From 943259f6b4af7bc77429fe44f75e263086dae68a Mon Sep 17 00:00:00 2001
From: SmileTheory <SmileTheory@gmail.com>
Date: Mon, 3 Nov 2014 19:10:35 -0800
Subject: [PATCH 01/14] OpenGL2: Support half floats for texcoords and vertex
 colors.

---
 code/renderergl2/tr_animation.c  |   2 +-
 code/renderergl2/tr_extensions.c |  23 ++++
 code/renderergl2/tr_extramath.c  |  61 +++++-----
 code/renderergl2/tr_init.c       |   2 +
 code/renderergl2/tr_local.h      |  11 +-
 code/renderergl2/tr_model.c      |  34 ++----
 code/renderergl2/tr_model_iqm.c  |   4 +-
 code/renderergl2/tr_shade_calc.c |   2 +-
 code/renderergl2/tr_surface.c    |  20 ++--
 code/renderergl2/tr_vbo.c        | 186 ++++++++++++++++++-------------
 10 files changed, 192 insertions(+), 153 deletions(-)

diff --git a/code/renderergl2/tr_animation.c b/code/renderergl2/tr_animation.c
index 7153556d88..74226752e3 100644
--- a/code/renderergl2/tr_animation.c
+++ b/code/renderergl2/tr_animation.c
@@ -412,7 +412,7 @@ void RB_MDRSurfaceAnim( mdrSurface_t *surface )
 		tess.xyz[baseVertex + j][1] = tempVert[1];
 		tess.xyz[baseVertex + j][2] = tempVert[2];
 
-		tess.normal[baseVertex + j] = R_VaoPackNormal(tempNormal);
+		R_VaoPackNormal((byte *)&tess.normal[baseVertex + j], tempNormal);
 
 		tess.texCoords[baseVertex + j][0][0] = v->texCoords[0];
 		tess.texCoords[baseVertex + j][0][1] = v->texCoords[1];
diff --git a/code/renderergl2/tr_extensions.c b/code/renderergl2/tr_extensions.c
index b4a8db5437..357d4795fb 100644
--- a/code/renderergl2/tr_extensions.c
+++ b/code/renderergl2/tr_extensions.c
@@ -709,4 +709,27 @@ void GLimp_InitExtraExtensions()
 		ri.Printf(PRINT_ALL, result[2], extension);
 	}
 
+	// GL_ARB_half_float_vertex
+	extension = "GL_ARB_half_float_vertex";
+	glRefConfig.packedTexcoordDataType = GL_FLOAT;
+	glRefConfig.packedTexcoordDataSize = sizeof(float) * 2;
+	glRefConfig.packedColorDataType    = GL_FLOAT;
+	glRefConfig.packedColorDataSize    = sizeof(float) * 4;
+	if( GLimp_HaveExtension( extension ) )
+	{
+		if (r_arb_half_float_vertex->integer)
+		{
+			glRefConfig.packedTexcoordDataType = GL_HALF_FLOAT;
+			glRefConfig.packedTexcoordDataSize = sizeof(uint16_t) * 2;
+			glRefConfig.packedColorDataType    = GL_HALF_FLOAT;
+			glRefConfig.packedColorDataSize    = sizeof(uint16_t) * 4;
+		}
+
+		ri.Printf(PRINT_ALL, result[r_arb_half_float_vertex->integer ? 1 : 0], extension);
+	}
+	else
+	{
+		ri.Printf(PRINT_ALL, result[2], extension);
+	}
+
 }
diff --git a/code/renderergl2/tr_extramath.c b/code/renderergl2/tr_extramath.c
index b844678f2e..da9a94d25d 100644
--- a/code/renderergl2/tr_extramath.c
+++ b/code/renderergl2/tr_extramath.c
@@ -199,42 +199,35 @@ int NextPowerOfTwo(int in)
 	return out;
 }
 
-unsigned short FloatToHalf(float in)
-{
-	unsigned short out;
-	
-	union
-	{
-		float f;
-		unsigned int i;
-	} f32;
-
-	int sign, inExponent, inFraction;
-	int outExponent, outFraction;
+union f32_u {
+	float f;
+	uint32_t i;
+	struct {
+		unsigned int fraction:23;
+		unsigned int exponent:8;
+		unsigned int sign:1;
+	} pack;
+};
+
+union f16_u {
+	uint16_t i;
+	struct {
+		unsigned int fraction:10;
+		unsigned int exponent:5;
+		unsigned int sign:1;
+	} pack;
+};
+
+uint16_t FloatToHalf(float in)
+{
+	union f32_u f32;
+	union f16_u f16;
 
 	f32.f = in;
 
-	sign       = (f32.i & 0x80000000) >> 31;
-	inExponent = (f32.i & 0x7F800000) >> 23;
-	inFraction =  f32.i & 0x007FFFFF;
-
-	outExponent = CLAMP(inExponent - 127, -15, 16) + 15;
-
-	outFraction = 0;
-	if (outExponent == 0x1F)
-	{
-		if (inExponent == 0xFF && inFraction != 0)
-			outFraction = 0x3FF;
-	}
-	else if (outExponent == 0x00)
-	{
-		if (inExponent == 0x00 && inFraction != 0)
-			outFraction = 0x3FF;
-	}
-	else
-		outFraction = inFraction >> 13;
-
-	out = (sign << 15) | (outExponent << 10) | outFraction;
+	f16.pack.exponent = CLAMP(f32.pack.exponent - 112, 0, 31);
+	f16.pack.fraction = f32.pack.fraction >> 13;
+	f16.pack.sign     = f32.pack.sign;
 
-	return out;
+	return f16.i;
 }
diff --git a/code/renderergl2/tr_init.c b/code/renderergl2/tr_init.c
index eb185305ad..e79d735d1a 100644
--- a/code/renderergl2/tr_init.c
+++ b/code/renderergl2/tr_init.c
@@ -101,6 +101,7 @@ cvar_t  *r_ext_multi_draw_arrays;
 cvar_t  *r_ext_framebuffer_object;
 cvar_t  *r_ext_texture_float;
 cvar_t  *r_arb_half_float_pixel;
+cvar_t  *r_arb_half_float_vertex;
 cvar_t  *r_ext_framebuffer_multisample;
 cvar_t  *r_arb_seamless_cube_map;
 cvar_t  *r_arb_vertex_type_2_10_10_10_rev;
@@ -1141,6 +1142,7 @@ void R_Register( void )
 	r_ext_framebuffer_object = ri.Cvar_Get( "r_ext_framebuffer_object", "1", CVAR_ARCHIVE | CVAR_LATCH);
 	r_ext_texture_float = ri.Cvar_Get( "r_ext_texture_float", "1", CVAR_ARCHIVE | CVAR_LATCH);
 	r_arb_half_float_pixel = ri.Cvar_Get( "r_arb_half_float_pixel", "1", CVAR_ARCHIVE | CVAR_LATCH);
+	r_arb_half_float_vertex = ri.Cvar_Get( "r_arb_half_float_vertex", "1", CVAR_ARCHIVE | CVAR_LATCH);
 	r_ext_framebuffer_multisample = ri.Cvar_Get( "r_ext_framebuffer_multisample", "0", CVAR_ARCHIVE | CVAR_LATCH);
 	r_arb_seamless_cube_map = ri.Cvar_Get( "r_arb_seamless_cube_map", "0", CVAR_ARCHIVE | CVAR_LATCH);
 	r_arb_vertex_type_2_10_10_10_rev = ri.Cvar_Get( "r_arb_vertex_type_2_10_10_10_rev", "1", CVAR_ARCHIVE | CVAR_LATCH);
diff --git a/code/renderergl2/tr_local.h b/code/renderergl2/tr_local.h
index 25517272ff..e4a5284ec3 100644
--- a/code/renderergl2/tr_local.h
+++ b/code/renderergl2/tr_local.h
@@ -1416,6 +1416,10 @@ typedef struct {
 	qboolean seamlessCubeMap;
 
 	GLenum packedNormalDataType;
+	GLenum packedTexcoordDataType;
+	GLenum packedColorDataType;
+	int packedTexcoordDataSize;
+	int packedColorDataSize;
 
 	qboolean floatLightmap;
 	qboolean vertexArrayObject;
@@ -1713,6 +1717,7 @@ extern  cvar_t  *r_ext_multi_draw_arrays;
 extern  cvar_t  *r_ext_framebuffer_object;
 extern  cvar_t  *r_ext_texture_float;
 extern  cvar_t  *r_arb_half_float_pixel;
+extern  cvar_t  *r_arb_half_float_vertex;
 extern  cvar_t  *r_ext_framebuffer_multisample;
 extern  cvar_t  *r_arb_seamless_cube_map;
 extern  cvar_t  *r_arb_vertex_type_2_10_10_10_rev;
@@ -2180,8 +2185,10 @@ VERTEX BUFFER OBJECTS
 ============================================================
 */
 
-uint32_t R_VaoPackTangent(vec4_t v);
-uint32_t R_VaoPackNormal(vec3_t v);
+int R_VaoPackTangent(byte *out, vec4_t v);
+int R_VaoPackNormal(byte *out, vec3_t v);
+int R_VaoPackTexCoord(byte *out, vec2_t st);
+int R_VaoPackColors(byte *out, vec4_t color);
 void R_VaoUnpackTangent(vec4_t v, uint32_t b);
 void R_VaoUnpackNormal(vec3_t v, uint32_t b);
 
diff --git a/code/renderergl2/tr_model.c b/code/renderergl2/tr_model.c
index fc2017692f..897e202e96 100644
--- a/code/renderergl2/tr_model.c
+++ b/code/renderergl2/tr_model.c
@@ -674,10 +674,10 @@ static qboolean R_LoadMD3(model_t * mod, int lod, void *buffer, int bufferSize,
 			{
 				// vertex animation, store texcoords first, then position/normal/tangents
 				offset_st      = 0;
-				offset_xyz     = surf->numVerts * sizeof(vec2_t);
+				offset_xyz     = surf->numVerts * glRefConfig.packedTexcoordDataSize;
 				offset_normal  = offset_xyz + sizeof(vec3_t);
 				offset_tangent = offset_normal + sizeof(uint32_t);
-				stride_st  = sizeof(vec2_t);
+				stride_st  = glRefConfig.packedTexcoordDataSize;
 				stride_xyz = sizeof(vec3_t) + sizeof(uint32_t);
 #ifdef USE_VERT_TANGENT_SPACE
 				stride_xyz += sizeof(uint32_t);
@@ -691,7 +691,7 @@ static qboolean R_LoadMD3(model_t * mod, int lod, void *buffer, int bufferSize,
 				// no animation, interleave everything
 				offset_xyz     = 0;
 				offset_st      = offset_xyz + sizeof(vec3_t);
-				offset_normal  = offset_st + sizeof(vec2_t);
+				offset_normal  = offset_st + glRefConfig.packedTexcoordDataSize;
 				offset_tangent = offset_normal + sizeof(uint32_t);
 #ifdef USE_VERT_TANGENT_SPACE
 				stride_xyz = offset_tangent + sizeof(uint32_t);
@@ -711,8 +711,7 @@ static qboolean R_LoadMD3(model_t * mod, int lod, void *buffer, int bufferSize,
 			{
 				st = surf->st;
 				for ( j = 0 ; j < surf->numVerts ; j++, st++ ) {
-					memcpy(data + dataOfs, &st->st, sizeof(vec2_t));
-					dataOfs += sizeof(vec2_t);
+					dataOfs += R_VaoPackTexCoord(data + dataOfs, st->st);
 				}
 
 				v = surf->verts;
@@ -722,16 +721,12 @@ static qboolean R_LoadMD3(model_t * mod, int lod, void *buffer, int bufferSize,
 					vec3_t nxt;
 					vec4_t tangent;
 #endif
-					uint32_t *p;
-
 					// xyz
 					memcpy(data + dataOfs, &v->xyz, sizeof(vec3_t));
 					dataOfs += sizeof(vec3_t);
 
