diff --git a/CMakeLists.txt b/CMakeLists.txt
index e1d8e77..867de59 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -14,7 +14,7 @@ project(
 file(GLOB _sources *.c)
 psn00bsdk_add_executable(ps1bsp STATIC ${_sources})
 
-psn00bsdk_target_incbin(ps1bsp PRIVATE tim_e1m1 atlas-e1m1.tim)
+psn00bsdk_target_incbin(ps1bsp PRIVATE tim_start atlas-start.tim)
 psn00bsdk_target_incbin(ps1bsp PRIVATE bsp_test test.ps1bsp)
 
 psn00bsdk_add_cd_image(
diff --git a/atlas-start.tim b/atlas-start.tim
new file mode 100644
index 0000000..70ec62b
Binary files /dev/null and b/atlas-start.tim differ
diff --git a/common.h b/common.h
index 527a20f..4daf011 100644
--- a/common.h
+++ b/common.h
@@ -10,6 +10,11 @@
 #include <psxgte.h>
 #include <psxgpu.h>
 
+typedef struct _STVECTOR {
+	int16_t vx, vy, vz;
+    int16_t u, v, pad;
+} STVECTOR;
+
 #define INLINE  __attribute__((always_inline)) inline
 
 #include "memory.h"
@@ -19,4 +24,6 @@ extern VECTOR cam_pos;
 extern SVECTOR cam_rot;
 extern u_short cam_leaf;
 
+extern u_short quake_clut;
+
 #endif  // __COMMON_H__
diff --git a/main.c b/main.c
index fb278dd..e773ebc 100644
--- a/main.c
+++ b/main.c
@@ -5,7 +5,7 @@
 #include "asset.h"
 #include "world.h"
 
-extern u_long tim_e1m1[];
+extern u_long tim_start[];
 extern u_long bsp_test[];
 
 VECTOR cam_pos = { 2176, 1152, 272 };	// START
@@ -15,6 +15,8 @@ u_short cam_leaf = 0;
 
 world_t world;
 
+u_short quake_clut;
+
 // BSP face rendering:
 // - Gather vertex data from face start index + length
 // - Store vertex data in scratchpad memory (should be plenty of space for one face)
@@ -29,7 +31,9 @@ void init(void)
 	display_init();
 	time_init();
 
-	//asset_loadTexture(tim_e1m1, NULL);
+	ps1texture_t maptex;
+	asset_loadTexture(tim_start, &maptex);
+	quake_clut = getClut(maptex.crect.x, maptex.crect.y);
 
 	world_load(bsp_test, &world);
 }
diff --git a/ps1bsp.h b/ps1bsp.h
index 74fd2cb..b355ae4 100755
--- a/ps1bsp.h
+++ b/ps1bsp.h
@@ -28,6 +28,7 @@ typedef struct
 {
 	u_short version;
 
+	ps1bsp_dentry_t textures;
 	ps1bsp_dentry_t vertices;
 	ps1bsp_dentry_t faces;
 	ps1bsp_dentry_t faceVertices;
@@ -41,17 +42,15 @@ typedef struct
 typedef struct
 {
 	unsigned char w, h;			// These may be necessary for scaling UVs, especially since we use a mix of mip0 and mip1 textures
-	int tpage;					// Texture page in PS1 VRAM (precalculated when generating the texture atlas)
-	short uoffs, voffs;			// Texture coordinate offset within the texture page
+	unsigned short tpage;		// Texture page in PS1 VRAM (precalculated when generating the texture atlas)
+	unsigned char uoffs, voffs;	// Texture coordinate offset within the texture page
 	unsigned short nextframe;	// If non-zero, the texture is animated and this points to the next texture in the sequence
 } ps1bsp_texture_t;
 
 // This matches the SVECTOR data type; we can use the extra padding to store some more data.
 typedef struct
 {
-	short x;
-	short y;
-	short z;
+	short x, y, z;
 	short pad;
 } ps1bsp_vertex_t;
 
@@ -60,8 +59,13 @@ typedef struct
 	unsigned short index;
 	unsigned short light;
 
