 iPxSample) -{ - return FFX_MIN16_F4(LoadHistory(iPxSample)); -} -#endif - - -#if FFX_FSR3UPSCALER_OPTION_REPROJECT_SAMPLERS_USE_DATA_HALF && FFX_HALF -DeclareCustomFetchBicubicSamplesMin16(FetchHistorySamples, WrapHistory) -DeclareCustomTextureSampleMin16(HistorySample, FFX_FSR3UPSCALER_GET_LANCZOS_SAMPLER1D(FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE), FetchHistorySamples) -#else -DeclareCustomFetchBicubicSamples(FetchHistorySamples, WrapHistory) -DeclareCustomTextureSample(HistorySample, FFX_FSR3UPSCALER_GET_LANCZOS_SAMPLER1D(FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE), FetchHistorySamples) -#endif - -FfxFloat32x4 WrapLockStatus(FfxInt32x2 iPxSample) -{ - FfxFloat32x4 fSample = FfxFloat32x4(LoadLockStatus(iPxSample), 0.0f, 0.0f); - return fSample; -} - -#if FFX_HALF -FFX_MIN16_F4 WrapLockStatus(FFX_MIN16_I2 iPxSample) -{ - FFX_MIN16_F4 fSample = FFX_MIN16_F4(LoadLockStatus(iPxSample), 0.0, 0.0); - - return fSample; -} -#endif - -#if 1 -#if FFX_FSR3UPSCALER_OPTION_REPROJECT_SAMPLERS_USE_DATA_HALF && FFX_HALF -DeclareCustomFetchBilinearSamplesMin16(FetchLockStatusSamples, WrapLockStatus) -DeclareCustomTextureSampleMin16(LockStatusSample, Bilinear, FetchLockStatusSamples) -#else -DeclareCustomFetchBilinearSamples(FetchLockStatusSamples, WrapLockStatus) -DeclareCustomTextureSample(LockStatusSample, Bilinear, FetchLockStatusSamples) -#endif -#else -#if FFX_FSR3UPSCALER_OPTION_REPROJECT_SAMPLERS_USE_DATA_HALF && FFX_HALF -DeclareCustomFetchBicubicSamplesMin16(FetchLockStatusSamples, WrapLockStatus) -DeclareCustomTextureSampleMin16(LockStatusSample, FFX_FSR3UPSCALER_GET_LANCZOS_SAMPLER1D(FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE), FetchLockStatusSamples) -#else -DeclareCustomFetchBicubicSamples(FetchLockStatusSamples, WrapLockStatus) -DeclareCustomTextureSample(LockStatusSample, FFX_FSR3UPSCALER_GET_LANCZOS_SAMPLER1D(FFX_FSR3UPSCALER_OPTION_REPROJECT_USE_LANCZOS_TYPE), FetchLockStatusSamples) -#endif -#endif - -FfxFloat32x2 GetMotionVector(FfxInt32x2 iPxHrPos, FfxFloat32x2 fHrUv) -{ -#if FFX_FSR3UPSCALER_OPTION_LOW_RESOLUTION_MOTION_VECTORS - FfxFloat32x2 fDilatedMotionVector = LoadDilatedMotionVector(FFX_MIN16_I2(fHrUv * RenderSize())); -#else - FfxFloat32x2 fDilatedMotionVector = LoadInputMotionVector(iPxHrPos); -#endif - - return fDilatedMotionVector; -} - -FfxBoolean IsUvInside(FfxFloat32x2 fUv) -{ - return (fUv.x >= 0.0f && fUv.x <= 1.0f) && (fUv.y >= 0.0f && fUv.y <= 1.0f); -} - -void ComputeReprojectedUVs(const AccumulationPassCommonParams params, FFX_PARAMETER_OUT FfxFloat32x2 fReprojectedHrUv, FFX_PARAMETER_OUT FfxBoolean bIsExistingSample) -{ - fReprojectedHrUv = params.fHrUv + params.fMotionVector; - - bIsExistingSample = IsUvInside(fReprojectedHrUv); -} - -void ReprojectHistoryColor(const AccumulationPassCommonParams params, FFX_PARAMETER_OUT FfxFloat32x3 fHistoryColor, FFX_PARAMETER_OUT FfxFloat32 fTemporalReactiveFactor, FFX_PARAMETER_OUT FfxBoolean bInMotionLastFrame) -{ - FfxFloat32x4 fHistory = HistorySample(params.fReprojectedHrUv, DisplaySize()); - - fHistoryColor = PrepareRgb(fHistory.rgb, Exposure(), PreviousFramePreExposure()); - - fHistoryColor = RGBToYCoCg(fHistoryColor); - - //Compute temporal reactivity info - fTemporalReactiveFactor = ffxSaturate(abs(fHistory.w)); - bInMotionLastFrame = (fHistory.w < 0.0f); -} - -LockState ReprojectHistoryLockStatus(const AccumulationPassCommonParams params, FFX_PARAMETER_OUT FfxFloat32x2 fReprojectedLockStatus) -{ - LockState state = { FFX_FALSE, FFX_FALSE }; - const FfxFloat32 fNewLockIntensity = LoadRwNewLocks(params.iPxHrPos); - state.NewLock = fNewLockIntensity > (127.0f / 255.0f); - - FfxFloat32 fInPlaceLockLifetime = state.NewLock ? fNewLockIntensity : 0; - - fReprojectedLockStatus = SampleLockStatus(params.fReprojectedHrUv); - - if (fReprojectedLockStatus[LOCK_LIFETIME_REMAINING] != FfxFloat32(0.0f)) { - state.WasLockedPrevFrame = true; - } - - return state; -} - -#endif //!defined( FFX_FSR3UPSCALER_REPROJECT_H ) diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reproject.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reproject.h.meta deleted file mode 100644 index 3fa39db..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_reproject.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 9d893016eebb2564f9a66b80afb0849f -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_resources.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_resources.h deleted file mode 100644 index d98cfcc..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_resources.h +++ /dev/null @@ -1,104 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#ifndef FFX_FSR3UPSCALER_RESOURCES_H -#define FFX_FSR3UPSCALER_RESOURCES_H - -#if defined(FFX_CPU) || defined(FFX_GPU) -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_NULL 0 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INPUT_OPAQUE_ONLY 1 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INPUT_COLOR 2 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INPUT_MOTION_VECTORS 3 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INPUT_DEPTH 4 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INPUT_EXPOSURE 5 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INPUT_REACTIVE_MASK 6 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INPUT_TRANSPARENCY_AND_COMPOSITION_MASK 7 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_RECONSTRUCTED_PREVIOUS_NEAREST_DEPTH 8 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_DILATED_MOTION_VECTORS 9 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_DILATED_DEPTH 10 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INTERNAL_UPSCALED_COLOR 11 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LOCK_STATUS 12 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_NEW_LOCKS 13 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_PREPARED_INPUT_COLOR 14 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LUMA_HISTORY 15 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_DEBUG_OUTPUT 16 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LANCZOS_LUT 17 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SPD_ATOMIC_COUNT 18 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_UPSCALED_OUTPUT 19 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_RCAS_INPUT 20 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LOCK_STATUS_1 21 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LOCK_STATUS_2 22 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INTERNAL_UPSCALED_COLOR_1 23 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INTERNAL_UPSCALED_COLOR_2 24 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INTERNAL_DEFAULT_REACTIVITY 25 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INTERNAL_DEFAULT_TRANSPARENCY_AND_COMPOSITION 26 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTITIER_UPSAMPLE_MAXIMUM_BIAS_LUT 27 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_DILATED_REACTIVE_MASKS 28 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE 29 // same as FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_0 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_0 29 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_1 30 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_2 31 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_3 32 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_4 33 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_5 34 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_6 35 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_7 36 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_8 37 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_9 38 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_10 39 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_11 40 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_12 41 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_INTERNAL_DEFAULT_EXPOSURE 42 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_AUTO_EXPOSURE 43 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_AUTOREACTIVE 44 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_AUTOCOMPOSITION_DEPRECATED 45 - -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_PREV_PRE_ALPHA_COLOR 46 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_PREV_POST_ALPHA_COLOR 47 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_PREV_PRE_ALPHA_COLOR_1 48 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_PREV_POST_ALPHA_COLOR_1 49 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_PREV_PRE_ALPHA_COLOR_2 50 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_PREV_POST_ALPHA_COLOR_2 51 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_PREVIOUS_DILATED_MOTION_VECTORS 52 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LUMA_HISTORY_1 53 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LUMA_HISTORY_2 54 -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_LOCK_INPUT_LUMA 55 - -// Shading change detection mip level setting, value must be in the range [FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_0, FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_12] -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_SHADING_CHANGE FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_4 -#define FFX_FSR3UPSCALER_SHADING_CHANGE_MIP_LEVEL (FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE_MIPMAP_SHADING_CHANGE - FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_SCENE_LUMINANCE) - -#define FFX_FSR3UPSCALER_RESOURCE_IDENTIFIER_COUNT 56 - -#define FFX_FSR3UPSCALER_CONSTANTBUFFER_IDENTIFIER_FSR3UPSCALER 0 -#define FFX_FSR3UPSCALER_CONSTANTBUFFER_IDENTIFIER_SPD 1 -#define FFX_FSR3UPSCALER_CONSTANTBUFFER_IDENTIFIER_RCAS 2 -#define FFX_FSR3UPSCALER_CONSTANTBUFFER_IDENTIFIER_GENREACTIVE 3 - -#define FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_APPLY_TONEMAP 1 -#define FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_APPLY_INVERSETONEMAP 2 -#define FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_APPLY_THRESHOLD 4 -#define FFX_FSR3UPSCALER_AUTOREACTIVEFLAGS_USE_COMPONENTS_MAX 8 - -#endif // #if defined(FFX_CPU) || defined(FFX_GPU) - -#endif //!defined( FFX_FSR3UPSCALER_RESOURCES_H ) diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_resources.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_resources.h.meta deleted file mode 100644 index 8e862df..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_resources.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: b5a95a38dcfaf3946a5095bbbc42939a -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_sample.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_sample.h deleted file mode 100644 index d33f70c..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_sample.h +++ /dev/null @@ -1,606 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#ifndef FFX_FSR3UPSCALER_SAMPLE_H -#define FFX_FSR3UPSCALER_SAMPLE_H - -// suppress warnings -#ifdef FFX_HLSL -#pragma warning(disable: 4008) // potentially divide by zero -#endif //FFX_HLSL - -struct FetchedBilinearSamples { - - FfxFloat32x4 fColor00; - FfxFloat32x4 fColor10; - - FfxFloat32x4 fColor01; - FfxFloat32x4 fColor11; -}; - -struct FetchedBicubicSamples { - - FfxFloat32x4 fColor00; - FfxFloat32x4 fColor10; - FfxFloat32x4 fColor20; - FfxFloat32x4 fColor30; - - FfxFloat32x4 fColor01; - FfxFloat32x4 fColor11; - FfxFloat32x4 fColor21; - FfxFloat32x4 fColor31; - - FfxFloat32x4 fColor02; - FfxFloat32x4 fColor12; - FfxFloat32x4 fColor22; - FfxFloat32x4 fColor32; - - FfxFloat32x4 fColor03; - FfxFloat32x4 fColor13; - FfxFloat32x4 fColor23; - FfxFloat32x4 fColor33; -}; - -#if FFX_HALF -struct FetchedBilinearSamplesMin16 { - - FFX_MIN16_F4 fColor00; - FFX_MIN16_F4 fColor10; - - FFX_MIN16_F4 fColor01; - FFX_MIN16_F4 fColor11; -}; - -struct FetchedBicubicSamplesMin16 { - - FFX_MIN16_F4 fColor00; - FFX_MIN16_F4 fColor10; - FFX_MIN16_F4 fColor20; - FFX_MIN16_F4 fColor30; - - FFX_MIN16_F4 fColor01; - FFX_MIN16_F4 fColor11; - FFX_MIN16_F4 fColor21; - FFX_MIN16_F4 fColor31; - - FFX_MIN16_F4 fColor02; - FFX_MIN16_F4 fColor12; - FFX_MIN16_F4 fColor22; - FFX_MIN16_F4 fColor32; - - FFX_MIN16_F4 fColor03; - FFX_MIN16_F4 fColor13; - FFX_MIN16_F4 fColor23; - FFX_MIN16_F4 fColor33; -}; -#else //FFX_HALF -#define FetchedBicubicSamplesMin16 FetchedBicubicSamples -#define FetchedBilinearSamplesMin16 FetchedBilinearSamples -#endif //FFX_HALF - -FfxFloat32x4 Linear(FfxFloat32x4 A, FfxFloat32x4 B, FfxFloat32 t) -{ - return A + (B - A) * t; -} - -FfxFloat32x4 Bilinear(FetchedBilinearSamples BilinearSamples, FfxFloat32x2 fPxFrac) -{ - FfxFloat32x4 fColorX0 = Linear(BilinearSamples.fColor00, BilinearSamples.fColor10, fPxFrac.x); - FfxFloat32x4 fColorX1 = Linear(BilinearSamples.fColor01, BilinearSamples.fColor11, fPxFrac.x); - FfxFloat32x4 fColorXY = Linear(fColorX0, fColorX1, fPxFrac.y); - return fColorXY; -} - -#if FFX_HALF -FFX_MIN16_F4 Linear(FFX_MIN16_F4 A, FFX_MIN16_F4 B, FFX_MIN16_F t) -{ - return A + (B - A) * t; -} - -FFX_MIN16_F4 Bilinear(FetchedBilinearSamplesMin16 BilinearSamples, FFX_MIN16_F2 fPxFrac) -{ - FFX_MIN16_F4 fColorX0 = Linear(BilinearSamples.fColor00, BilinearSamples.fColor10, fPxFrac.x); - FFX_MIN16_F4 fColorX1 = Linear(BilinearSamples.fColor01, BilinearSamples.fColor11, fPxFrac.x); - FFX_MIN16_F4 fColorXY = Linear(fColorX0, fColorX1, fPxFrac.y); - return fColorXY; -} -#endif - -FfxFloat32 Lanczos2NoClamp(FfxFloat32 x) -{ - const FfxFloat32 PI = 3.141592653589793f; // TODO: share SDK constants - return abs(x) < FSR3UPSCALER_EPSILON ? 1.f : (sin(PI * x) / (PI * x)) * (sin(0.5f * PI * x) / (0.5f * PI * x)); -} - -FfxFloat32 Lanczos2(FfxFloat32 x) -{ - x = ffxMin(abs(x), 2.0f); - return Lanczos2NoClamp(x); -} - -#if FFX_HALF - -#if 0 -FFX_MIN16_F Lanczos2NoClamp(FFX_MIN16_F x) -{ - const FFX_MIN16_F PI = FFX_MIN16_F(3.141592653589793f); // TODO: share SDK constants - return abs(x) < FFX_MIN16_F(FSR3UPSCALER_EPSILON) ? FFX_MIN16_F(1.f) : (sin(PI * x) / (PI * x)) * (sin(FFX_MIN16_F(0.5f) * PI * x) / (FFX_MIN16_F(0.5f) * PI * x)); -} -#endif - -FFX_MIN16_F Lanczos2(FFX_MIN16_F x) -{ - x = ffxMin(abs(x), FFX_MIN16_F(2.0f)); - return FFX_MIN16_F(Lanczos2NoClamp(x)); -} -#endif //FFX_HALF - -// FSR1 lanczos approximation. Input is x*x and must be <= 4. -FfxFloat32 Lanczos2ApproxSqNoClamp(FfxFloat32 x2) -{ - FfxFloat32 a = (2.0f / 5.0f) * x2 - 1; - FfxFloat32 b = (1.0f / 4.0f) * x2 - 1; - return ((25.0f / 16.0f) * a * a - (25.0f / 16.0f - 1)) * (b * b); -} - -#if FFX_HALF -FFX_MIN16_F Lanczos2ApproxSqNoClamp(FFX_MIN16_F x2) -{ - FFX_MIN16_F a = FFX_MIN16_F(2.0f / 5.0f) * x2 - FFX_MIN16_F(1); - FFX_MIN16_F b = FFX_MIN16_F(1.0f / 4.0f) * x2 - FFX_MIN16_F(1); - return (FFX_MIN16_F(25.0f / 16.0f) * a * a - FFX_MIN16_F(25.0f / 16.0f - 1)) * (b * b); -} -#endif //FFX_HALF - -FfxFloat32 Lanczos2ApproxSq(FfxFloat32 x2) -{ - x2 = ffxMin(x2, 4.0f); - return Lanczos2ApproxSqNoClamp(x2); -} - -#if FFX_HALF -FFX_MIN16_F Lanczos2ApproxSq(FFX_MIN16_F x2) -{ - x2 = ffxMin(x2, FFX_MIN16_F(4.0f)); - return Lanczos2ApproxSqNoClamp(x2); -} -#endif //FFX_HALF - -FfxFloat32 Lanczos2ApproxNoClamp(FfxFloat32 x) -{ - return Lanczos2ApproxSqNoClamp(x * x); -} - -#if FFX_HALF -FFX_MIN16_F Lanczos2ApproxNoClamp(FFX_MIN16_F x) -{ - return Lanczos2ApproxSqNoClamp(x * x); -} -#endif //FFX_HALF - -FfxFloat32 Lanczos2Approx(FfxFloat32 x) -{ - return Lanczos2ApproxSq(x * x); -} - -#if FFX_HALF -FFX_MIN16_F Lanczos2Approx(FFX_MIN16_F x) -{ - return Lanczos2ApproxSq(x * x); -} -#endif //FFX_HALF - -FfxFloat32 Lanczos2_UseLUT(FfxFloat32 x) -{ - return SampleLanczos2Weight(abs(x)); -} - -#if FFX_HALF -FFX_MIN16_F Lanczos2_UseLUT(FFX_MIN16_F x) -{ - return FFX_MIN16_F(SampleLanczos2Weight(abs(x))); -} -#endif //FFX_HALF - -FfxFloat32x4 Lanczos2_UseLUT(FfxFloat32x4 fColor0, FfxFloat32x4 fColor1, FfxFloat32x4 fColor2, FfxFloat32x4 fColor3, FfxFloat32 t) -{ - FfxFloat32 fWeight0 = Lanczos2_UseLUT(-1.f - t); - FfxFloat32 fWeight1 = Lanczos2_UseLUT(-0.f - t); - FfxFloat32 fWeight2 = Lanczos2_UseLUT(+1.f - t); - FfxFloat32 fWeight3 = Lanczos2_UseLUT(+2.f - t); - return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); -} -#if FFX_HALF -FFX_MIN16_F4 Lanczos2_UseLUT(FFX_MIN16_F4 fColor0, FFX_MIN16_F4 fColor1, FFX_MIN16_F4 fColor2, FFX_MIN16_F4 fColor3, FFX_MIN16_F t) -{ - FFX_MIN16_F fWeight0 = Lanczos2_UseLUT(FFX_MIN16_F(-1.f) - t); - FFX_MIN16_F fWeight1 = Lanczos2_UseLUT(FFX_MIN16_F(-0.f) - t); - FFX_MIN16_F fWeight2 = Lanczos2_UseLUT(FFX_MIN16_F(+1.f) - t); - FFX_MIN16_F fWeight3 = Lanczos2_UseLUT(FFX_MIN16_F(+2.f) - t); - return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); -} -#endif - -FfxFloat32x4 Lanczos2(FfxFloat32x4 fColor0, FfxFloat32x4 fColor1, FfxFloat32x4 fColor2, FfxFloat32x4 fColor3, FfxFloat32 t) -{ - FfxFloat32 fWeight0 = Lanczos2(-1.f - t); - FfxFloat32 fWeight1 = Lanczos2(-0.f - t); - FfxFloat32 fWeight2 = Lanczos2(+1.f - t); - FfxFloat32 fWeight3 = Lanczos2(+2.f - t); - return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); -} - -FfxFloat32x4 Lanczos2(FetchedBicubicSamples Samples, FfxFloat32x2 fPxFrac) -{ - FfxFloat32x4 fColorX0 = Lanczos2(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); - FfxFloat32x4 fColorX1 = Lanczos2(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); - FfxFloat32x4 fColorX2 = Lanczos2(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); - FfxFloat32x4 fColorX3 = Lanczos2(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); - FfxFloat32x4 fColorXY = Lanczos2(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); - - // Deringing - - // TODO: only use 4 by checking jitter - const FfxInt32 iDeringingSampleCount = 4; - const FfxFloat32x4 fDeringingSamples[4] = { - Samples.fColor11, - Samples.fColor21, - Samples.fColor12, - Samples.fColor22, - }; - - FfxFloat32x4 fDeringingMin = fDeringingSamples[0]; - FfxFloat32x4 fDeringingMax = fDeringingSamples[0]; - - FFX_UNROLL - for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) { - - fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); - fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); - } - - fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); - - return fColorXY; -} - -#if FFX_HALF -FFX_MIN16_F4 Lanczos2(FFX_MIN16_F4 fColor0, FFX_MIN16_F4 fColor1, FFX_MIN16_F4 fColor2, FFX_MIN16_F4 fColor3, FFX_MIN16_F t) -{ - FFX_MIN16_F fWeight0 = Lanczos2(FFX_MIN16_F(-1.f) - t); - FFX_MIN16_F fWeight1 = Lanczos2(FFX_MIN16_F(-0.f) - t); - FFX_MIN16_F fWeight2 = Lanczos2(FFX_MIN16_F(+1.f) - t); - FFX_MIN16_F fWeight3 = Lanczos2(FFX_MIN16_F(+2.f) - t); - return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); -} - -FFX_MIN16_F4 Lanczos2(FetchedBicubicSamplesMin16 Samples, FFX_MIN16_F2 fPxFrac) -{ - FFX_MIN16_F4 fColorX0 = Lanczos2(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); - FFX_MIN16_F4 fColorX1 = Lanczos2(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); - FFX_MIN16_F4 fColorX2 = Lanczos2(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); - FFX_MIN16_F4 fColorX3 = Lanczos2(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); - FFX_MIN16_F4 fColorXY = Lanczos2(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); - - // Deringing - - // TODO: only use 4 by checking jitter - const FfxInt32 iDeringingSampleCount = 4; - const FFX_MIN16_F4 fDeringingSamples[4] = { - Samples.fColor11, - Samples.fColor21, - Samples.fColor12, - Samples.fColor22, - }; - - FFX_MIN16_F4 fDeringingMin = fDeringingSamples[0]; - FFX_MIN16_F4 fDeringingMax = fDeringingSamples[0]; - - FFX_UNROLL - for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) - { - fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); - fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); - } - - fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); - - return fColorXY; -} -#endif //FFX_HALF - - -FfxFloat32x4 Lanczos2LUT(FetchedBicubicSamples Samples, FfxFloat32x2 fPxFrac) -{ - FfxFloat32x4 fColorX0 = Lanczos2_UseLUT(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); - FfxFloat32x4 fColorX1 = Lanczos2_UseLUT(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); - FfxFloat32x4 fColorX2 = Lanczos2_UseLUT(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); - FfxFloat32x4 fColorX3 = Lanczos2_UseLUT(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); - FfxFloat32x4 fColorXY = Lanczos2_UseLUT(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); - - // Deringing - - // TODO: only use 4 by checking jitter - const FfxInt32 iDeringingSampleCount = 4; - const FfxFloat32x4 fDeringingSamples[4] = { - Samples.fColor11, - Samples.fColor21, - Samples.fColor12, - Samples.fColor22, - }; - - FfxFloat32x4 fDeringingMin = fDeringingSamples[0]; - FfxFloat32x4 fDeringingMax = fDeringingSamples[0]; - - FFX_UNROLL - for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) { - - fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); - fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); - } - - fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); - - return fColorXY; -} - -#if FFX_HALF -FFX_MIN16_F4 Lanczos2LUT(FetchedBicubicSamplesMin16 Samples, FFX_MIN16_F2 fPxFrac) -{ - FFX_MIN16_F4 fColorX0 = Lanczos2_UseLUT(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); - FFX_MIN16_F4 fColorX1 = Lanczos2_UseLUT(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); - FFX_MIN16_F4 fColorX2 = Lanczos2_UseLUT(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); - FFX_MIN16_F4 fColorX3 = Lanczos2_UseLUT(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); - FFX_MIN16_F4 fColorXY = Lanczos2_UseLUT(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); - - // Deringing - - // TODO: only use 4 by checking jitter - const FfxInt32 iDeringingSampleCount = 4; - const FFX_MIN16_F4 fDeringingSamples[4] = { - Samples.fColor11, - Samples.fColor21, - Samples.fColor12, - Samples.fColor22, - }; - - FFX_MIN16_F4 fDeringingMin = fDeringingSamples[0]; - FFX_MIN16_F4 fDeringingMax = fDeringingSamples[0]; - - FFX_UNROLL - for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) - { - fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); - fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); - } - - fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); - - return fColorXY; -} -#endif //FFX_HALF - - - -FfxFloat32x4 Lanczos2Approx(FfxFloat32x4 fColor0, FfxFloat32x4 fColor1, FfxFloat32x4 fColor2, FfxFloat32x4 fColor3, FfxFloat32 t) -{ - FfxFloat32 fWeight0 = Lanczos2ApproxNoClamp(-1.f - t); - FfxFloat32 fWeight1 = Lanczos2ApproxNoClamp(-0.f - t); - FfxFloat32 fWeight2 = Lanczos2ApproxNoClamp(+1.f - t); - FfxFloat32 fWeight3 = Lanczos2ApproxNoClamp(+2.f - t); - return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); -} - -#if FFX_HALF -FFX_MIN16_F4 Lanczos2Approx(FFX_MIN16_F4 fColor0, FFX_MIN16_F4 fColor1, FFX_MIN16_F4 fColor2, FFX_MIN16_F4 fColor3, FFX_MIN16_F t) -{ - FFX_MIN16_F fWeight0 = Lanczos2ApproxNoClamp(FFX_MIN16_F(-1.f) - t); - FFX_MIN16_F fWeight1 = Lanczos2ApproxNoClamp(FFX_MIN16_F(-0.f) - t); - FFX_MIN16_F fWeight2 = Lanczos2ApproxNoClamp(FFX_MIN16_F(+1.f) - t); - FFX_MIN16_F fWeight3 = Lanczos2ApproxNoClamp(FFX_MIN16_F(+2.f) - t); - return (fWeight0 * fColor0 + fWeight1 * fColor1 + fWeight2 * fColor2 + fWeight3 * fColor3) / (fWeight0 + fWeight1 + fWeight2 + fWeight3); -} -#endif //FFX_HALF - -FfxFloat32x4 Lanczos2Approx(FetchedBicubicSamples Samples, FfxFloat32x2 fPxFrac) -{ - FfxFloat32x4 fColorX0 = Lanczos2Approx(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); - FfxFloat32x4 fColorX1 = Lanczos2Approx(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); - FfxFloat32x4 fColorX2 = Lanczos2Approx(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); - FfxFloat32x4 fColorX3 = Lanczos2Approx(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); - FfxFloat32x4 fColorXY = Lanczos2Approx(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); - - // Deringing - - // TODO: only use 4 by checking jitter - const FfxInt32 iDeringingSampleCount = 4; - const FfxFloat32x4 fDeringingSamples[4] = { - Samples.fColor11, - Samples.fColor21, - Samples.fColor12, - Samples.fColor22, - }; - - FfxFloat32x4 fDeringingMin = fDeringingSamples[0]; - FfxFloat32x4 fDeringingMax = fDeringingSamples[0]; - - FFX_UNROLL - for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) - { - fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); - fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); - } - - fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); - - return fColorXY; -} - -#if FFX_HALF -FFX_MIN16_F4 Lanczos2Approx(FetchedBicubicSamplesMin16 Samples, FFX_MIN16_F2 fPxFrac) -{ - FFX_MIN16_F4 fColorX0 = Lanczos2Approx(Samples.fColor00, Samples.fColor10, Samples.fColor20, Samples.fColor30, fPxFrac.x); - FFX_MIN16_F4 fColorX1 = Lanczos2Approx(Samples.fColor01, Samples.fColor11, Samples.fColor21, Samples.fColor31, fPxFrac.x); - FFX_MIN16_F4 fColorX2 = Lanczos2Approx(Samples.fColor02, Samples.fColor12, Samples.fColor22, Samples.fColor32, fPxFrac.x); - FFX_MIN16_F4 fColorX3 = Lanczos2Approx(Samples.fColor03, Samples.fColor13, Samples.fColor23, Samples.fColor33, fPxFrac.x); - FFX_MIN16_F4 fColorXY = Lanczos2Approx(fColorX0, fColorX1, fColorX2, fColorX3, fPxFrac.y); - - // Deringing - - // TODO: only use 4 by checking jitter - const FfxInt32 iDeringingSampleCount = 4; - const FFX_MIN16_F4 fDeringingSamples[4] = { - Samples.fColor11, - Samples.fColor21, - Samples.fColor12, - Samples.fColor22, - }; - - FFX_MIN16_F4 fDeringingMin = fDeringingSamples[0]; - FFX_MIN16_F4 fDeringingMax = fDeringingSamples[0]; - - FFX_UNROLL - for (FfxInt32 iSampleIndex = 1; iSampleIndex < iDeringingSampleCount; ++iSampleIndex) - { - fDeringingMin = ffxMin(fDeringingMin, fDeringingSamples[iSampleIndex]); - fDeringingMax = ffxMax(fDeringingMax, fDeringingSamples[iSampleIndex]); - } - - fColorXY = clamp(fColorXY, fDeringingMin, fDeringingMax); - - return fColorXY; -} -#endif - -// Clamp by offset direction. Assuming iPxSample is already in range and iPxOffset is compile time constant. -FfxInt32x2 ClampCoord(FfxInt32x2 iPxSample, FfxInt32x2 iPxOffset, FfxInt32x2 iTextureSize) -{ - FfxInt32x2 result = iPxSample + iPxOffset; - result.x = (iPxOffset.x < 0) ? ffxMax(result.x, 0) : result.x; - result.x = (iPxOffset.x > 0) ? ffxMin(result.x, iTextureSize.x - 1) : result.x; - result.y = (iPxOffset.y < 0) ? ffxMax(result.y, 0) : result.y; - result.y = (iPxOffset.y > 0) ? ffxMin(result.y, iTextureSize.y - 1) : result.y; - return result; -} -#if FFX_HALF -FFX_MIN16_I2 ClampCoord(FFX_MIN16_I2 iPxSample, FFX_MIN16_I2 iPxOffset, FFX_MIN16_I2 iTextureSize) -{ - FFX_MIN16_I2 result = iPxSample + iPxOffset; - result.x = (iPxOffset.x < FFX_MIN16_I(0)) ? ffxMax(result.x, FFX_MIN16_I(0)) : result.x; - result.x = (iPxOffset.x > FFX_MIN16_I(0)) ? ffxMin(result.x, iTextureSize.x - FFX_MIN16_I(1)) : result.x; - result.y = (iPxOffset.y < FFX_MIN16_I(0)) ? ffxMax(result.y, FFX_MIN16_I(0)) : result.y; - result.y = (iPxOffset.y > FFX_MIN16_I(0)) ? ffxMin(result.y, iTextureSize.y - FFX_MIN16_I(1)) : result.y; - return result; -} -#endif //FFX_HALF - - -#define DeclareCustomFetchBicubicSamplesWithType(SampleType, TextureType, AddrType, Name, LoadTexture) \ - SampleType Name(AddrType iPxSample, AddrType iTextureSize) \ - { \ - SampleType Samples; \ - \ - Samples.fColor00 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(-1, -1), iTextureSize))); \ - Samples.fColor10 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, -1), iTextureSize))); \ - Samples.fColor20 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, -1), iTextureSize))); \ - Samples.fColor30 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+2, -1), iTextureSize))); \ - \ - Samples.fColor01 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(-1, +0), iTextureSize))); \ - Samples.fColor11 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, +0), iTextureSize))); \ - Samples.fColor21 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, +0), iTextureSize))); \ - Samples.fColor31 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+2, +0), iTextureSize))); \ - \ - Samples.fColor02 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(-1, +1), iTextureSize))); \ - Samples.fColor12 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, +1), iTextureSize))); \ - Samples.fColor22 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, +1), iTextureSize))); \ - Samples.fColor32 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+2, +1), iTextureSize))); \ - \ - Samples.fColor03 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(-1, +2), iTextureSize))); \ - Samples.fColor13 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, +2), iTextureSize))); \ - Samples.fColor23 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, +2), iTextureSize))); \ - Samples.fColor33 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+2, +2), iTextureSize))); \ - \ - return Samples; \ - } - -#define DeclareCustomFetchBicubicSamples(Name, LoadTexture) \ - DeclareCustomFetchBicubicSamplesWithType(FetchedBicubicSamples, FfxFloat32x4, FfxInt32x2, Name, LoadTexture) - -#define DeclareCustomFetchBicubicSamplesMin16(Name, LoadTexture) \ - DeclareCustomFetchBicubicSamplesWithType(FetchedBicubicSamplesMin16, FFX_MIN16_F4, FfxInt32x2, Name, LoadTexture) - -#define DeclareCustomFetchBilinearSamplesWithType(SampleType, TextureType,AddrType, Name, LoadTexture) \ - SampleType Name(AddrType iPxSample, AddrType iTextureSize) \ - { \ - SampleType Samples; \ - Samples.fColor00 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, +0), iTextureSize))); \ - Samples.fColor10 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, +0), iTextureSize))); \ - Samples.fColor01 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+0, +1), iTextureSize))); \ - Samples.fColor11 = TextureType(LoadTexture(ClampCoord(iPxSample, AddrType(+1, +1), iTextureSize))); \ - return Samples; \ - } - -#define DeclareCustomFetchBilinearSamples(Name, LoadTexture) \ - DeclareCustomFetchBilinearSamplesWithType(FetchedBilinearSamples, FfxFloat32x4, FfxInt32x2, Name, LoadTexture) - -#define DeclareCustomFetchBilinearSamplesMin16(Name, LoadTexture) \ - DeclareCustomFetchBilinearSamplesWithType(FetchedBilinearSamplesMin16, FFX_MIN16_F4, FfxInt32x2, Name, LoadTexture) - -// BE CAREFUL: there is some precision issues and (3253, 125) leading to (3252.9989778, 125.001102) -// is common, so iPxSample can "jitter" -#define DeclareCustomTextureSample(Name, InterpolateSamples, FetchSamples) \ - FfxFloat32x4 Name(FfxFloat32x2 fUvSample, FfxInt32x2 iTextureSize) \ - { \ - FfxFloat32x2 fPxSample = (fUvSample * FfxFloat32x2(iTextureSize)) - FfxFloat32x2(0.5f, 0.5f); \ - /* Clamp base coords */ \ - fPxSample.x = ffxMax(0.0f, ffxMin(FfxFloat32(iTextureSize.x), fPxSample.x)); \ - fPxSample.y = ffxMax(0.0f, ffxMin(FfxFloat32(iTextureSize.y), fPxSample.y)); \ - /* */ \ - FfxInt32x2 iPxSample = FfxInt32x2(floor(fPxSample)); \ - FfxFloat32x2 fPxFrac = ffxFract(fPxSample); \ - FfxFloat32x4 fColorXY = FfxFloat32x4(InterpolateSamples(FetchSamples(iPxSample, iTextureSize), fPxFrac)); \ - return fColorXY; \ - } - -#define DeclareCustomTextureSampleMin16(Name, InterpolateSamples, FetchSamples) \ - FFX_MIN16_F4 Name(FfxFloat32x2 fUvSample, FfxInt32x2 iTextureSize) \ - { \ - FfxFloat32x2 fPxSample = (fUvSample * FfxFloat32x2(iTextureSize)) - FfxFloat32x2(0.5f, 0.5f); \ - /* Clamp base coords */ \ - fPxSample.x = ffxMax(0.0f, ffxMin(FfxFloat32(iTextureSize.x), fPxSample.x)); \ - fPxSample.y = ffxMax(0.0f, ffxMin(FfxFloat32(iTextureSize.y), fPxSample.y)); \ - /* */ \ - FfxInt32x2 iPxSample = FfxInt32x2(floor(fPxSample)); \ - FFX_MIN16_F2 fPxFrac = FFX_MIN16_F2(ffxFract(fPxSample)); \ - FFX_MIN16_F4 fColorXY = FFX_MIN16_F4(InterpolateSamples(FetchSamples(iPxSample, iTextureSize), fPxFrac)); \ - return fColorXY; \ - } - -#define FFX_FSR3UPSCALER_CONCAT_ID(x, y) x ## y -#define FFX_FSR3UPSCALER_CONCAT(x, y) FFX_FSR3UPSCALER_CONCAT_ID(x, y) -#define FFX_FSR3UPSCALER_SAMPLER_1D_0 Lanczos2 -#define FFX_FSR3UPSCALER_SAMPLER_1D_1 Lanczos2LUT -#define FFX_FSR3UPSCALER_SAMPLER_1D_2 Lanczos2Approx - -#define FFX_FSR3UPSCALER_GET_LANCZOS_SAMPLER1D(x) FFX_FSR3UPSCALER_CONCAT(FFX_FSR3UPSCALER_SAMPLER_1D_, x) - -#endif //!defined( FFX_FSR3UPSCALER_SAMPLE_H ) diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_sample.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_sample.h.meta deleted file mode 100644 index 8e3cf72..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_sample.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: dcb900c9deecd06419a8a4c10c305890 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_tcr_autogen.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_tcr_autogen.h deleted file mode 100644 index 2d446bb..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_tcr_autogen.h +++ /dev/null @@ -1,250 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2022-2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - -#define USE_YCOCG 1 - -#define fAutogenEpsilon 0.01f - -// EXPERIMENTAL - -FFX_MIN16_F ComputeAutoTC_01(FFX_MIN16_I2 uDispatchThreadId, FFX_MIN16_I2 iPrevIdx) -{ - FfxFloat32x3 colorPreAlpha = LoadOpaqueOnly(uDispatchThreadId); - FfxFloat32x3 colorPostAlpha = LoadInputColor(uDispatchThreadId); - FfxFloat32x3 colorPrevPreAlpha = LoadPrevPreAlpha(iPrevIdx); - FfxFloat32x3 colorPrevPostAlpha = LoadPrevPostAlpha(iPrevIdx); - -#if USE_YCOCG - colorPreAlpha = RGBToYCoCg(colorPreAlpha); - colorPostAlpha = RGBToYCoCg(colorPostAlpha); - colorPrevPreAlpha = RGBToYCoCg(colorPrevPreAlpha); - colorPrevPostAlpha = RGBToYCoCg(colorPrevPostAlpha); -#endif - - FfxFloat32x3 colorDeltaCurr = colorPostAlpha - colorPreAlpha; - FfxFloat32x3 colorDeltaPrev = colorPrevPostAlpha - colorPrevPreAlpha; - bool hasAlpha = any(FFX_GREATER_THAN(abs(colorDeltaCurr), FfxFloat32x3(fAutogenEpsilon, fAutogenEpsilon, fAutogenEpsilon))); - bool hadAlpha = any(FFX_GREATER_THAN(abs(colorDeltaPrev), FfxFloat32x3(fAutogenEpsilon, fAutogenEpsilon, fAutogenEpsilon))); - - FfxFloat32x3 X = colorPreAlpha; - FfxFloat32x3 Y = colorPostAlpha; - FfxFloat32x3 Z = colorPrevPreAlpha; - FfxFloat32x3 W = colorPrevPostAlpha; - - FFX_MIN16_F retVal = FFX_MIN16_F(ffxSaturate(dot(abs(abs(Y - X) - abs(W - Z)), FfxFloat32x3(1, 1, 1)))); - - // cleanup very small values - retVal = (retVal < TcThreshold()) ? FFX_MIN16_F(0.0f) : FFX_MIN16_F(1.f); - - return retVal; -} - -// works ok: thin edges -FFX_MIN16_F ComputeAutoTC_02(FFX_MIN16_I2 uDispatchThreadId, FFX_MIN16_I2 iPrevIdx) -{ - FfxFloat32x3 colorPreAlpha = LoadOpaqueOnly(uDispatchThreadId); - FfxFloat32x3 colorPostAlpha = LoadInputColor(uDispatchThreadId); - FfxFloat32x3 colorPrevPreAlpha = LoadPrevPreAlpha(iPrevIdx); - FfxFloat32x3 colorPrevPostAlpha = LoadPrevPostAlpha(iPrevIdx); - -#if USE_YCOCG - colorPreAlpha = RGBToYCoCg(colorPreAlpha); - colorPostAlpha = RGBToYCoCg(colorPostAlpha); - colorPrevPreAlpha = RGBToYCoCg(colorPrevPreAlpha); - colorPrevPostAlpha = RGBToYCoCg(colorPrevPostAlpha); -#endif - - FfxFloat32x3 colorDelta = colorPostAlpha - colorPreAlpha; - FfxFloat32x3 colorPrevDelta = colorPrevPostAlpha - colorPrevPreAlpha; - bool hasAlpha = any(FFX_GREATER_THAN(abs(colorDelta), FfxFloat32x3(fAutogenEpsilon, fAutogenEpsilon, fAutogenEpsilon))); - bool hadAlpha = any(FFX_GREATER_THAN(abs(colorPrevDelta), FfxFloat32x3(fAutogenEpsilon, fAutogenEpsilon, fAutogenEpsilon))); - - FfxFloat32x3 delta = colorPostAlpha - colorPreAlpha; //prev+1*d = post => d = color, alpha = - FfxFloat32x3 deltaPrev = colorPrevPostAlpha - colorPrevPreAlpha; - - FfxFloat32x3 X = colorPrevPreAlpha; - FfxFloat32x3 N = colorPreAlpha - colorPrevPreAlpha; - FfxFloat32x3 YAminusXA = colorPrevPostAlpha - colorPrevPreAlpha; - FfxFloat32x3 NminusNA = colorPostAlpha - colorPrevPostAlpha; - - FfxFloat32x3 A = (hasAlpha || hadAlpha) ? NminusNA / max(FfxFloat32x3(fAutogenEpsilon, fAutogenEpsilon, fAutogenEpsilon), N) : FfxFloat32x3(0, 0, 0); - - FFX_MIN16_F retVal = FFX_MIN16_F( max(max(A.x, A.y), A.z) ); - - // only pixels that have significantly changed in color shuold be considered - retVal = ffxSaturate(retVal * FFX_MIN16_F(length(colorPostAlpha - colorPrevPostAlpha)) ); - - return retVal; -} - -// This function computes the TransparencyAndComposition mask: -// This mask indicates pixels that should discard locks and apply color clamping. -// -// Typically this is the case for translucent pixels (that don't write depth values) or pixels where the correctness of -// the MVs can not be guaranteed (e.g. procedutal movement or vegetation that does not have MVs to reduce the cost during rasterization) -// Also, large changes in color due to changed lighting should be marked to remove locks on pixels with "old" lighting. -// -// This function takes a opaque only and a final texture and uses internal copies of those textures from the last frame. -// The function tries to determine where the color changes between opaque only and final image to determine the pixels that use transparency. -// Also it uses the previous frames and detects where the use of transparency changed to mark those pixels. -// Additionally it marks pixels where the color changed significantly in the opaque only image, e.g. due to lighting or texture animation. -// -// In the final step it stores the current textures in internal textures for the next frame - -FFX_MIN16_F ComputeTransparencyAndComposition(FFX_MIN16_I2 uDispatchThreadId, FFX_MIN16_I2 iPrevIdx) -{ - FFX_MIN16_F retVal = ComputeAutoTC_02(uDispatchThreadId, iPrevIdx); - - // [branch] - if (retVal > FFX_MIN16_F(0.01f)) - { - retVal = ComputeAutoTC_01(uDispatchThreadId, iPrevIdx); - } - return retVal; -} - -float computeSolidEdge(FFX_MIN16_I2 curPos, FFX_MIN16_I2 prevPos) -{ - float lum[9]; - int i = 0; - for (int y = -1; y < 2; ++y) - { - for (int x = -1; x < 2; ++x) - { - FfxFloat32x3 curCol = LoadOpaqueOnly(curPos + FFX_MIN16_I2(x, y)).rgb; - FfxFloat32x3 prevCol = LoadPrevPreAlpha(prevPos + FFX_MIN16_I2(x, y)).rgb; - lum[i++] = length(curCol - prevCol); - } - } - - //float gradX = abs(lum[3] - lum[4]) + abs(lum[5] - lum[4]); - //float gradY = abs(lum[1] - lum[4]) + abs(lum[7] - lum[4]); - - //return sqrt(gradX * gradX + gradY * gradY); - - float gradX = abs(lum[3] - lum[4]) * abs(lum[5] - lum[4]); - float