 					// normal
-					p = (uint32_t *)(data + dataOfs);
-					*p = R_VaoPackNormal(v->normal);
-					dataOfs += sizeof(uint32_t);
+					dataOfs += R_VaoPackNormal(data + dataOfs, v->normal);
 
 #ifdef USE_VERT_TANGENT_SPACE
 					CrossProduct(v->normal, v->tangent, nxt);
@@ -739,9 +734,7 @@ static qboolean R_LoadMD3(model_t * mod, int lod, void *buffer, int bufferSize,
 					tangent[3] = (DotProduct(nxt, v->bitangent) < 0.0f) ? -1.0f : 1.0f;
 
 					// tangent
-					p = (uint32_t *)(data + dataOfs);
-					*p = R_VaoPackTangent(tangent);
-					dataOfs += sizeof(uint32_t);
+					dataOfs += R_VaoPackTangent(data + dataOfs, tangent);
 #endif
 				}
 			}
@@ -755,20 +748,15 @@ static qboolean R_LoadMD3(model_t * mod, int lod, void *buffer, int bufferSize,
 					vec3_t nxt;
 					vec4_t tangent;
 #endif
-					uint32_t *p;
-
 					// xyz
 					memcpy(data + dataOfs, &v->xyz, sizeof(vec3_t));
 					dataOfs += sizeof(v->xyz);
 
 					// st
-					memcpy(data + dataOfs, &st->st, sizeof(vec2_t));
-					dataOfs += sizeof(st->st);
+					dataOfs += R_VaoPackTexCoord(data + dataOfs, st->st);
 
 					// normal
-					p = (uint32_t *)(data + dataOfs);
-					*p = R_VaoPackNormal(v->normal);
-					dataOfs += sizeof(uint32_t);
+					dataOfs += R_VaoPackNormal(data + dataOfs, v->normal);
 
 #ifdef USE_VERT_TANGENT_SPACE
 					CrossProduct(v->normal, v->tangent, nxt);
@@ -776,9 +764,7 @@ static qboolean R_LoadMD3(model_t * mod, int lod, void *buffer, int bufferSize,
 					tangent[3] = (DotProduct(nxt, v->bitangent) < 0.0f) ? -1.0f : 1.0f;
 
 					// tangent
-					p = (uint32_t *)(data + dataOfs);
-					*p = R_VaoPackTangent(tangent);
-					dataOfs += sizeof(uint32_t);
+					dataOfs += R_VaoPackTangent(data + dataOfs, tangent);
 #endif
 				}
 			}
@@ -807,7 +793,7 @@ static qboolean R_LoadMD3(model_t * mod, int lod, void *buffer, int bufferSize,
 			vaoSurf->vao->attribs[ATTR_INDEX_TANGENT ].count = 4;
 
 			vaoSurf->vao->attribs[ATTR_INDEX_POSITION].type = GL_FLOAT;
-			vaoSurf->vao->attribs[ATTR_INDEX_TEXCOORD].type = GL_FLOAT;
+			vaoSurf->vao->attribs[ATTR_INDEX_TEXCOORD].type = glRefConfig.packedTexcoordDataType;
 			vaoSurf->vao->attribs[ATTR_INDEX_NORMAL  ].type = glRefConfig.packedNormalDataType;
 			vaoSurf->vao->attribs[ATTR_INDEX_TANGENT ].type = glRefConfig.packedNormalDataType;
 
diff --git a/code/renderergl2/tr_model_iqm.c b/code/renderergl2/tr_model_iqm.c
index f0b2b38848..376e703092 100644
--- a/code/renderergl2/tr_model_iqm.c
+++ b/code/renderergl2/tr_model_iqm.c
@@ -1130,7 +1130,7 @@ void RB_IQMSurfaceAnim( surfaceType_t *surface ) {
 			normal[1] = DotProduct(&nrmMat[3], &data->normals[3*vtx]);
 			normal[2] = DotProduct(&nrmMat[6], &data->normals[3*vtx]);
 
-			*outNormal = R_VaoPackNormal(normal);
+			R_VaoPackNormal((byte *)outNormal++, normal);
 
 #ifdef USE_VERT_TANGENT_SPACE
 			tangent[0] = DotProduct(&nrmMat[0], &data->tangents[4*vtx]);
@@ -1138,7 +1138,7 @@ void RB_IQMSurfaceAnim( surfaceType_t *surface ) {
 			tangent[2] = DotProduct(&nrmMat[6], &data->tangents[4*vtx]);
 			tangent[3] = data->tangents[4*vtx+3];
 
-			*outTangent++ = R_VaoPackTangent(tangent);
+			R_VaoPackTangent((byte *)outTangent++, tangent);
 #endif
 		}
 
diff --git a/code/renderergl2/tr_shade_calc.c b/code/renderergl2/tr_shade_calc.c
index c5cc651ea3..bb16bb0d00 100644
--- a/code/renderergl2/tr_shade_calc.c
+++ b/code/renderergl2/tr_shade_calc.c
@@ -189,7 +189,7 @@ void RB_CalcDeformNormals( deformStage_t *ds ) {
 
 		VectorNormalizeFast( fNormal );
 
-		*normal = R_VaoPackNormal(fNormal);
+		R_VaoPackNormal((byte *)normal, fNormal);
 	}
 }
 
diff --git a/code/renderergl2/tr_surface.c b/code/renderergl2/tr_surface.c
index d4617ec482..2b0ae0f144 100644
--- a/code/renderergl2/tr_surface.c
+++ b/code/renderergl2/tr_surface.c
@@ -88,6 +88,7 @@ RB_AddQuadStampExt
 */
 void RB_AddQuadStampExt( vec3_t origin, vec3_t left, vec3_t up, float color[4], float s1, float t1, float s2, float t2 ) {
 	vec3_t		normal;
+	uint32_t    pNormal;
 	int			ndx;
 
 	RB_CHECKOVERFLOW( 4, 6 );
@@ -123,10 +124,11 @@ void RB_AddQuadStampExt( vec3_t origin, vec3_t left, vec3_t up, float color[4],
 	// constant normal all the way around
 	VectorSubtract( vec3_origin, backEnd.viewParms.or.axis[0], normal );
 
+	R_VaoPackNormal((byte *)&pNormal, normal);
 	tess.normal[ndx] =
 	tess.normal[ndx+1] =
 	tess.normal[ndx+2] =
-	tess.normal[ndx+3] = R_VaoPackNormal(normal);
+	tess.normal[ndx+3] = pNormal;
 
 	// standard square texture coordinates
 	VectorSet2(tess.texCoords[ndx  ][0], s1, t1);
@@ -346,7 +348,7 @@ static void RB_SurfaceVertsAndIndexes( int numVerts, srfVert_t *verts, int numIn
 		dv = verts;
 		normal = &tess.normal[ tess.numVertexes ];
 		for ( i = 0 ; i < numVerts ; i++, dv++, normal++ )
-			*normal = R_VaoPackNormal(dv->normal);
+			R_VaoPackNormal((byte *)normal, dv->normal);
 	}
 
 #ifdef USE_VERT_TANGENT_SPACE
@@ -355,7 +357,7 @@ static void RB_SurfaceVertsAndIndexes( int numVerts, srfVert_t *verts, int numIn
 		dv = verts;
 		tangent = &tess.tangent[ tess.numVertexes ];
 		for ( i = 0 ; i < numVerts ; i++, dv++, tangent++ )
-			*tangent = R_VaoPackTangent(dv->tangent);
+			R_VaoPackTangent((byte *)tangent, dv->tangent);
 	}
 #endif
 
@@ -388,7 +390,7 @@ static void RB_SurfaceVertsAndIndexes( int numVerts, srfVert_t *verts, int numIn
 		dv = verts;
 		lightdir = &tess.lightdir[ tess.numVertexes ];
 		for ( i = 0 ; i < numVerts ; i++, dv++, lightdir++ )
-			*lightdir = R_VaoPackNormal(dv->lightdir);
+			R_VaoPackNormal((byte *)lightdir, dv->lightdir);
 	}
 
 #if 0  // nothing even uses vertex dlightbits
@@ -1147,7 +1149,7 @@ static void LerpMeshVertexes_scalar(mdvSurface_t *surf, float backlerp)
 			VectorCopy(newVerts->xyz,    outXyz);
 			VectorCopy(newVerts->normal, normal);
 
-			*outNormal = R_VaoPackNormal(normal);
+			R_VaoPackNormal((byte *)outNormal, normal);
 
 			newVerts++;
 			outXyz += 4;
@@ -1172,7 +1174,7 @@ static void LerpMeshVertexes_scalar(mdvSurface_t *surf, float backlerp)
 			VectorLerp(newVerts->normal, oldVerts->normal, backlerp, normal);
 			VectorNormalize(normal);
 
-			*outNormal = R_VaoPackNormal(normal);
+			R_VaoPackNormal((byte *)outNormal, normal);
 
 			newVerts++;
 			oldVerts++;
@@ -1410,13 +1412,13 @@ static void RB_SurfaceGrid( srfBspSurface_t *srf ) {
 
 				if ( tess.shader->vertexAttribs & ATTR_NORMAL )
 				{
-					*normal++ = R_VaoPackNormal(dv->normal);
+					R_VaoPackNormal((byte *)normal++, dv->normal);
 				}
 
 #ifdef USE_VERT_TANGENT_SPACE
 				if ( tess.shader->vertexAttribs & ATTR_TANGENT )
 				{
-					*tangent++ = R_VaoPackTangent(dv->tangent);
+					R_VaoPackTangent((byte *)tangent++, dv->tangent);
 				}
 #endif
 				if ( tess.shader->vertexAttribs & ATTR_TEXCOORD )
@@ -1439,7 +1441,7 @@ static void RB_SurfaceGrid( srfBspSurface_t *srf ) {
 
 				if ( tess.shader->vertexAttribs & ATTR_LIGHTDIRECTION )
 				{
-					*lightdir++ = R_VaoPackNormal(dv->lightdir);
+					R_VaoPackNormal((byte *)lightdir++, dv->lightdir);
 				}
 
 				//*vDlightBits++ = dlightBits;
diff --git a/code/renderergl2/tr_vbo.c b/code/renderergl2/tr_vbo.c
index 981741fba7..8f9ba66fe7 100644
--- a/code/renderergl2/tr_vbo.c
+++ b/code/renderergl2/tr_vbo.c
@@ -44,81 +44,122 @@ union pack8_u {
 };
 
 
-uint32_t R_VaoPackTangent(vec4_t v)
+int R_VaoPackTangent(byte *out, vec4_t v)
 {
 	if (glRefConfig.packedNormalDataType == GL_INT_2_10_10_10_REV)
 	{
-		union pack10_u num;
+		union pack10_u *num = (union pack10_u *)out;
 
-		num.pack.x = v[0] * 511.0f;
-		num.pack.y = v[1] * 511.0f;
-		num.pack.z = v[2] * 511.0f;
-		num.pack.w = v[3];
+		num->pack.x = v[0] * 511.0f;
+		num->pack.y = v[1] * 511.0f;
+		num->pack.z = v[2] * 511.0f;
+		num->pack.w = v[3];
+	}
+	else
+	{
+		union pack8_u *num = (union pack8_u *)out;
+
+		num->pack.x = v[0] * 127.0f;
+		num->pack.y = v[1] * 127.0f;
+		num->pack.z = v[2] * 127.0f;
+		num->pack.w = v[3] * 127.0f;
+	}
+
+	return 4;
+}
+
+int R_VaoPackNormal(byte *out, vec3_t v)
+{
+	if (glRefConfig.packedNormalDataType == GL_INT_2_10_10_10_REV)
+	{
+		union pack10_u *num = (union pack10_u *)out;
+
+		num->pack.x = v[0] * 511.0f;
+		num->pack.y = v[1] * 511.0f;
+		num->pack.z = v[2] * 511.0f;
+		num->pack.w = 0;
+	}
+	else
+	{
+		union pack8_u *num = (union pack8_u *)out;
 
