diff --git a/Common/GPU/Vulkan/VulkanDebug.cpp b/Common/GPU/Vulkan/VulkanDebug.cpp
index b52e4396cb..83b3d974c0 100644
--- a/Common/GPU/Vulkan/VulkanDebug.cpp
+++ b/Common/GPU/Vulkan/VulkanDebug.cpp
@@ -93,4 +93,3 @@ VKAPI_ATTR VkBool32 VKAPI_CALL VulkanDebugUtilsCallback(
 	// keep that behavior here.
 	return false;
 }
-
diff --git a/GPU/Common/FragmentShaderGenerator.cpp b/GPU/Common/FragmentShaderGenerator.cpp
index e2088731af..04d8aa93ed 100644
--- a/GPU/Common/FragmentShaderGenerator.cpp
+++ b/GPU/Common/FragmentShaderGenerator.cpp
@@ -174,9 +174,7 @@ bool GenerateFragmentShader(const FShaderID &id, char *buffer, const ShaderLangu
 		WRITE(p, "layout (location = 1) %s in lowp vec4 v_color0;\n", shading);
 		if (lmode)
 			WRITE(p, "layout (location = 2) %s in lowp vec3 v_color1;\n", shading);
-		if (enableFog) {
-			WRITE(p, "layout (location = 3) in highp float v_fogdepth;\n");
-		}
+		WRITE(p, "layout (location = 3) in highp float v_fogdepth;\n");
 		if (doTexture) {
 			WRITE(p, "layout (location = 0) in highp vec3 v_texcoord;\n");
 		}
@@ -279,9 +277,7 @@ bool GenerateFragmentShader(const FShaderID &id, char *buffer, const ShaderLangu
 		if (lmode) {
 			WRITE(p, "  vec3 v_color1: COLOR1;\n");
 		}
-		if (enableFog) {
-			WRITE(p, "  float v_fogdepth: TEXCOORD1;\n");
-		}
+		WRITE(p, "  float v_fogdepth: TEXCOORD1;\n");
 		if (needFragCoord) {
 			if (compat.shaderLanguage == HLSL_D3D11) {
 				WRITE(p, "  vec4 pixelPos : SV_POSITION;\n");
@@ -394,8 +390,8 @@ bool GenerateFragmentShader(const FShaderID &id, char *buffer, const ShaderLangu
 		if (enableFog) {
 			*uniformMask |= DIRTY_FOGCOLOR;
 			WRITE(p, "uniform vec3 u_fogcolor;\n");
-			WRITE(p, "%s %s float v_fogdepth;\n", compat.varying_fs, highpFog ? "highp" : "mediump");
 		}
+		WRITE(p, "%s %s float v_fogdepth;\n", compat.varying_fs, highpFog ? "highp" : "mediump");
 		if (doTexture) {
 			WRITE(p, "%s %s vec3 v_texcoord;\n", compat.varying_fs, highpTexcoord ? "highp" : "mediump");
 		}
diff --git a/GPU/Common/ShaderId.cpp b/GPU/Common/ShaderId.cpp
index 5ed727a638..bb8dd54237 100644
--- a/GPU/Common/ShaderId.cpp
+++ b/GPU/Common/ShaderId.cpp
@@ -123,11 +123,11 @@ void ComputeVertexShaderID(VShaderID *id_out, u32 vertType, bool useHWTransform,
 
 		if (gstate.isLightingEnabled()) {
 			// doShadeMapping is stored as UVGenMode, and light type doesn't matter for shade mapping.
-			id.SetBits(VS_BIT_MATERIAL_UPDATE, 3, gstate.getMaterialUpdate());
 			id.SetBit(VS_BIT_LIGHTING_ENABLE);
 			if (gstate_c.Supports(GPU_USE_LIGHT_UBERSHADER)) {
 				id.SetBit(VS_BIT_LIGHT_UBERSHADER);
 			} else {
+				id.SetBits(VS_BIT_MATERIAL_UPDATE, 3, gstate.getMaterialUpdate());
 				// Light bits
 				for (int i = 0; i < 4; i++) {
 					bool chanEnabled = gstate.isLightChanEnabled(i) != 0;
@@ -248,6 +248,9 @@ std::string FragmentShaderDesc(const FShaderID &id) {
 	else if (id.Bit(FS_BIT_ALPHA_TEST)) desc << "AlphaTest " << alphaTestFuncs[id.Bits(FS_BIT_ALPHA_TEST_FUNC, 3)] << " ";
 	if (id.Bit(FS_BIT_COLOR_AGAINST_ZERO)) desc << "ColorTest0 " << alphaTestFuncs[id.Bits(FS_BIT_COLOR_TEST_FUNC, 2)] << " ";  // first 4 match;
 	else if (id.Bit(FS_BIT_COLOR_TEST)) desc << "ColorTest " << alphaTestFuncs[id.Bits(FS_BIT_COLOR_TEST_FUNC, 2)] << " ";  // first 4 match
+	if (id.Bit(FS_BIT_TEST_DISCARD_TO_ZERO)) desc << "TestDiscardToZero ";
+	if (id.Bit(FS_BIT_NO_DEPTH_CANNOT_DISCARD_STENCIL)) desc << "StencilDiscardWorkaround ";
+	if (id.Bits(FS_BIT_REPLACE_LOGIC_OP, 4) != GE_LOGIC_COPY) desc << "ReplaceLogic ";
 
