UnrealEngineUWP/Engine/Shaders/Common.usf

// Copyright 1998-2017 Epic Games, Inc. All Rights Reserved.

/*=============================================================================
	Common.usf: Common shader code
=============================================================================*/
  
#pragma once


#if PS4_PROFILE  
	// always #include PS4Common.usf so it can #define override anything in any other included file.
	#include "PS4/PS4Common.usf"
#endif
 
#if XBOXONE_PROFILE
	#include "XboxOne/XboxOneCommon.usf"
#endif

#if COMPILER_METAL
	// Helps with iteration when changing Metal shader code generation backend.
	#include "MetalCommon.usf"
#endif

#if VULKAN_PROFILE
	// Helps with iteration when changing Vulkan shader code generation backend.
	#include "VulkanCommon.usf"
#endif

#ifndef FORCE_FLOATS
#define FORCE_FLOATS 0
#endif

#if (!(COMPILER_GLSL_ES2 || COMPILER_GLSL_ES3_1) && !METAL_PROFILE) || FORCE_FLOATS
	// Always use floats when not using the ES2 compiler, because low precision modifiers are currently only tweaked for ES2, 
	// And we don't want potential side effects on other platforms
	#define half float
	#define half1 float1
	#define half2 float2
	#define half3 float3
	#define half4 float4
	#define half3x3 float3x3
	#define half4x4 float4x4 
	#define half4x3 float4x3 
	#define fixed float
	#define fixed1 float1
	#define fixed2 float2
	#define fixed3 float3
	#define fixed4 float4
	#define fixed3x3 float3x3
	#define fixed4x4 float4x4
	#define fixed4x3 float4x3
#endif

// These types are used for material translator generated code, or any functions the translated code can call
#if PIXELSHADER
	#define MaterialFloat half
	#define MaterialFloat2 half2
	#define MaterialFloat3 half3
	#define MaterialFloat4 half4
	#define MaterialFloat3x3 half3x3
	#define MaterialFloat4x4 half4x4 
	#define MaterialFloat4x3 half4x3 
#else
	// Material translated vertex shader code always uses floats, 
	// Because it's used for things like world position and UVs
	#define MaterialFloat float
	#define MaterialFloat2 float2
	#define MaterialFloat3 float3
	#define MaterialFloat4 float4
	#define MaterialFloat3x3 float3x3
	#define MaterialFloat4x4 float4x4 
	#define MaterialFloat4x3 float4x3 
#endif

#if COMPILER_GLSL_ES2 || COMPILER_METAL || COMPILER_SWITCH

	float determinant(float3x3 M)
	{    
		return	
			M[0][0] * (M[1][1] * M[2][2] - M[1][2] * M[2][1]) -
			M[1][0] * (M[0][1] * M[2][2] - M[0][2] * M[2][1]) +
			M[2][0] * (M[0][1] * M[1][2] - M[0][2] * M[1][1]);
	}

#endif   

#if POST_PROCESS_ALPHA
#define SceneColorLayout float4
#define CastFloat4ToSceneColorLayout(x) (x)
#define SetSceneColorLayoutToFloat4(dest,value) dest = (value)
#else
#define SceneColorLayout float3
#define CastFloat4ToSceneColorLayout(x) ((x).rgb)
#define SetSceneColorLayoutToFloat4(dest,value) dest.rgb = (value)
#endif

// Generated file that contains uniform buffer declarations needed by the shader being compiled 
#include "GeneratedUniformBuffers.usf" 

// to support the console command "r.InvalidateShaderCache"
#include "ShaderVersion.usf"
	
// uniform buffers specifics
#include "CommonViewUniformBuffer.usf"

#include "InstancedStereo.usf"

#include "Definitions.usf"

#define FEATURE_LEVEL_ES2	1
#define FEATURE_LEVEL_ES3_1	2
#define FEATURE_LEVEL_SM3	3
#define FEATURE_LEVEL_SM4	4
#define FEATURE_LEVEL_SM5	5
#define FEATURE_LEVEL_MAX	6

// A8 textures when sampled either have their component in R or A. D3D uses A and that is the default.
#define A8_SAMPLE_MASK .a
	
#if PS4_PROFILE

	#define FEATURE_LEVEL FEATURE_LEVEL_SM5

#elif SM5_PROFILE
	// SM5 = full dx11 features (high end UE4 rendering)
	#define FEATURE_LEVEL FEATURE_LEVEL_SM5
#elif SM4_PROFILE
	#define FEATURE_LEVEL FEATURE_LEVEL_SM4
#elif SWITCH_PROFILE || SWITCH_PROFILE_FORWARD
	#undef ES3_1_PROFILE