-		return num.i;
+		num->pack.x = v[0] * 127.0f;
+		num->pack.y = v[1] * 127.0f;
+		num->pack.z = v[2] * 127.0f;
+		num->pack.w = 0;
+	}
+
+	return 4;
+}
+
+int R_VaoPackTexCoord(byte *out, vec2_t st)
+{
+	if (glRefConfig.packedTexcoordDataType == GL_HALF_FLOAT)
+	{
+		uint16_t *num = (uint16_t *)out;
+
+		*num++ = FloatToHalf(st[0]);
+		*num++ = FloatToHalf(st[1]);
+
+		return sizeof(*num) * 2;
 	}
 	else
 	{
-		union pack8_u num;
+		float *num = (float *)out;
 
-		num.pack.x = v[0] * 127.0f;
-		num.pack.y = v[1] * 127.0f;
-		num.pack.z = v[2] * 127.0f;
-		num.pack.w = v[3] * 127.0f;
+		*num++ = st[0];
+		*num++ = st[1];
 
-		return num.i;
+		return sizeof(*num) * 2;
 	}
 }
 
-uint32_t R_VaoPackNormal(vec3_t v)
+int R_VaoPackColors(byte *out, vec4_t color)
 {
-	if (glRefConfig.packedNormalDataType == GL_INT_2_10_10_10_REV)
+	if (glRefConfig.packedTexcoordDataType == GL_HALF_FLOAT)
 	{
-		union pack10_u num;
+		uint16_t *num = (uint16_t *)out;
 
-		num.pack.x = v[0] * 511.0f;
-		num.pack.y = v[1] * 511.0f;
-		num.pack.z = v[2] * 511.0f;
-		num.pack.w = 0;
+		*num++ = FloatToHalf(color[0]);
+		*num++ = FloatToHalf(color[1]);
+		*num++ = FloatToHalf(color[2]);
+		*num++ = FloatToHalf(color[3]);
 
-		return num.i;
+		return sizeof(*num) * 4;
 	}
 	else
 	{
-		union pack8_u num;
+		float *num = (float *)out;
 
-		num.pack.x = v[0] * 127.0f;
-		num.pack.y = v[1] * 127.0f;
-		num.pack.z = v[2] * 127.0f;
-		num.pack.w = 0;
+		*num++ = color[0];
+		*num++ = color[1];
+		*num++ = color[2];
+		*num++ = color[3];
 
-		return num.i;
+		return sizeof(*num) * 4;
 	}
 }
 
+
 void R_VaoUnpackTangent(vec4_t v, uint32_t b)
 {
 	if (glRefConfig.packedNormalDataType == GL_INT_2_10_10_10_REV)
 	{
-		union pack10_u num;
+		union pack10_u *num = (union pack10_u *)&b;
 
-		num.i = b;
-
-		v[0] = num.pack.x / 511.0f;
-		v[1] = num.pack.y / 511.0f;
-		v[2] = num.pack.z / 511.0f;
-		v[3] = num.pack.w; 
+		v[0] = num->pack.x / 511.0f;
+		v[1] = num->pack.y / 511.0f;
+		v[2] = num->pack.z / 511.0f;
+		v[3] = num->pack.w; 
 	}
 	else
 	{
-		union pack8_u num;
-
-		num.i = b;
+		union pack8_u *num = (union pack8_u *)&b;
 
-		v[0] = num.pack.x / 127.0f;
-		v[1] = num.pack.y / 127.0f;
-		v[2] = num.pack.z / 127.0f;
-		v[3] = num.pack.w / 127.0f; 
+		v[0] = num->pack.x / 127.0f;
+		v[1] = num->pack.y / 127.0f;
+		v[2] = num->pack.z / 127.0f;
+		v[3] = num->pack.w / 127.0f; 
 	}
 }
 
@@ -126,23 +167,19 @@ void R_VaoUnpackNormal(vec3_t v, uint32_t b)
 {
 	if (glRefConfig.packedNormalDataType == GL_INT_2_10_10_10_REV)
 	{
-		union pack10_u num;
+		union pack10_u *num = (union pack10_u *)&b;
 
-		num.i = b;
-
-		v[0] = num.pack.x / 511.0f;
-		v[1] = num.pack.y / 511.0f;
-		v[2] = num.pack.z / 511.0f;
+		v[0] = num->pack.x / 511.0f;
+		v[1] = num->pack.y / 511.0f;
+		v[2] = num->pack.z / 511.0f;
 	}
 	else
 	{
-		union pack8_u num;
-
-		num.i = b;
+		union pack8_u *num = (union pack8_u *)&b;
 
-		v[0] = num.pack.x / 127.0f;
-		v[1] = num.pack.y / 127.0f;
-		v[2] = num.pack.z / 127.0f;
+		v[0] = num->pack.x / 127.0f;
+		v[1] = num->pack.y / 127.0f;
+		v[2] = num->pack.z / 127.0f;
 	}
 }
 
@@ -312,9 +349,9 @@ vao_t *R_CreateVao2(const char *name, int numVertexes, srfVert_t *verts, int num
 	vao->attribs[ATTR_INDEX_POSITION      ].type = GL_FLOAT;
 	vao->attribs[ATTR_INDEX_NORMAL        ].type = glRefConfig.packedNormalDataType;
 	vao->attribs[ATTR_INDEX_TANGENT       ].type = glRefConfig.packedNormalDataType;
-	vao->attribs[ATTR_INDEX_TEXCOORD      ].type = GL_FLOAT;
-	vao->attribs[ATTR_INDEX_LIGHTCOORD    ].type = GL_FLOAT;
-	vao->attribs[ATTR_INDEX_COLOR         ].type = GL_FLOAT;
+	vao->attribs[ATTR_INDEX_TEXCOORD      ].type = glRefConfig.packedTexcoordDataType;
+	vao->attribs[ATTR_INDEX_LIGHTCOORD    ].type = glRefConfig.packedTexcoordDataType;
+	vao->attribs[ATTR_INDEX_COLOR         ].type = glRefConfig.packedColorDataType;
 	vao->attribs[ATTR_INDEX_LIGHTDIRECTION].type = glRefConfig.packedNormalDataType;
 
 	vao->attribs[ATTR_INDEX_POSITION      ].normalized = GL_FALSE;
@@ -330,9 +367,9 @@ vao_t *R_CreateVao2(const char *name, int numVertexes, srfVert_t *verts, int num
 #ifdef USE_VERT_TANGENT_SPACE
 	vao->attribs[ATTR_INDEX_TANGENT       ].offset = dataSize; dataSize += sizeof(uint32_t);
 #endif
-	vao->attribs[ATTR_INDEX_TEXCOORD      ].offset = dataSize; dataSize += sizeof(verts[0].st);
-	vao->attribs[ATTR_INDEX_LIGHTCOORD    ].offset = dataSize; dataSize += sizeof(verts[0].lightmap);
-	vao->attribs[ATTR_INDEX_COLOR         ].offset = dataSize; dataSize += sizeof(verts[0].vertexColors);
+	vao->attribs[ATTR_INDEX_TEXCOORD      ].offset = dataSize; dataSize += glRefConfig.packedTexcoordDataSize;
+	vao->attribs[ATTR_INDEX_LIGHTCOORD    ].offset = dataSize; dataSize += glRefConfig.packedTexcoordDataSize;
+	vao->attribs[ATTR_INDEX_COLOR         ].offset = dataSize; dataSize += glRefConfig.packedColorDataSize;
 	vao->attribs[ATTR_INDEX_LIGHTDIRECTION].offset = dataSize; dataSize += sizeof(uint32_t);
 
 	vao->attribs[ATTR_INDEX_POSITION      ].stride = dataSize;
@@ -358,40 +395,29 @@ vao_t *R_CreateVao2(const char *name, int numVertexes, srfVert_t *verts, int num
 
 	for (i = 0; i < numVertexes; i++)
 	{
-		uint32_t *p;
-
 		// xyz
 		memcpy(data + dataOfs, &verts[i].xyz, sizeof(verts[i].xyz));
 		dataOfs += sizeof(verts[i].xyz);
 
 		// normal
-		p = (uint32_t *)(data + dataOfs);
-		*p = R_VaoPackNormal(verts[i].normal);
-		dataOfs += sizeof(uint32_t);
+		dataOfs += R_VaoPackNormal(data + dataOfs, verts[i].normal);
 
 #ifdef USE_VERT_TANGENT_SPACE
 		// tangent
-		p = (uint32_t *)(data + dataOfs);
-		*p = R_VaoPackTangent(verts[i].tangent);
-		dataOfs += sizeof(uint32_t);
+		dataOfs += R_VaoPackTangent(data + dataOfs, verts[i].tangent);
 #endif
 
-		// vertex texcoords
-		memcpy(data + dataOfs, &verts[i].st, sizeof(verts[i].st));
-		dataOfs += sizeof(verts[i].st);
+		// texcoords
+		dataOfs += R_VaoPackTexCoord(data + dataOfs, verts[i].st);
 
-		// feed vertex lightmap texcoords
-		memcpy(data + dataOfs, &verts[i].lightmap, sizeof(verts[i].lightmap));
-		dataOfs += sizeof(verts[i].lightmap);
+		// lightmap texcoords
+		dataOfs += R_VaoPackTexCoord(data + dataOfs, verts[i].lightmap);
 
-		// feed vertex colors
-		memcpy(data + dataOfs, &verts[i].vertexColors, sizeof(verts[i].vertexColors));
-		dataOfs += sizeof(verts[i].vertexColors);
+		// colors
+		dataOfs += R_VaoPackColors(data + dataOfs, verts[i].vertexColors);
 
-		// feed vertex light directions
-		p = (uint32_t *)(data + dataOfs);
-		*p = R_VaoPackNormal(verts[i].lightdir);
-		dataOfs += sizeof(uint32_t);
+		// light directions
+		dataOfs += R_VaoPackNormal(data + dataOfs, verts[i].lightdir);
 	}
 
 	vao->vertexesSize = dataSize;

From 1ad2b413efd36752a3b22ed44cbd442e58a123a9 Mon Sep 17 00:00:00 2001
From: SmileTheory <SmileTheory@gmail.com>
Date: Mon, 3 Nov 2014 22:18:17 -0800
Subject: [PATCH 02/14] Remove accidentally added increment.

---
 code/renderergl2/tr_model_iqm.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/code/renderergl2/tr_model_iqm.c b/code/renderergl2/tr_model_iqm.c
index 376e703092..3cabca0088 100644
--- a/code/renderergl2/tr_model_iqm.c
+++ b/code/renderergl2/tr_model_iqm.c
@@ -1130,7 +1130,7 @@ void RB_IQMSurfaceAnim( surfaceType_t *surface ) {
 			normal[1] = DotProduct(&nrmMat[3], &data->normals[3*vtx]);
 			normal[2] = DotProduct(&nrmMat[6], &data->normals[3*vtx]);
 
-			R_VaoPackNormal((byte *)outNormal++, normal);
+			R_VaoPackNormal((byte *)outNormal, normal);
 
 #ifdef USE_VERT_TANGENT_SPACE
 			tangent[0] = DotProduct(&nrmMat[0], &data->tangents[4*vtx]);

From 1d016e6ff5123c06a40c93adb1a5c4df3f697948 Mon Sep 17 00:00:00 2001
From: Zack Middleton <zturtleman@gmail.com>
Date: Fri, 7 Nov 2014 21:34:22 -0600
Subject: [PATCH 03/14] Clear window buffer when it's created

When starting the game in windowed mode, the window buffer used whatever
was on the screen before running the game. Kind of like you could see
through the window, but it doesn't update what happens behind it.

It makes it look like something is broken or non-responsive.

So clear the window opengl buffer to black.

