diff --git a/src/gl_canvas.c b/src/gl_canvas.c
index 7b0343e..86e47bd 100644
--- a/src/gl_canvas.c
+++ b/src/gl_canvas.c
@@ -44,9 +44,8 @@ typedef struct mg_gl_path_elt
 {
 	vec2 p[4];
 	int pathIndex;
-	int localEltIndex;
 	mg_gl_seg_kind kind;
-	u8 pad[4];
+
 } mg_gl_path_elt;
 
 enum {
@@ -105,7 +104,6 @@ typedef struct mg_gl_encoding_context
 	mg_gl_path* pathBufferData;
 	mg_gl_path_elt* elementBufferData;
 	int pathIndex;
-	int localEltIndex;
 	mg_primitive* primitive;
 	vec4 pathScreenExtents;
 	vec4 pathUserExtents;
@@ -195,8 +193,6 @@ void mg_gl_canvas_encode_element(mg_gl_encoding_context* context, mg_path_elt_ty
 			break;
 	}
 
-	glElt->localEltIndex = context->localEltIndex;
-
 	for(int i=0; i<count; i++)
 	{
 		mg_update_path_extents(&context->pathUserExtents, p[i]);
@@ -998,10 +994,9 @@ void mg_gl_render_batch(mg_gl_canvas_backend* backend,
 	glUseProgram(backend->raster);
 
 	glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 0, backend->pathBuffer);
-	glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 1, backend->segmentCountBuffer);
-	glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 2, backend->segmentBuffer);
-	glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 3, backend->tileOpBuffer);
-	glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 4, backend->screenTilesBuffer);
+	glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 1, backend->segmentBuffer);
+	glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 2, backend->tileOpBuffer);
+	glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 3, backend->screenTilesBuffer);
 
 	glUniform1f(0, scale);
 	glUniform1i(1, backend->msaaCount);
@@ -1126,8 +1121,6 @@ void mg_gl_canvas_render(mg_canvas_backend* interface,
 			    	(eltIndex < primitive->path.count) && (primitive->path.startIndex + eltIndex < eltCount);
 			    	eltIndex++)
 				{
-					context.localEltIndex = segCount;
-
 					mg_path_elt* elt = &pathElements[primitive->path.startIndex + eltIndex];
 
 					if(elt->type != MG_PATH_MOVE)
@@ -1453,12 +1446,12 @@ mg_canvas_backend* gl_canvas_backend_create(mg_wgl_surface* surface)
 
 		glGenBuffers(1, &backend->pathBuffer);
 		glBindBuffer(GL_SHADER_STORAGE_BUFFER, backend->pathBuffer);
-		glBufferData(GL_SHADER_STORAGE_BUFFER, MG_GL_PATH_BUFFER_SIZE, 0, GL_DYNAMIC_COPY);
+		glBufferData(GL_SHADER_STORAGE_BUFFER, MG_GL_PATH_BUFFER_SIZE, 0, GL_STREAM_DRAW);
 
 		//TODO change flags
 		glGenBuffers(1, &backend->elementBuffer);
 		glBindBuffer(GL_SHADER_STORAGE_BUFFER, backend->elementBuffer);
-		glBufferData(GL_SHADER_STORAGE_BUFFER, MG_GL_ELEMENT_BUFFER_SIZE, 0, GL_DYNAMIC_COPY);
+		glBufferData(GL_SHADER_STORAGE_BUFFER, MG_GL_ELEMENT_BUFFER_SIZE, 0, GL_STREAM_DRAW);
 
 		glGenBuffers(1, &backend->segmentBuffer);
 		glBindBuffer(GL_SHADER_STORAGE_BUFFER, backend->segmentBuffer);
diff --git a/src/glsl_shaders/common.glsl b/src/glsl_shaders/common.glsl
index b30adce..e0fcf1e 100644
--- a/src/glsl_shaders/common.glsl
+++ b/src/glsl_shaders/common.glsl
@@ -36,7 +36,6 @@ struct mg_gl_path_elt
 {
 	vec2 p[4];
 	int pathIndex;
-	int localEltIndex;
 	int kind;
 };
 