 	return desc.str();
 }
diff --git a/GPU/Common/ShaderUniforms.cpp b/GPU/Common/ShaderUniforms.cpp
index 8bf1f69356..f5cc8ca6c9 100644
--- a/GPU/Common/ShaderUniforms.cpp
+++ b/GPU/Common/ShaderUniforms.cpp
@@ -271,6 +271,8 @@ uint32_t PackLightControlBits() {
 	// Bit organization
 	// Bottom 4 bits are enable bits for each light.
 	// Then, for each light, comes 2 bits for "comp" and 2 bits for "type".
+	// At the end, at bit 20, we put the three material update bits.
+
 	uint32_t lightControl = 0;
 	for (int i = 0; i < 4; i++) {
 		if (gstate.isLightChanEnabled(i)) {
@@ -282,6 +284,9 @@ uint32_t PackLightControlBits() {
 		lightControl |= computation << (4 + i * 4);
 		lightControl |= type << (4 + i * 4 + 2);
 	}
+
+	lightControl |= gstate.getMaterialUpdate() << 20;
+
 	return lightControl;
 }
 
diff --git a/GPU/Common/VertexShaderGenerator.cpp b/GPU/Common/VertexShaderGenerator.cpp
index 8de59ab191..4b0a16977f 100644
--- a/GPU/Common/VertexShaderGenerator.cpp
+++ b/GPU/Common/VertexShaderGenerator.cpp
@@ -552,7 +552,7 @@ bool GenerateVertexShader(const VShaderID &id, char *buffer, const ShaderLanguag
 			if (enableLighting) {
 				WRITE(p, "uniform lowp vec4 u_ambient;\n");
 				*uniformMask |= DIRTY_AMBIENT;
-				if ((matUpdate & 2) == 0 || !hasColor) {
+				if (lightUberShader || (matUpdate & 2) == 0 || !hasColor) {
 					WRITE(p, "uniform lowp vec3 u_matdiffuse;\n");
 					*uniformMask |= DIRTY_MATDIFFUSE;
 				}
@@ -949,14 +949,25 @@ bool GenerateVertexShader(const VShaderID &id, char *buffer, const ShaderLanguag
 
 		// TODO: Declare variables for dots for shade mapping if needed.
 
-		const char *ambientStr = (matUpdate & 1) && hasColor ? "color0" : "u_matambientalpha";
-		const char *diffuseStr = (matUpdate & 2) && hasColor ? "color0.rgb" : "u_matdiffuse";
-		const char *specularStr = (matUpdate & 4) && hasColor ? "color0.rgb" : "u_matspecular.rgb";
+		const char *srcCol = "color0";
 		if (doBezier || doSpline) {
 			// TODO: Probably, should use hasColorTess but FF4 has a problem with drawing the background.
-			ambientStr = (matUpdate & 1) && hasColor ? "tess.col" : "u_matambientalpha";
-			diffuseStr = (matUpdate & 2) && hasColor ? "tess.col.rgb" : "u_matdiffuse";
-			specularStr = (matUpdate & 4) && hasColor ? "tess.col.rgb" : "u_matspecular.rgb";
+			srcCol = "tess.col";
+		}
+
+		if (lightUberShader && hasColor) {
+			p.F("  vec4 ambientColor = ((u_lightControl & (1u << 20u)) != 0u) ? %s : u_matambientalpha;\n", srcCol);
+			if (enableLighting) {
+				p.F("  vec3 diffuseColor = ((u_lightControl & (1u << 21u)) != 0u) ? %s.rgb : u_matdiffuse;\n", srcCol);
+				p.F("  vec3 specularColor = ((u_lightControl & (1u << 22u)) != 0u) ? %s.rgb : u_matspecular.rgb;\n", srcCol);
+			}
+		} else {
+			// This path also takes care of the lightUberShader && !hasColor path, because all comparisons fail.
+			p.F("  vec4 ambientColor = %s;\n", (matUpdate & 1) && hasColor ? srcCol : "u_matambientalpha");
+			if (enableLighting) {
+				p.F("  vec3 diffuseColor = %s.rgb;\n", (matUpdate & 2) && hasColor ? srcCol : "u_matdiffuse");
+				p.F("  vec3 specularColor = %s.rgb;\n", (matUpdate & 4) && hasColor ? srcCol : "u_matspecular");
+			}
 		}
 