	#if SWITCH_PROFILE
		#define FEATURE_LEVEL FEATURE_LEVEL_SM5
	#else
		#define FEATURE_LEVEL FEATURE_LEVEL_ES3_1
		// @todo switch: maybe all uses of this should check feature level not profile?
		#define ES3_1_PROFILE 1
	#endif


	// A8 textures when sampled have their component in R.
	#undef A8_SAMPLE_MASK
	#define A8_SAMPLE_MASK .r

	// switch doesn't have vertex attribute swizzling
	#define FCOLOR_COMPONENT_SWIZZLE .bgra

#elif VULKAN_PROFILE
	#define FEATURE_LEVEL FEATURE_LEVEL_ES3_1
#elif VULKAN_PROFILE_SM4
	#define FEATURE_LEVEL FEATURE_LEVEL_SM4
#elif VULKAN_PROFILE_SM5
	#define FEATURE_LEVEL FEATURE_LEVEL_SM5
#elif GL3_PROFILE || GL4_PROFILE

	#if GL3_PROFILE
		#define FEATURE_LEVEL FEATURE_LEVEL_SM4
	#elif GL4_PROFILE
		#define FEATURE_LEVEL FEATURE_LEVEL_SM5
	#endif

	// A8 textures when sampled have their component in R.
	#undef A8_SAMPLE_MASK
	#define A8_SAMPLE_MASK .r

	// hacks until the shader compiler supports those
	#if GL4_PROFILE
		#define class struct
	#endif
#elif METAL_PROFILE
	#define FEATURE_LEVEL  FEATURE_LEVEL_ES3_1
	// @todo metal: remove this and make sure all uses handle METAL_PROFILE
	#undef ES3_1_PROFILE
	#define ES3_1_PROFILE 1
	#define FCOLOR_COMPONENT_SWIZZLE .bgra
	#define STENCIL_COMPONENT_SWIZZLE .x
#elif METAL_ES2_PROFILE
	#define FEATURE_LEVEL  FEATURE_LEVEL_ES2
	// @todo metal: remove this and make sure all uses handle METAL_ES2_PROFILE
	#undef ES2_PROFILE
	#define ES2_PROFILE 1
	#define FCOLOR_COMPONENT_SWIZZLE .bgra
	#define STENCIL_COMPONENT_SWIZZLE .x
#elif METAL_MRT_PROFILE
	#define FEATURE_LEVEL FEATURE_LEVEL_SM4
	#define FCOLOR_COMPONENT_SWIZZLE .bgra
	#define STENCIL_COMPONENT_SWIZZLE .x
#elif METAL_SM4_PROFILE
	#define FEATURE_LEVEL FEATURE_LEVEL_SM4
	#define FCOLOR_COMPONENT_SWIZZLE .bgra
	#define STENCIL_COMPONENT_SWIZZLE .x
#elif METAL_SM5_PROFILE
	#define FEATURE_LEVEL FEATURE_LEVEL_SM5
	#define FCOLOR_COMPONENT_SWIZZLE .bgra
	#define STENCIL_COMPONENT_SWIZZLE .x
#elif ES2_PROFILE || ES3_1_PROFILE
	#if ES3_1_PROFILE
		#define FEATURE_LEVEL FEATURE_LEVEL_ES3_1
	#else
		//@todo ES3_1 GL
		#define FEATURE_LEVEL FEATURE_LEVEL_ES2
	#endif

	#if COMPILER_GLSL_ES2 || COMPILER_GLSL_ES3_1
		// Swizzle as we only support GL_BGRA on non-ES2 platforms that have that extension
		#define FCOLOR_COMPONENT_SWIZZLE .bgra  
	#else 
		#define FCOLOR_COMPONENT_SWIZZLE 
		#if COMPILER_GLSL
			// A8 textures when sampled have their component in R
			#undef A8_SAMPLE_MASK
			#define A8_SAMPLE_MASK .r
		#endif
	#endif
#else

	#error Add your platform here

	#define FEATURE_LEVEL FEATURE_LEVEL_MAX

#endif

#define METAL_TESSELLATION_PROFILE ((METAL_PROFILE || METAL_SM5_PROFILE) && USING_TESSELLATION)

// non-editor platforms generally never want development/editor features.
#define PLATFORM_SUPPORTS_DEVELOPMENT_SHADERS (!PS4_PROFILE && !XBOXONE_PROFILE && !ESDEFERRED_PROFILE && !SWITCH_PROFILE)
#define USE_DEVELOPMENT_SHADERS (COMPILE_SHADERS_FOR_DEVELOPMENT && PLATFORM_SUPPORTS_DEVELOPMENT_SHADERS)

#ifndef MOBILE_EMULATION
#define MOBILE_EMULATION ((FEATURE_LEVEL == FEATURE_LEVEL_ES2 || FEATURE_LEVEL == FEATURE_LEVEL_ES3_1) && (!(COMPILER_GLSL_ES3_1 || COMPILER_GLSL_ES2) && USE_DEVELOPMENT_SHADERS && (!(METAL_PROFILE) || MAC) && !VULKAN_PROFILE && !SWITCH_PROFILE))
#endif

