diff --git a/GPU/GLES/ShaderManager.cpp b/GPU/GLES/ShaderManager.cpp
index 1f23d2d776..6179e50262 100644
--- a/GPU/GLES/ShaderManager.cpp
+++ b/GPU/GLES/ShaderManager.cpp
@@ -270,6 +270,11 @@ void LinkedShader::updateUniforms() {
 			uvscaleoff[1] /= gstate_c.curTextureHeight;
 			uvscaleoff[2] /= gstate_c.curTextureWidth;
 			uvscaleoff[3] /= gstate_c.curTextureHeight;
+		} else {
+			uvscaleoff[0] *= 2.0f;
+			uvscaleoff[1] *= 2.0f;
+			uvscaleoff[2] *= 2.0f;
+			uvscaleoff[3] *= 2.0f;
 		}
 		glUniform4fv(u_uvscaleoffset, 1, uvscaleoff);
 	}
diff --git a/GPU/GLES/TransformPipeline.cpp b/GPU/GLES/TransformPipeline.cpp
index e57712e287..c922de8fa8 100644
--- a/GPU/GLES/TransformPipeline.cpp
+++ b/GPU/GLES/TransformPipeline.cpp
@@ -319,6 +319,7 @@ static const GlTypeInfo GLComp[] = {
 	{GL_UNSIGNED_SHORT, 2, GL_TRUE},// 	DEC_U16_2,
 	{GL_UNSIGNED_SHORT, 3, GL_TRUE},// 	DEC_U16_3,
 	{GL_UNSIGNED_SHORT, 4, GL_TRUE},// 	DEC_U16_4,
+	{GL_UNSIGNED_BYTE, 2, GL_FALSE},// 	DEC_U8A_2,
 	{GL_UNSIGNED_SHORT, 2, GL_FALSE},// 	DEC_U16A_2,
 };
 
diff --git a/GPU/GLES/VertexDecoder.cpp b/GPU/GLES/VertexDecoder.cpp
index ee6739714b..8bfc77a451 100644
--- a/GPU/GLES/VertexDecoder.cpp
+++ b/GPU/GLES/VertexDecoder.cpp
@@ -77,6 +77,7 @@ int DecFmtSize(u8 fmt) {
 	case DEC_U16_2: return 4;
 	case DEC_U16_3: return 8;
 	case DEC_U16_4: return 8;
+	case DEC_U8A_2: return 4;
 	case DEC_U16A_2: return 4;
 	default:
 		return 0;
@@ -143,18 +144,18 @@ void VertexDecoder::Step_WeightsFloat() const
 
 void VertexDecoder::Step_TcU8() const
 {
-	float *uv = (float *)(decoded_ + decFmt.uvoff);
+	u8 *uv = (u8*)(decoded_ + decFmt.uvoff);
 	const u8 *uvdata = (const u8*)(ptr_ + tcoff);
-	for (int j = 0; j < 2; j++)
-		uv[j] = (float)uvdata[j] / 128.0f;
+	uv[0] = uvdata[0];
+	uv[1] = uvdata[1];
 }
 
 void VertexDecoder::Step_TcU16() const
 {
-	float *uv = (float *)(decoded_ + decFmt.uvoff);
+	u16 *uv = (u16 *)(decoded_ + decFmt.uvoff);
 	const u16 *uvdata = (const u16*)(ptr_ + tcoff);
-	uv[0] = (float)uvdata[0] / 32768.0f;
-	uv[1] = (float)uvdata[1] / 32768.0f;
+	uv[0] = uvdata[0];
+	uv[1] = uvdata[1];
 }
 
 void VertexDecoder::Step_TcU16Through() const
@@ -169,15 +170,16 @@ void VertexDecoder::Step_TcFloat() const
 {
 	float *uv = (float *)(decoded_ + decFmt.uvoff);
 	const float *uvdata = (const float*)(ptr_ + tcoff);
-	memcpy(uv, uvdata, sizeof(float) * 2);
+	uv[0] = uvdata[0] * 0.5f;
+	uv[1] = uvdata[1] * 0.5f;
 }
 
 void VertexDecoder::Step_TcFloatThrough() const
 {
 	float *uv = (float *)(decoded_ + decFmt.uvoff);
 	const float *uvdata = (const float*)(ptr_ + tcoff);
-	uv[0] = uvdata[0];
-	uv[1] = uvdata[1];
+	uv[0] = uvdata[0] * 0.5f;
+	uv[1] = uvdata[1] * 0.5f;
 }
 
 void VertexDecoder::Step_Color565() const
@@ -588,11 +590,18 @@ void VertexDecoder::SetVertexType(u32 fmt) {
 
 		steps_[numSteps_++] = throughmode ? tcstep_through[tc] : tcstep[tc];
 
