Replace the shader uid system with a new one which quasi-automatically generates uids for shaders.

Currently used in the vertex shader only (had to fork lighting shaders for now).

3 files changed, 446 insertions, 275 deletions

diff --git a/Source/Core/VideoCommon/Src/Debugger.cpp b/Source/Core/VideoCommon/Src/Debugger.cpp
index 4aa25b72be..005f81c8e6 100644
--- a/Source/Core/VideoCommon/Src/Debugger.cpp
+++ b/Source/Core/VideoCommon/Src/Debugger.cpp
@@ -131,7 +131,7 @@ void GFXDebuggerBase::DumpVertexShader(const char* path)
 	sprintf(filename, "%sdump_vs.txt", path);
 
 	File::CreateEmptyFile(filename);
-	File::WriteStringToFile(true, GenerateVertexShaderCode(g_nativeVertexFmt->m_components, g_ActiveConfig.backend_info.APIType), filename);
+///	File::WriteStringToFile(true, GenerateVertexShaderCode(g_nativeVertexFmt->m_components, g_ActiveConfig.backend_info.APIType), filename);
 }
 
 void GFXDebuggerBase::DumpPixelShaderConstants(const char* path)
diff --git a/Source/Core/VideoCommon/Src/VertexShaderGen.cpp b/Source/Core/VideoCommon/Src/VertexShaderGen.cpp
index cad117c2c8..d7134b6dcb 100644
--- a/Source/Core/VideoCommon/Src/VertexShaderGen.cpp
+++ b/Source/Core/VideoCommon/Src/VertexShaderGen.cpp
@@ -26,149 +26,260 @@
 #include "VertexShaderGen.h"
 #include "VideoConfig.h"
 
-// Mash together all the inputs that contribute to the code of a generated vertex shader into
-// a unique identifier, basically containing all the bits. Yup, it's a lot ....
-void GetVertexShaderId(VERTEXSHADERUID *uid, u32 components)
-{
-	memset(uid->values, 0, sizeof(uid->values));
-	uid->values[0] = components |
-		(xfregs.numTexGen.numTexGens << 23) |
-		(xfregs.numChan.numColorChans << 27) |
-		(xfregs.dualTexTrans.enabled << 29);
+static char text[16768];
 
-	// TODO: If pixel lighting is enabled, do we even have to bother about storing lighting related registers here?
-	GetLightingShaderId(&uid->values[1]);
+enum GenOutput
+{
+	GO_ShaderCode,
+	GO_ShaderUid,
+};
+// TODO: Check if something goes wrong if the cached shaders used pixel lighting but it's disabled later??
+template<class T>
+void GenerateVSOutputStruct(T& object, u32 components, API_TYPE api_type)
+{
+	object.Write("struct VS_OUTPUT {\n");
+	object.Write("  float4 pos : POSITION;\n");
+	object.Write("  float4 colors_0 : COLOR0;\n");
+	object.Write("  float4 colors_1 : COLOR1;\n");
 
-	uid->values[2] |= (g_ActiveConfig.bEnablePixelLighting && g_ActiveConfig.backend_info.bSupportsPixelLighting) << 31;
-	u32 *pcurvalue = &uid->values[3];
-	for (unsigned int i = 0; i < xfregs.numTexGen.numTexGens; ++i) {
-		TexMtxInfo tinfo = xfregs.texMtxInfo[i];
-		if (tinfo.texgentype != XF_TEXGEN_EMBOSS_MAP)
-			tinfo.hex &= 0x7ff;
-		if (tinfo.texgentype != XF_TEXGEN_REGULAR)
-			tinfo.projection = 0;
-
-		u32 val = ((tinfo.hex >> 1) & 0x1ffff);
-		if (xfregs.dualTexTrans.enabled && tinfo.texgentype == XF_TEXGEN_REGULAR) {
-			// rewrite normalization and post index
-			val |= ((u32)xfregs.postMtxInfo[i].index << 17) | ((u32)xfregs.postMtxInfo[i].normalize << 23);
-		}
+	if (xfregs.numTexGen.numTexGens < 7)
+	{
+		for (unsigned int i = 0; i < xfregs.numTexGen.numTexGens; ++i)
+			object.Write("  float3 tex%d : TEXCOORD%d;\n", i, i);
 
-		switch (i & 3) {
-			case 0: pcurvalue[0] |= val; break;
-			case 1: pcurvalue[0] |= val << 24; pcurvalue[1] = val >> 8; ++pcurvalue; break;
-			case 2: pcurvalue[0] |= val << 16; pcurvalue[1] = val >> 16; ++pcurvalue; break;
-			case 3: pcurvalue[0] |= val << 8; ++pcurvalue; break;
+		object.Write("  float4 clipPos : TEXCOORD%d;\n", xfregs.numTexGen.numTexGens);
+///		if(g_ActiveConfig.bEnablePixelLighting && g_ActiveConfig.backend_info.bSupportsPixelLighting)
+///			object.Write("  float4 Normal : TEXCOORD%d;\n", xfregs.numTexGen.numTexGens + 1);
+	}
+	else
+	{
+		// clip position is in w of first 4 texcoords
+///		if(g_ActiveConfig.bEnablePixelLighting && g_ActiveConfig.backend_info.bSupportsPixelLighting)
+///		{
+///			for (int i = 0; i < 8; ++i)
+///				object.Write("  float4 tex%d : TEXCOORD%d;\n", i, i);
+///		}
+///		else
+		{
+			for (unsigned int i = 0; i < xfregs.numTexGen.numTexGens; ++i)
+				object.Write("  float%d tex%d : TEXCOORD%d;\n", i < 4 ? 4 : 3 , i, i);
 		}
 	}
+	object.Write("};\n");
 }
 