#ifndef MOBILE_FORCE_DEPTH_TEXTURE_READS
#define MOBILE_FORCE_DEPTH_TEXTURE_READS 0
#endif

// If we didn't request color component swizzling, just make it empty
#ifndef FCOLOR_COMPONENT_SWIZZLE
#define FCOLOR_COMPONENT_SWIZZLE
#endif

#ifndef STENCIL_COMPONENT_SWIZZLE
#define STENCIL_COMPONENT_SWIZZLE .g
#endif

// Whether the platform supports independent texture and samplers
// When enabled, different texture lookups can share samplers to allow more artist samplers in the base pass
// Ideally this would just be enabled for all SM4 and above feature level platforms
// @todo metal mrt: No reason this can't work with Metal, once cross compiler is fixed
#define SUPPORTS_INDEPENDENT_SAMPLERS (PS4_PROFILE || SM5_PROFILE || SM4_PROFILE || METAL_MRT_PROFILE || METAL_SM4_PROFILE || METAL_SM5_PROFILE)

// Whether the platform supports a global clip plane through SV_ClipDistance
// Ideally this would just be enabled for all SM4 and above feature level platforms, but not tested everywhere yet
#define PLATFORM_SUPPORTS_GLOBAL_CLIP_PLANE (PS4_PROFILE || SM5_PROFILE || SM4_PROFILE || METAL_PROFILE || METAL_MRT_PROFILE || METAL_SM4_PROFILE || METAL_SM5_PROFILE || GL4_PROFILE || GL3_PROFILE || VULKAN_PROFILE_SM4 || VULKAN_PROFILE_SM5)

// Using SV_ClipDistance has overhead (15% slower base pass in triangle bound test scene on PS4) so projects have to opt-in
#define USE_GLOBAL_CLIP_PLANE (PLATFORM_SUPPORTS_GLOBAL_CLIP_PLANE && PROJECT_ALLOW_GLOBAL_CLIP_PLANE)

// Whether the platform support pixel coverage on MSAA targets (SV_Coverage).
#define SUPPORTS_PIXEL_COVERAGE (FEATURE_LEVEL >= FEATURE_LEVEL_SM5 && !COMPILER_GLSL && !MOBILE_EMULATION)

// This would need to be a #define in GLSL to ignore the SamplerState, however, it is currently a function call in HLSL
// for type checking of the parameters - ironically the type checking is really only needed in GLSL!
MaterialFloat4 Texture1DSample(Texture1D Tex, SamplerState Sampler, float UV)
{
#if COMPUTESHADER
	return Tex.SampleLevel(Sampler, UV, 0);
#else
	return Tex.Sample(Sampler, UV);
#endif
}
MaterialFloat4 Texture2DSample(Texture2D Tex, SamplerState Sampler, float2 UV)
{
#if COMPUTESHADER
	return Tex.SampleLevel(Sampler, UV, 0);
#else
	return Tex.Sample(Sampler, UV);
#endif
}
MaterialFloat Texture2DSample_A8(Texture2D Tex, SamplerState Sampler, float2 UV)
{
#if COMPUTESHADER
	return Tex.SampleLevel(Sampler, UV, 0) A8_SAMPLE_MASK;
#else
	return Tex.Sample(Sampler, UV) A8_SAMPLE_MASK;
#endif
}
MaterialFloat4 Texture3DSample(Texture3D Tex, SamplerState Sampler, float3 UV)
{
#if COMPUTESHADER
	return Tex.SampleLevel(Sampler, UV, 0);
#else
	return Tex.Sample(Sampler, UV);
#endif
}
MaterialFloat4 TextureCubeSample(TextureCube Tex, SamplerState Sampler, float3 UV)
{
#if COMPUTESHADER
	return Tex.SampleLevel(Sampler, UV, 0);
#else
	return Tex.Sample(Sampler, UV);
#endif
}
MaterialFloat4 Texture1DSampleLevel(Texture1D Tex, SamplerState Sampler, float UV, MaterialFloat Mip)
{
	return Tex.SampleLevel(Sampler, UV, Mip);
}
MaterialFloat4 Texture2DSampleLevel(Texture2D Tex, SamplerState Sampler, float2 UV, MaterialFloat Mip)
{
	return Tex.SampleLevel(Sampler, UV, Mip);
}
MaterialFloat4 Texture2DSampleBias(Texture2D Tex, SamplerState Sampler, float2 UV, MaterialFloat MipBias)
{
#if COMPUTESHADER
	return Tex.SampleLevel(Sampler, UV, 0);
#else
	return Tex.SampleBias(Sampler, UV, MipBias);
#endif
}
MaterialFloat4 Texture2DSampleGrad(Texture2D Tex, SamplerState Sampler, float2 UV, MaterialFloat2 DDX, MaterialFloat2 DDY)
{
	return Tex.SampleGrad(Sampler, UV, DDX, DDY);
}
MaterialFloat4 Texture3DSampleLevel(Texture3D Tex, SamplerState Sampler, float3 UV, MaterialFloat Mip)
{
	return Tex.SampleLevel(Sampler, UV, Mip);
}
MaterialFloat4 TextureCubeSampleLevel(TextureCube Tex, SamplerState Sampler, float3 UV, MaterialFloat Mip)
{
	return Tex.SampleLevel(Sampler, UV, Mip);
}
MaterialFloat4 TextureCubeSampleBias(TextureCube Tex, SamplerState Sampler, float3 UV, MaterialFloat MipBias)
{
#if COMPUTESHADER
	return Tex.SampleLevel(Sampler, UV, 0);
#else
	return Tex.SampleBias(Sampler, UV, MipBias);
#endif
}
MaterialFloat4 TextureCubeSampleGrad(TextureCube Tex, SamplerState Sampler, float3 UV, MaterialFloat3 DDX, MaterialFloat3 DDY)
{
	return Tex.SampleGrad(Sampler, UV, DDX, DDY);
}