-		// All UV except through mode decode to DEC_FLOAT2 currently.
-		if (throughmode && (tc == (GE_VTYPE_TC_16BIT >> GE_VTYPE_TC_SHIFT)))
-			decFmt.uvfmt = DEC_U16A_2;
-		else
+		switch (tc) {
+		case GE_VTYPE_TC_8BIT >> GE_VTYPE_TC_SHIFT:
+			decFmt.uvfmt = throughmode ? DEC_U8A_2 : DEC_U8_2;
+			break;
+		case GE_VTYPE_TC_16BIT >> GE_VTYPE_TC_SHIFT:
+			decFmt.uvfmt = throughmode ? DEC_U16A_2 : DEC_U16_2;
+			break;
+		case GE_VTYPE_TC_FLOAT >> GE_VTYPE_TC_SHIFT:
 			decFmt.uvfmt = DEC_FLOAT_2;
+			break;
+		}
+
 		decFmt.uvoff = decOff;
 		decOff += DecFmtSize(decFmt.uvfmt);
 	}
diff --git a/GPU/GLES/VertexDecoder.h b/GPU/GLES/VertexDecoder.h
index a643370e63..09cbaafba4 100644
--- a/GPU/GLES/VertexDecoder.h
+++ b/GPU/GLES/VertexDecoder.h
@@ -42,6 +42,7 @@ enum {
 	DEC_U16_2,
 	DEC_U16_3,
 	DEC_U16_4,
+	DEC_U8A_2,
 	DEC_U16A_2,
 };
 
@@ -196,14 +197,14 @@ public:
 			{
 				const s16 *p = (s16 *)(data_ + decFmt_.posoff);
 				for (int i = 0; i < 3; i++)
-					pos[i] = p[i] / 32767.0f;
+					pos[i] = p[i] * (1.f / 32768.f);
 			}
 			break;
 		case DEC_S8_3:
 			{
 				const s8 *p = (s8 *)(data_ + decFmt_.posoff);
 				for (int i = 0; i < 3; i++)
-					pos[i] = p[i] / 127.0f;
+					pos[i] = p[i] * (1.f / 128.f);
 			}
 			break;
 		default:
@@ -221,14 +222,14 @@ public:
 			{
 				const s16 *p = (s16 *)(data_ + decFmt_.nrmoff);
 				for (int i = 0; i < 3; i++)
-					nrm[i] = p[i] / 32767.0f;
+					nrm[i] = p[i] * (1.f / 32768.f);
 			}
 			break;
 		case DEC_S8_3:
 			{
 				const s8 *p = (s8 *)(data_ + decFmt_.nrmoff);
 				for (int i = 0; i < 3; i++)
-					nrm[i] = p[i] / 127.0f;
+					nrm[i] = p[i] * (1.f / 128.0f);
 			}
 			break;
 		default:
@@ -238,9 +239,25 @@ public:
 	}
 