@@ -44,13 +43,12 @@ struct mg_gl_segment
 {
 	int kind;
 	int pathIndex;
-	int config; //TODO pack these
+	int config;
 	int windingIncrement;
 	vec4 box;
 	mat3 implicitMatrix;
-	float sign;
 	vec2 hullVertex;
-	int debugID;
+	float sign;
 };
 
 struct mg_gl_path_queue
@@ -62,10 +60,9 @@ struct mg_gl_path_queue
 struct mg_gl_tile_op
 {
 	int kind;
-	int index;
 	int next;
-	bool crossRight;
-	int windingOffset;
+	int index;
+	int windingOffsetOrCrossRight;
 };
 
 struct mg_gl_tile_queue
@@ -75,7 +72,6 @@ struct mg_gl_tile_queue
 	int last;
 };
 
-
 float ccw(vec2 a, vec2 b, vec2 c)
 {
 	return((b.x-a.x)*(c.y-a.y) - (b.y-a.y)*(c.x-a.x));
diff --git a/src/glsl_shaders/merge.glsl b/src/glsl_shaders/merge.glsl
index ca0b2e1..68302f4 100644
--- a/src/glsl_shaders/merge.glsl
+++ b/src/glsl_shaders/merge.glsl
@@ -82,7 +82,7 @@ void main()
 					tileOpBuffer.elements[pathOpIndex].kind = MG_GL_OP_START;
 					tileOpBuffer.elements[pathOpIndex].next = -1;
 					tileOpBuffer.elements[pathOpIndex].index = pathIndex;
-					tileOpBuffer.elements[pathOpIndex].windingOffset = windingOffset;
+					tileOpBuffer.elements[pathOpIndex].windingOffsetOrCrossRight = windingOffset;
 
 					vec4 clip = pathBuffer.elements[pathIndex].clip * scale;
 					vec4 tileBox = vec4(tileCoord.x, tileCoord.y, tileCoord.x+1, tileCoord.y+1);
@@ -114,7 +114,7 @@ void main()
 				tileOpBuffer.elements[pathOpIndex].kind = MG_GL_OP_START;
 				tileOpBuffer.elements[pathOpIndex].next = -1;
 				tileOpBuffer.elements[pathOpIndex].index = pathIndex;
-				tileOpBuffer.elements[pathOpIndex].windingOffset = windingOffset;
+				tileOpBuffer.elements[pathOpIndex].windingOffsetOrCrossRight = windingOffset;
 
 				if(lastOpIndex < 0)
 				{
diff --git a/src/glsl_shaders/raster.glsl b/src/glsl_shaders/raster.glsl
index a137db9..7963f05 100644
--- a/src/glsl_shaders/raster.glsl
+++ b/src/glsl_shaders/raster.glsl
@@ -9,22 +9,17 @@ layout(binding = 0) restrict readonly buffer pathBufferSSBO
 	mg_gl_path elements[];
 } pathBuffer;
 
-layout(binding = 1) restrict readonly buffer segmentCountBufferSSBO
-{
-	int elements[];
-} segmentCountBuffer;
-
-layout(binding = 2) restrict readonly buffer segmentBufferSSBO
+layout(binding = 1) restrict readonly buffer segmentBufferSSBO
 {
 	mg_gl_segment elements[];
 } segmentBuffer;
 
-layout(binding = 3) restrict readonly buffer tileOpBufferSSBO
+layout(binding = 2) restrict readonly buffer tileOpBufferSSBO
 {
 	mg_gl_tile_op elements[];
 } tileOpBuffer;
 
-layout(binding = 4) restrict readonly buffer screenTilesBufferSSBO
+layout(binding = 3) restrict readonly buffer screenTilesBufferSSBO
 {
 	int elements[];
 } screenTilesBuffer;
@@ -91,8 +86,17 @@ void main()
 		if(op.kind == MG_GL_OP_START)
 		{
 			vec4 clip = pathBuffer.elements[pathIndex].clip * scale;
-			vec4 pathColor = pathBuffer.elements[pathIndex].color;
-			pathColor.rgb *= pathColor.a;
+			vec4 nextColor = pathBuffer.elements[pathIndex].color;
+			nextColor.rgb *= nextColor.a;
+
+			if(useTexture != 0)
+			{
+				vec3 ph = vec3(centerCoord.xy, 1);
+				vec2 uv = (pathBuffer.elements[pathIndex].uvTransform * ph).xy;
+				vec4 texColor = texture(srcTexture, uv);
+				texColor.rgb *= texColor.a;
+				nextColor *= texColor;
+			}
 