 		bool diffuseIsZero = true;
@@ -964,7 +975,7 @@ bool GenerateVertexShader(const VShaderID &id, char *buffer, const ShaderLanguag
 		bool distanceNeeded = false;
 		bool anySpots = false;
 		if (enableLighting) {
-			WRITE(p, "  lowp vec4 lightSum0 = u_ambient * %s + vec4(u_matemissive, 0.0);\n", ambientStr);
+			p.C("  lowp vec4 lightSum0 = u_ambient * ambientColor + vec4(u_matemissive, 0.0);\n");
 
 			for (int i = 0; i < 4; i++) {
 				GELightType type = static_cast<GELightType>(id.Bits(VS_BIT_LIGHT0_TYPE + 4*i, 2));
@@ -1043,7 +1054,7 @@ bool GenerateVertexShader(const VShaderID &id, char *buffer, const ShaderLanguag
 				p.C("      lightScale = 1.0;\n");
 				p.C("      break;\n");
 				p.C("    }\n");
-				p.F("    diffuse = (u_lightdiffuse%i * %s) * max(ldot, 0.0);\n", i, diffuseStr);
+				p.F("    diffuse = (u_lightdiffuse%i * diffuseColor) * max(ldot, 0.0);\n", i);
 				p.C("    if (comp == 1u) {\n");  // do specular
 				p.C("      if (ldot >= 0.0) {\n");
 				p.C("        ldot = dot(normalize(toLight + vec3(0.0, 0.0, 1.0)), worldnormal);\n");
@@ -1053,10 +1064,10 @@ bool GenerateVertexShader(const VShaderID &id, char *buffer, const ShaderLanguag
 				p.C("          ldot = pow(max(ldot, 0.0), u_matspecular.a);\n");
 				p.C("        }\n");
 				p.C("        if (ldot > 0.0)\n");
-				p.F("          lightSum1 += u_lightspecular%i * %s * ldot * lightScale;\n", i, specularStr);
+				p.F("          lightSum1 += u_lightspecular%i * specularColor * ldot * lightScale;\n", i);
 				p.C("      }\n");
 				p.C("    }\n");
-				p.F("    lightSum0.rgb += (u_lightambient%i * %s.rgb + diffuse) * lightScale;\n", i, ambientStr);
+				p.F("    lightSum0.rgb += (u_lightambient%i * ambientColor.rgb + diffuse) * lightScale;\n", i);
 				p.C("  }\n");
 			}
 		} else {
@@ -1116,7 +1127,7 @@ bool GenerateVertexShader(const VShaderID &id, char *buffer, const ShaderLanguag
 					break;
 				}
 
-				p.F("  diffuse = (u_lightdiffuse%i * %s) * max(ldot, 0.0);\n", i, diffuseStr);
+				p.F("  diffuse = (u_lightdiffuse%i * diffuseColor) * max(ldot, 0.0);\n", i);
 				if (doSpecular) {
 					p.C("  if (ldot >= 0.0) {\n");
 					p.C("    ldot = dot(normalize(toLight + vec3(0.0, 0.0, 1.0)), worldnormal);\n");
@@ -1126,10 +1137,10 @@ bool GenerateVertexShader(const VShaderID &id, char *buffer, const ShaderLanguag
 					p.C("      ldot = pow(max(ldot, 0.0), u_matspecular.a);\n");
 					p.C("    }\n");
 					p.C("    if (ldot > 0.0)\n");
-					p.F("      lightSum1 += u_lightspecular%i * %s * ldot %s;\n", i, specularStr, timesLightScale);
+					p.F("      lightSum1 += u_lightspecular%i * specularColor * ldot %s;\n", i, timesLightScale);
 					p.C("  }\n");
 				}
-				p.F("  lightSum0.rgb += (u_lightambient%i * %s.rgb + diffuse)%s;\n", i, ambientStr, timesLightScale);
+				p.F("  lightSum0.rgb += (u_lightambient%i * ambientColor.rgb + diffuse)%s;\n", i, timesLightScale);
 			}
 		}
 