-void GetSafeVertexShaderId(VERTEXSHADERUIDSAFE *uid, u32 components)
+template<class T, GenOutput type>
+void _GenerateLightShader(T& object, int index, int litchan_index, const char* lightsName, int coloralpha)
 {
-	// Just store all used registers here without caring whether we need all bits or less.
-	memset(uid->values, 0, sizeof(uid->values));
-	u32* ptr = uid->values;
-	*ptr++ = components;
-	*ptr++ = xfregs.numTexGen.hex;
-	*ptr++ = xfregs.numChan.hex;
-	*ptr++ = xfregs.dualTexTrans.hex;
-
-	for (int i = 0; i < 2; ++i) {
-		*ptr++ = xfregs.color[i].hex;
-		*ptr++ = xfregs.alpha[i].hex;
-	}
-	*ptr++ = g_ActiveConfig.bEnablePixelLighting && g_ActiveConfig.backend_info.bSupportsPixelLighting;
-	for (unsigned int i = 0; i < 8; ++i) {
-		*ptr++ = xfregs.texMtxInfo[i].hex;
-		*ptr++ = xfregs.postMtxInfo[i].hex;
+#define SetUidField(name, value) if (type == GO_ShaderUid) { object.GetUidData().name = value; };
+	const LitChannel& chan = (litchan_index > 1) ? xfregs.alpha[litchan_index-2] : xfregs.color[litchan_index];
+	const char* swizzle = "xyzw";
+	if (coloralpha == 1 ) swizzle = "xyz";
+	else if (coloralpha == 2 ) swizzle = "w";
+
+	SetUidField(lit_chans[litchan_index].attnfunc, chan.attnfunc);
+	SetUidField(lit_chans[litchan_index].diffusefunc, chan.diffusefunc);
+	if (!(chan.attnfunc & 1)) {
+		// atten disabled
+		switch (chan.diffusefunc) {
+			case LIGHTDIF_NONE:
+				object.Write("lacc.%s += %s.lights[%d].col.%s;\n", swizzle, lightsName, index, swizzle);
+				break;
+			case LIGHTDIF_SIGN:
+			case LIGHTDIF_CLAMP:
+				object.Write("ldir = normalize(%s.lights[%d].pos.xyz - pos.xyz);\n", lightsName, index);
+				object.Write("lacc.%s += %sdot(ldir, _norm0)) * %s.lights[%d].col.%s;\n",
+					swizzle, chan.diffusefunc != LIGHTDIF_SIGN ? "max(0.0f," :"(", lightsName, index, swizzle);
+				break;
+			default: _assert_(0);
+		}
 	}
-	_assert_((ptr - uid->values) == uid->GetNumValues());
-}
-
-
-void ValidateVertexShaderIDs(API_TYPE api, VERTEXSHADERUIDSAFE old_id, const std::string& old_code, u32 components)
-{
-	if (!g_ActiveConfig.bEnableShaderDebugging)
-		return;
-
-	VERTEXSHADERUIDSAFE new_id;
-	GetSafeVertexShaderId(&new_id, components);
+	else { // spec and spot
+
+		if (chan.attnfunc == 3)
+		{ // spot
+			object.Write("ldir = %s.lights[%d].pos.xyz - pos.xyz;\n", lightsName, index);
+			object.Write("dist2 = dot(ldir, ldir);\n"
+						"dist = sqrt(dist2);\n"
+						"ldir = ldir / dist;\n"
+						"attn = max(0.0f, dot(ldir, %s.lights[%d].dir.xyz));\n", lightsName, index);
+			object.Write("attn = max(0.0f, dot(%s.lights[%d].cosatt.xyz, float3(1.0f, attn, attn*attn))) / dot(%s.lights[%d].distatt.xyz, float3(1.0f,dist,dist2));\n", lightsName, index, lightsName, index);
+		}
+		else if (chan.attnfunc == 1)
+		{ // specular
+			object.Write("ldir = normalize(%s.lights[%d].pos.xyz);\n", lightsName, index);
+			object.Write("attn = (dot(_norm0,ldir) >= 0.0f) ? max(0.0f, dot(_norm0, %s.lights[%d].dir.xyz)) : 0.0f;\n", lightsName, index);
+			object.Write("attn = max(0.0f, dot(%s.lights[%d].cosatt.xyz, float3(1,attn,attn*attn))) / dot(%s.lights[%d].distatt.xyz, float3(1,attn,attn*attn));\n", lightsName, index, lightsName, index);
+		}
 
-	if (!(old_id == new_id))
-	{
-		std::string new_code(GenerateVertexShaderCode(components, api));
-		if (old_code != new_code)
+		switch (chan.diffusefunc)
 		{
-			_assert_(old_id.GetNumValues() == new_id.GetNumValues());
-
-			char msg[8192];
-			char* ptr = msg;
-			ptr += sprintf(ptr, "Vertex shader IDs matched but unique IDs did not!\nUnique IDs (old <-> new):\n");
-			const int N = new_id.GetNumValues();
-			for (int i = 0; i < N/2; ++i)
-				ptr += sprintf(ptr, "%02d, %08X  %08X  |  %08X  %08X\n", 2*i, old_id.values[2*i], old_id.values[2*i+1],
-																			new_id.values[2*i], new_id.values[2*i+1]);
-			if (N % 2)
-				ptr += sprintf(ptr, "%02d, %08X  |  %08X\n", N-1, old_id.values[N-1], new_id.values[N-1]);
-				
-			static int num_failures = 0;
-			char szTemp[MAX_PATH];
-			sprintf(szTemp, "%svsuid_mismatch_%04i.txt", File::GetUserPath(D_DUMP_IDX).c_str(), num_failures++);
-			std::ofstream file(szTemp);
-			file << msg;
-			file << "\n\nOld shader code:\n" << old_code;
-			file << "\n\nNew shader code:\n" << new_code;
-			file.close();
-
-			PanicAlert("Unique pixel shader ID mismatch!\n\nReport this to the devs, along with the contents of %s.", szTemp);
+			case LIGHTDIF_NONE:
+				object.Write("lacc.%s += attn * %s.lights[%d].col.%s;\n", swizzle, lightsName, index, swizzle);
+				break;
+			case LIGHTDIF_SIGN:
+			case LIGHTDIF_CLAMP:
+				object.Write("lacc.%s += attn * %sdot(ldir, _norm0)) * %s.lights[%d].col.%s;\n",
+					swizzle,
+					chan.diffusefunc != LIGHTDIF_SIGN ? "max(0.0f," :"(",
+					lightsName,
+					index,
+					swizzle);
+				break;
+			default: _assert_(0);
 		}
 	}
+	object.Write("\n");
 }
 
+// vertex shader
+// lights/colors
+// materials name is I_MATERIALS in vs and I_PMATERIALS in ps
+// inColorName is color in vs and colors_ in ps
+// dest is o.colors_ in vs and colors_ in ps
+template<class T, GenOutput type>
+void _GenerateLightingShader(T& object, int components, const char* materialsName, const char* lightsName, const char* inColorName, const char* dest)
+{
+	for (unsigned int j = 0; j < xfregs.numChan.numColorChans; j++)
+	{
+		const LitChannel& color = xfregs.color[j];
+		const LitChannel& alpha = xfregs.alpha[j];
 
-static char text[16384];
+		object.Write("{\n");
 
-#define WRITE p+=sprintf
+		SetUidField(lit_chans[j].matsource, xfregs.color[j].matsource);
+		if (color.matsource) {// from vertex
+			if (components & (VB_HAS_COL0 << j))
+				object.Write("mat = %s%d;\n", inColorName, j);
+			else if (components & VB_HAS_COL0)
+				object.Write("mat = %s0;\n", inColorName);
+			else
+				object.Write("mat = float4(1.0f, 1.0f, 1.0f, 1.0f);\n");
+		}
+		else // from color
+			object.Write("mat = %s.C%d;\n", materialsName, j+2);
+
+		SetUidField(lit_chans[j].enablelighting, xfregs.color[j].enablelighting);
+		if (color.enablelighting) {
+			SetUidField(lit_chans[j].ambsource, xfregs.color[j].ambsource);
+			if (color.ambsource) { // from vertex
+				if (components & (VB_HAS_COL0<<j) )
+					object.Write("lacc = %s%d;\n", inColorName, j);
+				else if (components & VB_HAS_COL0 )
+					object.Write("lacc = %s0;\n", inColorName);
+				else
+					object.Write("lacc = float4(0.0f, 0.0f, 0.0f, 0.0f);\n");
+			}
+			else // from color
+				object.Write("lacc = %s.C%d;\n", materialsName, j);
+		}
+		else
+		{
+			object.Write("lacc = float4(1.0f, 1.0f, 1.0f, 1.0f);\n");
+		}
 
-char* GenerateVSOutputStruct(char* p, u32 components, API_TYPE api_type)
-{
-	WRITE(p, "struct VS_OUTPUT {\n");
-	WRITE(p, "  float4 pos : POSITION;\n");
-	WRITE(p, "  float4 colors_0 : COLOR0;\n");
-	WRITE(p, "  float4 colors_1 : COLOR1;\n");
+		// check if alpha is different
+		SetUidField(lit_chans[j+2].matsource, xfregs.alpha[j].matsource);
+		if (alpha.matsource != color.matsource) {
+			if (alpha.matsource) {// from vertex
+				if (components & (VB_HAS_COL0<<j))
+					object.Write("mat.w = %s%d.w;\n", inColorName, j);
+				else if (components & VB_HAS_COL0)
+					object.Write("mat.w = %s0.w;\n", inColorName);
+				else object.Write("mat.w = 1.0f;\n");
+			}
+			else // from color
+				object.Write("mat.w = %s.C%d.w;\n", materialsName, j+2);
+		}
 