 			for(int sampleIndex = 0; sampleIndex<sampleCount; sampleIndex++)
 			{
@@ -107,19 +111,10 @@ void main()
 					            ||(pathBuffer.elements[pathIndex].cmd == MG_GL_STROKE && (winding[sampleIndex] != 0));
 					if(filled)
 					{
-						vec4 nextColor = pathColor;
-						if(useTexture != 0)
-						{
-							vec3 ph = vec3(sampleCoord.xy, 1);
-							vec2 uv = (pathBuffer.elements[pathIndex].uvTransform * ph).xy;
-							vec4 texColor = texture(srcTexture, uv);
-							texColor.rgb *= texColor.a;
-							nextColor *= texColor;
-						}
 						color[sampleIndex] = color[sampleIndex]*(1-nextColor.a) + nextColor;
 					}
 				}
-				winding[sampleIndex] = op.windingOffset;
+				winding[sampleIndex] = op.windingOffsetOrCrossRight;
 			}
 			pathIndex = op.index;
 		}
@@ -139,7 +134,7 @@ void main()
 					winding[sampleIndex] += seg.windingIncrement;
 				}
 
-				if(op.crossRight)
+				if(op.windingOffsetOrCrossRight != 0)
 				{
 					if( (seg.config == MG_GL_BR || seg.config == MG_GL_TL)
 					  &&(sampleCoord.y > seg.box.w))
@@ -159,8 +154,17 @@ void main()
 	vec4 clip = pathBuffer.elements[pathIndex].clip * scale;
 
 	vec4 pixelColor = vec4(0);
-	vec4 pathColor = pathBuffer.elements[pathIndex].color;
-	pathColor.rgb *= pathColor.a;
+	vec4 nextColor = pathBuffer.elements[pathIndex].color;
+	nextColor.rgb *= nextColor.a;
+
+	if(useTexture != 0)
+	{
+		vec3 ph = vec3(centerCoord.xy, 1);
+		vec2 uv = (pathBuffer.elements[pathIndex].uvTransform * ph).xy;
+		vec4 texColor = texture(srcTexture, uv);
+		texColor.rgb *= texColor.a;
+		nextColor *= texColor;
+	}
 
 	for(int sampleIndex=0; sampleIndex<sampleCount; sampleIndex++)
 	{
@@ -175,15 +179,6 @@ void main()
 			            ||(pathBuffer.elements[pathIndex].cmd == MG_GL_STROKE && (winding[sampleIndex] != 0));
 			if(filled)
 			{
-				vec4 nextColor = pathColor;
-				if(useTexture != 0)
-				{
-					vec3 ph = vec3(sampleCoord.xy, 1);
-					vec2 uv = (pathBuffer.elements[pathIndex].uvTransform * ph).xy;
-					vec4 texColor = texture(srcTexture, uv);
-					texColor.rgb *= texColor.a;
-					nextColor *= texColor;
-				}
 				color[sampleIndex] = color[sampleIndex]*(1-nextColor.a) + nextColor;
 			}
 		}
diff --git a/src/glsl_shaders/segment_setup.glsl b/src/glsl_shaders/segment_setup.glsl
index 9fb505f..903eaf9 100644
--- a/src/glsl_shaders/segment_setup.glsl
+++ b/src/glsl_shaders/segment_setup.glsl
@@ -106,7 +106,7 @@ void bin_to_tiles(int segIndex)
 
 				tileOpBuffer.elements[tileOpIndex].kind = MG_GL_OP_SEGMENT;
 				tileOpBuffer.elements[tileOpIndex].index = segIndex;
-				tileOpBuffer.elements[tileOpIndex].crossRight = false;
+				tileOpBuffer.elements[tileOpIndex].windingOffsetOrCrossRight = 0;
 				tileOpBuffer.elements[tileOpIndex].next = -1;
 
 				int tileQueueIndex = pathQueue.tileQueues + y*pathArea.z + x;
@@ -126,7 +126,7 @@ void bin_to_tiles(int segIndex)
 				//NOTE: if the segment crosses the right boundary, mark it.
 				if(crossR)
 				{
-					tileOpBuffer.elements[tileOpIndex].crossRight = true;
+					tileOpBuffer.elements[tileOpIndex].windingOffsetOrCrossRight = 1;
 				}
 			}
 		}