Credit to theinvsblman for the code.
---
 code/sdl/sdl_glimp.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/code/sdl/sdl_glimp.c b/code/sdl/sdl_glimp.c
index 1f12fb72bc..df201686b1 100644
--- a/code/sdl/sdl_glimp.c
+++ b/code/sdl/sdl_glimp.c
@@ -462,6 +462,10 @@ static int GLimp_SetMode(int mode, qboolean fullscreen, qboolean noborder)
 			continue;
 		}
 
+		qglClearColor( 0, 0, 0, 1 );
+		qglClear( GL_COLOR_BUFFER_BIT );
+		SDL_GL_SwapWindow( SDL_window );
+
 		SDL_GL_SetSwapInterval( r_swapInterval->integer );
 
 		glConfig.colorBits = testColorBits;

From b1821e303daa2575cd232cc9923b6eaa1fa485a6 Mon Sep 17 00:00:00 2001
From: SmileTheory <SmileTheory@gmail.com>
Date: Mon, 10 Nov 2014 21:59:37 -0800
Subject: [PATCH 04/14] OpenGL2: remove lightmap support from generic glsl
 shader. This path was barely used and doing this compiles fewer shaders.

---
 code/renderergl2/glsl/generic_fp.glsl |  33 -----
 code/renderergl2/glsl/generic_vp.glsl |   9 +-
 code/renderergl2/tr_glsl.c            |  12 --
 code/renderergl2/tr_local.h           |   8 +-
 code/renderergl2/tr_shade.c           |  12 --
 code/renderergl2/tr_shader.c          | 167 --------------------------
 6 files changed, 3 insertions(+), 238 deletions(-)

diff --git a/code/renderergl2/glsl/generic_fp.glsl b/code/renderergl2/glsl/generic_fp.glsl
index aefa33c3fa..50db078547 100644
--- a/code/renderergl2/glsl/generic_fp.glsl
+++ b/code/renderergl2/glsl/generic_fp.glsl
@@ -1,45 +1,12 @@
 uniform sampler2D u_DiffuseMap;
 
-#if defined(USE_LIGHTMAP)
-uniform sampler2D u_LightMap;
-
-uniform int       u_Texture1Env;
-#endif
-
 varying vec2      var_DiffuseTex;
 
-#if defined(USE_LIGHTMAP)
-varying vec2      var_LightTex;
-#endif
-
 varying vec4      var_Color;
 
 
 void main()
 {
 	vec4 color  = texture2D(u_DiffuseMap, var_DiffuseTex);
-#if defined(USE_LIGHTMAP)
-	vec4 color2 = texture2D(u_LightMap, var_LightTex);
-  #if defined(RGBM_LIGHTMAP)
-	color2.rgb *= color2.a;
-	color2.a = 1.0;
-  #endif
-
-	if (u_Texture1Env == TEXENV_MODULATE)
-	{
-		color *= color2;
-	}
-	else if (u_Texture1Env == TEXENV_ADD)
-	{
-		color += color2;
-	}
-	else if (u_Texture1Env == TEXENV_REPLACE)
-	{
-		color = color2;
-	}
-	
-	//color = color * (u_Texture1Env.xxxx + color2 * u_Texture1Env.z) + color2 * u_Texture1Env.y;
-#endif
-
 	gl_FragColor = color * var_Color;
 }
diff --git a/code/renderergl2/glsl/generic_vp.glsl b/code/renderergl2/glsl/generic_vp.glsl
index 859b436383..bbe08e842c 100644
--- a/code/renderergl2/glsl/generic_vp.glsl
+++ b/code/renderergl2/glsl/generic_vp.glsl
@@ -9,7 +9,7 @@ attribute vec3 attr_Normal2;
 attribute vec4 attr_Color;
 attribute vec4 attr_TexCoord0;
 
-#if defined(USE_LIGHTMAP) || defined(USE_TCGEN)
+#if defined(USE_TCGEN)
 attribute vec4 attr_TexCoord1;
 #endif
 
@@ -57,9 +57,6 @@ uniform float  u_VertexLerp;
 #endif
 
 varying vec2   var_DiffuseTex;
-#if defined(USE_LIGHTMAP)
-varying vec2   var_LightTex;
-#endif
 varying vec4   var_Color;
 
 #if defined(USE_DEFORM_VERTEXES)
@@ -230,10 +227,6 @@ void main()
     var_DiffuseTex = tex;
 #endif
 
-#if defined(USE_LIGHTMAP)
-	var_LightTex = attr_TexCoord1.st;
-#endif
-
 #if defined(USE_RGBAGEN)
 	var_Color = CalcColor(position, normal);
 #else
diff --git a/code/renderergl2/tr_glsl.c b/code/renderergl2/tr_glsl.c
index 757bab3886..7dfea6a675 100644
--- a/code/renderergl2/tr_glsl.c
+++ b/code/renderergl2/tr_glsl.c
@@ -90,7 +90,6 @@ static uniformInfo_t uniformsInfo[] =
 
 	{ "u_DiffuseTexMatrix",  GLSL_VEC4 },
 	{ "u_DiffuseTexOffTurb", GLSL_VEC4 },
-	{ "u_Texture1Env",       GLSL_INT },
 
 	{ "u_TCGen0",        GLSL_INT },
 	{ "u_TCGen0Vector0", GLSL_VEC3 },
@@ -919,12 +918,6 @@ void GLSL_InitGPUShaders(void)
 		if (i & GENERICDEF_USE_RGBAGEN)
 			Q_strcat(extradefines, 1024, "#define USE_RGBAGEN\n");
 
-		if (i & GENERICDEF_USE_LIGHTMAP)
-			Q_strcat(extradefines, 1024, "#define USE_LIGHTMAP\n");
-
-		if (r_hdr->integer && !glRefConfig.floatLightmap)
-			Q_strcat(extradefines, 1024, "#define RGBM_LIGHTMAP\n");
-
 		if (!GLSL_InitGPUShader(&tr.genericShader[i], "generic", attribs, qtrue, extradefines, qtrue, fallbackShader_generic_vp, fallbackShader_generic_fp))
 		{
 			ri.Error(ERR_FATAL, "Could not load generic shader!");
@@ -1495,11 +1488,6 @@ shaderProgram_t *GLSL_GetGenericShaderProgram(int stage)
 		shaderAttribs |= GENERICDEF_USE_FOG;
 	}
 
-	if (pStage->bundle[1].image[0] && tess.shader->multitextureEnv)
-	{
-		shaderAttribs |= GENERICDEF_USE_LIGHTMAP;
-	}
-
 	switch (pStage->rgbGen)
 	{
 		case CGEN_LIGHTING_DIFFUSE:
diff --git a/code/renderergl2/tr_local.h b/code/renderergl2/tr_local.h
index e4a5284ec3..c21b7b4835 100644
--- a/code/renderergl2/tr_local.h
+++ b/code/renderergl2/tr_local.h
@@ -451,8 +451,6 @@ typedef struct shader_s {
 	float		portalRange;			// distance to fog out at
 	qboolean	isPortal;
 
-	int			multitextureEnv;		// 0, GL_MODULATE, GL_ADD (FIXME: put in stage)
-
 	cullType_t	cullType;				// CT_FRONT_SIDED, CT_BACK_SIDED, or CT_TWO_SIDED
 	qboolean	polygonOffset;			// set for decals and other items that must be offset 
 	qboolean	noMipMaps;				// for console fonts, 2D elements, etc.
@@ -563,9 +561,8 @@ enum
 	GENERICDEF_USE_VERTEX_ANIMATION = 0x0004,
 	GENERICDEF_USE_FOG              = 0x0008,
 	GENERICDEF_USE_RGBAGEN          = 0x0010,
-	GENERICDEF_USE_LIGHTMAP         = 0x0020,
-	GENERICDEF_ALL                  = 0x003F,
-	GENERICDEF_COUNT                = 0x0040,
+	GENERICDEF_ALL                  = 0x001F,
+	GENERICDEF_COUNT                = 0x0020,
 };
 
 enum
@@ -637,7 +634,6 @@ typedef enum
 
 	UNIFORM_DIFFUSETEXMATRIX,
 	UNIFORM_DIFFUSETEXOFFTURB,
-	UNIFORM_TEXTURE1ENV,
 
 	UNIFORM_TCGEN0,
 	UNIFORM_TCGEN0VECTOR0,
diff --git a/code/renderergl2/tr_shade.c b/code/renderergl2/tr_shade.c
index f70017e402..70b4353ed2 100644
--- a/code/renderergl2/tr_shade.c
+++ b/code/renderergl2/tr_shade.c
@@ -1351,16 +1351,6 @@ static void RB_IterateStagesGeneric( shaderCommands_t *input )
 		else if ( pStage->bundle[1].image[0] != 0 )
 		{
 			R_BindAnimatedImageToTMU( &pStage->bundle[0], 0 );
-
-			//
-			// lightmap/secondary pass
-			//
-			if ( r_lightmap->integer ) {
-				GLSL_SetUniformInt(sp, UNIFORM_TEXTURE1ENV, GL_REPLACE);
-			} else {
-				GLSL_SetUniformInt(sp, UNIFORM_TEXTURE1ENV, tess.shader->multitextureEnv);
-			}
-
 			R_BindAnimatedImageToTMU( &pStage->bundle[1], 1 );
 		}
 		else 
@@ -1369,8 +1359,6 @@ static void RB_IterateStagesGeneric( shaderCommands_t *input )
 			// set state
 			//
 			R_BindAnimatedImageToTMU( &pStage->bundle[0], 0 );
-
-			GLSL_SetUniformInt(sp, UNIFORM_TEXTURE1ENV, 0);
 		}
 
 		//
diff --git a/code/renderergl2/tr_shader.c b/code/renderergl2/tr_shader.c
index a8b13787d4..7ead2e57a7 100644
--- a/code/renderergl2/tr_shader.c
+++ b/code/renderergl2/tr_shader.c
@@ -2122,161 +2122,6 @@ static void ComputeVertexAttribs(void)
 	}
 }
 