-	if (xfregs.numTexGen.numTexGens < 7) {
-		for (unsigned int i = 0; i < xfregs.numTexGen.numTexGens; ++i)
-			WRITE(p, "  float3 tex%d : TEXCOORD%d;\n", i, i);
-		WRITE(p, "  float4 clipPos : TEXCOORD%d;\n", xfregs.numTexGen.numTexGens);
-		if(g_ActiveConfig.bEnablePixelLighting && g_ActiveConfig.backend_info.bSupportsPixelLighting)
-			WRITE(p, "  float4 Normal : TEXCOORD%d;\n", xfregs.numTexGen.numTexGens + 1);
-	} else {
-		// clip position is in w of first 4 texcoords
-		if(g_ActiveConfig.bEnablePixelLighting && g_ActiveConfig.backend_info.bSupportsPixelLighting)
+		SetUidField(lit_chans[j+2].enablelighting, xfregs.alpha[j].enablelighting);
+		if (alpha.enablelighting)
 		{
-			for (int i = 0; i < 8; ++i)
-				WRITE(p, "  float4 tex%d : TEXCOORD%d;\n", i, i);
+			SetUidField(lit_chans[j+2].ambsource, xfregs.alpha[j].ambsource);
+			if (alpha.ambsource) {// from vertex
+				if (components & (VB_HAS_COL0<<j) )
+					object.Write("lacc.w = %s%d.w;\n", inColorName, j);
+				else if (components & VB_HAS_COL0 )
+					object.Write("lacc.w = %s0.w;\n", inColorName);
+				else
+					object.Write("lacc.w = 0.0f;\n");
+			}
+			else // from color
+				object.Write("lacc.w = %s.C%d.w;\n", materialsName, j);
 		}
 		else
 		{
-			for (unsigned int i = 0; i < xfregs.numTexGen.numTexGens; ++i)
-				WRITE(p, "  float%d tex%d : TEXCOORD%d;\n", i < 4 ? 4 : 3 , i, i);
+			object.Write("lacc.w = 1.0f;\n");
 		}
-	}	
-	WRITE(p, "};\n");
 
-	return p;
+		if(color.enablelighting && alpha.enablelighting)
+		{
+			// both have lighting, test if they use the same lights
+			int mask = 0;
+			SetUidField(lit_chans[j].attnfunc, color.attnfunc);
+			SetUidField(lit_chans[j+2].attnfunc, alpha.attnfunc);
+			SetUidField(lit_chans[j].diffusefunc, color.diffusefunc);
+			SetUidField(lit_chans[j+2].diffusefunc, alpha.diffusefunc);
+			SetUidField(lit_chans[j].light_mask, color.GetFullLightMask());
+			SetUidField(lit_chans[j+2].light_mask, alpha.GetFullLightMask());
+			if(color.lightparams == alpha.lightparams)
+			{
+				mask = color.GetFullLightMask() & alpha.GetFullLightMask();
+				if(mask)
+				{
+					for (int i = 0; i < 8; ++i)
+					{
+						if (mask & (1<<i))
+						{
+							_GenerateLightShader<T,type>(object, i, j, lightsName, 3);
+						}
+					}
+				}
+			}
+
+			// no shared lights
+			for (int i = 0; i < 8; ++i)
+			{
+				if (!(mask&(1<<i)) && (color.GetFullLightMask() & (1<<i)))
+					_GenerateLightShader<T,type>(object, i, j, lightsName, 1);
+				if (!(mask&(1<<i)) && (alpha.GetFullLightMask() & (1<<i)))
+					_GenerateLightShader<T,type>(object, i, j+2, lightsName, 2);
+			}
+		}
+		else if (color.enablelighting || alpha.enablelighting)
+		{
+			// lights are disabled on one channel so process only the active ones
+			const LitChannel& workingchannel = color.enablelighting ? color : alpha;
+			const int lit_index = color.enablelighting ? j : (j+2);
+			int coloralpha = color.enablelighting ? 1 : 2;
+
+			SetUidField(lit_chans[lit_index].light_mask, workingchannel.GetFullLightMask());
+			for (int i = 0; i < 8; ++i)
+			{
+				if (workingchannel.GetFullLightMask() & (1<<i))
+					_GenerateLightShader<T,type>(object, i, lit_index, lightsName, coloralpha);
+			}
+		}
+		object.Write("%s%d = mat * saturate(lacc);\n", dest, j);
+		object.Write("}\n");
+	}
 }
 
-const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type)
+// TODO: Problem: this one uses copy constructors or sth for uids when returning...
+template<class T, GenOutput type>
+void GenerateShader(T& out, u32 components, API_TYPE api_type)
 {
-	setlocale(LC_NUMERIC, "C"); // Reset locale for compilation
-	text[sizeof(text) - 1] = 0x7C;  // canary
+#undef SetUidField
+#define SetUidField(name, value) if (type == GO_ShaderUid) {out.GetUidData().name = value; };
+
+	if (type == GO_ShaderCode)
+	{
+		out.SetBuffer(text);
+		setlocale(LC_NUMERIC, "C"); // Reset locale for compilation
+	}
+
+	///	text[sizeof(text) - 1] = 0x7C;  // canary
 
-	_assert_(bpmem.genMode.numtexgens == xfregs.numTexGen.numTexGens);
-	_assert_(bpmem.genMode.numcolchans == xfregs.numChan.numColorChans);
-	
 	bool is_d3d = (api_type & API_D3D9 || api_type == API_D3D11);
 	u32 lightMask = 0;
 	if (xfregs.numChan.numColorChans > 0)
@@ -176,141 +287,146 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type)
 	if (xfregs.numChan.numColorChans > 1)
 		lightMask |= xfregs.color[1].GetFullLightMask() | xfregs.alpha[1].GetFullLightMask();
 
-	char *p = text;
-	WRITE(p, "//Vertex Shader: comp:%x, \n", components);
-	WRITE(p, "typedef struct { float4 T0, T1, T2; float4 N0, N1, N2; } s_" I_POSNORMALMATRIX";\n"
-		"typedef struct { float4 t; } FLT4;\n"
-		"typedef struct { FLT4 T[24]; } s_" I_TEXMATRICES";\n"
-		"typedef struct { FLT4 T[64]; } s_" I_TRANSFORMMATRICES";\n"
-		"typedef struct { FLT4 T[32]; } s_" I_NORMALMATRICES";\n"
-		"typedef struct { FLT4 T[64]; } s_" I_POSTTRANSFORMMATRICES";\n"
-		"typedef struct { float4 col; float4 cosatt; float4 distatt; float4 pos; float4 dir; } Light;\n"
-		"typedef struct { Light lights[8]; } s_" I_LIGHTS";\n"
-		"typedef struct { float4 C0, C1, C2, C3; } s_" I_MATERIALS";\n"
-		"typedef struct { float4 T0, T1, T2, T3; } s_" I_PROJECTION";\n"
-		);
-
-	p = GenerateVSOutputStruct(p, components, api_type);
+	out.Write("//Vertex Shader: comp:%x, \n", components);
+	out.Write("typedef struct { float4 T0, T1, T2; float4 N0, N1, N2; } s_" I_POSNORMALMATRIX";\n"
+			"typedef struct { float4 t; } FLT4;\n"
+			"typedef struct { FLT4 T[24]; } s_" I_TEXMATRICES";\n"
+			"typedef struct { FLT4 T[64]; } s_" I_TRANSFORMMATRICES";\n"
+			"typedef struct { FLT4 T[32]; } s_" I_NORMALMATRICES";\n"
+			"typedef struct { FLT4 T[64]; } s_" I_POSTTRANSFORMMATRICES";\n"
+			"typedef struct { float4 col; float4 cosatt; float4 distatt; float4 pos; float4 dir; } Light;\n"
+			"typedef struct { Light lights[8]; } s_" I_LIGHTS";\n"
+			"typedef struct { float4 C0, C1, C2, C3; } s_" I_MATERIALS";\n"
+			"typedef struct { float4 T0, T1, T2, T3; } s_" I_PROJECTION";\n"
+			);
+
+///	p = GenerateVSOutputStruct(p, components, api_type);
+	GenerateVSOutputStruct(out, components, api_type);
 