gradY = abs(lum[1] - lum[4]) * abs(lum[7] - lum[4]); - - return sqrt(sqrt(gradX * gradY)); -} - -float computeAlphaEdge(FFX_MIN16_I2 curPos, FFX_MIN16_I2 prevPos) -{ - float lum[9]; - int i = 0; - for (int y = -1; y < 2; ++y) - { - for (int x = -1; x < 2; ++x) - { - FfxFloat32x3 curCol = abs(LoadInputColor(curPos + FFX_MIN16_I2(x, y)).rgb - LoadOpaqueOnly(curPos + FFX_MIN16_I2(x, y)).rgb); - FfxFloat32x3 prevCol = abs(LoadPrevPostAlpha(prevPos + FFX_MIN16_I2(x, y)).rgb - LoadPrevPreAlpha(prevPos + FFX_MIN16_I2(x, y)).rgb); - lum[i++] = length(curCol - prevCol); - } - } - - //float gradX = abs(lum[3] - lum[4]) + abs(lum[5] - lum[4]); - //float gradY = abs(lum[1] - lum[4]) + abs(lum[7] - lum[4]); - - //return sqrt(gradX * gradX + gradY * gradY); - - float gradX = abs(lum[3] - lum[4]) * abs(lum[5] - lum[4]); - float gradY = abs(lum[1] - lum[4]) * abs(lum[7] - lum[4]); - - return sqrt(sqrt(gradX * gradY)); -} - -FFX_MIN16_F ComputeAabbOverlap(FFX_MIN16_I2 uDispatchThreadId, FFX_MIN16_I2 iPrevIdx) -{ - FFX_MIN16_F retVal = FFX_MIN16_F(0.f); - - FfxFloat32x2 fMotionVector = LoadInputMotionVector(uDispatchThreadId); - FfxFloat32x3 colorPreAlpha = LoadOpaqueOnly(uDispatchThreadId); - FfxFloat32x3 colorPostAlpha = LoadInputColor(uDispatchThreadId); - FfxFloat32x3 colorPrevPreAlpha = LoadPrevPreAlpha(iPrevIdx); - FfxFloat32x3 colorPrevPostAlpha = LoadPrevPostAlpha(iPrevIdx); - -#if USE_YCOCG - colorPreAlpha = RGBToYCoCg(colorPreAlpha); - colorPostAlpha = RGBToYCoCg(colorPostAlpha); - colorPrevPreAlpha = RGBToYCoCg(colorPrevPreAlpha); - colorPrevPostAlpha = RGBToYCoCg(colorPrevPostAlpha); -#endif - FfxFloat32x3 minPrev = FFX_MIN16_F3(+1000.f, +1000.f, +1000.f); - FfxFloat32x3 maxPrev = FFX_MIN16_F3(-1000.f, -1000.f, -1000.f); - for (int y = -1; y < 2; ++y) - { - for (int x = -1; x < 2; ++x) - { - FfxFloat32x3 W = LoadPrevPostAlpha(iPrevIdx + FFX_MIN16_I2(x, y)); - -#if USE_YCOCG - W = RGBToYCoCg(W); -#endif - minPrev = min(minPrev, W); - maxPrev = max(maxPrev, W); - } - } - // instead of computing the overlap: simply count how many samples are outside - // set reactive based on that - FFX_MIN16_F count = FFX_MIN16_F(0.f); - for (int y = -1; y < 2; ++y) - { - for (int x = -1; x < 2; ++x) - { - FfxFloat32x3 Y = LoadInputColor(uDispatchThreadId + FFX_MIN16_I2(x, y)); - -#if USE_YCOCG - Y = RGBToYCoCg(Y); -#endif - count += ((Y.x < minPrev.x) || (Y.x > maxPrev.x)) ? FFX_MIN16_F(1.f) : FFX_MIN16_F(0.f); - count += ((Y.y < minPrev.y) || (Y.y > maxPrev.y)) ? FFX_MIN16_F(1.f) : FFX_MIN16_F(0.f); - count += ((Y.z < minPrev.z) || (Y.z > maxPrev.z)) ? FFX_MIN16_F(1.f) : FFX_MIN16_F(0.f); - } - } - retVal = count / FFX_MIN16_F(27.f); - - return retVal; -} - - -// This function computes the Reactive mask: -// We want pixels marked where the alpha portion of the frame changes a lot between neighbours -// Those pixels are expected to change quickly between frames, too. (e.g. small particles, reflections on curved surfaces...) -// As a result history would not be trustworthy. -// On the other hand we don't want pixels marked where pre-alpha has a large differnce, since those would profit from accumulation -// For mirrors we may assume the pre-alpha is pretty uniform color. -// -// This works well generally, but also marks edge pixels -FFX_MIN16_F ComputeReactive(FFX_MIN16_I2 uDispatchThreadId, FFX_MIN16_I2 iPrevIdx) -{ - // we only get here if alpha has a significant contribution and has changed since last frame. - FFX_MIN16_F retVal = FFX_MIN16_F(0.f); - - // mark pixels with huge variance in alpha as reactive - FFX_MIN16_F alphaEdge = FFX_MIN16_F(computeAlphaEdge(uDispatchThreadId, iPrevIdx)); - FFX_MIN16_F opaqueEdge = FFX_MIN16_F(computeSolidEdge(uDispatchThreadId, iPrevIdx)); - retVal = ffxSaturate(alphaEdge - opaqueEdge); - - // the above also marks edge pixels due to jitter, so we need to cancel those out - - - return retVal; -} diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_tcr_autogen.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_tcr_autogen.h.meta deleted file mode 100644 index bdeb970..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_tcr_autogen.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: f01d5a8fbd1f34a4ea8d971755a21b6c -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_upsample.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_upsample.h deleted file mode 100644 index 47e7ccf..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_upsample.h +++ /dev/null @@ -1,195 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -#ifndef FFX_FSR3UPSCALER_UPSAMPLE_H -#define FFX_FSR3UPSCALER_UPSAMPLE_H - -FFX_STATIC const FfxUInt32 iLanczos2SampleCount = 16; - -void Deringing(RectificationBox clippingBox, FFX_PARAMETER_INOUT FfxFloat32x3 fColor) -{ - fColor = clamp(fColor, clippingBox.aabbMin, clippingBox.aabbMax); -} -#if FFX_HALF -void Deringing(RectificationBoxMin16 clippingBox, FFX_PARAMETER_INOUT FFX_MIN16_F3 fColor) -{ - fColor = clamp(fColor, clippingBox.aabbMin, clippingBox.aabbMax); -} -#endif - -#ifndef FFX_FSR3UPSCALER_OPTION_UPSAMPLE_USE_LANCZOS_TYPE -#define FFX_FSR3UPSCALER_OPTION_UPSAMPLE_USE_LANCZOS_TYPE 2 // Approximate -#endif - -FfxFloat32 GetUpsampleLanczosWeight(FfxFloat32x2 fSrcSampleOffset, FfxFloat32 fKernelWeight) -{ - FfxFloat32x2 fSrcSampleOffsetBiased = fSrcSampleOffset * fKernelWeight.xx; -#if FFX_FSR3UPSCALER_OPTION_UPSAMPLE_USE_LANCZOS_TYPE == 0 // LANCZOS_TYPE_REFERENCE - FfxFloat32 fSampleWeight = Lanczos2(length(fSrcSampleOffsetBiased)); -#elif FFX_FSR3UPSCALER_OPTION_UPSAMPLE_USE_LANCZOS_TYPE == 1 // LANCZOS_TYPE_LUT - FfxFloat32 fSampleWeight = Lanczos2_UseLUT(length(fSrcSampleOffsetBiased)); -#elif FFX_FSR3UPSCALER_OPTION_UPSAMPLE_USE_LANCZOS_TYPE == 2 // LANCZOS_TYPE_APPROXIMATE - FfxFloat32 fSampleWeight = Lanczos2ApproxSq(dot(fSrcSampleOffsetBiased, fSrcSampleOffsetBiased)); -#else -#error "Invalid Lanczos type" -#endif - return fSampleWeight; -} - -#if FFX_HALF -FFX_MIN16_F GetUpsampleLanczosWeight(FFX_MIN16_F2 fSrcSampleOffset, FFX_MIN16_F fKernelWeight) -{ - FFX_MIN16_F2 fSrcSampleOffsetBiased = fSrcSampleOffset * fKernelWeight.xx; -#if FFX_FSR3UPSCALER_OPTION_UPSAMPLE_USE_LANCZOS_TYPE == 0 // LANCZOS_TYPE_REFERENCE - FFX_MIN16_F fSampleWeight = Lanczos2(length(fSrcSampleOffsetBiased)); -#elif FFX_FSR3UPSCALER_OPTION_UPSAMPLE_USE_LANCZOS_TYPE == 1 // LANCZOS_TYPE_LUT - FFX_MIN16_F fSampleWeight = Lanczos2_UseLUT(length(fSrcSampleOffsetBiased)); -#elif FFX_FSR3UPSCALER_OPTION_UPSAMPLE_USE_LANCZOS_TYPE == 2 // LANCZOS_TYPE_APPROXIMATE - FFX_MIN16_F fSampleWeight = Lanczos2ApproxSq(dot(fSrcSampleOffsetBiased, fSrcSampleOffsetBiased)); - - // To Test: Save reciproqual sqrt compute - // FfxFloat32 fSampleWeight = Lanczos2Sq_UseLUT(dot(fSrcSampleOffsetBiased, fSrcSampleOffsetBiased)); -#else -#error "Invalid Lanczos type" -#endif - return fSampleWeight; -} -#endif - -FfxFloat32 ComputeMaxKernelWeight() { - const FfxFloat32 fKernelSizeBias = 1.0f; - - FfxFloat32 fKernelWeight = FfxFloat32(1) + (FfxFloat32(1.0f) / FfxFloat32x2(DownscaleFactor()) - FfxFloat32(1)).x * FfxFloat32(fKernelSizeBias); - - return ffxMin(FfxFloat32(1.99f), fKernelWeight); -} - -FfxFloat32x4 ComputeUpsampledColorAndWeight(const AccumulationPassCommonParams params, - FFX_PARAMETER_INOUT RectificationBox clippingBox, FfxFloat32 fReactiveFactor) -{ - #if FFX_FSR3UPSCALER_OPTION_UPSAMPLE_SAMPLERS_USE_DATA_HALF && FFX_HALF - #include "ffx_fsr3upscaler_force16_begin.h" - #endif - // We compute a sliced lanczos filter with 2 lobes (other slices are accumulated temporaly) - FfxFloat32x2 fDstOutputPos = FfxFloat32x2(params.iPxHrPos) + FFX_BROADCAST_FLOAT32X2(0.5f); // Destination resolution output pixel center position - FfxFloat32x2 fSrcOutputPos = fDstOutputPos * DownscaleFactor(); // Source resolution output pixel center position - FfxInt32x2 iSrcInputPos = FfxInt32x2(floor(fSrcOutputPos)); // TODO: what about weird upscale factors... - - #if FFX_FSR3UPSCALER_OPTION_UPSAMPLE_SAMPLERS_USE_DATA_HALF && FFX_HALF - #include "ffx_fsr3upscaler_force16_end.h" - #endif - - FfxFloat32x3 fSamples[iLanczos2SampleCount]; - - FfxFloat32x2 fSrcUnjitteredPos = (FfxFloat32x2(iSrcInputPos) + FfxFloat32x2(0.5f, 0.5f)) - Jitter(); // This is the un-jittered position of the sample at offset 0,0 - - FfxInt32x2 offsetTL; - offsetTL.x = (fSrcUnjitteredPos.x > fSrcOutputPos.x) ? FfxInt32(-2) : FfxInt32(-1); - offsetTL.y = (fSrcUnjitteredPos.y > fSrcOutputPos.y) ? FfxInt32(-2) : FfxInt32(-1); - - //Load samples - // If fSrcUnjitteredPos.y > fSrcOutputPos.y, indicates offsetTL.y = -2, sample offset Y will be [-2, 1], clipbox will be rows [1, 3]. - // Flip row# for sampling offset in this case, so first 0~2 rows in the sampled array can always be used for computing the clipbox. - // This reduces branch or cmove on sampled colors, but moving this overhead to sample position / weight calculation time which apply to less values. - const FfxBoolean bFlipRow = fSrcUnjitteredPos.y > fSrcOutputPos.y; - const FfxBoolean bFlipCol = fSrcUnjitteredPos.x > fSrcOutputPos.x; - - FfxFloat32x2 fOffsetTL = FfxFloat32x2(offsetTL); - - FFX_UNROLL - for (FfxInt32 row = 0; row < 3; row++) { - - FFX_UNROLL - for (FfxInt32 col = 0; col < 3; col++) { - FfxInt32 iSampleIndex = col + (row << 2); - - FfxInt32x2 sampleColRow = FfxInt32x2(bFlipCol ? (3 - col) : col, bFlipRow ? (3 - row) : row); - FfxInt32x2 iSrcSamplePos = FfxInt32x2(iSrcInputPos) + offsetTL + sampleColRow; - - const FfxInt32x2 sampleCoord = ClampLoad(iSrcSamplePos, FfxInt32x2(0, 0), FfxInt32x2(RenderSize())); - - fSamples[iSampleIndex] = LoadPreparedInputColor(FfxInt32x2(sampleCoord)); - } - } - - FfxFloat32x4 fColorAndWeight = FfxFloat32x4(0.0f, 0.0f, 0.0f, 0.0f); - - FfxFloat32x2 fBaseSampleOffset = FfxFloat32x2(fSrcUnjitteredPos - fSrcOutputPos); - - // Identify how much of each upsampled color to be used for this frame - const FfxFloat32 fKernelReactiveFactor = ffxMax(fReactiveFactor, FfxFloat32(params.bIsNewSample)); - const FfxFloat32 fKernelBiasMax = ComputeMaxKernelWeight() * (1.0f - fKernelReactiveFactor); - - const FfxFloat32 fKernelBiasMin = ffxMax(1.0f, ((1.0f + fKernelBiasMax) * 0.3f)); - const FfxFloat32 fKernelBiasFactor = ffxMax(0.0f, ffxMax(0.25f * params.fDepthClipFactor, fKernelReactiveFactor)); - const FfxFloat32 fKernelBias = ffxLerp(fKernelBiasMax, fKernelBiasMin, fKernelBiasFactor); - - const FfxFloat32 fRectificationCurveBias = ffxLerp(-2.0f, -3.0f, ffxSaturate(params.fHrVelocity / 50.0f)); - - FFX_UNROLL - for (FfxInt32 row = 0; row < 3; row++) { - FFX_UNROLL - for (FfxInt32 col = 0; col < 3; col++) { - FfxInt32 iSampleIndex = col + (row << 2); - - const FfxInt32x2 sampleColRow = FfxInt32x2(bFlipCol ? (3 - col) : col, bFlipRow ? (3 - row) : row); - const FfxFloat32x2 fOffset = fOffsetTL + FfxFloat32x2(sampleColRow); - FfxFloat32x2 fSrcSampleOffset = fBaseSampleOffset + fOffset; - - FfxInt32x2 iSrcSamplePos = FfxInt32x2(iSrcInputPos) + FfxInt32x2(offsetTL) + sampleColRow; - - const FfxFloat32 fOnScreenFactor = FfxFloat32(IsOnScreen(FfxInt32x2(iSrcSamplePos), FfxInt32x2(RenderSize()))); - FfxFloat32 fSampleWeight = fOnScreenFactor * FfxFloat32(GetUpsampleLanczosWeight(fSrcSampleOffset, fKernelBias)); - - fColorAndWeight += FfxFloat32x4(fSamples[iSampleIndex] * fSampleWeight, fSampleWeight); - - // Update rectification box - { - const FfxFloat32 fSrcSampleOffsetSq = dot(fSrcSampleOffset, fSrcSampleOffset); - const FfxFloat32 fBoxSampleWeight = exp(fRectificationCurveBias * fSrcSampleOffsetSq); - - const FfxBoolean bInitialSample = (row == 0) && (col == 0); - RectificationBoxAddSample(bInitialSample, clippingBox, fSamples[iSampleIndex], fBoxSampleWeight); - } - } - } - - RectificationBoxComputeVarianceBoxData(clippingBox); - - fColorAndWeight.w *= FfxFloat32(fColorAndWeight.w > FSR3UPSCALER_EPSILON); - - if (fColorAndWeight.w > FSR3UPSCALER_EPSILON) { - // Normalize for deringing (we need to compare colors) - fColorAndWeight.xyz = fColorAndWeight.xyz / fColorAndWeight.w; - fColorAndWeight.w *= fUpsampleLanczosWeightScale; - - Deringing(clippingBox, fColorAndWeight.xyz); - } - - #if FFX_FSR3UPSCALER_OPTION_UPSAMPLE_SAMPLERS_USE_DATA_HALF && FFX_HALF - #include "ffx_fsr3upscaler_force16_end.h" - #endif - - return fColorAndWeight; -} - -#endif //!defined( FFX_FSR3UPSCALER_UPSAMPLE_H ) diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_upsample.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_upsample.h.meta deleted file mode 100644 index dc0a729..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/ffx_fsr3upscaler_upsample.h.meta +++ /dev/null @@ -1,65 +0,0 @@ -fileFormatVersion: 2 -guid: 3e7832c4a9154414f9eaa125acfe6cd5 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - DefaultValueInitialized: true - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/fsr1.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/fsr1.meta deleted file mode 100644 index 731c94f..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/fsr1.meta +++ /dev/null @@ -1,8 +0,0 @@ -fileFormatVersion: 2 -guid: 09438bc445e66204f970dc99ca8dae5a -folderAsset: yes -DefaultImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/fsr1/ffx_fsr1.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/fsr1/ffx_fsr1.h deleted file mode 100644 index e780995..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/fsr1/ffx_fsr1.h +++ /dev/null @@ -1,1252 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -/// @defgroup FfxGPUFsr1 FidelityFX FSR1 -/// FidelityFX Super Resolution 1 GPU documentation -/// -/// @ingroup FfxGPUEffects - -/// Setup required constant values for EASU (works on CPU or GPU). -/// -/// @param [out] con0 -/// @param [out] con1 -/// @param [out] con2 -/// @param [out] con3 -/// @param [in] inputViewportInPixelsX The rendered image resolution being upscaled in X dimension. -/// @param [in] inputViewportInPixelsY The rendered image resolution being upscaled in Y dimension. -/// @param [in] inputSizeInPixelsX The resolution of the resource containing the input image (useful for dynamic resolution) in X dimension. -/// @param [in] inputSizeInPixelsY The resolution of the resource containing the input image (useful for dynamic resolution) in Y dimension. -/// @param [in] outputSizeInPixelsX The display resolution which the input image gets upscaled to in X dimension. -/// @param [in] outputSizeInPixelsY The display resolution which the input image gets upscaled to in Y dimension. -/// -/// @ingroup FfxGPUFsr1 -FFX_STATIC void ffxFsrPopulateEasuConstants( - FFX_PARAMETER_INOUT FfxUInt32x4 con0, - FFX_PARAMETER_INOUT FfxUInt32x4 con1, - FFX_PARAMETER_INOUT FfxUInt32x4 con2, - FFX_PARAMETER_INOUT FfxUInt32x4 con3, - FFX_PARAMETER_IN FfxFloat32 inputViewportInPixelsX, - FFX_PARAMETER_IN FfxFloat32 inputViewportInPixelsY, - FFX_PARAMETER_IN FfxFloat32 inputSizeInPixelsX, - FFX_PARAMETER_IN FfxFloat32 inputSizeInPixelsY, - FFX_PARAMETER_IN FfxFloat32 outputSizeInPixelsX, - FFX_PARAMETER_IN FfxFloat32 outputSizeInPixelsY) -{ - // Output integer position to a pixel position in viewport. - con0[0] = ffxAsUInt32(inputViewportInPixelsX * ffxReciprocal(outputSizeInPixelsX)); - con0[1] = ffxAsUInt32(inputViewportInPixelsY * ffxReciprocal(outputSizeInPixelsY)); - con0[2] = ffxAsUInt32(FfxFloat32(0.5) * inputViewportInPixelsX * ffxReciprocal(outputSizeInPixelsX) - FfxFloat32(0.5)); - con0[3] = ffxAsUInt32(FfxFloat32(0.5) * inputViewportInPixelsY * ffxReciprocal(outputSizeInPixelsY) - FfxFloat32(0.5)); - - // Viewport pixel position to normalized image space. - // This is used to get upper-left of 'F' tap. - con1[0] = ffxAsUInt32(ffxReciprocal(inputSizeInPixelsX)); - con1[1] = ffxAsUInt32(ffxReciprocal(inputSizeInPixelsY)); - - // Centers of gather4, first offset from upper-left of 'F'. - // +---+---+ - // | | | - // +--(0)--+ - // | b | c | - // +---F---+---+---+ - // | e | f | g | h | - // +--(1)--+--(2)--+ - // | i | j | k | l | - // +---+---+---+---+ - // | n | o | - // +--(3)--+ - // | | | - // +---+---+ - con1[2] = ffxAsUInt32(FfxFloat32(1.0) * ffxReciprocal(inputSizeInPixelsX)); - con1[3] = ffxAsUInt32(FfxFloat32(-1.0) * ffxReciprocal(inputSizeInPixelsY)); - - // These are from (0) instead of 'F'. - con2[0] = ffxAsUInt32(FfxFloat32(-1.0) * ffxReciprocal(inputSizeInPixelsX)); - con2[1] = ffxAsUInt32(FfxFloat32(2.0) * ffxReciprocal(inputSizeInPixelsY)); - con2[2] = ffxAsUInt32(FfxFloat32(1.0) * ffxReciprocal(inputSizeInPixelsX)); - con2[3] = ffxAsUInt32(FfxFloat32(2.0) * ffxReciprocal(inputSizeInPixelsY)); - con3[0] = ffxAsUInt32(FfxFloat32(0.0) * ffxReciprocal(inputSizeInPixelsX)); - con3[1] = ffxAsUInt32(FfxFloat32(4.0) * ffxReciprocal(inputSizeInPixelsY)); - con3[2] = con3[3] = 0; -} - -/// Setup required constant values for EASU (works on CPU or GPU). -/// -/// @param [out] con0 -/// @param [out] con1 -/// @param [out] con2 -/// @param [out] con3 -/// @param [in] inputViewportInPixelsX The resolution of the input in the X dimension. -/// @param [in] inputViewportInPixelsY The resolution of the input in the Y dimension. -/// @param [in] inputSizeInPixelsX The input size in pixels in the X dimension. -/// @param [in] inputSizeInPixelsY The input size in pixels in the Y dimension. -/// @param [in] outputSizeInPixelsX The output size in pixels in the X dimension. -/// @param [in] outputSizeInPixelsY The output size in pixels in the Y dimension. -/// @param [in] inputOffsetInPixelsX The input image offset in the X dimension into the resource containing it (useful for dynamic resolution). -/// @param [in] inputOffsetInPixelsY The input image offset in the Y dimension into the resource containing it (useful for dynamic resolution). -/// -/// @ingroup FfxGPUFsr1 -FFX_STATIC void ffxFsrPopulateEasuConstantsOffset( - FFX_PARAMETER_INOUT FfxUInt32x4 con0, - FFX_PARAMETER_INOUT FfxUInt32x4 con1, - FFX_PARAMETER_INOUT FfxUInt32x4 con2, - FFX_PARAMETER_INOUT FfxUInt32x4 con3, - FFX_PARAMETER_IN FfxFloat32 inputViewportInPixelsX, - FFX_PARAMETER_IN FfxFloat32 inputViewportInPixelsY, - FFX_PARAMETER_IN FfxFloat32 inputSizeInPixelsX, - FFX_PARAMETER_IN FfxFloat32 inputSizeInPixelsY, - FFX_PARAMETER_IN FfxFloat32 outputSizeInPixelsX, - FFX_PARAMETER_IN FfxFloat32 outputSizeInPixelsY, - FFX_PARAMETER_IN FfxFloat32 inputOffsetInPixelsX, - FFX_PARAMETER_IN FfxFloat32 inputOffsetInPixelsY) -{ - ffxFsrPopulateEasuConstants( - con0, - con1, - con2, - con3, - inputViewportInPixelsX, - inputViewportInPixelsY, - inputSizeInPixelsX, - inputSizeInPixelsY, - outputSizeInPixelsX, - outputSizeInPixelsY); - - // override - con0[2] = ffxAsUInt32(FfxFloat32(0.5) * inputViewportInPixelsX * ffxReciprocal(outputSizeInPixelsX) - FfxFloat32(0.5) + inputOffsetInPixelsX); - con0[3] = ffxAsUInt32(FfxFloat32(0.5) * inputViewportInPixelsY * ffxReciprocal(outputSizeInPixelsY) - FfxFloat32(0.5) + inputOffsetInPixelsY); -} - -#if defined(FFX_GPU) && defined(FFX_FSR_EASU_FLOAT) -// Input callback prototypes, need to be implemented by calling shader -FfxFloat32x4 FsrEasuRF(FfxFloat32x2 p); -FfxFloat32x4 FsrEasuGF(FfxFloat32x2 p); -FfxFloat32x4 FsrEasuBF(FfxFloat32x2 p); - -// Filtering for a given tap for the scalar. -void