-typedef struct {
-	int		blendA;
-	int		blendB;
-
-	int		multitextureEnv;
-	int		multitextureBlend;
-} collapse_t;
-
-static collapse_t	collapse[] = {
-	{ 0, GLS_DSTBLEND_SRC_COLOR | GLS_SRCBLEND_ZERO,	
-		GL_MODULATE, 0 },
-
-	{ 0, GLS_DSTBLEND_ZERO | GLS_SRCBLEND_DST_COLOR,
-		GL_MODULATE, 0 },
-
-	{ GLS_DSTBLEND_ZERO | GLS_SRCBLEND_DST_COLOR, GLS_DSTBLEND_ZERO | GLS_SRCBLEND_DST_COLOR,
-		GL_MODULATE, GLS_DSTBLEND_ZERO | GLS_SRCBLEND_DST_COLOR },
-
-	{ GLS_DSTBLEND_SRC_COLOR | GLS_SRCBLEND_ZERO, GLS_DSTBLEND_ZERO | GLS_SRCBLEND_DST_COLOR,
-		GL_MODULATE, GLS_DSTBLEND_ZERO | GLS_SRCBLEND_DST_COLOR },
-
-	{ GLS_DSTBLEND_ZERO | GLS_SRCBLEND_DST_COLOR, GLS_DSTBLEND_SRC_COLOR | GLS_SRCBLEND_ZERO,
-		GL_MODULATE, GLS_DSTBLEND_ZERO | GLS_SRCBLEND_DST_COLOR },
-
-	{ GLS_DSTBLEND_SRC_COLOR | GLS_SRCBLEND_ZERO, GLS_DSTBLEND_SRC_COLOR | GLS_SRCBLEND_ZERO,
-		GL_MODULATE, GLS_DSTBLEND_ZERO | GLS_SRCBLEND_DST_COLOR },
-
-	{ 0, GLS_DSTBLEND_ONE | GLS_SRCBLEND_ONE,
-		GL_ADD, 0 },
-
-	{ GLS_DSTBLEND_ONE | GLS_SRCBLEND_ONE, GLS_DSTBLEND_ONE | GLS_SRCBLEND_ONE,
-		GL_ADD, GLS_DSTBLEND_ONE | GLS_SRCBLEND_ONE },
-#if 0
-	{ 0, GLS_DSTBLEND_ONE_MINUS_SRC_ALPHA | GLS_SRCBLEND_SRC_ALPHA,
-		GL_DECAL, 0 },
-#endif
-	{ -1 }
-};
-
-/*
-================
-CollapseMultitexture
-
-Attempt to combine two stages into a single multitexture stage
-FIXME: I think modulated add + modulated add collapses incorrectly
-=================
-*/
-static qboolean CollapseMultitexture( void ) {
-	int abits, bbits;
-	int i;
-	textureBundle_t tmpBundle;
-
-	if ( !qglActiveTextureARB ) {
-		return qfalse;
-	}
-
-	// make sure both stages are active
-	if ( !stages[0].active || !stages[1].active ) {
-		return qfalse;
-	}
-
-	// on voodoo2, don't combine different tmus
-	if ( glConfig.driverType == GLDRV_VOODOO ) {
-		if ( stages[0].bundle[0].image[0]->TMU ==
-			 stages[1].bundle[0].image[0]->TMU ) {
-			return qfalse;
-		}
-	}
-
-	abits = stages[0].stateBits;
-	bbits = stages[1].stateBits;
-
-	// make sure that both stages have identical state other than blend modes
-	if ( ( abits & ~( GLS_DSTBLEND_BITS | GLS_SRCBLEND_BITS | GLS_DEPTHMASK_TRUE ) ) !=
-		( bbits & ~( GLS_DSTBLEND_BITS | GLS_SRCBLEND_BITS | GLS_DEPTHMASK_TRUE ) ) ) {
-		return qfalse;
-	}
-
-	abits &= ( GLS_DSTBLEND_BITS | GLS_SRCBLEND_BITS );
-	bbits &= ( GLS_DSTBLEND_BITS | GLS_SRCBLEND_BITS );
-
-	// search for a valid multitexture blend function
-	for ( i = 0; collapse[i].blendA != -1 ; i++ ) {
-		if ( abits == collapse[i].blendA
-			&& bbits == collapse[i].blendB ) {
-			break;
-		}
-	}
-
-	// nothing found
-	if ( collapse[i].blendA == -1 ) {
-		return qfalse;
-	}
-
-	// GL_ADD is a separate extension
-	if ( collapse[i].multitextureEnv == GL_ADD && !glConfig.textureEnvAddAvailable ) {
-		return qfalse;
-	}
-
-	// make sure waveforms have identical parameters
-	if ( ( stages[0].rgbGen != stages[1].rgbGen ) ||
-		( stages[0].alphaGen != stages[1].alphaGen ) )  {
-		return qfalse;
-	}
-
-	// an add collapse can only have identity colors
-	if ( collapse[i].multitextureEnv == GL_ADD && stages[0].rgbGen != CGEN_IDENTITY ) {
-		return qfalse;
-	}
-
-	if ( stages[0].rgbGen == CGEN_WAVEFORM )
-	{
-		if ( memcmp( &stages[0].rgbWave,
-					 &stages[1].rgbWave,
-					 sizeof( stages[0].rgbWave ) ) )
-		{
-			return qfalse;
-		}
-	}
-	if ( stages[0].alphaGen == AGEN_WAVEFORM )
-	{
-		if ( memcmp( &stages[0].alphaWave,
-					 &stages[1].alphaWave,
-					 sizeof( stages[0].alphaWave ) ) )
-		{
-			return qfalse;
-		}
-	}
-
-
-	// make sure that lightmaps are in bundle 1 for 3dfx
-	if ( stages[0].bundle[0].isLightmap )
-	{
-		tmpBundle = stages[0].bundle[0];
-		stages[0].bundle[0] = stages[1].bundle[0];
-		stages[0].bundle[1] = tmpBundle;
-	}
-	else
-	{
-		stages[0].bundle[1] = stages[1].bundle[0];
-	}
-
-	// set the new blend state bits
-	shader.multitextureEnv = collapse[i].multitextureEnv;
-	stages[0].stateBits &= ~( GLS_DSTBLEND_BITS | GLS_SRCBLEND_BITS );
-	stages[0].stateBits |= collapse[i].multitextureBlend;
-
-	//
-	// move down subsequent shaders
-	//
-	memmove( &stages[1], &stages[2], sizeof( stages[0] ) * ( MAX_SHADER_STAGES - 2 ) );
-	Com_Memset( &stages[MAX_SHADER_STAGES-1], 0, sizeof( stages[0] ) );
-
-	return qtrue;
-}
 
 static void CollapseStagesToLightall(shaderStage_t *diffuse, 
 	shaderStage_t *normal, shaderStage_t *specular, shaderStage_t *lightmap, 
@@ -2607,9 +2452,6 @@ static int CollapseStagesToGLSL(void)
 		numStages++;
 	}
 
-	if (numStages == i && i >= 2 && CollapseMultitexture())
-		numStages--;
-
 	// convert any remaining lightmap stages to a lighting pass with a white texture
 	// only do this with r_sunlightMode non-zero, as it's only for correct shadows.
 	if (r_sunlightMode->integer && shader.numDeforms == 0)
@@ -3647,15 +3489,6 @@ void	R_ShaderList_f (void) {
 		} else {
 			ri.Printf (PRINT_ALL, "  ");
 		}
-		if ( shader->multitextureEnv == GL_ADD ) {
-			ri.Printf( PRINT_ALL, "MT(a) " );
-		} else if ( shader->multitextureEnv == GL_MODULATE ) {
-			ri.Printf( PRINT_ALL, "MT(m) " );
-		} else if ( shader->multitextureEnv == GL_DECAL ) {
-			ri.Printf( PRINT_ALL, "MT(d) " );
-		} else {
-			ri.Printf( PRINT_ALL, "      " );
-		}
 		if ( shader->explicitlyDefined ) {
 			ri.Printf( PRINT_ALL, "E " );
 		} else {

From d9e2184c1a9a1f3115202fb66d3fb29f6753f7ca Mon Sep 17 00:00:00 2001
From: SmileTheory <SmileTheory@gmail.com>
Date: Mon, 10 Nov 2014 22:11:36 -0800
Subject: [PATCH 05/14] OpenGL2: Add support for parallax occlusion mapping.

---
 code/renderergl2/glsl/lightall_fp.glsl | 14 +++++++++++++-
 code/renderergl2/tr_glsl.c             |  4 ++++
 opengl2-readme.txt                     |  3 ++-
 3 files changed, 19 insertions(+), 2 deletions(-)

diff --git a/code/renderergl2/glsl/lightall_fp.glsl b/code/renderergl2/glsl/lightall_fp.glsl
index f2da782105..d1155de464 100644
--- a/code/renderergl2/glsl/lightall_fp.glsl
+++ b/code/renderergl2/glsl/lightall_fp.glsl
@@ -100,6 +100,9 @@ float RayIntersectDisplaceMap(vec2 dp, vec2 ds, sampler2D normalMap)
 	// best match found (starts with last position 1.0)
 	float bestDepth = 1.0;
 
+	// texture depth at best depth
+	float texDepth = 0.0;
+
 	// search front to back for first point inside object
 	for(int i = 0; i < linearSearchSteps - 1; ++i)
 	{
@@ -109,11 +112,19 @@ float RayIntersectDisplaceMap(vec2 dp, vec2 ds, sampler2D normalMap)
 		
 		if(bestDepth > 0.996)		// if no depth found yet
 			if(depth >= t)
+			{
 				bestDepth = depth;	// store best depth
+				texDepth = t;
+			}
 	}
 
 	depth = bestDepth;
-	
+
+#if !defined (USE_RELIEFMAP)
+	float prevDepth = depth - size;
+	float prevTexDepth = SampleDepth(normalMap, dp + ds * prevDepth);
+	bestDepth -= size * (prevDepth - prevTexDepth) / (size - texDepth + prevTexDepth);
+#else
 	// recurse around first point (depth) for closest match
 	for(int i = 0; i < binarySearchSteps; ++i)
 	{
@@ -129,6 +140,7 @@ float RayIntersectDisplaceMap(vec2 dp, vec2 ds, sampler2D normalMap)
 
 		depth += size;
 	}
+#endif
 
 	return bestDepth;
 }
diff --git a/code/renderergl2/tr_glsl.c b/code/renderergl2/tr_glsl.c
index 7dfea6a675..f526b8c6e2 100644
--- a/code/renderergl2/tr_glsl.c
+++ b/code/renderergl2/tr_glsl.c
@@ -1079,7 +1079,11 @@ void GLSL_InitGPUShaders(void)
 #endif
 
 				if ((i & LIGHTDEF_USE_PARALLAXMAP) && !(i & LIGHTDEF_ENTITY) && r_parallaxMapping->integer)
+				{
 					Q_strcat(extradefines, 1024, "#define USE_PARALLAXMAP\n");
+					if (r_parallaxMapping->integer > 1)
+						Q_strcat(extradefines, 1024, "#define USE_RELIEFMAP\n");
+				}
 			}
 
 			if (r_specularMapping->integer)
diff --git a/opengl2-readme.txt b/opengl2-readme.txt
index 167d26f18d..5c4cb60b8a 100644
--- a/opengl2-readme.txt
+++ b/opengl2-readme.txt
@@ -187,7 +187,8 @@ Cvars for advanced material usage:
   r_parallaxMapping              - Enable parallax mapping for materials that
                                    support it.
                                      0 - No. (default)
-                                     1 - Yes.
+                                     1 - Use parallax occlusion mapping.
+                                     2 - Use relief mapping. (slower)
 
   r_baseSpecular                 - Set the specular reflectance of materials
                                    which don't include a specular map or

From 89b719ec8dc0e02af655c71c03c082a8068febf3 Mon Sep 17 00:00:00 2001
From: SmileTheory <SmileTheory@gmail.com>
Date: Fri, 14 Nov 2014 01:12:41 -0800
Subject: [PATCH 06/14] OpenGL2: Bit more parallax optimization.

---
 code/renderergl2/glsl/lightall_fp.glsl | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/code/renderergl2/glsl/lightall_fp.glsl b/code/renderergl2/glsl/lightall_fp.glsl
index d1155de464..d118278169 100644
--- a/code/renderergl2/glsl/lightall_fp.glsl
+++ b/code/renderergl2/glsl/lightall_fp.glsl
@@ -103,6 +103,9 @@ float RayIntersectDisplaceMap(vec2 dp, vec2 ds, sampler2D normalMap)
 	// texture depth at best depth
 	float texDepth = 0.0;
 
+	float prevT = SampleDepth(normalMap, dp);
+	float prevTexDepth = prevT;
+
 	// search front to back for first point inside object
 	for(int i = 0; i < linearSearchSteps - 1; ++i)
 	{
@@ -115,15 +118,16 @@ float RayIntersectDisplaceMap(vec2 dp, vec2 ds, sampler2D normalMap)
 			{
 				bestDepth = depth;	// store best depth
 				texDepth = t;
+				prevTexDepth = prevT;
 			}
+		prevT = t;
 	}
 
 	depth = bestDepth;
 
 #if !defined (USE_RELIEFMAP)
-	float prevDepth = depth - size;
-	float prevTexDepth = SampleDepth(normalMap, dp + ds * prevDepth);
-	bestDepth -= size * (prevDepth - prevTexDepth) / (size - texDepth + prevTexDepth);
+	float div = 1.0 / (1.0 + (prevTexDepth - texDepth) * float(linearSearchSteps));
+	bestDepth -= (depth - size - prevTexDepth) * div;
 #else
 	// recurse around first point (depth) for closest match
 	for(int i = 0; i < binarySearchSteps; ++i)
@@ -369,7 +373,7 @@ void main()
 	vec2 texCoords = var_TexCoords.xy;
 
 #if defined(USE_PARALLAXMAP)
-	vec3 offsetDir = normalize(E * tangentToWorld);
+	vec3 offsetDir = viewDir * tangentToWorld;
 
 	offsetDir.xy *= -u_NormalScale.a / offsetDir.z;
 

From 22bcda018b253d1d956dcaa9169e6febfbd81861 Mon Sep 17 00:00:00 2001
From: Zack Middleton <zturtleman@gmail.com>
Date: Sun, 16 Nov 2014 11:53:12 -0600
Subject: [PATCH 07/14] Correct a few OpenGL variable types