 	// uniforms
 
-	WRITE(p, "uniform s_" I_TRANSFORMMATRICES" " I_TRANSFORMMATRICES" : register(c%d);\n", C_TRANSFORMMATRICES);
-	WRITE(p, "uniform s_" I_TEXMATRICES" " I_TEXMATRICES" : register(c%d);\n", C_TEXMATRICES); // also using tex matrices
-	WRITE(p, "uniform s_" I_NORMALMATRICES" " I_NORMALMATRICES" : register(c%d);\n", C_NORMALMATRICES);
-	WRITE(p, "uniform s_" I_POSNORMALMATRIX" " I_POSNORMALMATRIX" : register(c%d);\n", C_POSNORMALMATRIX);
-	WRITE(p, "uniform s_" I_POSTTRANSFORMMATRICES" " I_POSTTRANSFORMMATRICES" : register(c%d);\n", C_POSTTRANSFORMMATRICES);
-	WRITE(p, "uniform s_" I_LIGHTS" " I_LIGHTS" : register(c%d);\n", C_LIGHTS);
-	WRITE(p, "uniform s_" I_MATERIALS" " I_MATERIALS" : register(c%d);\n", C_MATERIALS);
-	WRITE(p, "uniform s_" I_PROJECTION" " I_PROJECTION" : register(c%d);\n", C_PROJECTION);
-	WRITE(p, "uniform float4 " I_DEPTHPARAMS" : register(c%d);\n", C_DEPTHPARAMS);
-
-	WRITE(p, "VS_OUTPUT main(\n");	
-	
+	out.Write("uniform s_" I_TRANSFORMMATRICES" " I_TRANSFORMMATRICES" : register(c%d);\n", C_TRANSFORMMATRICES);
+	out.Write("uniform s_" I_TEXMATRICES" " I_TEXMATRICES" : register(c%d);\n", C_TEXMATRICES);
+	out.Write("uniform s_" I_NORMALMATRICES" " I_NORMALMATRICES" : register(c%d);\n", C_NORMALMATRICES);
+	out.Write("uniform s_" I_POSNORMALMATRIX" " I_POSNORMALMATRIX" : register(c%d);\n", C_POSNORMALMATRIX);
+	out.Write("uniform s_" I_POSTTRANSFORMMATRICES" " I_POSTTRANSFORMMATRICES" : register(c%d);\n", C_POSTTRANSFORMMATRICES);
+	out.Write("uniform s_" I_LIGHTS" " I_LIGHTS" : register(c%d);\n", C_LIGHTS);
+	out.Write("uniform s_" I_MATERIALS" " I_MATERIALS" : register(c%d);\n", C_MATERIALS);
+	out.Write("uniform s_" I_PROJECTION" " I_PROJECTION" : register(c%d);\n", C_PROJECTION);
+	out.Write("uniform float4 " I_DEPTHPARAMS" : register(c%d);\n", C_DEPTHPARAMS);
+
+	out.Write("VS_OUTPUT main(\n");
+
+	SetUidField(numTexGens, xfregs.numTexGen.numTexGens);
+	SetUidField(components, components);
 	// inputs
 	if (components & VB_HAS_NRM0)
-		WRITE(p, "  float3 rawnorm0 : NORMAL0,\n");
-	if (components & VB_HAS_NRM1) {
+		out.Write("  float3 rawnorm0 : NORMAL0,\n");
+	if (components & VB_HAS_NRM1)
+	{
 		if (is_d3d)
-			WRITE(p, "  float3 rawnorm1 : NORMAL1,\n");
+			out.Write("  float3 rawnorm1 : NORMAL1,\n");
 		else
-			WRITE(p, "  float3 rawnorm1 : ATTR%d,\n", SHADER_NORM1_ATTRIB);
+			out.Write("  float3 rawnorm1 : ATTR%d,\n", SHADER_NORM1_ATTRIB);
 	}
-	if (components & VB_HAS_NRM2) {
+	if (components & VB_HAS_NRM2)
+	{
 		if (is_d3d)
-			WRITE(p, "  float3 rawnorm2 : NORMAL2,\n");
+			out.Write("  float3 rawnorm2 : NORMAL2,\n");
 		else
-			WRITE(p, "  float3 rawnorm2 : ATTR%d,\n", SHADER_NORM2_ATTRIB);
+			out.Write("  float3 rawnorm2 : ATTR%d,\n", SHADER_NORM2_ATTRIB);
 	}
 	if (components & VB_HAS_COL0)
-		WRITE(p, "  float4 color0 : COLOR0,\n");
+		out.Write("  float4 color0 : COLOR0,\n");
 	if (components & VB_HAS_COL1)
-		WRITE(p, "  float4 color1 : COLOR1,\n");
+		out.Write("  float4 color1 : COLOR1,\n");
 	for (int i = 0; i < 8; ++i) {
 		u32 hastexmtx = (components & (VB_HAS_TEXMTXIDX0<<i));
 		if ((components & (VB_HAS_UV0<<i)) || hastexmtx)
-			WRITE(p, "  float%d tex%d : TEXCOORD%d,\n", hastexmtx ? 3 : 2, i, i);
+			out.Write("  float%d tex%d : TEXCOORD%d,\n", hastexmtx ? 3 : 2, i, i);
 	}
 	if (components & VB_HAS_POSMTXIDX) {
 		if (is_d3d)
 		{
-			WRITE(p, "  float4 blend_indices : BLENDINDICES,\n");
+			out.Write("  float4 blend_indices : BLENDINDICES,\n");
 		}
 		else
-			WRITE(p, "  float fposmtx : ATTR%d,\n", SHADER_POSMTX_ATTRIB);
+			out.Write("  float fposmtx : ATTR%d,\n", SHADER_POSMTX_ATTRIB);
 	}
-	WRITE(p, "  float4 rawpos : POSITION) {\n");
-	WRITE(p, "VS_OUTPUT o;\n");	
+	out.Write("  float4 rawpos : POSITION) {\n");
+	out.Write("VS_OUTPUT o;\n");
 
 	// transforms
 	if (components & VB_HAS_POSMTXIDX)
 	{
 		if (api_type & API_D3D9)
 		{
-			WRITE(p, "int4 indices = D3DCOLORtoUBYTE4(blend_indices);\n");
-			WRITE(p, "int posmtx = indices.x;\n");
+			out.Write("int4 indices = D3DCOLORtoUBYTE4(blend_indices);\n");
+			out.Write("int posmtx = indices.x;\n");
 		}
 		else if (api_type == API_D3D11)
 		{
-			WRITE(p, "int posmtx = blend_indices.x * 255.0f;\n");
+			out.Write("int posmtx = blend_indices.x * 255.0f;\n");
 		}
 		else
 		{
-			WRITE(p, "int posmtx = fposmtx;\n");
+			out.Write("int posmtx = fposmtx;\n");
 		}
 