fsrEasuTapFloat( - FFX_PARAMETER_INOUT FfxFloat32x3 accumulatedColor, // Accumulated color, with negative lobe. - FFX_PARAMETER_INOUT FfxFloat32 accumulatedWeight, // Accumulated weight. - FFX_PARAMETER_IN FfxFloat32x2 pixelOffset, // Pixel offset from resolve position to tap. - FFX_PARAMETER_IN FfxFloat32x2 gradientDirection, // Gradient direction. - FFX_PARAMETER_IN FfxFloat32x2 length, // Length. - FFX_PARAMETER_IN FfxFloat32 negativeLobeStrength, // Negative lobe strength. - FFX_PARAMETER_IN FfxFloat32 clippingPoint, // Clipping point. - FFX_PARAMETER_IN FfxFloat32x3 color) // Tap color. -{ - // Rotate offset by direction. - FfxFloat32x2 rotatedOffset; - rotatedOffset.x = (pixelOffset.x * (gradientDirection.x)) + (pixelOffset.y * gradientDirection.y); - rotatedOffset.y = (pixelOffset.x * (-gradientDirection.y)) + (pixelOffset.y * gradientDirection.x); - - // Anisotropy. - rotatedOffset *= length; - - // Compute distance^2. - FfxFloat32 distanceSquared = rotatedOffset.x * rotatedOffset.x + rotatedOffset.y * rotatedOffset.y; - - // Limit to the window as at corner, 2 taps can easily be outside. - distanceSquared = ffxMin(distanceSquared, clippingPoint); - - // Approximation of lancos2 without sin() or rcp(), or sqrt() to get x. - // (25/16 * (2/5 * x^2 - 1)^2 - (25/16 - 1)) * (1/4 * x^2 - 1)^2 - // |_______________________________________| |_______________| - // base window - // The general form of the 'base' is, - // (a*(b*x^2-1)^2-(a-1)) - // Where 'a=1/(2*b-b^2)' and 'b' moves around the negative lobe. - FfxFloat32 weightB = FfxFloat32(2.0 / 5.0) * distanceSquared + FfxFloat32(-1.0); - FfxFloat32 weightA = negativeLobeStrength * distanceSquared + FfxFloat32(-1.0); - weightB *= weightB; - weightA *= weightA; - weightB = FfxFloat32(25.0 / 16.0) * weightB + FfxFloat32(-(25.0 / 16.0 - 1.0)); - FfxFloat32 weight = weightB * weightA; - - // Do weighted average. - accumulatedColor += color * weight; - accumulatedWeight += weight; -} - -// Accumulate direction and length. -void fsrEasuSetFloat( - FFX_PARAMETER_INOUT FfxFloat32x2 direction, - FFX_PARAMETER_INOUT FfxFloat32 length, - FFX_PARAMETER_IN FfxFloat32x2 pp, - FFX_PARAMETER_IN FfxBoolean biS, - FFX_PARAMETER_IN FfxBoolean biT, - FFX_PARAMETER_IN FfxBoolean biU, - FFX_PARAMETER_IN FfxBoolean biV, - FFX_PARAMETER_IN FfxFloat32 lA, - FFX_PARAMETER_IN FfxFloat32 lB, - FFX_PARAMETER_IN FfxFloat32 lC, - FFX_PARAMETER_IN FfxFloat32 lD, - FFX_PARAMETER_IN FfxFloat32 lE) -{ - // Compute bilinear weight, branches factor out as predicates are compiler time immediates. - // s t - // u v - FfxFloat32 weight = FfxFloat32(0.0); - if (biS) - weight = (FfxFloat32(1.0) - pp.x) * (FfxFloat32(1.0) - pp.y); - if (biT) - weight = pp.x * (FfxFloat32(1.0) - pp.y); - if (biU) - weight = (FfxFloat32(1.0) - pp.x) * pp.y; - if (biV) - weight = pp.x * pp.y; - - // Direction is the '+' diff. - // a - // b c d - // e - // Then takes magnitude from abs average of both sides of 'c'. - // Length converts gradient reversal to 0, smoothly to non-reversal at 1, shaped, then adding horz and vert terms. - FfxFloat32 dc = lD - lC; - FfxFloat32 cb = lC - lB; - FfxFloat32 lengthX = max(abs(dc), abs(cb)); - lengthX = ffxApproximateReciprocal(lengthX); - FfxFloat32 directionX = lD - lB; - direction.x += directionX * weight; - lengthX = ffxSaturate(abs(directionX) * lengthX); - lengthX *= lengthX; - length += lengthX * weight; - - // Repeat for the y axis. - FfxFloat32 ec = lE - lC; - FfxFloat32 ca = lC - lA; - FfxFloat32 lengthY = max(abs(ec), abs(ca)); - lengthY = ffxApproximateReciprocal(lengthY); - FfxFloat32 directionY = lE - lA; - direction.y += directionY * weight; - lengthY = ffxSaturate(abs(directionY) * lengthY); - lengthY *= lengthY; - length += lengthY * weight; -} - -/// Apply edge-aware spatial upsampling using 32bit floating point precision calculations. -/// -/// @param [out] outPixel The computed color of a pixel. -/// @param [in] integerPosition Integer pixel position within the output. -/// @param [in] con0 The first constant value generated by ffxFsrPopulateEasuConstants. -/// @param [in] con1 The second constant value generated by ffxFsrPopulateEasuConstants. -/// @param [in] con2 The third constant value generated by ffxFsrPopulateEasuConstants. -/// @param [in] con3 The fourth constant value generated by ffxFsrPopulateEasuConstants. -/// -/// @ingroup FSR -void ffxFsrEasuFloat( - FFX_PARAMETER_OUT FfxFloat32x3 pix, - FFX_PARAMETER_IN FfxUInt32x2 ip, - FFX_PARAMETER_IN FfxUInt32x4 con0, - FFX_PARAMETER_IN FfxUInt32x4 con1, - FFX_PARAMETER_IN FfxUInt32x4 con2, - FFX_PARAMETER_IN FfxUInt32x4 con3) -{ - // Get position of 'f'. - FfxFloat32x2 pp = FfxFloat32x2(ip) * ffxAsFloat(con0.xy) + ffxAsFloat(con0.zw); - FfxFloat32x2 fp = floor(pp); - pp -= fp; - - // 12-tap kernel. - // b c - // e f g h - // i j k l - // n o - // Gather 4 ordering. - // a b - // r g - // For packed FP16, need either {rg} or {ab} so using the following setup for gather in all versions, - // a b <- unused (z) - // r g - // a b a b - // r g r g - // a b - // r g <- unused (z) - // Allowing dead-code removal to remove the 'z's. - FfxFloat32x2 p0 = fp * ffxAsFloat(con1.xy) + ffxAsFloat(con1.zw); - - // These are from p0 to avoid pulling two constants on pre-Navi hardware. - FfxFloat32x2 p1 = p0 + ffxAsFloat(con2.xy); - FfxFloat32x2 p2 = p0 + ffxAsFloat(con2.zw); - FfxFloat32x2 p3 = p0 + ffxAsFloat(con3.xy); - FfxFloat32x4 bczzR = FsrEasuRF(p0); - FfxFloat32x4 bczzG = FsrEasuGF(p0); - FfxFloat32x4 bczzB = FsrEasuBF(p0); - FfxFloat32x4 ijfeR = FsrEasuRF(p1); - FfxFloat32x4 ijfeG = FsrEasuGF(p1); - FfxFloat32x4 ijfeB = FsrEasuBF(p1); - FfxFloat32x4 klhgR = FsrEasuRF(p2); - FfxFloat32x4 klhgG = FsrEasuGF(p2); - FfxFloat32x4 klhgB = FsrEasuBF(p2); - FfxFloat32x4 zzonR = FsrEasuRF(p3); - FfxFloat32x4 zzonG = FsrEasuGF(p3); - FfxFloat32x4 zzonB = FsrEasuBF(p3); - - // Simplest multi-channel approximate luma possible (luma times 2, in 2 FMA/MAD). - FfxFloat32x4 bczzL = bczzB * ffxBroadcast4(0.5) + (bczzR * ffxBroadcast4(0.5) + bczzG); - FfxFloat32x4 ijfeL = ijfeB * ffxBroadcast4(0.5) + (ijfeR * ffxBroadcast4(0.5) + ijfeG); - FfxFloat32x4 klhgL = klhgB * ffxBroadcast4(0.5) + (klhgR * ffxBroadcast4(0.5) + klhgG); - FfxFloat32x4 zzonL = zzonB * ffxBroadcast4(0.5) + (zzonR * ffxBroadcast4(0.5) + zzonG); - - // Rename. - FfxFloat32 bL = bczzL.x; - FfxFloat32 cL = bczzL.y; - FfxFloat32 iL = ijfeL.x; - FfxFloat32 jL = ijfeL.y; - FfxFloat32 fL = ijfeL.z; - FfxFloat32 eL = ijfeL.w; - FfxFloat32 kL = klhgL.x; - FfxFloat32 lL = klhgL.y; - FfxFloat32 hL = klhgL.z; - FfxFloat32 gL = klhgL.w; - FfxFloat32 oL = zzonL.z; - FfxFloat32 nL = zzonL.w; - - // Accumulate for bilinear interpolation. - FfxFloat32x2 dir = ffxBroadcast2(0.0); - FfxFloat32 len = FfxFloat32(0.0); - fsrEasuSetFloat(dir, len, pp, FFX_TRUE, FFX_FALSE, FFX_FALSE, FFX_FALSE, bL, eL, fL, gL, jL); - fsrEasuSetFloat(dir, len, pp, FFX_FALSE, FFX_TRUE, FFX_FALSE, FFX_FALSE, cL, fL, gL, hL, kL); - fsrEasuSetFloat(dir, len, pp, FFX_FALSE, FFX_FALSE, FFX_TRUE, FFX_FALSE, fL, iL, jL, kL, nL); - fsrEasuSetFloat(dir, len, pp, FFX_FALSE, FFX_FALSE, FFX_FALSE, FFX_TRUE, gL, jL, kL, lL, oL); - - // Normalize with approximation, and cleanup close to zero. - FfxFloat32x2 dir2 = dir * dir; - FfxFloat32 dirR = dir2.x + dir2.y; - FfxBoolean zro = dirR < FfxFloat32(1.0 / 32768.0); - dirR = ffxApproximateReciprocalSquareRoot(dirR); - dirR = zro ? FfxFloat32(1.0) : dirR; - dir.x = zro ? FfxFloat32(1.0) : dir.x; - dir *= ffxBroadcast2(dirR); - - // Transform from {0 to 2} to {0 to 1} range, and shape with square. - len = len * FfxFloat32(0.5); - len *= len; - - // Stretch kernel {1.0 vert|horz, to sqrt(2.0) on diagonal}. - FfxFloat32 stretch = (dir.x * dir.x + dir.y * dir.y) * ffxApproximateReciprocal(max(abs(dir.x), abs(dir.y))); - - // Anisotropic length after rotation, - // x := 1.0 lerp to 'stretch' on edges - // y := 1.0 lerp to 2x on edges - FfxFloat32x2 len2 = FfxFloat32x2(FfxFloat32(1.0) + (stretch - FfxFloat32(1.0)) * len, FfxFloat32(1.0) + FfxFloat32(-0.5) * len); - - // Based on the amount of 'edge', - // the window shifts from +/-{sqrt(2.0) to slightly beyond 2.0}. - FfxFloat32 lob = FfxFloat32(0.5) + FfxFloat32((1.0 / 4.0 - 0.04) - 0.5) * len; - - // Set distance^2 clipping point to the end of the adjustable window. - FfxFloat32 clp = ffxApproximateReciprocal(lob); - - // Accumulation mixed with min/max of 4 nearest. - // b c - // e f g h - // i j k l - // n o - FfxFloat32x3 min4 = - ffxMin(ffxMin3(FfxFloat32x3(ijfeR.z, ijfeG.z, ijfeB.z), FfxFloat32x3(klhgR.w, klhgG.w, klhgB.w), FfxFloat32x3(ijfeR.y, ijfeG.y, ijfeB.y)), - FfxFloat32x3(klhgR.x, klhgG.x, klhgB.x)); - FfxFloat32x3 max4 = - max(ffxMax3(FfxFloat32x3(ijfeR.z, ijfeG.z, ijfeB.z), FfxFloat32x3(klhgR.w, klhgG.w, klhgB.w), FfxFloat32x3(ijfeR.y, ijfeG.y, ijfeB.y)), FfxFloat32x3(klhgR.x, klhgG.x, klhgB.x)); - - // Accumulation. - FfxFloat32x3 aC = ffxBroadcast3(0.0); - FfxFloat32 aW = FfxFloat32(0.0); - fsrEasuTapFloat(aC, aW, FfxFloat32x2(0.0, -1.0) - pp, dir, len2, lob, clp, FfxFloat32x3(bczzR.x, bczzG.x, bczzB.x)); // b - fsrEasuTapFloat(aC, aW, FfxFloat32x2(1.0, -1.0) - pp, dir, len2, lob, clp, FfxFloat32x3(bczzR.y, bczzG.y, bczzB.y)); // c - fsrEasuTapFloat(aC, aW, FfxFloat32x2(-1.0, 1.0) - pp, dir, len2, lob, clp, FfxFloat32x3(ijfeR.x, ijfeG.x, ijfeB.x)); // i - fsrEasuTapFloat(aC, aW, FfxFloat32x2(0.0, 1.0) - pp, dir, len2, lob, clp, FfxFloat32x3(ijfeR.y, ijfeG.y, ijfeB.y)); // j - fsrEasuTapFloat(aC, aW, FfxFloat32x2(0.0, 0.0) - pp, dir, len2, lob, clp, FfxFloat32x3(ijfeR.z, ijfeG.z, ijfeB.z)); // f - fsrEasuTapFloat(aC, aW, FfxFloat32x2(-1.0, 0.0) - pp, dir, len2, lob, clp, FfxFloat32x3(ijfeR.w, ijfeG.w, ijfeB.w)); // e - fsrEasuTapFloat(aC, aW, FfxFloat32x2(1.0, 1.0) - pp, dir, len2, lob, clp, FfxFloat32x3(klhgR.x, klhgG.x, klhgB.x)); // k - fsrEasuTapFloat(aC, aW, FfxFloat32x2(2.0, 1.0) - pp, dir, len2, lob, clp, FfxFloat32x3(klhgR.y, klhgG.y, klhgB.y)); // l - fsrEasuTapFloat(aC, aW, FfxFloat32x2(2.0, 0.0) - pp, dir, len2, lob, clp, FfxFloat32x3(klhgR.z, klhgG.z, klhgB.z)); // h - fsrEasuTapFloat(aC, aW, FfxFloat32x2(1.0, 0.0) - pp, dir, len2, lob, clp, FfxFloat32x3(klhgR.w, klhgG.w, klhgB.w)); // g - fsrEasuTapFloat(aC, aW, FfxFloat32x2(1.0, 2.0) - pp, dir, len2, lob, clp, FfxFloat32x3(zzonR.z, zzonG.z, zzonB.z)); // o - fsrEasuTapFloat(aC, aW, FfxFloat32x2(0.0, 2.0) - pp, dir, len2, lob, clp, FfxFloat32x3(zzonR.w, zzonG.w, zzonB.w)); // n - - // Normalize and dering. - pix = ffxMin(max4, max(min4, aC * ffxBroadcast3(rcp(aW)))); -} -#endif // #if defined(FFX_GPU) && defined(FFX_FSR_EASU_FLOAT) - -#if defined(FFX_GPU) && FFX_HALF == 1 && defined(FFX_FSR_EASU_HALF) -// Input callback prototypes, need to be implemented by calling shader -FfxFloat16x4 FsrEasuRH(FfxFloat32x2 p); -FfxFloat16x4 FsrEasuGH(FfxFloat32x2 p); -FfxFloat16x4 FsrEasuBH(FfxFloat32x2 p); - -// This runs 2 taps in parallel. -void FsrEasuTapH( - FFX_PARAMETER_INOUT FfxFloat16x2 aCR, - FFX_PARAMETER_INOUT FfxFloat16x2 aCG, - FFX_PARAMETER_INOUT FfxFloat16x2 aCB, - FFX_PARAMETER_INOUT FfxFloat16x2 aW, - FFX_PARAMETER_IN FfxFloat16x2 offX, - FFX_PARAMETER_IN FfxFloat16x2 offY, - FFX_PARAMETER_IN FfxFloat16x2 dir, - FFX_PARAMETER_IN FfxFloat16x2 len, - FFX_PARAMETER_IN FfxFloat16 lob, - FFX_PARAMETER_IN FfxFloat16 clp, - FFX_PARAMETER_IN FfxFloat16x2 cR, - FFX_PARAMETER_IN FfxFloat16x2 cG, - FFX_PARAMETER_IN FfxFloat16x2 cB) -{ - FfxFloat16x2 vX, vY; - vX = offX * dir.xx + offY * dir.yy; - vY = offX * (-dir.yy) + offY * dir.xx; - vX *= len.x; - vY *= len.y; - FfxFloat16x2 d2 = vX * vX + vY * vY; - d2 = min(d2, FFX_BROADCAST_FLOAT16X2(clp)); - FfxFloat16x2 wB = FFX_BROADCAST_FLOAT16X2(2.0 / 5.0) * d2 + FFX_BROADCAST_FLOAT16X2(-1.0); - FfxFloat16x2 wA = FFX_BROADCAST_FLOAT16X2(lob) * d2 + FFX_BROADCAST_FLOAT16X2(-1.0); - wB *= wB; - wA *= wA; - wB = FFX_BROADCAST_FLOAT16X2(25.0 / 16.0) * wB + FFX_BROADCAST_FLOAT16X2(-(25.0 / 16.0 - 1.0)); - FfxFloat16x2 w = wB * wA; - aCR += cR * w; - aCG += cG * w; - aCB += cB * w; - aW += w; -} - -// This runs 2 taps in parallel. -void FsrEasuSetH( - FFX_PARAMETER_INOUT FfxFloat16x2 dirPX, - FFX_PARAMETER_INOUT FfxFloat16x2 dirPY, - FFX_PARAMETER_INOUT FfxFloat16x2 lenP, - FFX_PARAMETER_IN FfxFloat16x2 pp, - FFX_PARAMETER_IN FfxBoolean biST, - FFX_PARAMETER_IN FfxBoolean biUV, - FFX_PARAMETER_IN FfxFloat16x2 lA, - FFX_PARAMETER_IN FfxFloat16x2 lB, - FFX_PARAMETER_IN FfxFloat16x2 lC, - FFX_PARAMETER_IN FfxFloat16x2 lD, - FFX_PARAMETER_IN FfxFloat16x2 lE) -{ - FfxFloat16x2 w = FFX_BROADCAST_FLOAT16X2(0.0); - - if (biST) - w = (FfxFloat16x2(1.0, 0.0) + FfxFloat16x2(-pp.x, pp.x)) * FFX_BROADCAST_FLOAT16X2(FFX_BROADCAST_FLOAT16(1.0) - pp.y); - - if (biUV) - w = (FfxFloat16x2(1.0, 0.0) + FfxFloat16x2(-pp.x, pp.x)) * FFX_BROADCAST_FLOAT16X2(pp.y); - - // ABS is not free in the packed FP16 path. - FfxFloat16x2 dc = lD - lC; - FfxFloat16x2 cb = lC - lB; - FfxFloat16x2 lenX = max(abs(dc), abs(cb)); - lenX = ffxReciprocalHalf(lenX); - - FfxFloat16x2 dirX = lD - lB; - dirPX += dirX * w; - lenX = FfxFloat16x2(ffxSaturate(abs(dirX) * lenX)); - lenX *= lenX; - lenP += lenX * w; - FfxFloat16x2 ec = lE - lC; - FfxFloat16x2 ca = lC - lA; - FfxFloat16x2 lenY = max(abs(ec), abs(ca)); - lenY = ffxReciprocalHalf(lenY); - FfxFloat16x2 dirY = lE - lA; - dirPY += dirY * w; - lenY = FfxFloat16x2(ffxSaturate(abs(dirY) * lenY)); - lenY *= lenY; - lenP += lenY * w; -} - -void FsrEasuH( - FFX_PARAMETER_OUT FfxFloat16x3 pix, - FFX_PARAMETER_IN FfxUInt32x2 ip, - FFX_PARAMETER_IN FfxUInt32x4 con0, - FFX_PARAMETER_IN FfxUInt32x4 con1, - FFX_PARAMETER_IN FfxUInt32x4 con2, - FFX_PARAMETER_IN FfxUInt32x4 con3) -{ - FfxFloat32x2 pp = FfxFloat32x2(ip) * ffxAsFloat(con0.xy) + ffxAsFloat(con0.zw); - FfxFloat32x2 fp = floor(pp); - pp -= fp; - FfxFloat16x2 ppp = FfxFloat16x2(pp); - - FfxFloat32x2 p0 = fp * ffxAsFloat(con1.xy) + ffxAsFloat(con1.zw); - FfxFloat32x2 p1 = p0 + ffxAsFloat(con2.xy); - FfxFloat32x2 p2 = p0 + ffxAsFloat(con2.zw); - FfxFloat32x2 p3 = p0 + ffxAsFloat(con3.xy); - FfxFloat16x4 bczzR = FsrEasuRH(p0); - FfxFloat16x4 bczzG = FsrEasuGH(p0); - FfxFloat16x4 bczzB = FsrEasuBH(p0); - FfxFloat16x4 ijfeR = FsrEasuRH(p1); - FfxFloat16x4 ijfeG = FsrEasuGH(p1); - FfxFloat16x4 ijfeB = FsrEasuBH(p1); - FfxFloat16x4 klhgR = FsrEasuRH(p2); - FfxFloat16x4 klhgG = FsrEasuGH(p2); - FfxFloat16x4 klhgB = FsrEasuBH(p2); - FfxFloat16x4 zzonR = FsrEasuRH(p3); - FfxFloat16x4 zzonG = FsrEasuGH(p3); - FfxFloat16x4 zzonB = FsrEasuBH(p3); - - FfxFloat16x4 bczzL = bczzB * FFX_BROADCAST_FLOAT16X4(0.5) + (bczzR * FFX_BROADCAST_FLOAT16X4(0.5) + bczzG); - FfxFloat16x4 ijfeL = ijfeB * FFX_BROADCAST_FLOAT16X4(0.5) + (ijfeR * FFX_BROADCAST_FLOAT16X4(0.5) + ijfeG); - FfxFloat16x4 klhgL = klhgB * FFX_BROADCAST_FLOAT16X4(0.5) + (klhgR * FFX_BROADCAST_FLOAT16X4(0.5) + klhgG); - FfxFloat16x4 zzonL = zzonB * FFX_BROADCAST_FLOAT16X4(0.5) + (zzonR * FFX_BROADCAST_FLOAT16X4(0.5) + zzonG); - FfxFloat16 bL = bczzL.x; - FfxFloat16 cL = bczzL.y; - FfxFloat16 iL = ijfeL.x; - FfxFloat16 jL = ijfeL.y; - FfxFloat16 fL = ijfeL.z; - FfxFloat16 eL = ijfeL.w; - FfxFloat16 kL = klhgL.x; - FfxFloat16 lL = klhgL.y; - FfxFloat16 hL = klhgL.z; - FfxFloat16 gL = klhgL.w; - FfxFloat16 oL = zzonL.z; - FfxFloat16 nL = zzonL.w; - - // This part is different, accumulating 2 taps in parallel. - FfxFloat16x2 dirPX = FFX_BROADCAST_FLOAT16X2(0.0); - FfxFloat16x2 dirPY = FFX_BROADCAST_FLOAT16X2(0.0); - FfxFloat16x2 lenP = FFX_BROADCAST_FLOAT16X2(0.0); - FsrEasuSetH(dirPX, - dirPY, - lenP, - ppp, - FfxBoolean(true), - FfxBoolean(false), - FfxFloat16x2(bL, cL), - FfxFloat16x2(eL, fL), - FfxFloat16x2(fL, gL), - FfxFloat16x2(gL, hL), - FfxFloat16x2(jL, kL)); - FsrEasuSetH(dirPX, - dirPY, - lenP, - ppp, - FfxBoolean(false), - FfxBoolean(true), - FfxFloat16x2(fL, gL), - FfxFloat16x2(iL, jL), - FfxFloat16x2(jL, kL), - FfxFloat16x2(kL, lL), - FfxFloat16x2(nL, oL)); - FfxFloat16x2 dir = FfxFloat16x2(dirPX.r + dirPX.g, dirPY.r + dirPY.g); - FfxFloat16 len = lenP.r + lenP.g; - - FfxFloat16x2 dir2 = dir * dir; - FfxFloat16 dirR = dir2.x + dir2.y; - FfxUInt32 zro = FfxUInt32(dirR < FFX_BROADCAST_FLOAT16(1.0 / 32768.0)); - dirR = ffxApproximateReciprocalSquareRootHalf(dirR); - dirR = (zro > 0) ? FFX_BROADCAST_FLOAT16(1.0) : dirR; - dir.x = (zro > 0) ? FFX_BROADCAST_FLOAT16(1.0) : dir.x; - dir *= FFX_BROADCAST_FLOAT16X2(dirR); - len = len * FFX_BROADCAST_FLOAT16(0.5); - len *= len; - FfxFloat16 stretch = (dir.x * dir.x + dir.y * dir.y) * ffxApproximateReciprocalHalf(max(abs(dir.x), abs(dir.y))); - FfxFloat16x2 len2 = - FfxFloat16x2(FFX_BROADCAST_FLOAT16(1.0) + (stretch - FFX_BROADCAST_FLOAT16(1.0)) * len, FFX_BROADCAST_FLOAT16(1.0) + FFX_BROADCAST_FLOAT16(-0.5) * len); - FfxFloat16 lob = FFX_BROADCAST_FLOAT16(0.5) + FFX_BROADCAST_FLOAT16((1.0 / 4.0 - 0.04) - 0.5) * len; - FfxFloat16 clp = ffxApproximateReciprocalHalf(lob); - - // FP16 is different, using packed trick to do min and max in same operation. - FfxFloat16x2 bothR = - max(max(FfxFloat16x2(-ijfeR.z, ijfeR.z), FfxFloat16x2(-klhgR.w, klhgR.w)), max(FfxFloat16x2(-ijfeR.y, ijfeR.y), FfxFloat16x2(-klhgR.x, klhgR.x))); - FfxFloat16x2 bothG = - max(max(FfxFloat16x2(-ijfeG.z, ijfeG.z), FfxFloat16x2(-klhgG.w, klhgG.w)), max(FfxFloat16x2(-ijfeG.y, ijfeG.y), FfxFloat16x2(-klhgG.x, klhgG.x))); - FfxFloat16x2 bothB = - max(max(FfxFloat16x2(-ijfeB.z, ijfeB.z), FfxFloat16x2(-klhgB.w, klhgB.w)), max(FfxFloat16x2(-ijfeB.y, ijfeB.y), FfxFloat16x2(-klhgB.x, klhgB.x))); - - // This part is different for FP16, working pairs of taps at a time. - FfxFloat16x2 pR = FFX_BROADCAST_FLOAT16X2(0.0); - FfxFloat16x2 pG = FFX_BROADCAST_FLOAT16X2(0.0); - FfxFloat16x2 pB = FFX_BROADCAST_FLOAT16X2(0.0); - FfxFloat16x2 pW = FFX_BROADCAST_FLOAT16X2(0.0); - FsrEasuTapH(pR, pG, pB, pW, FfxFloat16x2(0.0, 1.0) - ppp.xx, FfxFloat16x2(-1.0, -1.0) - ppp.yy, dir, len2, lob, clp, bczzR.xy, bczzG.xy, bczzB.xy); - FsrEasuTapH(pR, pG, pB, pW, FfxFloat16x2(-1.0, 0.0) - ppp.xx, FfxFloat16x2(1.0, 1.0) - ppp.yy, dir, len2, lob, clp, ijfeR.xy, ijfeG.xy, ijfeB.xy); - FsrEasuTapH(pR, pG, pB, pW, FfxFloat16x2(0.0, -1.0) - ppp.xx, FfxFloat16x2(0.0, 0.0) - ppp.yy, dir, len2, lob, clp, ijfeR.zw, ijfeG.zw, ijfeB.zw); - FsrEasuTapH(pR, pG, pB, pW, FfxFloat16x2(1.0, 2.0) - ppp.xx, FfxFloat16x2(1.0, 1.0) - ppp.yy, dir, len2, lob, clp, klhgR.xy, klhgG.xy, klhgB.xy); - FsrEasuTapH(pR, pG, pB, pW, FfxFloat16x2(2.0, 1.0) - ppp.xx, FfxFloat16x2(0.0, 0.0) - ppp.yy, dir, len2, lob, clp, klhgR.zw, klhgG.zw, klhgB.zw); - FsrEasuTapH(pR, pG, pB, pW, FfxFloat16x2(1.0, 0.0) - ppp.xx, FfxFloat16x2(2.0, 2.0) - ppp.yy, dir, len2, lob, clp, zzonR.zw, zzonG.zw, zzonB.zw); - FfxFloat16x3 aC = FfxFloat16x3(pR.x + pR.y, pG.x + pG.y, pB.x + pB.y); - FfxFloat16 aW = pW.x + pW.y; - - // Slightly different for FP16 version due to combined min and max. - pix = min(FfxFloat16x3(bothR.y, bothG.y, bothB.y), max(-FfxFloat16x3(bothR.x, bothG.x, bothB.x), aC * FFX_BROADCAST_FLOAT16X3(ffxReciprocalHalf(aW)))); -} -#endif // #if defined(FFX_GPU) && defined(FFX_HALF) && defined(FFX_FSR_EASU_HALF) - -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//_____________________________________________________________/\_______________________________________________________________ -//============================================================================================================================== -// -// FSR - [RCAS] ROBUST CONTRAST ADAPTIVE SHARPENING -// -//------------------------------------------------------------------------------------------------------------------------------ -// CAS uses a simplified mechanism to convert local contrast into a variable amount of sharpness. -// RCAS uses a more exact mechanism, solving for the maximum local sharpness possible before clipping. -// RCAS also has a built in process to limit sharpening of what it detects as possible noise. -// RCAS sharper does not support scaling, as it should be applied after EASU scaling. -// Pass EASU output straight into RCAS, no color conversions necessary. -//------------------------------------------------------------------------------------------------------------------------------ -// RCAS is based on the following logic. -// RCAS uses a 5 tap filter in a cross pattern (same as CAS), -// w n -// w 1 w for taps w m e -// w s -// Where 'w' is the negative lobe weight. -// output = (w*(n+e+w+s)+m)/(4*w+1) -// RCAS solves for 'w' by seeing where the signal might clip out of the {0 to 1} input range, -// 0 == (w*(n+e+w+s)+m)/(4*w+1) -> w = -m/(n+e+w+s) -// 1 == (w*(n+e+w+s)+m)/(4*w+1) -> w = (1-m)/(n+e+w+s-4*1) -// Then chooses the 'w' which results in no clipping, limits 'w', and multiplies by the 'sharp' amount. -// This solution above has issues with MSAA input as the steps along the gradient cause edge detection issues. -// So RCAS uses 4x the maximum and 4x the minimum (depending on equation)in place of the individual taps. -// As well as switching from 'm' to either the minimum or maximum (depending on side), to help in energy conservation. -// This stabilizes RCAS. -// RCAS does a simple highpass which is normalized against the local contrast then shaped, -// 0.25 -// 0.25 -1 0.25 -// 0.25 -// This is used as a noise detection filter, to reduce the effect of RCAS on grain, and focus on real edges. -// -// GLSL example for the required callbacks : -// -// FfxFloat16x4 FsrRcasLoadH(FfxInt16x2 p){return FfxFloat16x4(imageLoad(imgSrc,FfxInt32x2(p)));} -// void FsrRcasInputH(inout FfxFloat16 r,inout FfxFloat16 g,inout FfxFloat16 b) -// { -// //do any simple input color conversions here or leave empty if none needed -// } -// -// FsrRcasCon need to be called from the CPU or GPU to set up constants. -// Including a GPU example here, the 'con' value would be stored out to a constant buffer. -// -// FfxUInt32x4 con; -// FsrRcasCon(con, -// 0.0); // The scale is {0.0 := maximum sharpness, to N>0, where N is the number of stops (halving) of the reduction of sharpness}. -// --------------- -// RCAS sharpening supports a CAS-like pass-through alpha via, -// #define FSR_RCAS_PASSTHROUGH_ALPHA 1 -// RCAS also supports a define to enable a more expensive path to avoid some sharpening of noise. -// Would suggest it is better to apply film grain after RCAS sharpening (and after scaling) instead of using this define, -// #define FSR_RCAS_DENOISE 1 -//============================================================================================================================== -// This is set at the limit of providing unnatural results for sharpening. -#define FSR_RCAS_LIMIT (0.25-(1.0/16.0)) -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//_____________________________________________________________/\_______________________________________________________________ -//============================================================================================================================== -// CONSTANT SETUP -//============================================================================================================================== -// Call to setup required constant values (works on CPU or GPU). - FFX_STATIC void FsrRcasCon(FfxUInt32x4 con, - // The scale is {0.0 := maximum, to N>0, where N is the number of stops (halving) of the reduction of sharpness}. - FfxFloat32 sharpness) - { - // Transform from stops to linear value. - sharpness = exp2(-sharpness); - FfxFloat32x2 hSharp = {sharpness, sharpness}; - con[0] = ffxAsUInt32(sharpness); - con[1] = packHalf2x16(hSharp); - con[2] = 0; - con[3] = 0; - } - //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//_____________________________________________________________/\_______________________________________________________________ -//============================================================================================================================== -// NON-PACKED 32-BIT VERSION -//============================================================================================================================== -#if defined(FFX_GPU)&&defined(FSR_RCAS_F) - // Input callback prototypes that need to be implemented by calling shader - FfxFloat32x4 FsrRcasLoadF(FfxInt32x2 p); - void FsrRcasInputF(inout FfxFloat32 r,inout FfxFloat32 g,inout FfxFloat32 b); -//------------------------------------------------------------------------------------------------------------------------------ - void FsrRcasF(out FfxFloat32 pixR, // Output values, non-vector so port between RcasFilter() and RcasFilterH() is easy. - out FfxFloat32 pixG, - out FfxFloat32 pixB, -#ifdef FSR_RCAS_PASSTHROUGH_ALPHA - out FfxFloat32 pixA, -#endif - FfxUInt32x2 ip, // Integer pixel position in output. - FfxUInt32x4 con) - { // Constant generated by RcasSetup(). - // Algorithm uses minimal 3x3 pixel neighborhood. - // b - // d e f - // h - FfxInt32x2 sp = FfxInt32x2(ip); - FfxFloat32x3 b = FsrRcasLoadF(sp + FfxInt32x2(0, -1)).rgb; - FfxFloat32x3 d = FsrRcasLoadF(sp + FfxInt32x2(-1, 0)).rgb; -#ifdef FSR_RCAS_PASSTHROUGH_ALPHA - FfxFloat32x4 ee = FsrRcasLoadF(sp); - FfxFloat32x3 e = ee.rgb; - pixA = ee.a; -#else - FfxFloat32x3 e = FsrRcasLoadF(sp).rgb; -#endif - FfxFloat32x3 f = FsrRcasLoadF(sp + FfxInt32x2(1, 0)).rgb; - FfxFloat32x3 h = FsrRcasLoadF(sp + FfxInt32x2(0, 1)).rgb; - // Rename (32-bit) or regroup (16-bit). - FfxFloat32 bR = b.r; - FfxFloat32 bG = b.g; - FfxFloat32 bB = b.b; - FfxFloat32 dR = d.r; - FfxFloat32 dG = d.g; - FfxFloat32 dB = d.b; - FfxFloat32 eR = e.r; - FfxFloat32 eG = e.g; - FfxFloat32 eB = e.b; - FfxFloat32 fR = f.r; - FfxFloat32 fG = f.g; - FfxFloat32 fB = f.b; - FfxFloat32 hR = h.r; - FfxFloat32 hG = h.g; - FfxFloat32 hB = h.b; - // Run optional input transform. - FsrRcasInputF(bR, bG, bB); - FsrRcasInputF(dR, dG, dB); - FsrRcasInputF(eR, eG, eB); - FsrRcasInputF(fR, fG, fB); - FsrRcasInputF(hR, hG, hB); - // Luma times 2. - FfxFloat32 bL = bB * FfxFloat32(0.5) + (bR * FfxFloat32(0.5) + bG); - FfxFloat32 dL = dB * FfxFloat32(0.5) + (dR * FfxFloat32(0.5) + dG); - FfxFloat32 eL = eB * FfxFloat32(0.5) + (eR * FfxFloat32(0.5) + eG); - FfxFloat32 fL = fB * FfxFloat32(0.5) + (fR * FfxFloat32(0.5) + fG); - FfxFloat32 hL = hB * FfxFloat32(0.5) + (hR * FfxFloat32(0.5) + hG); - // Noise detection. - FfxFloat32 nz = FfxFloat32(0.25) * bL + FfxFloat32(0.25) * dL + FfxFloat32(0.25) * fL + FfxFloat32(0.25) * hL - eL; - nz = ffxSaturate(abs(nz) * ffxApproximateReciprocalMedium(ffxMax3(ffxMax3(bL, dL, eL), fL, hL) - ffxMin3(ffxMin3(bL, dL, eL), fL, hL))); - nz = FfxFloat32(-0.5) * nz + FfxFloat32(1.0); - // Min and max of ring. - FfxFloat32 mn4R = ffxMin(ffxMin3(bR, dR, fR), hR); - FfxFloat32 mn4G = ffxMin(ffxMin3(bG, dG, fG), hG); - FfxFloat32 mn4B = ffxMin(ffxMin3(bB, dB, fB), hB); - FfxFloat32 mx4R = max(ffxMax3(bR, dR, fR), hR); - FfxFloat32 mx4G = max(ffxMax3(bG, dG, fG), hG); - FfxFloat32 mx4B = max(ffxMax3(bB, dB, fB), hB); - // Immediate constants for peak range. - FfxFloat32x2 peakC = FfxFloat32x2(1.0, -1.0 * 4.0); - // Limiters, these need to be high precision RCPs. - FfxFloat32 hitMinR = mn4R * rcp(FfxFloat32(4.0) * mx4R); - FfxFloat32 hitMinG = mn4G * rcp(FfxFloat32(4.0) * mx4G); - FfxFloat32 hitMinB = mn4B * rcp(FfxFloat32(4.0) * mx4B); - FfxFloat32 hitMaxR = (peakC.x - mx4R) * rcp(FfxFloat32(4.0) * mn4R + peakC.y); - FfxFloat32 hitMaxG = (peakC.x - mx4G) * rcp(FfxFloat32(4.0) * mn4G + peakC.y); - FfxFloat32 hitMaxB = (peakC.x - mx4B) * rcp(FfxFloat32(4.0) * mn4B + peakC.y); - FfxFloat32 lobeR = max(-hitMinR, hitMaxR); - FfxFloat32 lobeG = max(-hitMinG, hitMaxG); - FfxFloat32 lobeB = max(-hitMinB, hitMaxB); - FfxFloat32 lobe = max(FfxFloat32(-FSR_RCAS_LIMIT), ffxMin(ffxMax3(lobeR, lobeG, lobeB), FfxFloat32(0.0))) * ffxAsFloat - (con.x); - // Apply noise removal. -#ifdef FSR_RCAS_DENOISE - lobe *= nz; -#endif - // Resolve, which needs the medium precision rcp approximation to avoid visible tonality changes. - FfxFloat32 rcpL = ffxApproximateReciprocalMedium(FfxFloat32(4.0) * lobe + FfxFloat32(1.0)); - pixR = (lobe * bR + lobe * dR + lobe * hR + lobe * fR + eR) * rcpL; - pixG = (lobe * bG + lobe * dG + lobe * hG + lobe * fG + eG) * rcpL; - pixB = (lobe * bB + lobe * dB + lobe * hB + lobe * fB + eB) * rcpL; - return; - } -#endif -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//_____________________________________________________________/\_______________________________________________________________ -//============================================================================================================================== -// NON-PACKED 16-BIT VERSION -//============================================================================================================================== -#if defined(FFX_GPU) && FFX_HALF == 1 && defined(FSR_RCAS_H) - // Input callback prototypes that need to be implemented by calling shader - FfxFloat16x4 FsrRcasLoadH(FfxInt16x2 p); - void FsrRcasInputH(inout FfxFloat16 r,inout FfxFloat16 g,inout FfxFloat16 b); -//------------------------------------------------------------------------------------------------------------------------------ - void FsrRcasH( - out FfxFloat16 pixR, // Output values, non-vector so port between RcasFilter() and RcasFilterH() is easy. - out FfxFloat16 pixG, - out FfxFloat16 pixB, - #ifdef FSR_RCAS_PASSTHROUGH_ALPHA - out FfxFloat16 pixA, - #endif - FfxUInt32x2 ip, // Integer pixel position in output. - FfxUInt32x4 con){ // Constant generated by RcasSetup(). - // Sharpening algorithm uses minimal 3x3 pixel neighborhood. - // b - // d e f - // h - FfxInt16x2 sp=FfxInt16x2(ip); - FfxFloat16x3 b=FsrRcasLoadH(sp+FfxInt16x2( 0,-1)).rgb; - FfxFloat16x3 d=FsrRcasLoadH(sp+FfxInt16x2(-1, 0)).rgb; - #ifdef FSR_RCAS_PASSTHROUGH_ALPHA - FfxFloat16x4 ee=FsrRcasLoadH(sp); - FfxFloat16x3 e=ee.rgb;pixA=ee.a; - #else - FfxFloat16x3 e=FsrRcasLoadH(sp).rgb; - #endif - FfxFloat16x3 f=FsrRcasLoadH(sp+FfxInt16x2( 1, 0)).rgb; - FfxFloat16x3 h=FsrRcasLoadH(sp+FfxInt16x2( 0, 1)).rgb; - // Rename (32-bit) or regroup (16-bit). - FfxFloat16 bR=b.r; - FfxFloat16 bG=b.g; - FfxFloat16 bB=b.b; - FfxFloat16 dR=d.r; - FfxFloat16 dG=d.g; - FfxFloat16 dB=d.b; - FfxFloat16 eR=e.r; - FfxFloat16 eG=e.g; - FfxFloat16 eB=e.b; - FfxFloat16 fR=f.r; - FfxFloat16 fG=f.g; - FfxFloat16 fB=f.b; - FfxFloat16 hR=h.r; - FfxFloat16 hG=h.g; - FfxFloat16 hB=h.b; - // Run optional input transform. - FsrRcasInputH(bR,bG,bB); - FsrRcasInputH(dR,dG,dB); - FsrRcasInputH(eR,eG,eB); - FsrRcasInputH(fR,fG,fB); - FsrRcasInputH(hR,hG,hB); - // Luma times 2. - FfxFloat16 bL=bB*FFX_BROADCAST_FLOAT16(0.5)+(bR*FFX_BROADCAST_FLOAT16(0.5)+bG); - FfxFloat16 dL=dB*FFX_BROADCAST_FLOAT16(0.5)+(dR*FFX_BROADCAST_FLOAT16(0.5)+dG); - FfxFloat16 eL=eB*FFX_BROADCAST_FLOAT16(0.5)+(eR*FFX_BROADCAST_FLOAT16(0.5)+eG); - FfxFloat16 fL=fB*FFX_BROADCAST_FLOAT16(0.5)+(fR*FFX_BROADCAST_FLOAT16(0.5)+fG); - FfxFloat16 hL=hB*FFX_BROADCAST_FLOAT16(0.5)+(hR*FFX_BROADCAST_FLOAT16(0.5)+hG); - // Noise detection. - FfxFloat16 nz=FFX_BROADCAST_FLOAT16(0.25)*bL+FFX_BROADCAST_FLOAT16(0.25)*dL+FFX_BROADCAST_FLOAT16(0.25)*fL+FFX_BROADCAST_FLOAT16(0.25)*hL-eL; - nz=FfxFloat16(ffxSaturate(abs(nz)*ffxApproximateReciprocalMediumHalf(ffxMax3Half(ffxMax3Half(bL,dL,eL),fL,hL)-ffxMin3Half(ffxMin3Half(bL,dL,eL),fL,hL)))); - nz=FFX_BROADCAST_FLOAT16(-0.5)*nz+FFX_BROADCAST_FLOAT16(1.0); - // Min and max of ring. - FfxFloat16 mn4R=min(ffxMin3Half(bR,dR,fR),hR); - FfxFloat16 mn4G=min(ffxMin3Half(bG,dG,fG),hG); - FfxFloat16 mn4B=min(ffxMin3Half(bB,dB,fB),hB); - FfxFloat16 mx4R=max(ffxMax3Half(bR,dR,fR),hR); - FfxFloat16 mx4G=max(ffxMax3Half(bG,dG,fG),hG); - FfxFloat16 mx4B=max(ffxMax3Half(bB,dB,fB),hB); - // Immediate constants for peak range. - FfxFloat16x2 peakC=FfxFloat16x2(1.0,-1.0*4.0); - // Limiters, these need to be high precision RCPs. - FfxFloat16 hitMinR=mn4R*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16(4.0)*mx4R); - FfxFloat16 hitMinG=mn4G*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16(4.0)*mx4G); - FfxFloat16 hitMinB=mn4B*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16(4.0)*mx4B); - FfxFloat16 hitMaxR=(peakC.x-mx4R)*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16(4.0)*mn4R+peakC.y); - FfxFloat16 hitMaxG=(peakC.x-mx4G)*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16(4.0)*mn4G+peakC.y); - FfxFloat16 hitMaxB=(peakC.x-mx4B)*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16(4.0)*mn4B+peakC.y); - FfxFloat16 lobeR=max(-hitMinR,hitMaxR); - FfxFloat16 lobeG=max(-hitMinG,hitMaxG); - FfxFloat16 lobeB=max(-hitMinB,hitMaxB); - FfxFloat16 lobe=max(FFX_BROADCAST_FLOAT16(-FSR_RCAS_LIMIT),min(ffxMax3Half(lobeR,lobeG,lobeB),FFX_BROADCAST_FLOAT16(0.0)))*FFX_UINT32_TO_FLOAT16X2(con.y).x; - // Apply noise removal. - #ifdef FSR_RCAS_DENOISE - lobe*=nz; - #endif - // Resolve, which needs the medium precision rcp approximation to avoid visible tonality changes. - FfxFloat16 rcpL=ffxApproximateReciprocalMediumHalf(FFX_BROADCAST_FLOAT16(4.0)*lobe+FFX_BROADCAST_FLOAT16(1.0)); - pixR=(lobe*bR+lobe*dR+lobe*hR+lobe*fR+eR)*rcpL; - pixG=(lobe*bG+lobe*dG+lobe*hG+lobe*fG+eG)*rcpL; - pixB=(lobe*bB+lobe*dB+lobe*hB+lobe*fB+eB)*rcpL; -} -#endif -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//_____________________________________________________________/\_______________________________________________________________ -//============================================================================================================================== -// PACKED 16-BIT VERSION -//============================================================================================================================== -#if defined(FFX_GPU)&& FFX_HALF == 1 && defined(FSR_RCAS_HX2) - // Input callback prototypes that need to be implemented by the calling shader - FfxFloat16x4 FsrRcasLoadHx2(FfxInt16x2 p); - void FsrRcasInputHx2(inout FfxFloat16x2 r,inout FfxFloat16x2 g,inout FfxFloat16x2 b); -//------------------------------------------------------------------------------------------------------------------------------ - // Can be used to convert from packed Structures of Arrays to Arrays of Structures for store. - void FsrRcasDepackHx2(out FfxFloat16x4 pix0,out FfxFloat16x4 pix1,FfxFloat16x2 pixR,FfxFloat16x2 pixG,FfxFloat16x2 pixB){ - #ifdef FFX_HLSL - // Invoke a slower path for DX only, since it won't allow uninitialized values. - pix0.a=pix1.a=0.0; - #endif - pix0.rgb=FfxFloat16x3(pixR.x,pixG.x,pixB.x); - pix1.rgb=FfxFloat16x3(pixR.y,pixG.y,pixB.y);} -//------------------------------------------------------------------------------------------------------------------------------ - void FsrRcasHx2( - // Output values are for 2 8x8 tiles in a 16x8 region. - // pix.x = left 8x8 tile - // pix.y = right 8x8 tile - // This enables later processing to easily be packed as well. - out FfxFloat16x2 pixR, - out FfxFloat16x2 pixG, - out FfxFloat16x2 pixB, - #ifdef FSR_RCAS_PASSTHROUGH_ALPHA - out FfxFloat16x2 pixA, - #endif - FfxUInt32x2 ip, // Integer pixel position in output. - FfxUInt32x4 con){ // Constant generated by RcasSetup(). - // No scaling algorithm uses minimal 3x3 pixel neighborhood. - FfxInt16x2 sp0=FfxInt16x2(ip); - FfxFloat16x3 b0=FsrRcasLoadHx2(sp0+FfxInt16x2( 0,-1)).rgb; - FfxFloat16x3 d0=FsrRcasLoadHx2(sp0+FfxInt16x2(-1, 0)).rgb; - #ifdef FSR_RCAS_PASSTHROUGH_ALPHA - FfxFloat16x4 ee0=FsrRcasLoadHx2(sp0); - FfxFloat16x3 e0=ee0.rgb;pixA.r=ee0.a; - #else - FfxFloat16x3 e0=FsrRcasLoadHx2(sp0).rgb; - #endif - FfxFloat16x3 f0=FsrRcasLoadHx2(sp0+FfxInt16x2( 1, 0)).rgb; - FfxFloat16x3 h0=FsrRcasLoadHx2(sp0+FfxInt16x2( 0, 1)).rgb; - FfxInt16x2 sp1=sp0+FfxInt16x2(8,0); - FfxFloat16x3 b1=FsrRcasLoadHx2(sp1+FfxInt16x2( 0,-1)).rgb; - FfxFloat16x3 d1=FsrRcasLoadHx2(sp1+FfxInt16x2(-1, 0)).rgb; - #ifdef FSR_RCAS_PASSTHROUGH_ALPHA - FfxFloat16x4 ee1=FsrRcasLoadHx2(sp1); - FfxFloat16x3 e1=ee1.rgb;pixA.g=ee1.a; - #else - FfxFloat16x3 e1=FsrRcasLoadHx2(sp1).rgb; - #endif - FfxFloat16x3 f1=FsrRcasLoadHx2(sp1+FfxInt16x2( 1, 0)).rgb; - FfxFloat16x3 h1=FsrRcasLoadHx2(sp1+FfxInt16x2( 0, 1)).rgb; - // Arrays of Structures to Structures of Arrays conversion. - FfxFloat16x2 bR=FfxFloat16x2(b0.r,b1.r); - FfxFloat16x2 bG=FfxFloat16x2(b0.g,b1.g); - FfxFloat16x2 bB=FfxFloat16x2(b0.b,b1.b); - FfxFloat16x2 dR=FfxFloat16x2(d0.r,d1.r); - FfxFloat16x2 dG=FfxFloat16x2(d0.g,d1.g); - FfxFloat16x2 dB=FfxFloat16x2(d0.b,d1.b); - FfxFloat16x2 eR=FfxFloat16x2(e0.r,e1.r); - FfxFloat16x2 eG=FfxFloat16x2(e0.g,e1.g); - FfxFloat16x2 eB=FfxFloat16x2(e0.b,e1.b); - FfxFloat16x2 fR=FfxFloat16x2(f0.r,f1.r); - FfxFloat16x2 fG=FfxFloat16x2(f0.g,f1.g); - FfxFloat16x2 fB=FfxFloat16x2(f0.b,f1.b); - FfxFloat16x2 hR=FfxFloat16x2(h0.r,h1.r); - FfxFloat16x2 hG=FfxFloat16x2(h0.g,h1.g); - FfxFloat16x2 hB=FfxFloat16x2(h0.b,h1.b); - // Run optional input transform. - FsrRcasInputHx2(bR,bG,bB); - FsrRcasInputHx2(dR,dG,dB); - FsrRcasInputHx2(eR,eG,eB); - FsrRcasInputHx2(fR,fG,fB); - FsrRcasInputHx2(hR,hG,hB); - // Luma times 2. - FfxFloat16x2 bL=bB*FFX_BROADCAST_FLOAT16X2(0.5)+(bR*FFX_BROADCAST_FLOAT16X2(0.5)+bG); - FfxFloat16x2 dL=dB*FFX_BROADCAST_FLOAT16X2(0.5)+(dR*FFX_BROADCAST_FLOAT16X2(0.5)+dG); - FfxFloat16x2 eL=eB*FFX_BROADCAST_FLOAT16X2(0.5)+(eR*FFX_BROADCAST_FLOAT16X2(0.5)+eG); - FfxFloat16x2 fL=fB*FFX_BROADCAST_FLOAT16X2(0.5)+(fR*FFX_BROADCAST_FLOAT16X2(0.5)+fG); - FfxFloat16x2 hL=hB*FFX_BROADCAST_FLOAT16X2(0.5)+(hR*FFX_BROADCAST_FLOAT16X2(0.5)+hG); - // Noise detection. - FfxFloat16x2 nz=FFX_BROADCAST_FLOAT16X2(0.25)*bL+FFX_BROADCAST_FLOAT16X2(0.25)*dL+FFX_BROADCAST_FLOAT16X2(0.25)*fL+FFX_BROADCAST_FLOAT16X2(0.25)*hL-eL; - nz=ffxSaturate(abs(nz)*ffxApproximateReciprocalMediumHalf(ffxMax3Half(ffxMax3Half(bL,dL,eL),fL,hL)-ffxMin3Half(ffxMin3Half(bL,dL,eL),fL,hL))); - nz=FFX_BROADCAST_FLOAT16X2(-0.5)*nz+FFX_BROADCAST_FLOAT16X2(1.0); - // Min and max of ring. - FfxFloat16x2 mn4R=min(ffxMin3Half(bR,dR,fR),hR); - FfxFloat16x2 mn4G=min(ffxMin3Half(bG,dG,fG),hG); - FfxFloat16x2 mn4B=min(ffxMin3Half(bB,dB,fB),hB); - FfxFloat16x2 mx4R=max(ffxMax3Half(bR,dR,fR),hR); - FfxFloat16x2 mx4G=max(ffxMax3Half(bG,dG,fG),hG); - FfxFloat16x2 mx4B=max(ffxMax3Half(bB,dB,fB),hB); - // Immediate constants for peak range. - FfxFloat16x2 peakC=FfxFloat16x2(1.0,-1.0*4.0); - // Limiters, these need to be high precision RCPs. - FfxFloat16x2 