Affectly no change for desktop OpenGL.
Use correct types for OpenGLES support.
---
 code/renderergl1/tr_backend.c | 2 +-
 code/renderergl1/tr_shade.c   | 4 ++--
 code/renderergl1/tr_surface.c | 3 ++-
 code/renderergl2/tr_backend.c | 2 +-
 4 files changed, 6 insertions(+), 5 deletions(-)

diff --git a/code/renderergl1/tr_backend.c b/code/renderergl1/tr_backend.c
index 788f780165..a11224561c 100644
--- a/code/renderergl1/tr_backend.c
+++ b/code/renderergl1/tr_backend.c
@@ -474,7 +474,7 @@ void RB_BeginDrawingView (void) {
 	// clip to the plane of the portal
 	if ( backEnd.viewParms.isPortal ) {
 		float	plane[4];
-		double	plane2[4];
+		GLdouble	plane2[4];
 
 		plane[0] = backEnd.viewParms.portalPlane.normal[0];
 		plane[1] = backEnd.viewParms.portalPlane.normal[1];
diff --git a/code/renderergl1/tr_shade.c b/code/renderergl1/tr_shade.c
index 08648b2e8a..9bdf1e3953 100644
--- a/code/renderergl1/tr_shade.c
+++ b/code/renderergl1/tr_shade.c
@@ -421,7 +421,7 @@ static void ProjectDlightTexture_altivec( void ) {
 	byte	clipBits[SHADER_MAX_VERTEXES];
 	float	texCoordsArray[SHADER_MAX_VERTEXES][2];
 	byte	colorArray[SHADER_MAX_VERTEXES][4];
-	unsigned	hitIndexes[SHADER_MAX_INDEXES];
+	glIndex_t	hitIndexes[SHADER_MAX_INDEXES];
 	int		numIndexes;
 	float	scale;
 	float	radius;
@@ -593,7 +593,7 @@ static void ProjectDlightTexture_scalar( void ) {
 	byte	clipBits[SHADER_MAX_VERTEXES];
 	float	texCoordsArray[SHADER_MAX_VERTEXES][2];
 	byte	colorArray[SHADER_MAX_VERTEXES][4];
-	unsigned	hitIndexes[SHADER_MAX_INDEXES];
+	glIndex_t	hitIndexes[SHADER_MAX_INDEXES];
 	int		numIndexes;
 	float	scale;
 	float	radius;
diff --git a/code/renderergl1/tr_surface.c b/code/renderergl1/tr_surface.c
index 1879fe2c02..d318df1db5 100644
--- a/code/renderergl1/tr_surface.c
+++ b/code/renderergl1/tr_surface.c
@@ -907,7 +907,8 @@ RB_SurfaceFace
 */
 static void RB_SurfaceFace( srfSurfaceFace_t *surf ) {
 	int			i;
-	unsigned	*indices, *tessIndexes;
+	unsigned	*indices;
+	glIndex_t	*tessIndexes;
 	float		*v;
 	float		*normal;
 	int			ndx;
diff --git a/code/renderergl2/tr_backend.c b/code/renderergl2/tr_backend.c
index bb5e57e7be..3b297b8d75 100644
--- a/code/renderergl2/tr_backend.c
+++ b/code/renderergl2/tr_backend.c
@@ -548,7 +548,7 @@ void RB_BeginDrawingView (void) {
 	if ( backEnd.viewParms.isPortal ) {
 #if 0
 		float	plane[4];
-		double	plane2[4];
+		GLdouble	plane2[4];
 
 		plane[0] = backEnd.viewParms.portalPlane.normal[0];
 		plane[1] = backEnd.viewParms.portalPlane.normal[1];

From 60d28722effd3447538650f607a937c333c08886 Mon Sep 17 00:00:00 2001
From: Zack Middleton <zturtleman@gmail.com>
Date: Sun, 16 Nov 2014 11:58:52 -0600
Subject: [PATCH 08/14] Remove unfinished OpenGL display list code

It seems unlikely anyone is going to do anything with this aside from stub
it out in OpenGLES ports.
---
 code/renderergl1/tr_local.h   | 6 ------
 code/renderergl1/tr_surface.c | 9 +--------
 code/renderergl2/tr_local.h   | 6 ------
 code/renderergl2/tr_surface.c | 7 -------
 4 files changed, 1 insertion(+), 27 deletions(-)

diff --git a/code/renderergl1/tr_local.h b/code/renderergl1/tr_local.h
index c0c62f03f1..47d8ca81ee 100644
--- a/code/renderergl1/tr_local.h
+++ b/code/renderergl1/tr_local.h
@@ -477,7 +477,6 @@ typedef enum {
 	SF_IQM,
 	SF_FLARE,
 	SF_ENTITY,				// beams, rails, lightning, etc that can be determined by entity
-	SF_DISPLAY_LIST,
 
 	SF_NUM_SURFACE_TYPES,
 	SF_MAX = 0x7fffffff			// ensures that sizeof( surfaceType_t ) == sizeof( int )
@@ -503,11 +502,6 @@ typedef struct srfPoly_s {
 	polyVert_t		*verts;
 } srfPoly_t;
 
-typedef struct srfDisplayList_s {
-	surfaceType_t	surfaceType;
-	int				listNum;
-} srfDisplayList_t;
-
 
 typedef struct srfFlare_s {
 	surfaceType_t	surfaceType;
diff --git a/code/renderergl1/tr_surface.c b/code/renderergl1/tr_surface.c
index d318df1db5..79852f051d 100644
--- a/code/renderergl1/tr_surface.c
+++ b/code/renderergl1/tr_surface.c
@@ -1218,12 +1218,6 @@ static void RB_SurfaceFlare(srfFlare_t *surf)
 		RB_AddFlare(surf, tess.fogNum, surf->origin, surf->color, surf->normal);
 }
 
-static void RB_SurfaceDisplayList( srfDisplayList_t *surf ) {
-	// all apropriate state must be set in RB_BeginSurface
-	// this isn't implemented yet...
-	qglCallList( surf->listNum );
-}
-
 static void RB_SurfaceSkip( void *surf ) {
 }
 
@@ -1239,6 +1233,5 @@ void (*rb_surfaceTable[SF_NUM_SURFACE_TYPES])( void *) = {
 	(void(*)(void*))RB_MDRSurfaceAnim,		// SF_MDR,
 	(void(*)(void*))RB_IQMSurfaceAnim,		// SF_IQM,
 	(void(*)(void*))RB_SurfaceFlare,		// SF_FLARE,
-	(void(*)(void*))RB_SurfaceEntity,		// SF_ENTITY
-	(void(*)(void*))RB_SurfaceDisplayList		// SF_DISPLAY_LIST
+	(void(*)(void*))RB_SurfaceEntity		// SF_ENTITY
 };
diff --git a/code/renderergl2/tr_local.h b/code/renderergl2/tr_local.h
index c21b7b4835..a346a8c70c 100644
--- a/code/renderergl2/tr_local.h
+++ b/code/renderergl2/tr_local.h
@@ -848,7 +848,6 @@ typedef enum {
 	SF_IQM,
 	SF_FLARE,
 	SF_ENTITY,				// beams, rails, lightning, etc that can be determined by entity
-	SF_DISPLAY_LIST,
 	SF_VAO_MESH,
 	SF_VAO_MDVMESH,
 
@@ -877,11 +876,6 @@ typedef struct srfPoly_s {
 	polyVert_t		*verts;
 } srfPoly_t;
 
-typedef struct srfDisplayList_s {
-	surfaceType_t	surfaceType;
-	int				listNum;
-} srfDisplayList_t;
-
 
 typedef struct srfFlare_s {
 	surfaceType_t	surfaceType;
diff --git a/code/renderergl2/tr_surface.c b/code/renderergl2/tr_surface.c
index 2b0ae0f144..872208e468 100644
--- a/code/renderergl2/tr_surface.c
+++ b/code/renderergl2/tr_surface.c
@@ -1658,12 +1658,6 @@ void RB_SurfaceVaoMdvMesh(srfVaoMdvMesh_t * surface)
 	glState.vertexAnimation = qfalse;
 }
 
-static void RB_SurfaceDisplayList( srfDisplayList_t *surf ) {
-	// all apropriate state must be set in RB_BeginSurface
-	// this isn't implemented yet...
-	qglCallList( surf->listNum );
-}
-
 static void RB_SurfaceSkip( void *surf ) {
 }
 
@@ -1680,7 +1674,6 @@ void (*rb_surfaceTable[SF_NUM_SURFACE_TYPES])( void *) = {
 	(void(*)(void*))RB_IQMSurfaceAnim,		// SF_IQM,
 	(void(*)(void*))RB_SurfaceFlare,		// SF_FLARE,
 	(void(*)(void*))RB_SurfaceEntity,		// SF_ENTITY
-	(void(*)(void*))RB_SurfaceDisplayList,		// SF_DISPLAY_LIST
 	(void(*)(void*))RB_SurfaceVaoMesh,	    // SF_VAO_MESH,
 	(void(*)(void*))RB_SurfaceVaoMdvMesh,   // SF_VAO_MDVMESH
 };

From d06deb41c80159b7d16bee77a7cdd1b197d610e2 Mon Sep 17 00:00:00 2001
From: Zack Middleton <zturtleman@gmail.com>
Date: Sun, 16 Nov 2014 12:47:49 -0600
Subject: [PATCH 09/14] Don't set fog image border color

GL1's R_CreateImage sets GL texture to 0 before it ends, so border color is not
applied to the fog image. GL_CLAMP is not used for fog image (in either renderer),
so it would presumably not be used even if applied to the fog image.
---
 code/renderergl1/tr_image.c | 12 ------------
 code/renderergl2/tr_image.c | 11 -----------
 2 files changed, 23 deletions(-)

diff --git a/code/renderergl1/tr_image.c b/code/renderergl1/tr_image.c
index f8a8899172..39f3e57f35 100644
--- a/code/renderergl1/tr_image.c
+++ b/code/renderergl1/tr_image.c
@@ -899,7 +899,6 @@ image_t *R_CreateImage( const char *name, byte *pic, int width, int height,
 	qglTexParameterf( GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, glWrapClampMode );
 	qglTexParameterf( GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, glWrapClampMode );
 
-	// FIXME: this stops fog from setting border color?
 	glState.currenttextures[glState.currenttmu] = 0;
 	qglBindTexture( GL_TEXTURE_2D, 0 );
 
@@ -1168,7 +1167,6 @@ static void R_CreateFogImage( void ) {
 	int		x,y;
 	byte	*data;
 	float	d;
-	float	borderColor[4];
 
 	data = ri.Hunk_AllocateTempMemory( FOG_S * FOG_T * 4 );
 
@@ -1183,18 +1181,8 @@ static void R_CreateFogImage( void ) {
 			data[(y*FOG_S+x)*4+3] = 255*d;
 		}
 	}
-	// standard openGL clamping doesn't really do what we want -- it includes
-	// the border color at the edges.  OpenGL 1.2 has clamp-to-edge, which does
-	// what we want.
 	tr.fogImage = R_CreateImage("*fog", (byte *)data, FOG_S, FOG_T, IMGTYPE_COLORALPHA, IMGFLAG_CLAMPTOEDGE, 0 );
 	ri.Hunk_FreeTempMemory( data );
-
-	borderColor[0] = 1.0;
-	borderColor[1] = 1.0;
-	borderColor[2] = 1.0;
-	borderColor[3] = 1;
-
-	qglTexParameterfv( GL_TEXTURE_2D, GL_TEXTURE_BORDER_COLOR, borderColor );
 }
 
 /*
diff --git a/code/renderergl2/tr_image.c b/code/renderergl2/tr_image.c
index cb42ee052f..e95b559f5c 100644
--- a/code/renderergl2/tr_image.c
+++ b/code/renderergl2/tr_image.c
@@ -2641,7 +2641,6 @@ static void R_CreateFogImage( void ) {
 	int		x,y;
 	byte	*data;
 	float	d;
-	float	borderColor[4];
 
 	data = ri.Hunk_AllocateTempMemory( FOG_S * FOG_T * 4 );
 