-		WRITE(p, "float4 pos = float4(dot(" I_TRANSFORMMATRICES".T[posmtx].t, rawpos), dot(" I_TRANSFORMMATRICES".T[posmtx+1].t, rawpos), dot(" I_TRANSFORMMATRICES".T[posmtx+2].t, rawpos), 1);\n");		
+		out.Write("float4 pos = float4(dot(" I_TRANSFORMMATRICES".T[posmtx].t, rawpos), dot(" I_TRANSFORMMATRICES".T[posmtx+1].t, rawpos), dot(" I_TRANSFORMMATRICES".T[posmtx+2].t, rawpos), 1);\n");
 
 		if (components & VB_HAS_NRMALL) {
-			WRITE(p, "int normidx = posmtx >= 32 ? (posmtx-32) : posmtx;\n");
-			WRITE(p, "float3 N0 = " I_NORMALMATRICES".T[normidx].t.xyz, N1 = " I_NORMALMATRICES".T[normidx+1].t.xyz, N2 = " I_NORMALMATRICES".T[normidx+2].t.xyz;\n");
+			out.Write("int normidx = posmtx >= 32 ? (posmtx-32) : posmtx;\n");
+			out.Write("float3 N0 = " I_NORMALMATRICES".T[normidx].t.xyz, N1 = " I_NORMALMATRICES".T[normidx+1].t.xyz, N2 = " I_NORMALMATRICES".T[normidx+2].t.xyz;\n");
 		}
 
 		if (components & VB_HAS_NRM0)
-			WRITE(p, "float3 _norm0 = normalize(float3(dot(N0, rawnorm0), dot(N1, rawnorm0), dot(N2, rawnorm0)));\n");
+			out.Write("float3 _norm0 = normalize(float3(dot(N0, rawnorm0), dot(N1, rawnorm0), dot(N2, rawnorm0)));\n");
 		if (components & VB_HAS_NRM1)
-			WRITE(p, "float3 _norm1 = float3(dot(N0, rawnorm1), dot(N1, rawnorm1), dot(N2, rawnorm1));\n");
+			out.Write("float3 _norm1 = float3(dot(N0, rawnorm1), dot(N1, rawnorm1), dot(N2, rawnorm1));\n");
 		if (components & VB_HAS_NRM2)
-			WRITE(p, "float3 _norm2 = float3(dot(N0, rawnorm2), dot(N1, rawnorm2), dot(N2, rawnorm2));\n");
+			out.Write("float3 _norm2 = float3(dot(N0, rawnorm2), dot(N1, rawnorm2), dot(N2, rawnorm2));\n");
 	}
 	else
 	{
-		WRITE(p, "float4 pos = float4(dot(" I_POSNORMALMATRIX".T0, rawpos), dot(" I_POSNORMALMATRIX".T1, rawpos), dot(" I_POSNORMALMATRIX".T2, rawpos), 1.0f);\n");
+		out.Write("float4 pos = float4(dot(" I_POSNORMALMATRIX".T0, rawpos), dot(" I_POSNORMALMATRIX".T1, rawpos), dot(" I_POSNORMALMATRIX".T2, rawpos), 1.0f);\n");
 		if (components & VB_HAS_NRM0)
-			WRITE(p, "float3 _norm0 = normalize(float3(dot(" I_POSNORMALMATRIX".N0.xyz, rawnorm0), dot(" I_POSNORMALMATRIX".N1.xyz, rawnorm0), dot(" I_POSNORMALMATRIX".N2.xyz, rawnorm0)));\n");
+			out.Write("float3 _norm0 = normalize(float3(dot(" I_POSNORMALMATRIX".N0.xyz, rawnorm0), dot(" I_POSNORMALMATRIX".N1.xyz, rawnorm0), dot(" I_POSNORMALMATRIX".N2.xyz, rawnorm0)));\n");
 		if (components & VB_HAS_NRM1)
-			WRITE(p, "float3 _norm1 = float3(dot(" I_POSNORMALMATRIX".N0.xyz, rawnorm1), dot(" I_POSNORMALMATRIX".N1.xyz, rawnorm1), dot(" I_POSNORMALMATRIX".N2.xyz, rawnorm1));\n");
+			out.Write("float3 _norm1 = float3(dot(" I_POSNORMALMATRIX".N0.xyz, rawnorm1), dot(" I_POSNORMALMATRIX".N1.xyz, rawnorm1), dot(" I_POSNORMALMATRIX".N2.xyz, rawnorm1));\n");
 		if (components & VB_HAS_NRM2)
-			WRITE(p, "float3 _norm2 = float3(dot(" I_POSNORMALMATRIX".N0.xyz, rawnorm2), dot(" I_POSNORMALMATRIX".N1.xyz, rawnorm2), dot(" I_POSNORMALMATRIX".N2.xyz, rawnorm2));\n");
+			out.Write("float3 _norm2 = float3(dot(" I_POSNORMALMATRIX".N0.xyz, rawnorm2), dot(" I_POSNORMALMATRIX".N1.xyz, rawnorm2), dot(" I_POSNORMALMATRIX".N2.xyz, rawnorm2));\n");
 	}
 
 	if (!(components & VB_HAS_NRM0))
-		WRITE(p, "float3 _norm0 = float3(0.0f, 0.0f, 0.0f);\n");
+		out.Write("float3 _norm0 = float3(0.0f, 0.0f, 0.0f);\n");
 
-	
 
-	WRITE(p, "o.pos = float4(dot(" I_PROJECTION".T0, pos), dot(" I_PROJECTION".T1, pos), dot(" I_PROJECTION".T2, pos), dot(" I_PROJECTION".T3, pos));\n");
 
-	WRITE(p, "float4 mat, lacc;\n"
-	"float3 ldir, h;\n"
-	"float dist, dist2, attn;\n");
+	out.Write("o.pos = float4(dot(" I_PROJECTION".T0, pos), dot(" I_PROJECTION".T1, pos), dot(" I_PROJECTION".T2, pos), dot(" I_PROJECTION".T3, pos));\n");
 
+	out.Write("float4 mat, lacc;\n"
+			"float3 ldir, h;\n"
+			"float dist, dist2, attn;\n");
+
+	SetUidField(numColorChans, xfregs.numChan.numColorChans);
 	if(xfregs.numChan.numColorChans == 0)
 	{
 		if (components & VB_HAS_COL0)
-			WRITE(p, "o.colors_0 = color0;\n");
+			out.Write("o.colors_0 = color0;\n");
 		else
-			WRITE(p, "o.colors_0 = float4(1.0f, 1.0f, 1.0f, 1.0f);\n");		
+			out.Write("o.colors_0 = float4(1.0f, 1.0f, 1.0f, 1.0f);\n");
 	}
 
 	// TODO: This probably isn't necessary if pixel lighting is enabled.
-	p = GenerateLightingShader(p, components, I_MATERIALS, I_LIGHTS, "color", "o.colors_");
+	_GenerateLightingShader<T,type>(out, components, I_MATERIALS, I_LIGHTS, "color", "o.colors_");
 
 	if(xfregs.numChan.numColorChans < 2)
 	{
 		if (components & VB_HAS_COL1)
-			WRITE(p, "o.colors_1 = color1;\n");
+			out.Write("o.colors_1 = color1;\n");
 		else
-			WRITE(p, "o.colors_1 = o.colors_0;\n");		
+			out.Write("o.colors_1 = o.colors_0;\n");
 	}
 	// special case if only pos and tex coord 0 and tex coord input is AB11
 	// donko - this has caused problems in some games. removed for now.
@@ -322,21 +438,22 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type)
 		*/
 
 	// transform texcoords
-	WRITE(p, "float4 coord = float4(0.0f, 0.0f, 1.0f, 1.0f);\n");
+	out.Write("float4 coord = float4(0.0f, 0.0f, 1.0f, 1.0f);\n");
 	for (unsigned int i = 0; i < xfregs.numTexGen.numTexGens; ++i) {
 		TexMtxInfo& texinfo = xfregs.texMtxInfo[i];
 