hitMinR=mn4R*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16X2(4.0)*mx4R); - FfxFloat16x2 hitMinG=mn4G*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16X2(4.0)*mx4G); - FfxFloat16x2 hitMinB=mn4B*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16X2(4.0)*mx4B); - FfxFloat16x2 hitMaxR=(peakC.x-mx4R)*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16X2(4.0)*mn4R+peakC.y); - FfxFloat16x2 hitMaxG=(peakC.x-mx4G)*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16X2(4.0)*mn4G+peakC.y); - FfxFloat16x2 hitMaxB=(peakC.x-mx4B)*ffxReciprocalHalf(FFX_BROADCAST_FLOAT16X2(4.0)*mn4B+peakC.y); - FfxFloat16x2 lobeR=max(-hitMinR,hitMaxR); - FfxFloat16x2 lobeG=max(-hitMinG,hitMaxG); - FfxFloat16x2 lobeB=max(-hitMinB,hitMaxB); - FfxFloat16x2 lobe=max(FFX_BROADCAST_FLOAT16X2(-FSR_RCAS_LIMIT),min(ffxMax3Half(lobeR,lobeG,lobeB),FFX_BROADCAST_FLOAT16X2(0.0)))*FFX_BROADCAST_FLOAT16X2(FFX_UINT32_TO_FLOAT16X2(con.y).x); - // Apply noise removal. - #ifdef FSR_RCAS_DENOISE - lobe*=nz; - #endif - // Resolve, which needs the medium precision rcp approximation to avoid visible tonality changes. - FfxFloat16x2 rcpL=ffxApproximateReciprocalMediumHalf(FFX_BROADCAST_FLOAT16X2(4.0)*lobe+FFX_BROADCAST_FLOAT16X2(1.0)); - pixR=(lobe*bR+lobe*dR+lobe*hR+lobe*fR+eR)*rcpL; - pixG=(lobe*bG+lobe*dG+lobe*hG+lobe*fG+eG)*rcpL; - pixB=(lobe*bB+lobe*dB+lobe*hB+lobe*fB+eB)*rcpL;} -#endif -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//_____________________________________________________________/\_______________________________________________________________ -//============================================================================================================================== -// -// FSR - [LFGA] LINEAR FILM GRAIN APPLICATOR -// -//------------------------------------------------------------------------------------------------------------------------------ -// Adding output-resolution film grain after scaling is a good way to mask both rendering and scaling artifacts. -// Suggest using tiled blue noise as film grain input, with peak noise frequency set for a specific look and feel. -// The 'Lfga*()' functions provide a convenient way to introduce grain. -// These functions limit grain based on distance to signal limits. -// This is done so that the grain is temporally energy preserving, and thus won't modify image tonality. -// Grain application should be done in a linear colorspace. -// The grain should be temporally changing, but have a temporal sum per pixel that adds to zero (non-biased). -//------------------------------------------------------------------------------------------------------------------------------ -// Usage, -// FsrLfga*( -// color, // In/out linear colorspace color {0 to 1} ranged. -// grain, // Per pixel grain texture value {-0.5 to 0.5} ranged, input is 3-channel to support colored grain. -// amount); // Amount of grain (0 to 1} ranged. -//------------------------------------------------------------------------------------------------------------------------------ -// Example if grain texture is monochrome: 'FsrLfgaF(color,ffxBroadcast3(grain),amount)' -//============================================================================================================================== -#if defined(FFX_GPU) - // Maximum grain is the minimum distance to the signal limit. - void FsrLfgaF(inout FfxFloat32x3 c, FfxFloat32x3 t, FfxFloat32 a) - { - c += (t * ffxBroadcast3(a)) * ffxMin(ffxBroadcast3(1.0) - c, c); - } -#endif -//============================================================================================================================== -#if defined(FFX_GPU)&& FFX_HALF == 1 - // Half precision version (slower). - void FsrLfgaH(inout FfxFloat16x3 c, FfxFloat16x3 t, FfxFloat16 a) - { - c += (t * FFX_BROADCAST_FLOAT16X3(a)) * min(FFX_BROADCAST_FLOAT16X3(1.0) - c, c); - } - //------------------------------------------------------------------------------------------------------------------------------ - // Packed half precision version (faster). - void FsrLfgaHx2(inout FfxFloat16x2 cR,inout FfxFloat16x2 cG,inout FfxFloat16x2 cB,FfxFloat16x2 tR,FfxFloat16x2 tG,FfxFloat16x2 tB,FfxFloat16 a){ - cR+=(tR*FFX_BROADCAST_FLOAT16X2(a))*min(FFX_BROADCAST_FLOAT16X2(1.0)-cR,cR);cG+=(tG*FFX_BROADCAST_FLOAT16X2(a))*min(FFX_BROADCAST_FLOAT16X2(1.0)-cG,cG);cB+=(tB*FFX_BROADCAST_FLOAT16X2(a))*min(FFX_BROADCAST_FLOAT16X2(1.0)-cB,cB);} -#endif -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//_____________________________________________________________/\_______________________________________________________________ -//============================================================================================================================== -// -// FSR - [SRTM] SIMPLE REVERSIBLE TONE-MAPPER -// -//------------------------------------------------------------------------------------------------------------------------------ -// This provides a way to take linear HDR color {0 to FP16_MAX} and convert it into a temporary {0 to 1} ranged post-tonemapped linear. -// The tonemapper preserves RGB ratio, which helps maintain HDR color bleed during filtering. -//------------------------------------------------------------------------------------------------------------------------------ -// Reversible tonemapper usage, -// FsrSrtm*(color); // {0 to FP16_MAX} converted to {0 to 1}. -// FsrSrtmInv*(color); // {0 to 1} converted into {0 to 32768, output peak safe for FP16}. -//============================================================================================================================== -#if defined(FFX_GPU) - void FsrSrtmF(inout FfxFloat32x3 c) - { - c *= ffxBroadcast3(rcp(ffxMax3(c.r, c.g, c.b) + FfxFloat32(1.0))); - } - // The extra max solves the c=1.0 case (which is a /0). - void FsrSrtmInvF(inout FfxFloat32x3 c){c*=ffxBroadcast3(rcp(max(FfxFloat32(1.0/32768.0),FfxFloat32(1.0)-ffxMax3(c.r,c.g,c.b))));} -#endif -//============================================================================================================================== -#if defined(FFX_GPU )&& FFX_HALF == 1 - void FsrSrtmH(inout FfxFloat16x3 c) - { - c *= FFX_BROADCAST_FLOAT16X3(ffxReciprocalHalf(ffxMax3Half(c.r, c.g, c.b) + FFX_BROADCAST_FLOAT16(1.0))); - } - void FsrSrtmInvH(inout FfxFloat16x3 c) - { - c *= FFX_BROADCAST_FLOAT16X3(ffxReciprocalHalf(max(FFX_BROADCAST_FLOAT16(1.0 / 32768.0), FFX_BROADCAST_FLOAT16(1.0) - ffxMax3Half(c.r, c.g, c.b)))); - } - //------------------------------------------------------------------------------------------------------------------------------ - void FsrSrtmHx2(inout FfxFloat16x2 cR, inout FfxFloat16x2 cG, inout FfxFloat16x2 cB) - { - FfxFloat16x2 rcp = ffxReciprocalHalf(ffxMax3Half(cR, cG, cB) + FFX_BROADCAST_FLOAT16X2(1.0)); - cR *= rcp; - cG *= rcp; - cB *= rcp; - } - void FsrSrtmInvHx2(inout FfxFloat16x2 cR,inout FfxFloat16x2 cG,inout FfxFloat16x2 cB) - { - FfxFloat16x2 rcp=ffxReciprocalHalf(max(FFX_BROADCAST_FLOAT16X2(1.0/32768.0),FFX_BROADCAST_FLOAT16X2(1.0)-ffxMax3Half(cR,cG,cB))); - cR*=rcp; - cG*=rcp; - cB*=rcp; - } -#endif -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//_____________________________________________________________/\_______________________________________________________________ -//============================================================================================================================== -// -// FSR - [TEPD] TEMPORAL ENERGY PRESERVING DITHER -// -//------------------------------------------------------------------------------------------------------------------------------ -// Temporally energy preserving dithered {0 to 1} linear to gamma 2.0 conversion. -// Gamma 2.0 is used so that the conversion back to linear is just to square the color. -// The conversion comes in 8-bit and 10-bit modes, designed for output to 8-bit UNORM or 10:10:10:2 respectively. -// Given good non-biased temporal blue noise as dither input, -// the output dither will temporally conserve energy. -// This is done by choosing the linear nearest step point instead of perceptual nearest. -// See code below for details. -//------------------------------------------------------------------------------------------------------------------------------ -// DX SPEC RULES FOR FLOAT->UNORM 8-BIT CONVERSION -// =============================================== -// - Output is 'FfxUInt32(floor(saturate(n)*255.0+0.5))'. -// - Thus rounding is to nearest. -// - NaN gets converted to zero. -// - INF is clamped to {0.0 to 1.0}. -//============================================================================================================================== -#if defined(FFX_GPU) - // Hand tuned integer position to dither value, with more values than simple checkerboard. - // Only 32-bit has enough precision for this compddation. - // Output is {0 to <1}. - FfxFloat32 FsrTepdDitF(FfxUInt32x2 p, FfxUInt32 f) - { - FfxFloat32 x = FfxFloat32(p.x + f); - FfxFloat32 y = FfxFloat32(p.y); - // The 1.61803 golden ratio. - FfxFloat32 a = FfxFloat32((1.0 + ffxSqrt(5.0f)) / 2.0); - // Number designed to provide a good visual pattern. - FfxFloat32 b = FfxFloat32(1.0 / 3.69); - x = x * a + (y * b); - return ffxFract(x); - } - //------------------------------------------------------------------------------------------------------------------------------ - // This version is 8-bit gamma 2.0. - // The 'c' input is {0 to 1}. - // Output is {0 to 1} ready for image store. - void FsrTepdC8F(inout FfxFloat32x3 c, FfxFloat32 dit) - { - FfxFloat32x3 n = ffxSqrt(c); - n = floor(n * ffxBroadcast3(255.0)) * ffxBroadcast3(1.0 / 255.0); - FfxFloat32x3 a = n * n; - FfxFloat32x3 b = n + ffxBroadcast3(1.0 / 255.0); - b = b * b; - // Ratio of 'a' to 'b' required to produce 'c'. - // ffxApproximateReciprocal() won't work here (at least for very high dynamic ranges). - // ffxApproximateReciprocalMedium() is an IADD,FMA,MUL. - FfxFloat32x3 r = (c - b) * ffxApproximateReciprocalMedium(a - b); - // Use the ratio as a cutoff to choose 'a' or 'b'. - // ffxIsGreaterThanZero() is a MUL. - c = ffxSaturate(n + ffxIsGreaterThanZero(ffxBroadcast3(dit) - r) * ffxBroadcast3(1.0 / 255.0)); - } - //------------------------------------------------------------------------------------------------------------------------------ - // This version is 10-bit gamma 2.0. - // The 'c' input is {0 to 1}. - // Output is {0 to 1} ready for image store. - void FsrTepdC10F(inout FfxFloat32x3 c, FfxFloat32 dit) - { - FfxFloat32x3 n = ffxSqrt(c); - n = floor(n * ffxBroadcast3(1023.0)) * ffxBroadcast3(1.0 / 1023.0); - FfxFloat32x3 a = n * n; - FfxFloat32x3 b = n + ffxBroadcast3(1.0 / 1023.0); - b = b * b; - FfxFloat32x3 r = (c - b) * ffxApproximateReciprocalMedium(a - b); - c = ffxSaturate(n + ffxIsGreaterThanZero(ffxBroadcast3(dit) - r) * ffxBroadcast3(1.0 / 1023.0)); - } -#endif -//============================================================================================================================== -#if defined(FFX_GPU)&& FFX_HALF == 1 - FfxFloat16 FsrTepdDitH(FfxUInt32x2 p, FfxUInt32 f) - { - FfxFloat32 x = FfxFloat32(p.x + f); - FfxFloat32 y = FfxFloat32(p.y); - FfxFloat32 a = FfxFloat32((1.0 + ffxSqrt(5.0f)) / 2.0); - FfxFloat32 b = FfxFloat32(1.0 / 3.69); - x = x * a + (y * b); - return FfxFloat16(ffxFract(x)); - } - //------------------------------------------------------------------------------------------------------------------------------ - void FsrTepdC8H(inout FfxFloat16x3 c, FfxFloat16 dit) - { - FfxFloat16x3 n = sqrt(c); - n = floor(n * FFX_BROADCAST_FLOAT16X3(255.0)) * FFX_BROADCAST_FLOAT16X3(1.0 / 255.0); - FfxFloat16x3 a = n * n; - FfxFloat16x3 b = n + FFX_BROADCAST_FLOAT16X3(1.0 / 255.0); - b = b * b; - FfxFloat16x3 r = (c - b) * ffxApproximateReciprocalMediumHalf(a - b); - c = FfxFloat16x3(ffxSaturate(n + ffxIsGreaterThanZeroHalf(FFX_BROADCAST_FLOAT16X3(dit) - r) * FFX_BROADCAST_FLOAT16X3(1.0 / 255.0))); - } - //------------------------------------------------------------------------------------------------------------------------------ - void FsrTepdC10H(inout FfxFloat16x3 c, FfxFloat16 dit) - { - FfxFloat16x3 n = sqrt(c); - n = floor(n * FFX_BROADCAST_FLOAT16X3(1023.0)) * FFX_BROADCAST_FLOAT16X3(1.0 / 1023.0); - FfxFloat16x3 a = n * n; - FfxFloat16x3 b = n + FFX_BROADCAST_FLOAT16X3(1.0 / 1023.0); - b = b * b; - FfxFloat16x3 r = (c - b) * ffxApproximateReciprocalMediumHalf(a - b); - c = FfxFloat16x3(ffxSaturate(n + ffxIsGreaterThanZeroHalf(FFX_BROADCAST_FLOAT16X3(dit) - r) * FFX_BROADCAST_FLOAT16X3(1.0 / 1023.0))); - } - //============================================================================================================================== - // This computes dither for positions 'p' and 'p+{8,0}'. - FfxFloat16x2 FsrTepdDitHx2(FfxUInt32x2 p, FfxUInt32 f) - { - FfxFloat32x2 x; - x.x = FfxFloat32(p.x + f); - x.y = x.x + FfxFloat32(8.0); - FfxFloat32 y = FfxFloat32(p.y); - FfxFloat32 a = FfxFloat32((1.0 + ffxSqrt(5.0f)) / 2.0); - FfxFloat32 b = FfxFloat32(1.0 / 3.69); - x = x * ffxBroadcast2(a) + ffxBroadcast2(y * b); - return FfxFloat16x2(ffxFract(x)); - } - //------------------------------------------------------------------------------------------------------------------------------ - void FsrTepdC8Hx2(inout FfxFloat16x2 cR, inout FfxFloat16x2 cG, inout FfxFloat16x2 cB, FfxFloat16x2 dit) - { - FfxFloat16x2 nR = sqrt(cR); - FfxFloat16x2 nG = sqrt(cG); - FfxFloat16x2 nB = sqrt(cB); - nR = floor(nR * FFX_BROADCAST_FLOAT16X2(255.0)) * FFX_BROADCAST_FLOAT16X2(1.0 / 255.0); - nG = floor(nG * FFX_BROADCAST_FLOAT16X2(255.0)) * FFX_BROADCAST_FLOAT16X2(1.0 / 255.0); - nB = floor(nB * FFX_BROADCAST_FLOAT16X2(255.0)) * FFX_BROADCAST_FLOAT16X2(1.0 / 255.0); - FfxFloat16x2 aR = nR * nR; - FfxFloat16x2 aG = nG * nG; - FfxFloat16x2 aB = nB * nB; - FfxFloat16x2 bR = nR + FFX_BROADCAST_FLOAT16X2(1.0 / 255.0); - bR = bR * bR; - FfxFloat16x2 bG = nG + FFX_BROADCAST_FLOAT16X2(1.0 / 255.0); - bG = bG * bG; - FfxFloat16x2 bB = nB + FFX_BROADCAST_FLOAT16X2(1.0 / 255.0); - bB = bB * bB; - FfxFloat16x2 rR = (cR - bR) * ffxApproximateReciprocalMediumHalf(aR - bR); - FfxFloat16x2 rG = (cG - bG) * ffxApproximateReciprocalMediumHalf(aG - bG); - FfxFloat16x2 rB = (cB - bB) * ffxApproximateReciprocalMediumHalf(aB - bB); - cR = FfxFloat16x2(ffxSaturate(nR + ffxIsGreaterThanZeroHalf(dit - rR) * FFX_BROADCAST_FLOAT16X2(1.0 / 255.0))); - cG = FfxFloat16x2(ffxSaturate(nG + ffxIsGreaterThanZeroHalf(dit - rG) * FFX_BROADCAST_FLOAT16X2(1.0 / 255.0))); - cB = FfxFloat16x2(ffxSaturate(nB + ffxIsGreaterThanZeroHalf(dit - rB) * FFX_BROADCAST_FLOAT16X2(1.0 / 255.0))); - } - //------------------------------------------------------------------------------------------------------------------------------ - void FsrTepdC10Hx2(inout FfxFloat16x2 cR,inout FfxFloat16x2 cG,inout FfxFloat16x2 cB,FfxFloat16x2 dit){ - FfxFloat16x2 nR=sqrt(cR); - FfxFloat16x2 nG=sqrt(cG); - FfxFloat16x2 nB=sqrt(cB); - nR=floor(nR*FFX_BROADCAST_FLOAT16X2(1023.0))*FFX_BROADCAST_FLOAT16X2(1.0/1023.0); - nG=floor(nG*FFX_BROADCAST_FLOAT16X2(1023.0))*FFX_BROADCAST_FLOAT16X2(1.0/1023.0); - nB=floor(nB*FFX_BROADCAST_FLOAT16X2(1023.0))*FFX_BROADCAST_FLOAT16X2(1.0/1023.0); - FfxFloat16x2 aR=nR*nR; - FfxFloat16x2 aG=nG*nG; - FfxFloat16x2 aB=nB*nB; - FfxFloat16x2 bR=nR+FFX_BROADCAST_FLOAT16X2(1.0/1023.0);bR=bR*bR; - FfxFloat16x2 bG=nG+FFX_BROADCAST_FLOAT16X2(1.0/1023.0);bG=bG*bG; - FfxFloat16x2 bB=nB+FFX_BROADCAST_FLOAT16X2(1.0/1023.0);bB=bB*bB; - FfxFloat16x2 rR=(cR-bR)*ffxApproximateReciprocalMediumHalf(aR-bR); - FfxFloat16x2 rG=(cG-bG)*ffxApproximateReciprocalMediumHalf(aG-bG); - FfxFloat16x2 rB=(cB-bB)*ffxApproximateReciprocalMediumHalf(aB-bB); - cR=FfxFloat16x2(ffxSaturate(nR+ffxIsGreaterThanZeroHalf(dit-rR)*FFX_BROADCAST_FLOAT16X2(1.0/1023.0))); - cG=FfxFloat16x2(ffxSaturate(nG+ffxIsGreaterThanZeroHalf(dit-rG)*FFX_BROADCAST_FLOAT16X2(1.0/1023.0))); - cB=FfxFloat16x2(ffxSaturate(nB + ffxIsGreaterThanZeroHalf(dit - rB) * FFX_BROADCAST_FLOAT16X2(1.0 / 1023.0))); -} -#endif diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/fsr1/ffx_fsr1.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/fsr1/ffx_fsr1.h.meta deleted file mode 100644 index 3c97f69..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/fsr1/ffx_fsr1.h.meta +++ /dev/null @@ -1,67 +0,0 @@ -fileFormatVersion: 2 -guid: 628e23510f46ef44bbf0035ce9a63be0 -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - CPU: AnyCPU - DefaultValueInitialized: true - OS: AnyOS - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/spd.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/spd.meta deleted file mode 100644 index 0b775af..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/spd.meta +++ /dev/null @@ -1,8 +0,0 @@ -fileFormatVersion: 2 -guid: 0f03de1579ac3294595ae4f40106b7a2 -folderAsset: yes -DefaultImporter: - externalObjects: {} - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/spd/ffx_spd.h b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/spd/ffx_spd.h deleted file mode 100644 index 6441419..