-	// TODO: add texture uv's
-	// TODO: add sampled texture color * light, for untextured gouraud shaded drawing at range
+	unsigned char u, v;		// TODO: make into unsigned short, clamp/mask/modulo to u_char at run-time. So we can build tiling polygons later.
+	
+	// Sampled texture color * light, for untextured gouraud shaded drawing at range
+	unsigned char a : 1;
+	unsigned char r : 5;
+	unsigned char g : 5;
+	unsigned char b : 5;
 } ps1bsp_facevertex_t;
 
 typedef struct
@@ -70,7 +74,9 @@ typedef struct
 	unsigned short side;
 
 	unsigned short firstFaceVertex;
-	unsigned short numFaceVertices;
+	unsigned char numFaceVertices;
+
+	unsigned char textureId;
 
 	SVECTOR center;
 
@@ -120,7 +126,7 @@ typedef struct
 typedef struct
 {
 	unsigned short length;
-	char message[];
+	char message[1];
 } ps1bsp_message_t;
 
 #ifdef __cplusplus
diff --git a/test.ps1bsp b/test.ps1bsp
index 3b0d9cb..49a0e86 100755
Binary files a/test.ps1bsp and b/test.ps1bsp differ
diff --git a/world.c b/world.c
index 30b4a70..0898208 100644
--- a/world.c
+++ b/world.c
@@ -39,6 +39,7 @@ void world_load(const u_long *data, world_t *world)
 
     ps1bsp_header_t* header = (ps1bsp_header_t*)bytes;
 
+    LOAD_CHUNK(ps1bsp_texture_t, world->textures, world->numTextures, bytes, header->textures);
     LOAD_CHUNK(ps1bsp_vertex_t, world->vertices, world->numVertices, bytes, header->vertices);
     LOAD_CHUNK(ps1bsp_face_t, world->faces, world->numFaces, bytes, header->faces);
     LOAD_CHUNK(ps1bsp_facevertex_t, world->faceVertices, world->numFaceVertices, bytes, header->faceVertices);
@@ -203,6 +204,70 @@ static INLINE void drawface_quad_strip(const ps1bsp_face_t *face, SVECTOR *vecs)
     }
 }
 
+static INLINE void drawface_quad_strip_tex(const ps1bsp_face_t *face, STVECTOR *vecs, u_short tpage)
+{
+    int p;
+
+    if (!mem_checkprim(sizeof(POLY_GT4), face->numFaceVertices))
+        return;
+
+    // Draw the face as a quad strip
+    const STVECTOR *v0, *v1, *v2, *v3;
+    const STVECTOR *head = vecs;
+    const STVECTOR *tail = vecs + face->numFaceVertices;
+
+    // Initialize the first two vertices
+    v2 = --tail;
+    v3 = head++;
+
+    // Normally a quad strip would have (N-2)/2 quads, but we might end up with a sole triangle at the end which will be drawn as a collapsed quad
+    u_char numQuads = (face->numFaceVertices - 1) / 2;
+    for (u_char quadIdx = 0; quadIdx < numQuads; ++quadIdx)
+    {
+        v0 = v2;
+        v1 = v3;
+        v2 = --tail;
+        v3 = head++;
+
+        // Naively draw the quad with GTE, nothing special or optimized about this
+        gte_ldv3(v0, v1, v2);
+        gte_rtpt();     // Rotation, translation, perspective projection
+
+        // Average Z for depth sorting and culling
+        gte_avsz3();
+        gte_stotz(&p);
+        short depth = p >> 2;
+        if (depth <= 0 || depth >= OTLEN)
+            continue;
+
+        // Draw a flat-shaded untextured colored quad
+        POLY_GT4 *poly = (POLY_GT4*)mem_prim(sizeof(POLY_GT4));
+        setPolyGT4(poly);
+        gte_stsxy0(&poly->x0);
+        gte_stsxy1(&poly->x1);
+        gte_stsxy2(&poly->x2);
+
+        // Transform the fourth vertex to complete the quad
+        gte_ldv0(v3);
+        gte_rtps();
+        gte_stsxy(&poly->x3);
+
+        // Texture UVs
+        setUV4(poly, v0->u, v0->v, v1->u, v1->v, v2->u, v2->v, v3->u, v3->v);
+        poly->clut = quake_clut;
+        poly->tpage = tpage;
+
+        // Vertex color lighting
+        poly->r0 = poly->g0 = poly->b0 = (uint8_t)v0->pad;
+        poly->r1 = poly->g1 = poly->b1 = (uint8_t)v1->pad;
+        poly->r2 = poly->g2 = poly->b2 = (uint8_t)v2->pad;
+        poly->r3 = poly->g3 = poly->b3 = (uint8_t)v3->pad;
+
+        addPrim(curOT + depth, poly);
+        ++polyCount;
+    }
+}
+
 static INLINE short world_pointPlaneDist(const VECTOR *point, const ps1bsp_plane_t *plane)
 {
     // Make use of axis-aligned planes to skip the need for a dot product
@@ -237,22 +302,47 @@ static void world_drawface(const world_t *world, const ps1bsp_face_t *face)
     if ((dot >= 0) ^ face->side)
         return;
 