@@ -2656,18 +2655,8 @@ static void R_CreateFogImage( void ) {
 			data[(y*FOG_S+x)*4+3] = 255*d;
 		}
 	}
-	// standard openGL clamping doesn't really do what we want -- it includes
-	// the border color at the edges.  OpenGL 1.2 has clamp-to-edge, which does
-	// what we want.
 	tr.fogImage = R_CreateImage("*fog", (byte *)data, FOG_S, FOG_T, IMGTYPE_COLORALPHA, IMGFLAG_CLAMPTOEDGE, 0 );
 	ri.Hunk_FreeTempMemory( data );
-
-	borderColor[0] = 1.0;
-	borderColor[1] = 1.0;
-	borderColor[2] = 1.0;
-	borderColor[3] = 1;
-
-	qglTexParameterfv( GL_TEXTURE_2D, GL_TEXTURE_BORDER_COLOR, borderColor );
 }
 
 /*

From c787cf3aef0daa32478f90a5f43fb794ff6b5313 Mon Sep 17 00:00:00 2001
From: Zack Middleton <zturtleman@gmail.com>
Date: Sun, 16 Nov 2014 21:21:15 -0600
Subject: [PATCH 10/14] Fix stencil shadows not drawing if has 500 or more
 vertexes

Stencil shadow is not drawn if a mesh, or multiple meshes with the same
entity and shader, have more than 500 vertexes. The issue is caused by storing
the projected positions in the tess vertex buffer. Use a new array instead.
---
 code/renderergl1/tr_shadows.c | 20 ++++++++------------
 code/renderergl2/tr_shadows.c | 20 ++++++++------------
 2 files changed, 16 insertions(+), 24 deletions(-)

diff --git a/code/renderergl1/tr_shadows.c b/code/renderergl1/tr_shadows.c
index 3582ae4479..13fbfdf74e 100644
--- a/code/renderergl1/tr_shadows.c
+++ b/code/renderergl1/tr_shadows.c
@@ -44,6 +44,7 @@ typedef struct {
 static	edgeDef_t	edgeDefs[SHADER_MAX_VERTEXES][MAX_EDGE_DEFS];
 static	int			numEdgeDefs[SHADER_MAX_VERTEXES];
 static	int			facing[SHADER_MAX_INDEXES/3];
+static	vec3_t		shadowXyz[SHADER_MAX_VERTEXES];
 
 void R_AddEdgeDef( int i1, int i2, int facing ) {
 	int		c;
@@ -80,13 +81,13 @@ void R_RenderShadowEdges( void ) {
 
 		qglBegin( GL_TRIANGLE_STRIP );
 		qglVertex3fv( tess.xyz[ i1 ] );
-		qglVertex3fv( tess.xyz[ i1 + tess.numVertexes ] );
+		qglVertex3fv( shadowXyz[ i1 ] );
 		qglVertex3fv( tess.xyz[ i2 ] );
-		qglVertex3fv( tess.xyz[ i2 + tess.numVertexes ] );
+		qglVertex3fv( shadowXyz[ i2 ] );
 		qglVertex3fv( tess.xyz[ i3 ] );
-		qglVertex3fv( tess.xyz[ i3 + tess.numVertexes ] );
+		qglVertex3fv( shadowXyz[ i3 ] );
 		qglVertex3fv( tess.xyz[ i1 ] );
-		qglVertex3fv( tess.xyz[ i1 + tess.numVertexes ] );
+		qglVertex3fv( shadowXyz[ i1 ] );
 		qglEnd();
 	}
 #else
@@ -126,9 +127,9 @@ void R_RenderShadowEdges( void ) {
 			if ( hit[ 1 ] == 0 ) {
 				qglBegin( GL_TRIANGLE_STRIP );
 				qglVertex3fv( tess.xyz[ i ] );
-				qglVertex3fv( tess.xyz[ i + tess.numVertexes ] );
+				qglVertex3fv( shadowXyz[ i ] );
 				qglVertex3fv( tess.xyz[ i2 ] );
-				qglVertex3fv( tess.xyz[ i2 + tess.numVertexes ] );
+				qglVertex3fv( shadowXyz[ i2 ] );
 				qglEnd();
 				c_edges++;
 			} else {
@@ -157,11 +158,6 @@ void RB_ShadowTessEnd( void ) {
 	vec3_t	lightDir;
 	GLboolean rgba[4];
 
-	// we can only do this if we have enough space in the vertex buffers
-	if ( tess.numVertexes >= SHADER_MAX_VERTEXES / 2 ) {
-		return;
-	}
-
 	if ( glConfig.stencilBits < 4 ) {
 		return;
 	}
@@ -170,7 +166,7 @@ void RB_ShadowTessEnd( void ) {
 
 	// project vertexes away from light direction
 	for ( i = 0 ; i < tess.numVertexes ; i++ ) {
-		VectorMA( tess.xyz[i], -512, lightDir, tess.xyz[i+tess.numVertexes] );
+		VectorMA( tess.xyz[i], -512, lightDir, shadowXyz[i] );
 	}
 
 	// decide which triangles face the light
diff --git a/code/renderergl2/tr_shadows.c b/code/renderergl2/tr_shadows.c
index 3582ae4479..13fbfdf74e 100644
--- a/code/renderergl2/tr_shadows.c
+++ b/code/renderergl2/tr_shadows.c
@@ -44,6 +44,7 @@ typedef struct {
 static	edgeDef_t	edgeDefs[SHADER_MAX_VERTEXES][MAX_EDGE_DEFS];
 static	int			numEdgeDefs[SHADER_MAX_VERTEXES];
 static	int			facing[SHADER_MAX_INDEXES/3];
+static	vec3_t		shadowXyz[SHADER_MAX_VERTEXES];
 
 void R_AddEdgeDef( int i1, int i2, int facing ) {
 	int		c;
@@ -80,13 +81,13 @@ void R_RenderShadowEdges( void ) {
 
 		qglBegin( GL_TRIANGLE_STRIP );
 		qglVertex3fv( tess.xyz[ i1 ] );
-		qglVertex3fv( tess.xyz[ i1 + tess.numVertexes ] );
+		qglVertex3fv( shadowXyz[ i1 ] );
 		qglVertex3fv( tess.xyz[ i2 ] );
-		qglVertex3fv( tess.xyz[ i2 + tess.numVertexes ] );
+		qglVertex3fv( shadowXyz[ i2 ] );
 		qglVertex3fv( tess.xyz[ i3 ] );
-		qglVertex3fv( tess.xyz[ i3 + tess.numVertexes ] );
+		qglVertex3fv( shadowXyz[ i3 ] );
 		qglVertex3fv( tess.xyz[ i1 ] );
-		qglVertex3fv( tess.xyz[ i1 + tess.numVertexes ] );
+		qglVertex3fv( shadowXyz[ i1 ] );
 		qglEnd();
 	}
 #else
@@ -126,9 +127,9 @@ void R_RenderShadowEdges( void ) {
 			if ( hit[ 1 ] == 0 ) {
 				qglBegin( GL_TRIANGLE_STRIP );
 				qglVertex3fv( tess.xyz[ i ] );
-				qglVertex3fv( tess.xyz[ i + tess.numVertexes ] );
+				qglVertex3fv( shadowXyz[ i ] );
 				qglVertex3fv( tess.xyz[ i2 ] );
-				qglVertex3fv( tess.xyz[ i2 + tess.numVertexes ] );
+				qglVertex3fv( shadowXyz[ i2 ] );
 				qglEnd();
 				c_edges++;
 			} else {
@@ -157,11 +158,6 @@ void RB_ShadowTessEnd( void ) {
 	vec3_t	lightDir;
 	GLboolean rgba[4];
 
-	// we can only do this if we have enough space in the vertex buffers
-	if ( tess.numVertexes >= SHADER_MAX_VERTEXES / 2 ) {
-		return;
-	}
-
 	if ( glConfig.stencilBits < 4 ) {
 		return;
 	}
@@ -170,7 +166,7 @@ void RB_ShadowTessEnd( void ) {
 
 	// project vertexes away from light direction
 	for ( i = 0 ; i < tess.numVertexes ; i++ ) {
-		VectorMA( tess.xyz[i], -512, lightDir, tess.xyz[i+tess.numVertexes] );
+		VectorMA( tess.xyz[i], -512, lightDir, shadowXyz[i] );
 	}
 
 	// decide which triangles face the light

From fd23249357c3efee4d7b1661dab229445191088a Mon Sep 17 00:00:00 2001
From: SmileTheory <SmileTheory@gmail.com>
Date: Fri, 21 Nov 2014 16:12:53 -0800
Subject: [PATCH 11/14] OpenGL2: Ensure tess VAO is bound before using it.

---
 code/renderergl2/tr_surface.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/code/renderergl2/tr_surface.c b/code/renderergl2/tr_surface.c
index 872208e468..c5b1b0f8d3 100644
--- a/code/renderergl2/tr_surface.c
+++ b/code/renderergl2/tr_surface.c
@@ -91,6 +91,8 @@ void RB_AddQuadStampExt( vec3_t origin, vec3_t left, vec3_t up, float color[4],
 	uint32_t    pNormal;
 	int			ndx;
 
+	RB_CheckVao(tess.vao);
+
 	RB_CHECKOVERFLOW( 4, 6 );
 
 	ndx = tess.numVertexes;
@@ -283,6 +285,8 @@ static void RB_SurfacePolychain( srfPoly_t *p ) {
 	int		i;
 	int		numv;
 
+	RB_CheckVao(tess.vao);
+
 	RB_CHECKOVERFLOW( p->numVerts, 3*(p->numVerts - 2) );
 
 	// fan triangles into the tess array
@@ -626,6 +630,8 @@ static void DoRailCore( const vec3_t start, const vec3_t end, const vec3_t up, f
 	int			vbase;
 	float		t = len / 256.0f;
 
+	RB_CheckVao(tess.vao);
+
 	RB_CHECKOVERFLOW( 4, 6 );
 
 	vbase = tess.numVertexes;
@@ -707,6 +713,8 @@ static void DoRailDiscs( int numSegs, const vec3_t start, const vec3_t dir, cons
 		}
 	}
 
+	RB_CheckVao(tess.vao);
+
 	for ( i = 0; i < numSegs; i++ )
 	{
 		int j;
@@ -1218,6 +1226,8 @@ static void RB_SurfaceMesh(mdvSurface_t *surface) {
 		backlerp = backEnd.currentEntity->e.backlerp;
 	}
 
+	RB_CheckVao(tess.vao);
+
 	RB_CHECKOVERFLOW( surface->numVerts, surface->numIndexes );
 
 	LerpMeshVertexes (surface, backlerp);
@@ -1325,6 +1335,8 @@ static void RB_SurfaceGrid( srfBspSurface_t *srf ) {
 		return;
 	}
 
+	RB_CheckVao(tess.vao);
+
 	dlightBits = srf->dlightBits;
 	tess.dlightBits |= dlightBits;
 

From 8c7fedb1feb5f3b046c6e640d8f4da1ad8ec7477 Mon Sep 17 00:00:00 2001
From: SmileTheory <SmileTheory@gmail.com>
Date: Sun, 30 Nov 2014 21:50:33 -0800
Subject: [PATCH 12/14] OpenGL2: Fix face culling.

