Parameters =
{
	StructBuffer boneMatrices : auto("BoneMatrices");
};

Technique 
#ifdef USE_BLEND_SHAPES
	: base("SkinnedMorphVertexInput") =
#else
	: base("SkinnedVertexInput") =
#endif
{
	Language = "HLSL11";

	Pass =
	{
		Common = 
		{
			struct VStoFS
			{
				float4 position : SV_Position;
				float2 uv0 : TEXCOORD0;
				
				float3 tangentToWorldZ : NORMAL; // Note: Half-precision could be used
				float4 tangentToWorldX : TANGENT; // Note: Half-precision could be used
			};
		};

		Vertex =
		{
			StructuredBuffer<float4> boneMatrices;
		
			struct VertexInput
			{
				float3 position : POSITION;
				float3 normal : NORMAL; // Note: Half-precision could be used
				float4 tangent : TANGENT; // Note: Half-precision could be used
				float2 uv0 : TEXCOORD0;
				uint4 blendIndices : BLENDINDICES;
				float4 blendWeights : BLENDWEIGHT;
				
				#ifdef USE_BLEND_SHAPES
					float3 deltaPosition : POSITION1;
					float4 deltaNormal : NORMAL1;
				#endif
			};
			
			struct VertexIntermediate
			{
				float3x4 blendMatrix;
			
				float3 worldNormal; // Note: Half-precision could be used
				float4 worldTangent; // Note: Half-precision could be used
			};
			
			float3x4 getBoneMatrix(uint idx)
			{
				float4 row0 = boneMatrices[idx * 3 + 0];
				float4 row1 = boneMatrices[idx * 3 + 1];
				float4 row2 = boneMatrices[idx * 3 + 2];
				
				return float3x4(row0, row1, row2);
			}
			
			float3x4 getBlendMatrix(VertexInput input)
			{
				float3x4 result = input.blendWeights.x * getBoneMatrix(input.blendIndices.x);
				result += input.blendWeights.y * getBoneMatrix(input.blendIndices.y);
				result += input.blendWeights.z * getBoneMatrix(input.blendIndices.z);
				result += input.blendWeights.w * getBoneMatrix(input.blendIndices.w);
				
				return result;
			}
			
			float3x3 getSkinnedTangentToLocal(VertexInput input, float3x4 blendMatrix, out float tangentSign)
			{
				tangentSign = input.tangent.w * 2.0f - 1.0f;
			
				float3 normal = input.normal * 2.0f - 1.0f;
				float3 tangent = input.tangent.xyz * 2.0f - 1.0f;
				
				#ifdef USE_BLEND_SHAPES
					float3 deltaNormal = (input.deltaNormal.xyz * 2.0f - 1.0f) * 2.0f;
					normal = normalize(normal + deltaNormal * input.deltaNormal.w);
					tangent = normalize(tangent - dot(tangent, normal) * normal);
				#endif
				
				normal = mul(blendMatrix, float4(normal, 0.0f)).xyz;
				tangent = mul(blendMatrix, float4(tangent, 0.0f)).xyz;
				
				float3 bitangent = cross(normal, tangent) * tangentSign;
				tangentSign *= gWorldDeterminantSign;
				
				float3x3 result = float3x3(tangent, bitangent, normal);
				result = transpose(result);
												
				return result;
			}
			
			VertexIntermediate getVertexIntermediate(VertexInput input)
			{
				VertexIntermediate result;
				
				result.blendMatrix = getBlendMatrix(input);
				
				float tangentSign;
				float3x3 tangentToLocal = getSkinnedTangentToLocal(input, result.blendMatrix, tangentSign);
				float3x3 tangentToWorld = mul((float3x3)gMatWorldNoScale, tangentToLocal);
				
				result.worldNormal = float3(tangentToWorld._m02_m12_m22); // Normal basis vector
				result.worldTangent = float4(tangentToWorld._m00_m10_m20, tangentSign); // Tangent basis vector
				
				return result;
			}
			
			float4 getVertexWorldPosition(VertexInput input, VertexIntermediate intermediate)
			{
				#ifdef USE_BLEND_SHAPES
					float4 position = float4(input.position + input.deltaPosition, 1.0f);
				#else
					float4 position = float4(input.position, 1.0f);
				#endif
			
				position = float4(mul(intermediate.blendMatrix, position), 1.0f);
				return mul(gMatWorld, position);
			}
			
			void populateVertexOutput(VertexInput input, VertexIntermediate intermediate, inout VStoFS result)
			{
				result.uv0 = input.uv0;
				
				result.tangentToWorldZ = intermediate.worldNormal;
				result.tangentToWorldX = intermediate.worldTangent;
			}
		};
	};
};