-    SVECTOR *vecs = (SVECTOR*)(scratchpad + 256);
-
-    // Copy this face's vertices into scratch RAM for fast reuse
-    // TODO: this is the main performance bottleneck right now!
-    ps1bsp_facevertex_t *faceVertex = &world->faceVertices[face->firstFaceVertex];
-    for (int vertIdx = 0; vertIdx < face->numFaceVertices; ++vertIdx, ++faceVertex)
+    if (0)
     {
-        const ps1bsp_vertex_t *vert = &world->vertices[faceVertex->index];
-        vecs[vertIdx] = *((SVECTOR*)vert);
-        vecs[vertIdx].pad = faceVertex->light;
-    }
+        // Draw untextured, vertex colored polygons
+        SVECTOR *vecs = (SVECTOR*)(scratchpad + 256);
+
+        // Copy this face's vertices into scratch RAM for fast reuse
+        // TODO: this is the main performance bottleneck right now!
+        ps1bsp_facevertex_t *faceVertex = &world->faceVertices[face->firstFaceVertex];
+        SVECTOR *curVec = vecs;
+        for (int vertIdx = 0; vertIdx < face->numFaceVertices; ++vertIdx, ++faceVertex, ++curVec)
+        {
+            const ps1bsp_vertex_t *vert = &world->vertices[faceVertex->index];
+            *curVec = *((SVECTOR*)vert);
+            curVec->pad = faceVertex->light;
+        }
 
-    if (face->numFaceVertices == 3) // Special case: draw single triangles using the simplest method
-        drawface_triangle_fan(face, vecs);
+        if (face->numFaceVertices == 3) // Special case: draw single triangles using the simplest method
+            drawface_triangle_fan(face, vecs);
+        else
+            drawface_quad_strip(face, vecs);
+    }
     else
-        drawface_quad_strip(face, vecs);
+    {
+        // Draw textured, vertex colored polygons
+        STVECTOR *vecs = (STVECTOR*)(scratchpad + 256);
+
+        // Copy this face's vertices into scratch RAM for fast reuse
+        ps1bsp_facevertex_t *faceVertex = &world->faceVertices[face->firstFaceVertex];
+        ps1bsp_texture_t *faceTexture = &world->textures[face->textureId];
+        STVECTOR *curVec = vecs;
+        for (int vertIdx = 0; vertIdx < face->numFaceVertices; ++vertIdx, ++faceVertex, ++curVec)
+        {
+            const ps1bsp_vertex_t *vert = &world->vertices[faceVertex->index];
+            *((SVECTOR*)curVec) = *((SVECTOR*)vert);
+            curVec->u = faceVertex->u + faceTexture->uoffs;
+            curVec->v = faceVertex->v + faceTexture->voffs;
+            curVec->pad = faceVertex->light;
+        }
+
+        drawface_quad_strip_tex(face, vecs, faceTexture->tpage);
+    }
 }
 
 static void world_drawnode(const world_t *world, short nodeIdx, u_char *pvs)
diff --git a/world.h b/world.h
index fe23b3a..0f2bf6e 100644
--- a/world.h
+++ b/world.h
@@ -5,6 +5,9 @@
 
 typedef struct
 {
+    u_short numTextures;
+    ps1bsp_texture_t *textures;
+
     u_short numVertices;
     ps1bsp_vertex_t *vertices;