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/spd/ffx_spd.h +++ /dev/null @@ -1,1009 +0,0 @@ -// This file is part of the FidelityFX SDK. -// -// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved. -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// The above copyright notice and this permission notice shall be included in -// all copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN -// THE SOFTWARE. - - -/// @defgroup FfxGPUSpd FidelityFX SPD -/// FidelityFX Single Pass Downsampler 2.0 GPU documentation -/// -/// @ingroup FfxGPUEffects - -/// Setup required constant values for SPD (CPU). -/// -/// @param [out] dispatchThreadGroupCountXY CPU side: dispatch thread group count xy. z is number of slices of the input texture -/// @param [out] workGroupOffset GPU side: pass in as constant -/// @param [out] numWorkGroupsAndMips GPU side: pass in as constant -/// @param [in] rectInfo left, top, width, height -/// @param [in] mips optional: if -1, calculate based on rect width and height -/// -/// @ingroup FfxGPUSpd -#if defined(FFX_CPU) -FFX_STATIC void ffxSpdSetup(FfxUInt32x2 dispatchThreadGroupCountXY, - FfxUInt32x2 workGroupOffset, - FfxUInt32x2 numWorkGroupsAndMips, - FfxUInt32x4 rectInfo, - FfxInt32 mips) -{ - // determines the offset of the first tile to downsample based on - // left (rectInfo[0]) and top (rectInfo[1]) of the subregion. - workGroupOffset[0] = rectInfo[0] / 64; - workGroupOffset[1] = rectInfo[1] / 64; - - FfxUInt32 endIndexX = (rectInfo[0] + rectInfo[2] - 1) / 64; // rectInfo[0] = left, rectInfo[2] = width - FfxUInt32 endIndexY = (rectInfo[1] + rectInfo[3] - 1) / 64; // rectInfo[1] = top, rectInfo[3] = height - - // we only need to dispatch as many thread groups as tiles we need to downsample - // number of tiles per slice depends on the subregion to downsample - dispatchThreadGroupCountXY[0] = endIndexX + 1 - workGroupOffset[0]; - dispatchThreadGroupCountXY[1] = endIndexY + 1 - workGroupOffset[1]; - - // number of thread groups per slice - numWorkGroupsAndMips[0] = (dispatchThreadGroupCountXY[0]) * (dispatchThreadGroupCountXY[1]); - - if (mips >= 0) - { - numWorkGroupsAndMips[1] = FfxUInt32(mips); - } - else - { - // calculate based on rect width and height - FfxUInt32 resolution = ffxMax(rectInfo[2], rectInfo[3]); - numWorkGroupsAndMips[1] = FfxUInt32((ffxMin(floor(log2(FfxFloat32(resolution))), FfxFloat32(12)))); - } -} - -/// Setup required constant values for SPD (CPU). -/// -/// @param [out] dispatchThreadGroupCountXY CPU side: dispatch thread group count xy. z is number of slices of the input texture -/// @param [out] workGroupOffset GPU side: pass in as constant -/// @param [out] numWorkGroupsAndMips GPU side: pass in as constant -/// @param [in] rectInfo left, top, width, height -/// -/// @ingroup FfxGPUSpd -FFX_STATIC void ffxSpdSetup(FfxUInt32x2 dispatchThreadGroupCountXY, - FfxUInt32x2 workGroupOffset, - FfxUInt32x2 numWorkGroupsAndMips, - FfxUInt32x4 rectInfo) -{ - ffxSpdSetup(dispatchThreadGroupCountXY, workGroupOffset, numWorkGroupsAndMips, rectInfo, -1); -} -#endif // #if defined(FFX_CPU) - - -//============================================================================================================================== -// NON-PACKED VERSION -//============================================================================================================================== -#if defined(FFX_GPU) -#if defined(FFX_SPD_PACKED_ONLY) -// Avoid compiler errors by including default implementations of these callbacks. -FfxFloat32x4 SpdLoadSourceImage(FfxInt32x2 p, FfxUInt32 slice) -{ - return FfxFloat32x4(0.0, 0.0, 0.0, 0.0); -} - -FfxFloat32x4 SpdLoad(FfxInt32x2 p, FfxUInt32 slice) -{ - return FfxFloat32x4(0.0, 0.0, 0.0, 0.0); -} -void SpdStore(FfxInt32x2 p, FfxFloat32x4 value, FfxUInt32 mip, FfxUInt32 slice) -{ -} -FfxFloat32x4 SpdLoadIntermediate(FfxUInt32 x, FfxUInt32 y) -{ - return FfxFloat32x4(0.0, 0.0, 0.0, 0.0); -} -void SpdStoreIntermediate(FfxUInt32 x, FfxUInt32 y, FfxFloat32x4 value) -{ -} -FfxFloat32x4 SpdReduce4(FfxFloat32x4 v0, FfxFloat32x4 v1, FfxFloat32x4 v2, FfxFloat32x4 v3) -{ - return FfxFloat32x4(0.0, 0.0, 0.0, 0.0); -} -#endif // #if FFX_SPD_PACKED_ONLY - -//_____________________________________________________________/\_______________________________________________________________ - -void ffxSpdWorkgroupShuffleBarrier() -{ - FFX_GROUP_MEMORY_BARRIER(); -} - -// Only last active workgroup should proceed -bool SpdExitWorkgroup(FfxUInt32 numWorkGroups, FfxUInt32 localInvocationIndex, FfxUInt32 slice) -{ - // global atomic counter - if (localInvocationIndex == 0) - { - SpdIncreaseAtomicCounter(slice); - } - - ffxSpdWorkgroupShuffleBarrier(); - return (SpdGetAtomicCounter() != (numWorkGroups - 1)); -} - -// User defined: FfxFloat32x4 SpdReduce4(FfxFloat32x4 v0, FfxFloat32x4 v1, FfxFloat32x4 v2, FfxFloat32x4 v3); -FfxFloat32x4 SpdReduceQuad(FfxFloat32x4 v) -{ -#if defined(FFX_GLSL) && !defined(FFX_SPD_NO_WAVE_OPERATIONS) - - FfxFloat32x4 v0 = v; - FfxFloat32x4 v1 = subgroupQuadSwapHorizontal(v); - FfxFloat32x4 v2 = subgroupQuadSwapVertical(v); - FfxFloat32x4 v3 = subgroupQuadSwapDiagonal(v); - return SpdReduce4(v0, v1, v2, v3); - -#elif defined(FFX_HLSL) && !defined(FFX_SPD_NO_WAVE_OPERATIONS) - - // requires SM6.0 - FfxUInt32 quad = WaveGetLaneIndex() & (~0x3); - FfxFloat32x4 v0 = v; - FfxFloat32x4 v1 = WaveReadLaneAt(v, quad | 1); - FfxFloat32x4 v2 = WaveReadLaneAt(v, quad | 2); - FfxFloat32x4 v3 = WaveReadLaneAt(v, quad | 3); - return SpdReduce4(v0, v1, v2, v3); -/* - // if SM6.0 is not available, you can use the AMD shader intrinsics - // the AMD shader intrinsics are available in AMD GPU Services (AGS) library: - // https://gpuopen.com/amd-gpu-services-ags-library/ - // works for DX11 - FfxFloat32x4 v0 = v; - FfxFloat32x4 v1; - v1.x = AmdExtD3DShaderIntrinsics_SwizzleF(v.x, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); - v1.y = AmdExtD3DShaderIntrinsics_SwizzleF(v.y, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); - v1.z = AmdExtD3DShaderIntrinsics_SwizzleF(v.z, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); - v1.w = AmdExtD3DShaderIntrinsics_SwizzleF(v.w, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); - FfxFloat32x4 v2; - v2.x = AmdExtD3DShaderIntrinsics_SwizzleF(v.x, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); - v2.y = AmdExtD3DShaderIntrinsics_SwizzleF(v.y, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); - v2.z = AmdExtD3DShaderIntrinsics_SwizzleF(v.z, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); - v2.w = AmdExtD3DShaderIntrinsics_SwizzleF(v.w, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); - FfxFloat32x4 v3; - v3.x = AmdExtD3DShaderIntrinsics_SwizzleF(v.x, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); - v3.y = AmdExtD3DShaderIntrinsics_SwizzleF(v.y, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); - v3.z = AmdExtD3DShaderIntrinsics_SwizzleF(v.z, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); - v3.w = AmdExtD3DShaderIntrinsics_SwizzleF(v.w, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); - return SpdReduce4(v0, v1, v2, v3); - */ -#endif - return v; -} - -FfxFloat32x4 SpdReduceIntermediate(FfxUInt32x2 i0, FfxUInt32x2 i1, FfxUInt32x2 i2, FfxUInt32x2 i3) -{ - FfxFloat32x4 v0 = SpdLoadIntermediate(i0.x, i0.y); - FfxFloat32x4 v1 = SpdLoadIntermediate(i1.x, i1.y); - FfxFloat32x4 v2 = SpdLoadIntermediate(i2.x, i2.y); - FfxFloat32x4 v3 = SpdLoadIntermediate(i3.x, i3.y); - return SpdReduce4(v0, v1, v2, v3); -} - -FfxFloat32x4 SpdReduceLoad4(FfxUInt32x2 i0, FfxUInt32x2 i1, FfxUInt32x2 i2, FfxUInt32x2 i3, FfxUInt32 slice) -{ - FfxFloat32x4 v0 = SpdLoad(FfxInt32x2(i0), slice); - FfxFloat32x4 v1 = SpdLoad(FfxInt32x2(i1), slice); - FfxFloat32x4 v2 = SpdLoad(FfxInt32x2(i2), slice); - FfxFloat32x4 v3 = SpdLoad(FfxInt32x2(i3), slice); - return SpdReduce4(v0, v1, v2, v3); -} - -FfxFloat32x4 SpdReduceLoad4(FfxUInt32x2 base, FfxUInt32 slice) -{ - return SpdReduceLoad4(FfxUInt32x2(base + FfxUInt32x2(0, 0)), FfxUInt32x2(base + FfxUInt32x2(0, 1)), FfxUInt32x2(base + FfxUInt32x2(1, 0)), FfxUInt32x2(base + FfxUInt32x2(1, 1)), slice); -} - -FfxFloat32x4 SpdReduceLoadSourceImage4(FfxUInt32x2 i0, FfxUInt32x2 i1, FfxUInt32x2 i2, FfxUInt32x2 i3, FfxUInt32 slice) -{ - FfxFloat32x4 v0 = SpdLoadSourceImage(FfxInt32x2(i0), slice); - FfxFloat32x4 v1 = SpdLoadSourceImage(FfxInt32x2(i1), slice); - FfxFloat32x4 v2 = SpdLoadSourceImage(FfxInt32x2(i2), slice); - FfxFloat32x4 v3 = SpdLoadSourceImage(FfxInt32x2(i3), slice); - return SpdReduce4(v0, v1, v2, v3); -} - -FfxFloat32x4 SpdReduceLoadSourceImage(FfxUInt32x2 base, FfxUInt32 slice) -{ -#if defined(SPD_LINEAR_SAMPLER) - return SpdLoadSourceImage(FfxInt32x2(base), slice); -#else - return SpdReduceLoadSourceImage4(FfxUInt32x2(base + FfxUInt32x2(0, 0)), FfxUInt32x2(base + FfxUInt32x2(0, 1)), FfxUInt32x2(base + FfxUInt32x2(1, 0)), FfxUInt32x2(base + FfxUInt32x2(1, 1)), slice); -#endif -} - -void SpdDownsampleMips_0_1_Intrinsics(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) -{ - FfxFloat32x4 v[4]; - - FfxInt32x2 tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2); - FfxInt32x2 pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y); - v[0] = SpdReduceLoadSourceImage(tex, slice); - SpdStore(pix, v[0], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y); - v[1] = SpdReduceLoadSourceImage(tex, slice); - SpdStore(pix, v[1], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2 + 32); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y + 16); - v[2] = SpdReduceLoadSourceImage(tex, slice); - SpdStore(pix, v[2], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2 + 32); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y + 16); - v[3] = SpdReduceLoadSourceImage(tex, slice); - SpdStore(pix, v[3], 0, slice); - - if (mip <= 1) - return; - - v[0] = SpdReduceQuad(v[0]); - v[1] = SpdReduceQuad(v[1]); - v[2] = SpdReduceQuad(v[2]); - v[3] = SpdReduceQuad(v[3]); - - if ((localInvocationIndex % 4) == 0) - { - SpdStore(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2, y / 2), v[0], 1, slice); - SpdStoreIntermediate(x / 2, y / 2, v[0]); - - SpdStore(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2 + 8, y / 2), v[1], 1, slice); - SpdStoreIntermediate(x / 2 + 8, y / 2, v[1]); - - SpdStore(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2, y / 2 + 8), v[2], 1, slice); - SpdStoreIntermediate(x / 2, y / 2 + 8, v[2]); - - SpdStore(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2 + 8, y / 2 + 8), v[3], 1, slice); - SpdStoreIntermediate(x / 2 + 8, y / 2 + 8, v[3]); - } -} - -void SpdDownsampleMips_0_1_LDS(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) -{ - FfxFloat32x4 v[4]; - - FfxInt32x2 tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2); - FfxInt32x2 pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y); - v[0] = SpdReduceLoadSourceImage(tex, slice); - SpdStore(pix, v[0], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y); - v[1] = SpdReduceLoadSourceImage(tex, slice); - SpdStore(pix, v[1], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2 + 32); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y + 16); - v[2] = SpdReduceLoadSourceImage(tex, slice); - SpdStore(pix, v[2], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2 + 32); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y + 16); - v[3] = SpdReduceLoadSourceImage(tex, slice); - SpdStore(pix, v[3], 0, slice); - - if (mip <= 1) - return; - - for (FfxUInt32 i = 0; i < 4; i++) - { - SpdStoreIntermediate(x, y, v[i]); - ffxSpdWorkgroupShuffleBarrier(); - if (localInvocationIndex < 64) - { - v[i] = SpdReduceIntermediate(FfxUInt32x2(x * 2 + 0, y * 2 + 0), FfxUInt32x2(x * 2 + 1, y * 2 + 0), FfxUInt32x2(x * 2 + 0, y * 2 + 1), FfxUInt32x2(x * 2 + 1, y * 2 + 1)); - SpdStore(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x + (i % 2) * 8, y + (i / 2) * 8), v[i], 1, slice); - } - ffxSpdWorkgroupShuffleBarrier(); - } - - if (localInvocationIndex < 64) - { - SpdStoreIntermediate(x + 0, y + 0, v[0]); - SpdStoreIntermediate(x + 8, y + 0, v[1]); - SpdStoreIntermediate(x + 0, y + 8, v[2]); - SpdStoreIntermediate(x + 8, y + 8, v[3]); - } -} - -void SpdDownsampleMips_0_1(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) -{ -#if defined(FFX_SPD_NO_WAVE_OPERATIONS) - SpdDownsampleMips_0_1_LDS(x, y, workGroupID, localInvocationIndex, mip, slice); -#else - SpdDownsampleMips_0_1_Intrinsics(x, y, workGroupID, localInvocationIndex, mip, slice); -#endif -} - - -void SpdDownsampleMip_2(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) -{ -#if defined(FFX_SPD_NO_WAVE_OPERATIONS) - if (localInvocationIndex < 64) - { - FfxFloat32x4 v = SpdReduceIntermediate(FfxUInt32x2(x * 2 + 0, y * 2 + 0), FfxUInt32x2(x * 2 + 1, y * 2 + 0), FfxUInt32x2(x * 2 + 0, y * 2 + 1), FfxUInt32x2(x * 2 + 1, y * 2 + 1)); - SpdStore(FfxInt32x2(workGroupID.xy * 8) + FfxInt32x2(x, y), v, mip, slice); - // store to LDS, try to reduce bank conflicts - // x 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 - // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - // 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 x - // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - // x 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 - // ... - // x 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 - SpdStoreIntermediate(x * 2 + y % 2, y * 2, v); - } -#else - FfxFloat32x4 v = SpdLoadIntermediate(x, y); - v = SpdReduceQuad(v); - // quad index 0 stores result - if (localInvocationIndex % 4 == 0) - { - SpdStore(FfxInt32x2(workGroupID.xy * 8) + FfxInt32x2(x / 2, y / 2), v, mip, slice); - SpdStoreIntermediate(x + (y / 2) % 2, y, v); - } -#endif -} - -void SpdDownsampleMip_3(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) -{ -#if defined(FFX_SPD_NO_WAVE_OPERATIONS) - if (localInvocationIndex < 16) - { - // x 0 x 0 - // 0 0 0 0 - // 0 x 0 x - // 0 0 0 0 - FfxFloat32x4 v = - SpdReduceIntermediate(FfxUInt32x2(x * 4 + 0 + 0, y * 4 + 0), FfxUInt32x2(x * 4 + 2 + 0, y * 4 + 0), FfxUInt32x2(x * 4 + 0 + 1, y * 4 + 2), FfxUInt32x2(x * 4 + 2 + 1, y * 4 + 2)); - SpdStore(FfxInt32x2(workGroupID.xy * 4) + FfxInt32x2(x, y), v, mip, slice); - // store to LDS - // x 0 0 0 x 0 0 0 x 0 0 0 x 0 0 0 - // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - // 0 x 0 0 0 x 0 0 0 x 0 0 0 x 0 0 - // ... - // 0 0 x 0 0 0 x 0 0 0 x 0 0 0 x 0 - // ... - // 0 0 0 x 0 0 0 x 0 0 0 x 0 0 0 x - // ... - SpdStoreIntermediate(x * 4 + y, y * 4, v); - } -#else - if (localInvocationIndex < 64) - { - FfxFloat32x4 v = SpdLoadIntermediate(x * 2 + y % 2, y * 2); - v = SpdReduceQuad(v); - // quad index 0 stores result - if (localInvocationIndex % 4 == 0) - { - SpdStore(FfxInt32x2(workGroupID.xy * 4) + FfxInt32x2(x / 2, y / 2), v, mip, slice); - SpdStoreIntermediate(x * 2 + y / 2, y * 2, v); - } - } -#endif -} - -void SpdDownsampleMip_4(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) -{ -#if defined(FFX_SPD_NO_WAVE_OPERATIONS) - if (localInvocationIndex < 4) - { - // x 0 0 0 x 0 0 0 - // ... - // 0 x 0 0 0 x 0 0 - FfxFloat32x4 v = SpdReduceIntermediate(FfxUInt32x2(x * 8 + 0 + 0 + y * 2, y * 8 + 0), - FfxUInt32x2(x * 8 + 4 + 0 + y * 2, y * 8 + 0), - FfxUInt32x2(x * 8 + 0 + 1 + y * 2, y * 8 + 4), - FfxUInt32x2(x * 8 + 4 + 1 + y * 2, y * 8 + 4)); - SpdStore(FfxInt32x2(workGroupID.xy * 2) + FfxInt32x2(x, y), v, mip, slice); - // store to LDS - // x x x x 0 ... - // 0 ... - SpdStoreIntermediate(x + y * 2, 0, v); - } -#else - if (localInvocationIndex < 16) - { - FfxFloat32x4 v = SpdLoadIntermediate(x * 4 + y, y * 4); - v = SpdReduceQuad(v); - // quad index 0 stores result - if (localInvocationIndex % 4 == 0) - { - SpdStore(FfxInt32x2(workGroupID.xy * 2) + FfxInt32x2(x / 2, y / 2), v, mip, slice); - SpdStoreIntermediate(x / 2 + y, 0, v); - } - } -#endif -} - -void SpdDownsampleMip_5(FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) -{ -#if defined(FFX_SPD_NO_WAVE_OPERATIONS) - if (localInvocationIndex < 1) - { - // x x x x 0 ... - // 0 ... - FfxFloat32x4 v = SpdReduceIntermediate(FfxUInt32x2(0, 0), FfxUInt32x2(1, 0), FfxUInt32x2(2, 0), FfxUInt32x2(3, 0)); - SpdStore(FfxInt32x2(workGroupID.xy), v, mip, slice); - } -#else - if (localInvocationIndex < 4) - { - FfxFloat32x4 v = SpdLoadIntermediate(localInvocationIndex, 0); - v = SpdReduceQuad(v); - // quad index 0 stores result - if (localInvocationIndex % 4 == 0) - { - SpdStore(FfxInt32x2(workGroupID.xy), v, mip, slice); - } - } -#endif -} - -void SpdDownsampleMips_6_7(FfxUInt32 x, FfxUInt32 y, FfxUInt32 mips, FfxUInt32 slice) -{ - FfxInt32x2 tex = FfxInt32x2(x * 4 + 0, y * 4 + 0); - FfxInt32x2 pix = FfxInt32x2(x * 2 + 0, y * 2 + 0); - FfxFloat32x4 v0 = SpdReduceLoad4(tex, slice); - SpdStore(pix, v0, 6, slice); - - tex = FfxInt32x2(x * 4 + 2, y * 4 + 0); - pix = FfxInt32x2(x * 2 + 1, y * 2 + 0); - FfxFloat32x4 v1 = SpdReduceLoad4(tex, slice); - SpdStore(pix, v1, 6, slice); - - tex = FfxInt32x2(x * 4 + 0, y * 4 + 2); - pix = FfxInt32x2(x * 2 + 0, y * 2 + 1); - FfxFloat32x4 v2 = SpdReduceLoad4(tex, slice); - SpdStore(pix, v2, 6, slice); - - tex = FfxInt32x2(x * 4 + 2, y * 4 + 2); - pix = FfxInt32x2(x * 2 + 1, y * 2 + 1); - FfxFloat32x4 v3 = SpdReduceLoad4(tex, slice); - SpdStore(pix, v3, 6, slice); - - if (mips <= 7) - return; - // no barrier needed, working on values only from the same thread - - FfxFloat32x4 v = SpdReduce4(v0, v1, v2, v3); - SpdStore(FfxInt32x2(x, y), v, 7, slice); - SpdStoreIntermediate(x, y, v); -} - -void SpdDownsampleNextFour(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 baseMip, FfxUInt32 mips, FfxUInt32 slice) -{ - if (mips <= baseMip) - return; - ffxSpdWorkgroupShuffleBarrier(); - SpdDownsampleMip_2(x, y, workGroupID, localInvocationIndex, baseMip, slice); - - if (mips <= baseMip + 1) - return; - ffxSpdWorkgroupShuffleBarrier(); - SpdDownsampleMip_3(x, y, workGroupID, localInvocationIndex, baseMip + 1, slice); - - if (mips <= baseMip + 2) - return; - ffxSpdWorkgroupShuffleBarrier(); - SpdDownsampleMip_4(x, y, workGroupID, localInvocationIndex, baseMip + 2, slice); - - if (mips <= baseMip + 3) - return; - ffxSpdWorkgroupShuffleBarrier(); - SpdDownsampleMip_5(workGroupID, localInvocationIndex, baseMip + 3, slice); -} - -/// Downsamples a 64x64 tile based on the work group id. -/// If after downsampling it's the last active thread group, computes the remaining MIP levels. -/// -/// @param [in] workGroupID index of the work group / thread group -/// @param [in] localInvocationIndex index of the thread within the thread group in 1D -/// @param [in] mips the number of total MIP levels to compute for the input texture -/// @param [in] numWorkGroups the total number of dispatched work groups / thread groups for this slice -/// @param [in] slice the slice of the input texture -/// -/// @ingroup FfxGPUSpd -void SpdDownsample(FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mips, FfxUInt32 numWorkGroups, FfxUInt32 slice) -{ - // compute MIP level 0 and 1 - FfxUInt32x2 sub_xy = ffxRemapForWaveReduction(localInvocationIndex % 64); - FfxUInt32 x = sub_xy.x + 8 * ((localInvocationIndex >> 6) % 2); - FfxUInt32 y = sub_xy.y + 8 * ((localInvocationIndex >> 7)); - SpdDownsampleMips_0_1(x, y, workGroupID, localInvocationIndex, mips, slice); - - // compute MIP level 2, 3, 4, 5 - SpdDownsampleNextFour(x, y, workGroupID, localInvocationIndex, 2, mips, slice); - - if (mips <= 6) - return; - - // increase the global atomic counter for the given slice and check if it's the last remaining thread group: - // terminate if not, continue if yes. - if (SpdExitWorkgroup(numWorkGroups, localInvocationIndex, slice)) - return; - - // reset the global atomic counter back to 0 for the next spd dispatch - SpdResetAtomicCounter(slice); - - // After mip 5 there is only a single workgroup left that downsamples the remaining up to 64x64 texels. - // compute MIP level 6 and 7 - SpdDownsampleMips_6_7(x, y, mips, slice); - - // compute MIP level 8, 9, 10, 11 - SpdDownsampleNextFour(x, y, FfxUInt32x2(0, 0), localInvocationIndex, 8, mips, slice); -} -/// Downsamples a 64x64 tile based on the work group id and work group offset. -/// If after downsampling it's the last active thread group, computes the remaining MIP levels. -/// -/// @param [in] workGroupID index of the work group / thread group -/// @param [in] localInvocationIndex index of the thread within the thread group in 1D -/// @param [in] mips the number of total MIP levels to compute for the input texture -/// @param [in] numWorkGroups the total number of dispatched work groups / thread groups for this slice -/// @param [in] slice the slice of the input texture -/// @param [in] workGroupOffset the work group offset. it's (0,0) in case the entire input texture is downsampled. -/// -/// @ingroup FfxGPUSpd -void SpdDownsample(FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mips, FfxUInt32 numWorkGroups, FfxUInt32 slice, FfxUInt32x2 workGroupOffset) -{ - SpdDownsample(workGroupID + workGroupOffset, localInvocationIndex, mips, numWorkGroups, slice); -} - -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// -//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// - -//============================================================================================================================== -// PACKED