-		WRITE(p, "{\n");
-		WRITE(p, "coord = float4(0.0f, 0.0f, 1.0f, 1.0f);\n");
+		out.Write("{\n");
+		out.Write("coord = float4(0.0f, 0.0f, 1.0f, 1.0f);\n");
+		SetUidField(texMtxInfo[i].sourcerow, xfregs.texMtxInfo[i].sourcerow);
 		switch (texinfo.sourcerow) {
 		case XF_SRCGEOM_INROW:
 			_assert_( texinfo.inputform == XF_TEXINPUT_ABC1 );
-			WRITE(p, "coord = rawpos;\n"); // pos.w is 1
+			out.Write("coord = rawpos;\n"); // pos.w is 1
 			break;
 		case XF_SRCNORMAL_INROW:
 			if (components & VB_HAS_NRM0) {
 				_assert_( texinfo.inputform == XF_TEXINPUT_ABC1 );
-				WRITE(p, "coord = float4(rawnorm0.xyz, 1.0f);\n");
+				out.Write("coord = float4(rawnorm0.xyz, 1.0f);\n");
 			}
 			break;
 		case XF_SRCCOLORS_INROW:
@@ -345,72 +462,79 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type)
 		case XF_SRCBINORMAL_T_INROW:
 			if (components & VB_HAS_NRM1) {
 				_assert_( texinfo.inputform == XF_TEXINPUT_ABC1 );
-				WRITE(p, "coord = float4(rawnorm1.xyz, 1.0f);\n");
+				out.Write("coord = float4(rawnorm1.xyz, 1.0f);\n");
 			}
 			break;
 		case XF_SRCBINORMAL_B_INROW:
 			if (components & VB_HAS_NRM2) {
 				_assert_( texinfo.inputform == XF_TEXINPUT_ABC1 );
-				WRITE(p, "coord = float4(rawnorm2.xyz, 1.0f);\n");
+				out.Write("coord = float4(rawnorm2.xyz, 1.0f);\n");
 			}
 			break;
 		default:
 			_assert_(texinfo.sourcerow <= XF_SRCTEX7_INROW);
 			if (components & (VB_HAS_UV0<<(texinfo.sourcerow - XF_SRCTEX0_INROW)) )
-				WRITE(p, "coord = float4(tex%d.x, tex%d.y, 1.0f, 1.0f);\n", texinfo.sourcerow - XF_SRCTEX0_INROW, texinfo.sourcerow - XF_SRCTEX0_INROW);
+				out.Write("coord = float4(tex%d.x, tex%d.y, 1.0f, 1.0f);\n", texinfo.sourcerow - XF_SRCTEX0_INROW, texinfo.sourcerow - XF_SRCTEX0_INROW);
 			break;
 		}
 
 		// first transformation
+		SetUidField(texMtxInfo[i].texgentype, xfregs.texMtxInfo[i].texgentype);
 		switch (texinfo.texgentype) {
 			case XF_TEXGEN_EMBOSS_MAP: // calculate tex coords into bump map
 
 				if (components & (VB_HAS_NRM1|VB_HAS_NRM2)) {
 					// transform the light dir into tangent space
-					WRITE(p, "ldir = normalize(" I_LIGHTS".lights[%d].pos.xyz - pos.xyz);\n", texinfo.embosslightshift);
-					WRITE(p, "o.tex%d.xyz = o.tex%d.xyz + float3(dot(ldir, _norm1), dot(ldir, _norm2), 0.0f);\n", i, texinfo.embosssourceshift);
+					SetUidField(texMtxInfo[i].embosslightshift, xfregs.texMtxInfo[i].embosslightshift);
+					SetUidField(texMtxInfo[i].embosssourceshift, xfregs.texMtxInfo[i].embosssourceshift);
+					out.Write("ldir = normalize(" I_LIGHTS".lights[%d].pos.xyz - pos.xyz);\n", texinfo.embosslightshift);
+					out.Write("o.tex%d.xyz = o.tex%d.xyz + float3(dot(ldir, _norm1), dot(ldir, _norm2), 0.0f);\n", i, texinfo.embosssourceshift);
 				}
 				else
 				{
 					_assert_(0); // should have normals
-					WRITE(p, "o.tex%d.xyz = o.tex%d.xyz;\n", i, texinfo.embosssourceshift);
+					SetUidField(texMtxInfo[i].embosssourceshift, xfregs.texMtxInfo[i].embosssourceshift);
+					out.Write("o.tex%d.xyz = o.tex%d.xyz;\n", i, texinfo.embosssourceshift);
 				}
 
 				break;
 			case XF_TEXGEN_COLOR_STRGBC0:
 				_assert_(texinfo.sourcerow == XF_SRCCOLORS_INROW);
-				WRITE(p, "o.tex%d.xyz = float3(o.colors_0.x, o.colors_0.y, 1);\n", i);
+				out.Write("o.tex%d.xyz = float3(o.colors_0.x, o.colors_0.y, 1);\n", i);
 				break;
 			case XF_TEXGEN_COLOR_STRGBC1:
 				_assert_(texinfo.sourcerow == XF_SRCCOLORS_INROW);
-				WRITE(p, "o.tex%d.xyz = float3(o.colors_1.x, o.colors_1.y, 1);\n", i);
+				out.Write("o.tex%d.xyz = float3(o.colors_1.x, o.colors_1.y, 1);\n", i);
 				break;
 			case XF_TEXGEN_REGULAR:
 			default:
+				SetUidField(texMtxInfo[i].projection, xfregs.texMtxInfo[i].projection);
 				if (components & (VB_HAS_TEXMTXIDX0<<i)) {
 					if (texinfo.projection == XF_TEXPROJ_STQ)
-						WRITE(p, "o.tex%d.xyz = float3(dot(coord, " I_TRANSFORMMATRICES".T[tex%d.z].t), dot(coord, " I_TRANSFORMMATRICES".T[tex%d.z+1].t), dot(coord, " I_TRANSFORMMATRICES".T[tex%d.z+2].t));\n", i, i, i, i);
+						out.Write("o.tex%d.xyz = float3(dot(coord, " I_TRANSFORMMATRICES".T[tex%d.z].t), dot(coord, " I_TRANSFORMMATRICES".T[tex%d.z+1].t), dot(coord, " I_TRANSFORMMATRICES".T[tex%d.z+2].t));\n", i, i, i, i);
 					else {
-						WRITE(p, "o.tex%d.xyz = float3(dot(coord, " I_TRANSFORMMATRICES".T[tex%d.z].t), dot(coord, " I_TRANSFORMMATRICES".T[tex%d.z+1].t), 1);\n", i, i, i);
+						out.Write("o.tex%d.xyz = float3(dot(coord, " I_TRANSFORMMATRICES".T[tex%d.z].t), dot(coord, " I_TRANSFORMMATRICES".T[tex%d.z+1].t), 1);\n", i, i, i);
 					}
 				}
 				else {
 					if (texinfo.projection == XF_TEXPROJ_STQ)
-						WRITE(p, "o.tex%d.xyz = float3(dot(coord, " I_TEXMATRICES".T[%d].t), dot(coord, " I_TEXMATRICES".T[%d].t), dot(coord, " I_TEXMATRICES".T[%d].t));\n", i, 3*i, 3*i+1, 3*i+2);
+						out.Write("o.tex%d.xyz = float3(dot(coord, " I_TEXMATRICES".T[%d].t), dot(coord, " I_TEXMATRICES".T[%d].t), dot(coord, " I_TEXMATRICES".T[%d].t));\n", i, 3*i, 3*i+1, 3*i+2);
 					else
-						WRITE(p, "o.tex%d.xyz = float3(dot(coord, " I_TEXMATRICES".T[%d].t), dot(coord, " I_TEXMATRICES".T[%d].t), 1);\n", i, 3*i, 3*i+1);
+						out.Write("o.tex%d.xyz = float3(dot(coord, " I_TEXMATRICES".T[%d].t), dot(coord, " I_TEXMATRICES".T[%d].t), 1);\n", i, 3*i, 3*i+1);
 				}
 				break;
 		}
 