Technique
#ifdef USE_BLEND_SHAPES
	: base("SkinnedMorphVertexInput") =
#else
	: base("SkinnedVertexInput") =
#endif
{
	Language = "GLSL";
	
	Pass =
	{
		Common = 
		{
			varying vec2 uv0;
			varying vec3 tangentToWorldZ;
			varying vec4 tangentToWorldX;	
		};
	
		Vertex =
		{			
			in vec3 bs_position;
			in vec3 bs_normal;
			in vec4 bs_tangent;
			in vec2 bs_texcoord0;
		
			in uvec4 bs_blendindices;
			in vec4 bs_blendweights;
				
			#ifdef USE_BLEND_SHAPES
				in vec3 bs_position1;
				in vec4 bs_normal1;
			#endif
			
			uniform samplerBuffer boneMatrices;
		
			struct VertexIntermediate
			{
				mat4x3 blendMatrix;
			
				vec3 worldNormal;
				vec4 worldTangent;
			};
		
			out gl_PerVertex
			{
				vec4 gl_Position;
			};
			
			void getBoneMatrix(uint idx, out mat4x3 result)
			{
				mat3x4 temp;
			
				temp[0] = texelFetch(boneMatrices, idx * 3 + 0);
				temp[1] = texelFetch(boneMatrices, idx * 3 + 1);
				temp[2] = texelFetch(boneMatrices, idx * 3 + 2);
				
				result = transpose(temp);				
			}
			
			void getBlendMatrix(out mat4x3 result)
			{
				mat4x3 boneMatrix;
				
				getBoneMatrix(bs_blendindices.x, out boneMatrix);
				result = bs_blendweights.x * boneMatrix;
				
				getBoneMatrix(bs_blendindices.y, out boneMatrix);
				result += bs_blendweights.y * boneMatrix;
				
				getBoneMatrix(bs_blendindices.z, out boneMatrix);
				result += bs_blendweights.z * boneMatrix;
				
				getBoneMatrix(bs_blendindices.w, out boneMatrix);
				result += bs_blendweights.w * boneMatrix;
			}
			
			void getSkinnedTangentToLocal(mat4x3 blendMatrix, out float tangentSign, out mat3x3 tangentToLocal)
			{
				tangentSign = bs_tangent.w * 2.0f - 1.0f;
			
				vec3 normal = bs_normal * 2.0f - 1.0f;
				vec3 tangent = bs_tangent.xyz * 2.0f - 1.0f;
				
				#ifdef USE_BLEND_SHAPES
					vec3 deltaNormal = (bs_normal1.xyz * 2.0f - 1.0f) * 2.0f;
					normal = normalize(normal + deltaNormal * bs_normal1.w);
					tangent = normalize(tangent - dot(tangent, normal) * normal);
				#endif
				
				normal = (blendMatrix * vec4(normal, 0.0f)).xyz;
				tangent = (blendMatrix * vec4(tangent, 0.0f)).xyz;
				
				vec3 bitangent = cross(normal, tangent) * tangentSign;
				tangentSign *= gWorldDeterminantSign;
				
				tangentToLocal[0] = tangent.xyz;
				tangentToLocal[1] = bitangent;
				tangentToLocal[2] = normal;
			}			
			
			void getVertexIntermediate(out VertexIntermediate result)
			{
				getBlendMatrix(result.blendMatrix);
				
				float tangentSign;
				mat3 tangentToLocal;
				getSkinnedTangentToLocal(tangentSign, tangentToLocal);
				
				mat3 tangentToWorld = mat3(gMatWorldNoScale) * tangentToLocal;
				result.worldNormal = tangentToWorld[2]; // Normal basis vector
				result.worldTangent = vec4(tangentToWorld[0].xyz, tangentSign); // Tangent basis vector				
			}
			
			void getVertexWorldPosition(VertexIntermediate intermediate, out vec4 result)
			{
				#ifdef USE_BLEND_SHAPES
					vec4 position = vec4(bs_position + bs_position1, 1.0f);
				#else
					vec4 position = vec4(bs_position, 1.0f);
				#endif
			
				position = vec4(intermediate.blendMatrix * position, 1.0f);
				result = gMatWorld * position;
			}
			
			void populateVertexOutput(VertexIntermediate intermediate)
			{
				uv0 = bs_texcoord0;
				
				tangentToWorldZ = intermediate.worldNormal;
				tangentToWorldX = intermediate.worldTangent;
			}
		};
	};
};