diff --git a/GPU/GPUCommon.cpp b/GPU/GPUCommon.cpp
index 00ac51f9c1..56dbbbcf59 100644
--- a/GPU/GPUCommon.cpp
+++ b/GPU/GPUCommon.cpp
@@ -111,7 +111,7 @@ const CommonCommandTableEntry commonCommandTable[] = {
 	{ GE_CMD_LIGHTTYPE1, FLAG_FLUSHBEFOREONCHANGE, DIRTY_VERTEXSHADER_STATE | DIRTY_LIGHT1 },
 	{ GE_CMD_LIGHTTYPE2, FLAG_FLUSHBEFOREONCHANGE, DIRTY_VERTEXSHADER_STATE | DIRTY_LIGHT2 },
 	{ GE_CMD_LIGHTTYPE3, FLAG_FLUSHBEFOREONCHANGE, DIRTY_VERTEXSHADER_STATE | DIRTY_LIGHT3 },
-	{ GE_CMD_MATERIALUPDATE, FLAG_FLUSHBEFOREONCHANGE, DIRTY_VERTEXSHADER_STATE | DIRTY_FRAGMENTSHADER_STATE },  // TODO: This should not need to dirty the fragment shader?
+	{ GE_CMD_MATERIALUPDATE, FLAG_FLUSHBEFOREONCHANGE, DIRTY_VERTEXSHADER_STATE },
 
 	// These change both shaders so need flushing.
 	{ GE_CMD_LIGHTMODE, FLAG_FLUSHBEFOREONCHANGE, DIRTY_VERTEXSHADER_STATE | DIRTY_FRAGMENTSHADER_STATE },
@@ -465,6 +465,14 @@ void GPUCommon::UpdateCmdInfo() {
 			cmdInfo_[GE_CMD_LIGHTTYPE0 + i].AddDirty(DIRTY_VERTEXSHADER_STATE);
 		}
 	}
+
+	if (gstate_c.Supports(GPU_USE_LIGHT_UBERSHADER)) {
+		cmdInfo_[GE_CMD_MATERIALUPDATE].RemoveDirty(DIRTY_VERTEXSHADER_STATE);
+		cmdInfo_[GE_CMD_MATERIALUPDATE].AddDirty(DIRTY_LIGHT_CONTROL);
+	} else {
+		cmdInfo_[GE_CMD_MATERIALUPDATE].RemoveDirty(DIRTY_LIGHT_CONTROL);
+		cmdInfo_[GE_CMD_MATERIALUPDATE].AddDirty(DIRTY_VERTEXSHADER_STATE);
+	}
 }
 
 void GPUCommon::BeginHostFrame() {
diff --git a/GPU/Vulkan/ShaderManagerVulkan.cpp b/GPU/Vulkan/ShaderManagerVulkan.cpp
index acdeba7b4a..6c39280a82 100644
--- a/GPU/Vulkan/ShaderManagerVulkan.cpp
+++ b/GPU/Vulkan/ShaderManagerVulkan.cpp
@@ -376,7 +376,7 @@ VulkanFragmentShader *ShaderManagerVulkan::GetFragmentShaderFromModule(VkShaderM
 // instantaneous.
 
 #define CACHE_HEADER_MAGIC 0xff51f420 
-#define CACHE_VERSION 27
+#define CACHE_VERSION 28
 struct VulkanCacheHeader {
 	uint32_t magic;
 	uint32_t version;
diff --git a/Windows/GEDebugger/TabState.cpp b/Windows/GEDebugger/TabState.cpp
index 63a7199ed6..0a50b4fd19 100644
--- a/Windows/GEDebugger/TabState.cpp
+++ b/Windows/GEDebugger/TabState.cpp
@@ -447,7 +447,7 @@ void FormatStateRow(wchar_t *dest, const TabStateRow &info, u32 value, bool enab
 				swprintf(dest, 255, L"%S", texformats[value]);
 			}
 			else if ((value & 0xF) < (u32)ARRAY_SIZE(texformats)) {
-				swprintf(dest, 255, L"%S (extra bits %06x)", texformats[value & 0xF], value);
+				swprintf(dest, 255, L"%S (extra bits %06x)", texformats[value & 0xF], value & ~0xF);
 			}
 			else {
 				swprintf(dest, 255, L"%06x", value);