+		SetUidField(dualTexTrans.enabled, xfregs.dualTexTrans.enabled);
 		if (xfregs.dualTexTrans.enabled && texinfo.texgentype == XF_TEXGEN_REGULAR) { // only works for regular tex gen types?
 			const PostMtxInfo& postInfo = xfregs.postMtxInfo[i];
 
+			SetUidField(postMtxInfo[i].index, xfregs.postMtxInfo[i].index);
 			int postidx = postInfo.index;
-			WRITE(p, "float4 P0 = " I_POSTTRANSFORMMATRICES".T[%d].t;\n"
-				"float4 P1 = " I_POSTTRANSFORMMATRICES".T[%d].t;\n"
-				"float4 P2 = " I_POSTTRANSFORMMATRICES".T[%d].t;\n",
-				postidx&0x3f, (postidx+1)&0x3f, (postidx+2)&0x3f);
+			out.Write("float4 P0 = " I_POSTTRANSFORMMATRICES".T[%d].t;\n"
+					"float4 P1 = " I_POSTTRANSFORMMATRICES".T[%d].t;\n"
+					"float4 P2 = " I_POSTTRANSFORMMATRICES".T[%d].t;\n",
+					postidx&0x3f, (postidx+1)&0x3f, (postidx+2)&0x3f);
 
 			if (texGenSpecialCase) {
 				// no normalization
@@ -418,70 +542,71 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type)
 				// q of output is unknown
 
 				// multiply by postmatrix
-				WRITE(p, "o.tex%d.xyz = float3(dot(P0.xy, o.tex%d.xy) + P0.z + P0.w, dot(P1.xy, o.tex%d.xy) + P1.z + P1.w, 0.0f);\n", i, i, i);
+				out.Write("o.tex%d.xyz = float3(dot(P0.xy, o.tex%d.xy) + P0.z + P0.w, dot(P1.xy, o.tex%d.xy) + P1.z + P1.w, 0.0f);\n", i, i, i);
 			}
 			else
 			{
+				SetUidField(postMtxInfo[i].normalize, xfregs.postMtxInfo[i].normalize);
 				if (postInfo.normalize)
-					WRITE(p, "o.tex%d.xyz = normalize(o.tex%d.xyz);\n", i, i);
+					out.Write("o.tex%d.xyz = normalize(o.tex%d.xyz);\n", i, i);
 
 				// multiply by postmatrix
-				WRITE(p, "o.tex%d.xyz = float3(dot(P0.xyz, o.tex%d.xyz) + P0.w, dot(P1.xyz, o.tex%d.xyz) + P1.w, dot(P2.xyz, o.tex%d.xyz) + P2.w);\n", i, i, i, i);
+				out.Write("o.tex%d.xyz = float3(dot(P0.xyz, o.tex%d.xyz) + P0.w, dot(P1.xyz, o.tex%d.xyz) + P1.w, dot(P2.xyz, o.tex%d.xyz) + P2.w);\n", i, i, i, i);
 			}
 		}
 
-		WRITE(p, "}\n");
+		out.Write("}\n");
 	}
 
 	// clipPos/w needs to be done in pixel shader, not here
 	if (xfregs.numTexGen.numTexGens < 7) {
-		WRITE(p, "o.clipPos = float4(pos.x,pos.y,o.pos.z,o.pos.w);\n");
+		out.Write("o.clipPos = float4(pos.x,pos.y,o.pos.z,o.pos.w);\n");
 	} else {
-		WRITE(p, "o.tex0.w = pos.x;\n");
-		WRITE(p, "o.tex1.w = pos.y;\n");
-		WRITE(p, "o.tex2.w = o.pos.z;\n");
-		WRITE(p, "o.tex3.w = o.pos.w;\n");
+		out.Write("o.tex0.w = pos.x;\n");
+		out.Write("o.tex1.w = pos.y;\n");
+		out.Write("o.tex2.w = o.pos.z;\n");
+		out.Write("o.tex3.w = o.pos.w;\n");
 	}
 
-	if(g_ActiveConfig.bEnablePixelLighting && g_ActiveConfig.backend_info.bSupportsPixelLighting)
+/*	if(g_ActiveConfig.bEnablePixelLighting && g_ActiveConfig.backend_info.bSupportsPixelLighting)
 	{
 		if (xfregs.numTexGen.numTexGens < 7) {
-			WRITE(p, "o.Normal = float4(_norm0.x,_norm0.y,_norm0.z,pos.z);\n");
+			out.Write("o.Normal = float4(_norm0.x,_norm0.y,_norm0.z,pos.z);\n");
 		} else {
-			WRITE(p, "o.tex4.w = _norm0.x;\n");
-			WRITE(p, "o.tex5.w = _norm0.y;\n");
-			WRITE(p, "o.tex6.w = _norm0.z;\n");
+			out.Write("o.tex4.w = _norm0.x;\n");
+			out.Write("o.tex5.w = _norm0.y;\n");
+			out.Write("o.tex6.w = _norm0.z;\n");
 			if (xfregs.numTexGen.numTexGens < 8)
-				WRITE(p, "o.tex7 = pos.xyzz;\n");
+				out.Write("o.tex7 = pos.xyzz;\n");
 			else
-				WRITE(p, "o.tex7.w = pos.z;\n");
-		}		
+				out.Write("o.tex7.w = pos.z;\n");
+		}
 		if (components & VB_HAS_COL0)
-			WRITE(p, "o.colors_0 = color0;\n");		
+			out.Write("o.colors_0 = color0;\n");
 
 		if (components & VB_HAS_COL1)
-			WRITE(p, "o.colors_1 = color1;\n");
-	}
+			out.Write("o.colors_1 = color1;\n");
+	}*/
 
 	//write the true depth value, if the game uses depth textures pixel shaders will override with the correct values
 	//if not early z culling will improve speed
 	// TODO: Can probably be dropped?
 	if (is_d3d)
 	{
-		WRITE(p, "o.pos.z = " I_DEPTHPARAMS".x * o.pos.w + o.pos.z * " I_DEPTHPARAMS".y;\n");
+		out.Write("o.pos.z = " I_DEPTHPARAMS".x * o.pos.w + o.pos.z * " I_DEPTHPARAMS".y;\n");
 	}
 	else
 	{
 	    // this results in a scale from -1..0 to -1..1 after perspective
 	    // divide
-	    WRITE(p, "o.pos.z = o.pos.w + o.pos.z * 2.0f;\n");
+	    out.Write("o.pos.z = o.pos.w + o.pos.z * 2.0f;\n");
 
 	    // Sonic Unleashed puts its final rendering at the near or
 	    // far plane of the viewing frustrum(actually box, they use
 	    // orthogonal projection for that), and we end up putting it
 	    // just beyond, and the rendering gets clipped away. (The
 	    // primitive gets dropped)
-	    WRITE(p, "o.pos.z = o.pos.z * 1048575.0f/1048576.0f;\n");
+	    out.Write("o.pos.z = o.pos.z * 1048575.0f/1048576.0f;\n");
 