 	void ReadUV(float uv[2]) {
+		const u8 *b = (const u8 *)(data_ + decFmt_.uvoff);
+		const u16 *s = (const u16 *)(data_ + decFmt_.uvoff);
+		const float *f = (const float *)(data_ + decFmt_.uvoff);
 		switch (decFmt_.uvfmt) {
+		case DEC_U8_2:
+			uv[0] = b[0] * (1.f / 128.f);
+			uv[1] = b[1] * (1.f / 128.f);
+			break;
+
+		case DEC_U16_2:
+			uv[0] = s[0] * (1.f / 32768.f);
+			uv[1] = s[1] * (1.f / 32768.f);
+			break;
+
 		case DEC_FLOAT_2:
-			memcpy(uv, data_ + decFmt_.uvoff, 8); break;
+			uv[0] = f[0] * 2.0f;
+			uv[1] = f[1] * 2.0f;
+			break;
+
 		case DEC_U16A_2:
 			{
 				const u16 *p = (const u16 *)(data_ + decFmt_.uvoff);
@@ -260,7 +277,7 @@ public:
 			{
 				const u8 *p = (const u8 *)(data_ + decFmt_.c0off);
 				for (int i = 0; i < 4; i++)
-					color[i] = p[i] / 255.0f;
+					color[i] = p[i] * (1.f / 255.f);
 			}
 			break;
 		case DEC_FLOAT_4:
@@ -277,7 +294,7 @@ public:
 			{
 				const u8 *p = (const u8 *)(data_ + decFmt_.c1off);
 				for (int i = 0; i < 3; i++)
-					color[i] = p[i] / 255.0f;
+					color[i] = p[i] * (1.f / 255.f);
 			}
 			break;
 		case DEC_FLOAT_4:
@@ -300,14 +317,14 @@ public:
 			for (int i = 0; i <= decFmt_.w0fmt - DEC_FLOAT_1; i++)
 				weights[i] = f[i] * 2.0;
 			break;
-		case DEC_U8_1: weights[0] = p[0] / 128.f; break;
-		case DEC_U8_2: for (int i = 0; i < 2; i++) weights[i] = p[i] / 128.f; break;
-		case DEC_U8_3: for (int i = 0; i < 3; i++) weights[i] = p[i] / 128.f; break;
-		case DEC_U8_4: for (int i = 0; i < 4; i++) weights[i] = p[i] / 128.f; break;
-		case DEC_U16_1: weights[0] = s[0] / 32768.f; break;
-		case DEC_U16_2: for (int i = 0; i < 2; i++) weights[i] = s[i] / 32768.f; break;
-		case DEC_U16_3: for (int i = 0; i < 3; i++) weights[i] = s[i] / 32768.f; break;
-		case DEC_U16_4: for (int i = 0; i < 4; i++) weights[i] = s[i] / 32768.f; break;
+		case DEC_U8_1: weights[0] = p[0] * (1.f / 128.f); break;
+		case DEC_U8_2: for (int i = 0; i < 2; i++) weights[i] = p[i] * (1.f / 128.f); break;
+		case DEC_U8_3: for (int i = 0; i < 3; i++) weights[i] = p[i] * (1.f / 128.f); break;
+		case DEC_U8_4: for (int i = 0; i < 4; i++) weights[i] = p[i] * (1.f / 128.f); break;
+		case DEC_U16_1: weights[0] = s[0] * (1.f / 32768.f); break;
+		case DEC_U16_2: for (int i = 0; i < 2; i++) weights[i] = s[i] * (1.f / 32768.f); break;
+		case DEC_U16_3: for (int i = 0; i < 3; i++) weights[i] = s[i] * (1.f / 32768.f); break;
+		case DEC_U16_4: for (int i = 0; i < 4; i++) weights[i] = s[i] * (1.f / 32768.f); break;
 		default:
 			ERROR_LOG(G3D, "Reader: Unsupported W0 Format");
 			break;
@@ -327,10 +344,10 @@ public:
 			for (int i = 0; i <= decFmt_.w1fmt - DEC_FLOAT_1; i++)
 				weights[i+4] = f[i] * 2.0;
 			break;
-		case DEC_U16_1: weights[4] = s[0] / 32768.f; break;
-		case DEC_U16_2: for (int i = 0; i < 2; i++) weights[i+4] = s[i] / 32768.f; break;
-		case DEC_U16_3: for (int i = 0; i < 3; i++) weights[i+4] = s[i] / 32768.f; break;
-		case DEC_U16_4: for (int i = 0; i < 4; i++) weights[i+4] = s[i] / 32768.f; break;
+		case DEC_U16_1: weights[4] = s[0] * (1.f / 32768.f); break;
+		case DEC_U16_2: for (int i = 0; i < 2; i++) weights[i+4] = s[i] * (1.f / 32768.f); break;
+		case DEC_U16_3: for (int i = 0; i < 3; i++) weights[i+4] = s[i] * (1.f / 32768.f); break;
+		case DEC_U16_4: for (int i = 0; i < 4; i++) weights[i+4] = s[i]  * (1.f / 32768.f); break;
 		default:
 			ERROR_LOG(G3D, "Reader: Unsupported W1 Format");
 			break;