---
 code/renderergl2/tr_backend.c | 3 ---
 code/renderergl2/tr_init.c    | 2 ++
 2 files changed, 2 insertions(+), 3 deletions(-)

diff --git a/code/renderergl2/tr_backend.c b/code/renderergl2/tr_backend.c
index 3b297b8d75..e74f8f2480 100644
--- a/code/renderergl2/tr_backend.c
+++ b/code/renderergl2/tr_backend.c
@@ -538,9 +538,6 @@ void RB_BeginDrawingView (void) {
 		backEnd.isHyperspace = qfalse;
 	}
 
-	glState.faceCulling = -1;		// force face culling to set next time
-	glState.faceCullFront = -1;     // same as above
-
 	// we will only draw a sun if there was sky rendered in this view
 	backEnd.skyRenderedThisView = qfalse;
 
diff --git a/code/renderergl2/tr_init.c b/code/renderergl2/tr_init.c
index e79d735d1a..525dfba0cc 100644
--- a/code/renderergl2/tr_init.c
+++ b/code/renderergl2/tr_init.c
@@ -946,6 +946,8 @@ void GL_SetDefaultState( void )
 	//
 	glState.glStateBits = GLS_DEPTHTEST_DISABLE | GLS_DEPTHMASK_TRUE;
 	glState.storedGlState = 0;
+	glState.faceCulling = CT_TWO_SIDED;
+	glState.faceCullFront = qtrue;
 
 	glState.currentProgram = 0;
 	qglUseProgramObjectARB(0);

From 08ddb99732444a33bc5d5616116f4ee49f57de5f Mon Sep 17 00:00:00 2001
From: Zack Middleton <zturtleman@gmail.com>
Date: Mon, 1 Dec 2014 20:23:22 -0600
Subject: [PATCH 13/14] Fix saving/loading glyph 255 in RegisterFont

The glyph for character 255 (lower case y with two dots above it) was
rendered, but it's glyph information was not stored in fontInfo_t and
not saved into .dat file (including the ones in Team Arena).

Attempting to load it from existing .dat font files is fine because
shader name is "" and gets 0 handle. The handle was already 0 anyway.
---
 code/renderercommon/tr_font.c | 17 +++++++++++------
 1 file changed, 11 insertions(+), 6 deletions(-)

diff --git a/code/renderercommon/tr_font.c b/code/renderercommon/tr_font.c
index 4b641bac4a..0f89d5ad68 100644
--- a/code/renderercommon/tr_font.c
+++ b/code/renderercommon/tr_font.c
@@ -397,7 +397,7 @@ void RE_RegisterFont(const char *fontName, int pointSize, fontInfo_t *font) {
 
 //		Com_Memcpy(font, faceData, sizeof(fontInfo_t));
 		Q_strncpyz(font->name, name, sizeof(font->name));
-		for (i = GLYPH_START; i < GLYPH_END; i++) {
+		for (i = GLYPH_START; i <= GLYPH_END; i++) {
 			font->glyphs[i].glyph = RE_RegisterShaderNoMip(font->glyphs[i].shaderName);
 		}
 		Com_Memcpy(&registeredFont[registeredFontCount++], font, sizeof(fontInfo_t));
@@ -445,7 +445,7 @@ void RE_RegisterFont(const char *fontName, int pointSize, fontInfo_t *font) {
 
 	maxHeight = 0;
 
-	for (i = GLYPH_START; i < GLYPH_END; i++) {
+	for (i = GLYPH_START; i <= GLYPH_END; i++) {
 		RE_ConstructGlyphInfo(out, &xOut, &yOut, &maxHeight, face, (unsigned char)i, qtrue);
 	}
 
@@ -455,11 +455,16 @@ void RE_RegisterFont(const char *fontName, int pointSize, fontInfo_t *font) {
 	lastStart = i;
 	imageNumber = 0;
 
-	while ( i <= GLYPH_END ) {
+	while ( i <= GLYPH_END + 1 ) {
 
-		glyph = RE_ConstructGlyphInfo(out, &xOut, &yOut, &maxHeight, face, (unsigned char)i, qfalse);
+		if ( i == GLYPH_END + 1 ) {
+			// upload/save current image buffer
+			xOut = yOut = -1;
+		} else {
+			glyph = RE_ConstructGlyphInfo(out, &xOut, &yOut, &maxHeight, face, (unsigned char)i, qfalse);
+		}
 
-		if (xOut == -1 || yOut == -1 || i == GLYPH_END)  {
+		if (xOut == -1 || yOut == -1)  {
 			// ran out of room
 			// we need to create an image from the bitmap, set all the handles in the glyphs to this point
 			// 
@@ -504,7 +509,7 @@ void RE_RegisterFont(const char *fontName, int pointSize, fontInfo_t *font) {
 			xOut = 0;
 			yOut = 0;
 			ri.Free(imageBuff);
-			if(i == GLYPH_END)
+			if ( i == GLYPH_END + 1 )
 				i++;
 		} else {
 			Com_Memcpy(&font->glyphs[i], glyph, sizeof(glyphInfo_t));

From b21a59af8c1cc58947f8b5ee29e55b14b8321238 Mon Sep 17 00:00:00 2001
From: Zack Middleton <zturtleman@gmail.com>
Date: Mon, 1 Dec 2014 21:53:06 -0600
Subject: [PATCH 14/14] Fix negative glyph index in Team Arena text functions

Team Arena's text functions cast signed char values to int and use as an array index.
This works fine for values 0 to 127, but not for -128 to -1 which are a negative array index.
Instead use "character & 255" like client and original Q3 ui/cgame string drawing code.
---
 code/cgame/cg_draw.c    | 12 +++---------
 code/cgame/cg_newdraw.c |  2 +-
 code/ui/ui_main.c       | 12 ++++++------
 3 files changed, 10 insertions(+), 16 deletions(-)

diff --git a/code/cgame/cg_draw.c b/code/cgame/cg_draw.c
index 178c967926..e9c7d4da84 100644
--- a/code/cgame/cg_draw.c
+++ b/code/cgame/cg_draw.c
@@ -49,8 +49,6 @@ int CG_Text_Width(const char *text, float scale, int limit) {
 	float out;
 	glyphInfo_t *glyph;
 	float useScale;
-// FIXME: see ui_main.c, same problem
-//	const unsigned char *s = text;
 	const char *s = text;
 	fontInfo_t *font = &cgDC.Assets.textFont;
 	if (scale <= cg_smallFont.value) {
@@ -71,7 +69,7 @@ int CG_Text_Width(const char *text, float scale, int limit) {
 				s += 2;
 				continue;
 			} else {
-				glyph = &font->glyphs[(int)*s]; // TTimo: FIXME: getting nasty warnings without the cast, hopefully this doesn't break the VM build
+				glyph = &font->glyphs[*s & 255];
 				out += glyph->xSkip;
 				s++;
 				count++;
@@ -86,8 +84,6 @@ int CG_Text_Height(const char *text, float scale, int limit) {
 	float max;
 	glyphInfo_t *glyph;
 	float useScale;
-// TTimo: FIXME
-//	const unsigned char *s = text;
 	const char *s = text;
 	fontInfo_t *font = &cgDC.Assets.textFont;
 	if (scale <= cg_smallFont.value) {
@@ -108,7 +104,7 @@ int CG_Text_Height(const char *text, float scale, int limit) {
 				s += 2;
 				continue;
 			} else {
-				glyph = &font->glyphs[(int)*s]; // TTimo: FIXME: getting nasty warnings without the cast, hopefully this doesn't break the VM build
+				glyph = &font->glyphs[*s & 255];
 	      if (max < glyph->height) {
 		      max = glyph->height;
 			  }
@@ -141,8 +137,6 @@ void CG_Text_Paint(float x, float y, float scale, vec4_t color, const char *text
 	}
 	useScale = scale * font->glyphScale;
   if (text) {
-// TTimo: FIXME
-//		const unsigned char *s = text;
 		const char *s = text;
 		trap_R_SetColor( color );
 		memcpy(&newColor[0], &color[0], sizeof(vec4_t));
@@ -152,7 +146,7 @@ void CG_Text_Paint(float x, float y, float scale, vec4_t color, const char *text
 		}
 		count = 0;
 		while (s && *s && count < len) {
-			glyph = &font->glyphs[(int)*s]; // TTimo: FIXME: getting nasty warnings without the cast, hopefully this doesn't break the VM build
+			glyph = &font->glyphs[*s & 255];
       //int yadj = Assets.textFont.glyphs[text[i]].bottom + Assets.textFont.glyphs[text[i]].top;
       //float yadj = scale * (Assets.textFont.glyphs[text[i]].imageHeight - Assets.textFont.glyphs[text[i]].height);
 			if ( Q_IsColorString( s ) ) {
diff --git a/code/cgame/cg_newdraw.c b/code/cgame/cg_newdraw.c
index 986a40d2e4..dc3ed42ad9 100644
--- a/code/cgame/cg_newdraw.c
+++ b/code/cgame/cg_newdraw.c
@@ -1214,7 +1214,7 @@ static void CG_Text_Paint_Limit(float *maxX, float x, float y, float scale, vec4
 		}
 		count = 0;
 		while (s && *s && count < len) {
-			glyph = &font->glyphs[(int)*s]; // TTimo: FIXME: getting nasty warnings without the cast, hopefully this doesn't break the VM build
+			glyph = &font->glyphs[*s & 255];
 			if ( Q_IsColorString( s ) ) {
 				memcpy( newColor, g_color_table[ColorIndex(*(s+1))], sizeof( newColor ) );
 				newColor[3] = color[3];
diff --git a/code/ui/ui_main.c b/code/ui/ui_main.c
index 6cdaa12c84..7d06e18014 100644
--- a/code/ui/ui_main.c
+++ b/code/ui/ui_main.c
@@ -284,7 +284,7 @@ int Text_Width(const char *text, float scale, int limit) {
 				s += 2;
 				continue;
 			} else {
-				glyph = &font->glyphs[(int)*s];
+				glyph = &font->glyphs[*s & 255];
 				out += glyph->xSkip;
 				s++;
 				count++;
@@ -319,7 +319,7 @@ int Text_Height(const char *text, float scale, int limit) {
 				s += 2;
 				continue;
 			} else {
-				glyph = &font->glyphs[(int)*s]; // TTimo: FIXME: getting nasty warnings without the cast, hopefully this doesn't break the VM build
+				glyph = &font->glyphs[*s & 255];
 	      if (max < glyph->height) {
 		      max = glyph->height;
 			  }
@@ -361,7 +361,7 @@ void Text_Paint(float x, float y, float scale, vec4_t color, const char *text, f
 		}
 		count = 0;
 		while (s && *s && count < len) {
-			glyph = &font->glyphs[(int)*s]; // TTimo: FIXME: getting nasty warnings without the cast, hopefully this doesn't break the VM build
+			glyph = &font->glyphs[*s & 255];
       //int yadj = Assets.textFont.glyphs[text[i]].bottom + Assets.textFont.glyphs[text[i]].top;
       //float yadj = scale * (Assets.textFont.glyphs[text[i]].imageHeight - Assets.textFont.glyphs[text[i]].height);
 			if ( Q_IsColorString( s ) ) {
@@ -429,9 +429,9 @@ void Text_PaintWithCursor(float x, float y, float scale, vec4_t color, const cha
 			len = limit;
 		}
 		count = 0;
-		glyph2 = &font->glyphs[ (int) cursor];
+		glyph2 = &font->glyphs[cursor & 255];
 		while (s && *s && count < len) {
-			glyph = &font->glyphs[(int)*s]; // TTimo: FIXME: getting nasty warnings without the cast, hopefully this doesn't break the VM build
+			glyph = &font->glyphs[*s & 255];
       //int yadj = Assets.textFont.glyphs[text[i]].bottom + Assets.textFont.glyphs[text[i]].top;
       //float yadj = scale * (Assets.textFont.glyphs[text[i]].imageHeight - Assets.textFont.glyphs[text[i]].height);
 			if ( Q_IsColorString( s ) ) {
@@ -528,7 +528,7 @@ static void Text_Paint_Limit(float *maxX, float x, float y, float scale, vec4_t
 		}
 		count = 0;
 		while (s && *s && count < len) {
-			glyph = &font->glyphs[(int)*s]; // TTimo: FIXME: getting nasty warnings without the cast, hopefully this doesn't break the VM build
+			glyph = &font->glyphs[*s & 255];
 			if ( Q_IsColorString( s ) ) {
 				memcpy( newColor, g_color_table[ColorIndex(*(s+1))], sizeof( newColor ) );
 				newColor[3] = color[3];