 	    // the next steps of the OGL pipeline are:
 	    // (x_c,y_c,z_c,w_c) = o.pos  //switch to OGL spec terminology
@@ -489,7 +614,7 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type)
 	    // (x_d,y_d,z_d) = (x_c,y_c,z_c)/w_c//perspective divide
 	    // z_w = (f-n)/2*z_d + (n+f)/2
 	    // z_w now contains the value to go to the 0..1 depth buffer
-	    
+
 	    //trying to get the correct semantic while not using glDepthRange
 	    //seems to get rather complicated
 	}
@@ -498,14 +623,24 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type)
 	{
 		// D3D9 is addressing pixel centers instead of pixel boundaries in clip space.
 		// Thus we need to offset the final position by half a pixel
-		WRITE(p, "o.pos = o.pos + float4(" I_DEPTHPARAMS".z, " I_DEPTHPARAMS".w, 0.f, 0.f);\n");
+		out.Write("o.pos = o.pos + float4(" I_DEPTHPARAMS".z, " I_DEPTHPARAMS".w, 0.f, 0.f);\n");
 	}
 
-	WRITE(p, "return o;\n}\n");
+	out.Write("return o;\n}\n");
+
 
+///	if (text[sizeof(text) - 1] != 0x7C)
+///		PanicAlert("VertexShader generator - buffer too small, canary has been eaten!");
+	if (type == GO_ShaderCode)
+		setlocale(LC_NUMERIC, ""); // restore locale
+}
 
-	if (text[sizeof(text) - 1] != 0x7C)
-		PanicAlert("VertexShader generator - buffer too small, canary has been eaten!");
-	setlocale(LC_NUMERIC, ""); // restore locale
-	return text;
+void GenerateShaderUid(ShaderUid& object, u32 components, API_TYPE api_type)
+{
+	return GenerateShader<ShaderUid, GO_ShaderUid>(object, components, api_type);
+}
+
+void GenerateShaderCode(ShaderCode& object, u32 components, API_TYPE api_type)
+{
+	return GenerateShader<ShaderCode, GO_ShaderCode>(object, components, api_type);
 }
diff --git a/Source/Core/VideoCommon/Src/VertexShaderGen.h b/Source/Core/VideoCommon/Src/VertexShaderGen.h
index cb253a9b6c..11bb5c29f6 100644
--- a/Source/Core/VideoCommon/Src/VertexShaderGen.h
+++ b/Source/Core/VideoCommon/Src/VertexShaderGen.h
@@ -18,6 +18,7 @@
 #ifndef GCOGL_VERTEXSHADER_H
 #define GCOGL_VERTEXSHADER_H
 
+#include <stdarg.h>
 #include "XFMemory.h"
 #include "VideoCommon.h"
 
@@ -48,71 +49,106 @@
 #define C_DEPTHPARAMS           (C_POSTTRANSFORMMATRICES + 64)
 #define C_VENVCONST_END			(C_DEPTHPARAMS + 4)
 
-template<bool safe>
-class _VERTEXSHADERUID
+// TODO: Need packing?
+struct uid_data
 {
-#define NUM_VSUID_VALUES_SAFE 25
-public:
-	u32 values[safe ? NUM_VSUID_VALUES_SAFE : 9];
+	u32 components;
+	u32 numColorChans : 2;
+	u32 numTexGens : 4;
+
+	struct {
+		u32 projection : 1; // XF_TEXPROJ_X
+		u32 inputform : 2; // XF_TEXINPUT_X
+		u32 texgentype : 3; // XF_TEXGEN_X
+		u32 sourcerow : 5; // XF_SRCGEOM_X
+		u32 embosssourceshift : 3; // what generated texcoord to use
+		u32 embosslightshift : 3; // light index that is used
+	} texMtxInfo[8];
+	struct {
+		u32 index : 6; // base row of dual transform matrix
+		u32 normalize : 1; // normalize before send operation
+	} postMtxInfo[8];
+	struct {
+		u32 enabled : 1;
+	} dualTexTrans;
+	struct {
+		u32 matsource : 1;
+		u32 enablelighting : 1;
+		u32 ambsource : 1;
+        u32 diffusefunc : 2;
+        u32 attnfunc : 2;
+		u32 light_mask : 8;
+	} lit_chans[4];
+};
 
-	_VERTEXSHADERUID()
-	{
-	}
 
-	_VERTEXSHADERUID(const _VERTEXSHADERUID& r)
+class ShaderUid
+{
+public:
+	ShaderUid()
 	{
-		for (size_t i = 0; i < sizeof(values) / sizeof(u32); ++i) 
-			values[i] = r.values[i]; 
+		memset(values, 0, sizeof(values));
 	}
 
-	int GetNumValues() const 
+	void Write(const char* fmt, ...) {}
+	const char* GetBuffer() { return NULL; }
+	void SetBuffer(char* buffer) { }
+
+	bool operator == (const ShaderUid& obj) const
 	{
-		if (safe) return NUM_VSUID_VALUES_SAFE;
-		else return (((values[0] >> 23) & 0xf) * 3 + 3) / 4 + 3; // numTexGens*3/4+1
+		return memcmp(this->values, obj.values, sizeof(values)) == 0;
 	}
 
-	bool operator <(const _VERTEXSHADERUID& _Right) const
+	// TODO: Store last frame used and order by that? makes much more sense anyway...
+	bool operator < (const ShaderUid& obj) const
 	{
-		if (values[0] < _Right.values[0])
-			return true;
-		else if (values[0] > _Right.values[0])
-			return false;
-		int N = GetNumValues();
-		for (int i = 1; i < N; ++i) 
+		for (int i = 0; i < 24; ++i)
 		{
-			if (values[i] < _Right.values[i])
+			if (this->values[i] < obj.values[i])
 				return true;
-			else if (values[i] > _Right.values[i])
+			else if (this->values[i] > obj.values[i])
 				return false;
 		}
 		return false;
 	}
 
-	bool operator ==(const _VERTEXSHADERUID& _Right) const
+	uid_data& GetUidData() { return data; }
+
+private:
+	union
 	{
-		if (values[0] != _Right.values[0])
-			return false;
-		int N = GetNumValues();
-		for (int i = 1; i < N; ++i)
-		{
-			if (values[i] != _Right.values[i])
-				return false;
-		}
-		return true;
-	}
+		uid_data data;
+		u32 values[24]; // TODO: Length?
+	};
 };
-typedef _VERTEXSHADERUID<false> VERTEXSHADERUID;
-typedef _VERTEXSHADERUID<true> VERTEXSHADERUIDSAFE;
 
+class ShaderCode
+{
+public:
+	ShaderCode() : buf(NULL), write_ptr(NULL)
+	{
+
+	}
 
-// components is included in the uid.
-char* GenerateVSOutputStruct(char* p, u32 components, API_TYPE api_type);
-const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type);
+	void Write(const char* fmt, ...)
+	{
+		va_list arglist;
+		va_start(arglist, fmt);
+		write_ptr += vsprintf(write_ptr, fmt, arglist);
+		va_end(arglist);
+	}
+
+	const char* GetBuffer() { return buf; }
+	void SetBuffer(char* buffer) { buf = buffer; write_ptr = buffer; }
+	uid_data& GetUidData() { return *(uid_data*)NULL; }
+
+private:
+	const char* buf;
+	char* write_ptr;
+};
 
-void GetVertexShaderId(VERTEXSHADERUID *uid, u32 components);
-void GetSafeVertexShaderId(VERTEXSHADERUIDSAFE *uid, u32 components);
+void GenerateShaderUid(ShaderUid& object, u32 components, API_TYPE api_type);
+void GenerateShaderCode(ShaderCode& object, u32 components, API_TYPE api_type);
 
-// Used to make sure that our optimized vertex shader IDs don't lose any possible shader code changes
-void ValidateVertexShaderIDs(API_TYPE api, VERTEXSHADERUIDSAFE old_id, const std::string& old_code, u32 components);
 
 #endif // GCOGL_VERTEXSHADER_H