VERSION -//============================================================================================================================== - -#if FFX_HALF - -FfxFloat16x4 SpdReduceQuadH(FfxFloat16x4 v) -{ -#if defined(FFX_GLSL) && !defined(FFX_SPD_NO_WAVE_OPERATIONS) - FfxFloat16x4 v0 = v; - FfxFloat16x4 v1 = subgroupQuadSwapHorizontal(v); - FfxFloat16x4 v2 = subgroupQuadSwapVertical(v); - FfxFloat16x4 v3 = subgroupQuadSwapDiagonal(v); - return SpdReduce4H(v0, v1, v2, v3); -#elif defined(FFX_HLSL) && !defined(FFX_SPD_NO_WAVE_OPERATIONS) - // requires SM6.0 - FfxUInt32 quad = WaveGetLaneIndex() & (~0x3); - FfxFloat16x4 v0 = v; - FfxFloat16x4 v1 = WaveReadLaneAt(v, quad | 1); - FfxFloat16x4 v2 = WaveReadLaneAt(v, quad | 2); - FfxFloat16x4 v3 = WaveReadLaneAt(v, quad | 3); - return SpdReduce4H(v0, v1, v2, v3); -/* - // if SM6.0 is not available, you can use the AMD shader intrinsics - // the AMD shader intrinsics are available in AMD GPU Services (AGS) library: - // https://gpuopen.com/amd-gpu-services-ags-library/ - // works for DX11 - FfxFloat16x4 v0 = v; - FfxFloat16x4 v1; - v1.x = AmdExtD3DShaderIntrinsics_SwizzleF(v.x, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); - v1.y = AmdExtD3DShaderIntrinsics_SwizzleF(v.y, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); - v1.z = AmdExtD3DShaderIntrinsics_SwizzleF(v.z, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); - v1.w = AmdExtD3DShaderIntrinsics_SwizzleF(v.w, AmdExtD3DShaderIntrinsicsSwizzle_SwapX1); - FfxFloat16x4 v2; - v2.x = AmdExtD3DShaderIntrinsics_SwizzleF(v.x, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); - v2.y = AmdExtD3DShaderIntrinsics_SwizzleF(v.y, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); - v2.z = AmdExtD3DShaderIntrinsics_SwizzleF(v.z, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); - v2.w = AmdExtD3DShaderIntrinsics_SwizzleF(v.w, AmdExtD3DShaderIntrinsicsSwizzle_SwapX2); - FfxFloat16x4 v3; - v3.x = AmdExtD3DShaderIntrinsics_SwizzleF(v.x, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); - v3.y = AmdExtD3DShaderIntrinsics_SwizzleF(v.y, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); - v3.z = AmdExtD3DShaderIntrinsics_SwizzleF(v.z, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); - v3.w = AmdExtD3DShaderIntrinsics_SwizzleF(v.w, AmdExtD3DShaderIntrinsicsSwizzle_ReverseX4); - return SpdReduce4H(v0, v1, v2, v3); - */ -#endif - return FfxFloat16x4(0.0, 0.0, 0.0, 0.0); -} - -FfxFloat16x4 SpdReduceIntermediateH(FfxUInt32x2 i0, FfxUInt32x2 i1, FfxUInt32x2 i2, FfxUInt32x2 i3) -{ - FfxFloat16x4 v0 = SpdLoadIntermediateH(i0.x, i0.y); - FfxFloat16x4 v1 = SpdLoadIntermediateH(i1.x, i1.y); - FfxFloat16x4 v2 = SpdLoadIntermediateH(i2.x, i2.y); - FfxFloat16x4 v3 = SpdLoadIntermediateH(i3.x, i3.y); - return SpdReduce4H(v0, v1, v2, v3); -} - -FfxFloat16x4 SpdReduceLoad4H(FfxUInt32x2 i0, FfxUInt32x2 i1, FfxUInt32x2 i2, FfxUInt32x2 i3, FfxUInt32 slice) -{ - FfxFloat16x4 v0 = SpdLoadH(FfxInt32x2(i0), slice); - FfxFloat16x4 v1 = SpdLoadH(FfxInt32x2(i1), slice); - FfxFloat16x4 v2 = SpdLoadH(FfxInt32x2(i2), slice); - FfxFloat16x4 v3 = SpdLoadH(FfxInt32x2(i3), slice); - return SpdReduce4H(v0, v1, v2, v3); -} - -FfxFloat16x4 SpdReduceLoad4H(FfxUInt32x2 base, FfxUInt32 slice) -{ - return SpdReduceLoad4H(FfxUInt32x2(base + FfxUInt32x2(0, 0)), FfxUInt32x2(base + FfxUInt32x2(0, 1)), FfxUInt32x2(base + FfxUInt32x2(1, 0)), FfxUInt32x2(base + FfxUInt32x2(1, 1)), slice); -} - -FfxFloat16x4 SpdReduceLoadSourceImage4H(FfxUInt32x2 i0, FfxUInt32x2 i1, FfxUInt32x2 i2, FfxUInt32x2 i3, FfxUInt32 slice) -{ - FfxFloat16x4 v0 = SpdLoadSourceImageH(FfxInt32x2(i0), slice); - FfxFloat16x4 v1 = SpdLoadSourceImageH(FfxInt32x2(i1), slice); - FfxFloat16x4 v2 = SpdLoadSourceImageH(FfxInt32x2(i2), slice); - FfxFloat16x4 v3 = SpdLoadSourceImageH(FfxInt32x2(i3), slice); - return SpdReduce4H(v0, v1, v2, v3); -} - -FfxFloat16x4 SpdReduceLoadSourceImageH(FfxUInt32x2 base, FfxUInt32 slice) -{ -#if defined(SPD_LINEAR_SAMPLER) - return SpdLoadSourceImageH(FfxInt32x2(base), slice); -#else - return SpdReduceLoadSourceImage4H(FfxUInt32x2(base + FfxUInt32x2(0, 0)), FfxUInt32x2(base + FfxUInt32x2(0, 1)), FfxUInt32x2(base + FfxUInt32x2(1, 0)), FfxUInt32x2(base + FfxUInt32x2(1, 1)), slice); -#endif -} - -void SpdDownsampleMips_0_1_IntrinsicsH(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mips, FfxUInt32 slice) -{ - FfxFloat16x4 v[4]; - - FfxInt32x2 tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2); - FfxInt32x2 pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y); - v[0] = SpdReduceLoadSourceImageH(tex, slice); - SpdStoreH(pix, v[0], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y); - v[1] = SpdReduceLoadSourceImageH(tex, slice); - SpdStoreH(pix, v[1], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2 + 32); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y + 16); - v[2] = SpdReduceLoadSourceImageH(tex, slice); - SpdStoreH(pix, v[2], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2 + 32); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y + 16); - v[3] = SpdReduceLoadSourceImageH(tex, slice); - SpdStoreH(pix, v[3], 0, slice); - - if (mips <= 1) - return; - - v[0] = SpdReduceQuadH(v[0]); - v[1] = SpdReduceQuadH(v[1]); - v[2] = SpdReduceQuadH(v[2]); - v[3] = SpdReduceQuadH(v[3]); - - if ((localInvocationIndex % 4) == 0) - { - SpdStoreH(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2, y / 2), v[0], 1, slice); - SpdStoreIntermediateH(x / 2, y / 2, v[0]); - - SpdStoreH(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2 + 8, y / 2), v[1], 1, slice); - SpdStoreIntermediateH(x / 2 + 8, y / 2, v[1]); - - SpdStoreH(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2, y / 2 + 8), v[2], 1, slice); - SpdStoreIntermediateH(x / 2, y / 2 + 8, v[2]); - - SpdStoreH(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x / 2 + 8, y / 2 + 8), v[3], 1, slice); - SpdStoreIntermediateH(x / 2 + 8, y / 2 + 8, v[3]); - } -} - -void SpdDownsampleMips_0_1_LDSH(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mips, FfxUInt32 slice) -{ - FfxFloat16x4 v[4]; - - FfxInt32x2 tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2); - FfxInt32x2 pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y); - v[0] = SpdReduceLoadSourceImageH(tex, slice); - SpdStoreH(pix, v[0], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y); - v[1] = SpdReduceLoadSourceImageH(tex, slice); - SpdStoreH(pix, v[1], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2, y * 2 + 32); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x, y + 16); - v[2] = SpdReduceLoadSourceImageH(tex, slice); - SpdStoreH(pix, v[2], 0, slice); - - tex = FfxInt32x2(workGroupID.xy * 64) + FfxInt32x2(x * 2 + 32, y * 2 + 32); - pix = FfxInt32x2(workGroupID.xy * 32) + FfxInt32x2(x + 16, y + 16); - v[3] = SpdReduceLoadSourceImageH(tex, slice); - SpdStoreH(pix, v[3], 0, slice); - - if (mips <= 1) - return; - - for (FfxUInt32 i = 0; i < 4; i++) - { - SpdStoreIntermediateH(x, y, v[i]); - ffxSpdWorkgroupShuffleBarrier(); - if (localInvocationIndex < 64) - { - v[i] = SpdReduceIntermediateH(FfxUInt32x2(x * 2 + 0, y * 2 + 0), FfxUInt32x2(x * 2 + 1, y * 2 + 0), FfxUInt32x2(x * 2 + 0, y * 2 + 1), FfxUInt32x2(x * 2 + 1, y * 2 + 1)); - SpdStoreH(FfxInt32x2(workGroupID.xy * 16) + FfxInt32x2(x + (i % 2) * 8, y + (i / 2) * 8), v[i], 1, slice); - } - ffxSpdWorkgroupShuffleBarrier(); - } - - if (localInvocationIndex < 64) - { - SpdStoreIntermediateH(x + 0, y + 0, v[0]); - SpdStoreIntermediateH(x + 8, y + 0, v[1]); - SpdStoreIntermediateH(x + 0, y + 8, v[2]); - SpdStoreIntermediateH(x + 8, y + 8, v[3]); - } -} - -void SpdDownsampleMips_0_1H(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mips, FfxUInt32 slice) -{ -#if defined(FFX_SPD_NO_WAVE_OPERATIONS) - SpdDownsampleMips_0_1_LDSH(x, y, workGroupID, localInvocationIndex, mips, slice); -#else - SpdDownsampleMips_0_1_IntrinsicsH(x, y, workGroupID, localInvocationIndex, mips, slice); -#endif -} - - -void SpdDownsampleMip_2H(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) -{ -#if defined(FFX_SPD_NO_WAVE_OPERATIONS) - if (localInvocationIndex < 64) - { - FfxFloat16x4 v = SpdReduceIntermediateH(FfxUInt32x2(x * 2 + 0, y * 2 + 0), FfxUInt32x2(x * 2 + 1, y * 2 + 0), FfxUInt32x2(x * 2 + 0, y * 2 + 1), FfxUInt32x2(x * 2 + 1, y * 2 + 1)); - SpdStoreH(FfxInt32x2(workGroupID.xy * 8) + FfxInt32x2(x, y), v, mip, slice); - // store to LDS, try to reduce bank conflicts - // x 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 - // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - // 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 x - // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - // x 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 - // ... - // x 0 x 0 x 0 x 0 x 0 x 0 x 0 x 0 - SpdStoreIntermediateH(x * 2 + y % 2, y * 2, v); - } -#else - FfxFloat16x4 v = SpdLoadIntermediateH(x, y); - v = SpdReduceQuadH(v); - // quad index 0 stores result - if (localInvocationIndex % 4 == 0) - { - SpdStoreH(FfxInt32x2(workGroupID.xy * 8) + FfxInt32x2(x / 2, y / 2), v, mip, slice); - SpdStoreIntermediateH(x + (y / 2) % 2, y, v); - } -#endif -} - -void SpdDownsampleMip_3H(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) -{ -#if defined(FFX_SPD_NO_WAVE_OPERATIONS) - if (localInvocationIndex < 16) - { - // x 0 x 0 - // 0 0 0 0 - // 0 x 0 x - // 0 0 0 0 - FfxFloat16x4 v = - SpdReduceIntermediateH(FfxUInt32x2(x * 4 + 0 + 0, y * 4 + 0), FfxUInt32x2(x * 4 + 2 + 0, y * 4 + 0), FfxUInt32x2(x * 4 + 0 + 1, y * 4 + 2), FfxUInt32x2(x * 4 + 2 + 1, y * 4 + 2)); - SpdStoreH(FfxInt32x2(workGroupID.xy * 4) + FfxInt32x2(x, y), v, mip, slice); - // store to LDS - // x 0 0 0 x 0 0 0 x 0 0 0 x 0 0 0 - // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - // 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - // 0 x 0 0 0 x 0 0 0 x 0 0 0 x 0 0 - // ... - // 0 0 x 0 0 0 x 0 0 0 x 0 0 0 x 0 - // ... - // 0 0 0 x 0 0 0 x 0 0 0 x 0 0 0 x - // ... - SpdStoreIntermediateH(x * 4 + y, y * 4, v); - } -#else - if (localInvocationIndex < 64) - { - FfxFloat16x4 v = SpdLoadIntermediateH(x * 2 + y % 2, y * 2); - v = SpdReduceQuadH(v); - // quad index 0 stores result - if (localInvocationIndex % 4 == 0) - { - SpdStoreH(FfxInt32x2(workGroupID.xy * 4) + FfxInt32x2(x / 2, y / 2), v, mip, slice); - SpdStoreIntermediateH(x * 2 + y / 2, y * 2, v); - } - } -#endif -} - -void SpdDownsampleMip_4H(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) -{ -#if defined(FFX_SPD_NO_WAVE_OPERATIONS) - if (localInvocationIndex < 4) - { - // x 0 0 0 x 0 0 0 - // ... - // 0 x 0 0 0 x 0 0 - FfxFloat16x4 v = SpdReduceIntermediateH(FfxUInt32x2(x * 8 + 0 + 0 + y * 2, y * 8 + 0), - FfxUInt32x2(x * 8 + 4 + 0 + y * 2, y * 8 + 0), - FfxUInt32x2(x * 8 + 0 + 1 + y * 2, y * 8 + 4), - FfxUInt32x2(x * 8 + 4 + 1 + y * 2, y * 8 + 4)); - SpdStoreH(FfxInt32x2(workGroupID.xy * 2) + FfxInt32x2(x, y), v, mip, slice); - // store to LDS - // x x x x 0 ... - // 0 ... - SpdStoreIntermediateH(x + y * 2, 0, v); - } -#else - if (localInvocationIndex < 16) - { - FfxFloat16x4 v = SpdLoadIntermediateH(x * 4 + y, y * 4); - v = SpdReduceQuadH(v); - // quad index 0 stores result - if (localInvocationIndex % 4 == 0) - { - SpdStoreH(FfxInt32x2(workGroupID.xy * 2) + FfxInt32x2(x / 2, y / 2), v, mip, slice); - SpdStoreIntermediateH(x / 2 + y, 0, v); - } - } -#endif -} - -void SpdDownsampleMip_5H(FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mip, FfxUInt32 slice) -{ -#if defined(FFX_SPD_NO_WAVE_OPERATIONS) - if (localInvocationIndex < 1) - { - // x x x x 0 ... - // 0 ... - FfxFloat16x4 v = SpdReduceIntermediateH(FfxUInt32x2(0, 0), FfxUInt32x2(1, 0), FfxUInt32x2(2, 0), FfxUInt32x2(3, 0)); - SpdStoreH(FfxInt32x2(workGroupID.xy), v, mip, slice); - } -#else - if (localInvocationIndex < 4) - { - FfxFloat16x4 v = SpdLoadIntermediateH(localInvocationIndex, 0); - v = SpdReduceQuadH(v); - // quad index 0 stores result - if (localInvocationIndex % 4 == 0) - { - SpdStoreH(FfxInt32x2(workGroupID.xy), v, mip, slice); - } - } -#endif -} - -void SpdDownsampleMips_6_7H(FfxUInt32 x, FfxUInt32 y, FfxUInt32 mips, FfxUInt32 slice) -{ - FfxInt32x2 tex = FfxInt32x2(x * 4 + 0, y * 4 + 0); - FfxInt32x2 pix = FfxInt32x2(x * 2 + 0, y * 2 + 0); - FfxFloat16x4 v0 = SpdReduceLoad4H(tex, slice); - SpdStoreH(pix, v0, 6, slice); - - tex = FfxInt32x2(x * 4 + 2, y * 4 + 0); - pix = FfxInt32x2(x * 2 + 1, y * 2 + 0); - FfxFloat16x4 v1 = SpdReduceLoad4H(tex, slice); - SpdStoreH(pix, v1, 6, slice); - - tex = FfxInt32x2(x * 4 + 0, y * 4 + 2); - pix = FfxInt32x2(x * 2 + 0, y * 2 + 1); - FfxFloat16x4 v2 = SpdReduceLoad4H(tex, slice); - SpdStoreH(pix, v2, 6, slice); - - tex = FfxInt32x2(x * 4 + 2, y * 4 + 2); - pix = FfxInt32x2(x * 2 + 1, y * 2 + 1); - FfxFloat16x4 v3 = SpdReduceLoad4H(tex, slice); - SpdStoreH(pix, v3, 6, slice); - - if (mips < 8) - return; - // no barrier needed, working on values only from the same thread - - FfxFloat16x4 v = SpdReduce4H(v0, v1, v2, v3); - SpdStoreH(FfxInt32x2(x, y), v, 7, slice); - SpdStoreIntermediateH(x, y, v); -} - -void SpdDownsampleNextFourH(FfxUInt32 x, FfxUInt32 y, FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 baseMip, FfxUInt32 mips, FfxUInt32 slice) -{ - if (mips <= baseMip) - return; - ffxSpdWorkgroupShuffleBarrier(); - SpdDownsampleMip_2H(x, y, workGroupID, localInvocationIndex, baseMip, slice); - - if (mips <= baseMip + 1) - return; - ffxSpdWorkgroupShuffleBarrier(); - SpdDownsampleMip_3H(x, y, workGroupID, localInvocationIndex, baseMip + 1, slice); - - if (mips <= baseMip + 2) - return; - ffxSpdWorkgroupShuffleBarrier(); - SpdDownsampleMip_4H(x, y, workGroupID, localInvocationIndex, baseMip + 2, slice); - - if (mips <= baseMip + 3) - return; - ffxSpdWorkgroupShuffleBarrier(); - SpdDownsampleMip_5H(workGroupID, localInvocationIndex, baseMip + 3, slice); -} - -/// Downsamples a 64x64 tile based on the work group id and work group offset. -/// If after downsampling it's the last active thread group, computes the remaining MIP levels. -/// Uses half types. -/// -/// @param [in] workGroupID index of the work group / thread group -/// @param [in] localInvocationIndex index of the thread within the thread group in 1D -/// @param [in] mips the number of total MIP levels to compute for the input texture -/// @param [in] numWorkGroups the total number of dispatched work groups / thread groups for this slice -/// @param [in] slice the slice of the input texture -/// -/// @ingroup FfxGPUSpd -void SpdDownsampleH(FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mips, FfxUInt32 numWorkGroups, FfxUInt32 slice) -{ - FfxUInt32x2 sub_xy = ffxRemapForWaveReduction(localInvocationIndex % 64); - FfxUInt32 x = sub_xy.x + 8 * ((localInvocationIndex >> 6) % 2); - FfxUInt32 y = sub_xy.y + 8 * ((localInvocationIndex >> 7)); - - // compute MIP level 0 and 1 - SpdDownsampleMips_0_1H(x, y, workGroupID, localInvocationIndex, mips, slice); - - // compute MIP level 2, 3, 4, 5 - SpdDownsampleNextFourH(x, y, workGroupID, localInvocationIndex, 2, mips, slice); - - if (mips < 7) - return; - - // increase the global atomic counter for the given slice and check if it's the last remaining thread group: - // terminate if not, continue if yes. - if (SpdExitWorkgroup(numWorkGroups, localInvocationIndex, slice)) - return; - - // reset the global atomic counter back to 0 for the next spd dispatch - SpdResetAtomicCounter(slice); - - // After mip 5 there is only a single workgroup left that downsamples the remaining up to 64x64 texels. - // compute MIP level 6 and 7 - SpdDownsampleMips_6_7H(x, y, mips, slice); - - // compute MIP level 8, 9, 10, 11 - SpdDownsampleNextFourH(x, y, FfxUInt32x2(0, 0), localInvocationIndex, 8, mips, slice); -} - -/// Downsamples a 64x64 tile based on the work group id and work group offset. -/// If after downsampling it's the last active thread group, computes the remaining MIP levels. -/// Uses half types. -/// -/// @param [in] workGroupID index of the work group / thread group -/// @param [in] localInvocationIndex index of the thread within the thread group in 1D -/// @param [in] mips the number of total MIP levels to compute for the input texture -/// @param [in] numWorkGroups the total number of dispatched work groups / thread groups for this slice -/// @param [in] slice the slice of the input texture -/// @param [in] workGroupOffset the work group offset. it's (0,0) in case the entire input texture is downsampled. -/// -/// @ingroup FfxGPUSpd -void SpdDownsampleH(FfxUInt32x2 workGroupID, FfxUInt32 localInvocationIndex, FfxUInt32 mips, FfxUInt32 numWorkGroups, FfxUInt32 slice, FfxUInt32x2 workGroupOffset) -{ - SpdDownsampleH(workGroupID + workGroupOffset, localInvocationIndex, mips, numWorkGroups, slice); -} - -#endif // #if FFX_HALF -#endif // #if defined(FFX_GPU) diff --git a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/spd/ffx_spd.h.meta b/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/spd/ffx_spd.h.meta deleted file mode 100644 index 2741ab1..0000000 --- a/com.unity.postprocessing/PostProcessing/Shaders/FSR3/shaders/fsr3upscaler/spd/ffx_spd.h.meta +++ /dev/null @@ -1,67 +0,0 @@ -fileFormatVersion: 2 -guid: face65176ee3b82498bd0b8fed0ddacd -PluginImporter: - externalObjects: {} - serializedVersion: 2 - iconMap: {} - executionOrder: {} - defineConstraints: [] - isPreloaded: 0 - isOverridable: 0 - isExplicitlyReferenced: 0 - validateReferences: 1 - platformData: - - first: - : Any - second: - enabled: 0 - settings: - Exclude Editor: 1 - Exclude GameCoreScarlett: 1 - Exclude GameCoreXboxOne: 1 - Exclude Linux64: 1 - Exclude OSXUniversal: 1 - Exclude PS4: 1 - Exclude PS5: 1 - Exclude Win: 1 - Exclude Win64: 1 - - first: - Any: - second: - enabled: 0 - settings: {} - - first: - Editor: Editor - second: - enabled: 0 - settings: - CPU: AnyCPU - DefaultValueInitialized: true - OS: AnyOS - - first: - Standalone: Linux64 - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: OSXUniversal - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win - second: - enabled: 0 - settings: - CPU: None - - first: - Standalone: Win64 - second: - enabled: 0 - settings: - CPU: None - userData: - assetBundleName: - assetBundleVariant: diff --git a/com.unity.postprocessing/package.json b/com.unity.postprocessing/package.json deleted file mode 100644 index 6d0688d..0000000 --- a/com.unity.postprocessing/package.json +++ /dev/null @@ -1,11 +0,0 @@ -{ - "name": "com.unity.postprocessing", - "version": "3.4.0", - "displayName": "Post Processing", - "unity": "2019.4", - "unityRelease": "19f1", - "description": "The post-processing stack (v2) comes with a collection of effects and image filters you can apply to your cameras to improve the visuals of your games.", - "dependencies": { - "com.unity.modules.physics": "1.0.0" - } -} diff --git a/images/comparison-fsr3.1.png b/images/comparison-fsr3.1.png new file mode 100644 index 0000000..c800c57 Binary files /dev/null and b/images/comparison-fsr3.1.png differ diff --git a/images/comparison.png b/images/comparison.png deleted file mode 100644 index e379f8c..0000000 Binary files a/images/comparison.png and /dev/null differ