//converts an input 1d to 2d position. Useful for locating z frames that have been laid out in a 2d grid like a flipbook.
float2 Tile1Dto2D(float xsize, float idx)
{
	float2 xyidx = 0;
	xyidx.y = floor(idx / xsize);
	xyidx.x = idx - xsize * xyidx.y;

	return xyidx;
}

// return a pseudovolume texture sample.
// useful for simulating 3D texturing with a 2D texture or as a texture flipbook with lerped transitions
// treats 2d layout of frames a 3d texture and performs bilinear filtering by blending with an offset Z frame.
// @param Tex          = Input Texture Object storing Volume Data
// @param inPos        = Input float3 for Position, 0-1
// @param xsize        = Input float for num frames in x,y directions
// @param numFrames    = Input float for num total frames
// @param mipmode      = Sampling mode: 0 = use miplevel, 1 = use UV computed gradients, 2 = Use gradients (default=0)
// @param miplevel     = MIP level to use in mipmode=0 (default 0)
// @param InDDX, InDDY = Texture gradients in mipmode=2
float4 PseudoVolumeTexture(Texture2D Tex, SamplerState TexSampler, float3 inPos, float xsize, float numframes,
	uint mipmode = 0, float miplevel = 0, float2 InDDX = 0, float2 InDDY = 0)
{
	float zframe = ceil(inPos.z * numframes);
	float zphase = frac(inPos.z * numframes);

	float2 uv = frac(inPos.xy) / xsize;

	float2 curframe = Tile1Dto2D(xsize, zframe) / xsize;
	float2 nextframe = Tile1Dto2D(xsize, zframe + 1) / xsize;

	float4 sampleA = 0, sampleB = 0;
	switch (mipmode)
	{
	case 0: // Mip level
		sampleA = Tex.SampleLevel(TexSampler, uv + curframe, miplevel);
		sampleB = Tex.SampleLevel(TexSampler, uv + nextframe, miplevel);
		break;
	case 1: // Gradients automatic from UV
		sampleA = Texture2DSample(Tex, TexSampler, uv + curframe);
		sampleB = Texture2DSample(Tex, TexSampler, uv + nextframe);
		break;
	case 2: // Deriviatives provided
		sampleA = Tex.SampleGrad(TexSampler, uv + curframe,  InDDX, InDDY);
		sampleB = Tex.SampleGrad(TexSampler, uv + nextframe, InDDX, InDDY);
		break;
	default:
		break;
	}

	return lerp(sampleA, sampleB, zphase);
}

#if FEATURE_LEVEL >= FEATURE_LEVEL_SM5 // Cubemap arrays are not supported in SM4 feature level
	MaterialFloat4 TextureCubeArraySampleLevel(TextureCubeArray Tex, SamplerState Sampler, float3 UV, float ArrayIndex, MaterialFloat Mip)
	{
		return Tex.SampleLevel(Sampler, float4(UV, ArrayIndex), Mip);
	}
#endif

void Bicubic2DCatmullRom( in float2 UV, in float2 Size, out float2 Sample[3], out float2 Weight[3] )
{
	const float2 InvSize = 1.0 / Size;
 
	UV *= Size;

	float2 tc = floor( UV - 0.5 ) + 0.5;
	float2 f = UV - tc;
	float2 f2 = f * f;
	float2 f3 = f2 * f;

	float2 w0 = f2 - 0.5 * (f3 + f);
	float2 w1 = 1.5 * f3 - 2.5 * f2 + 1;
	float2 w3 = 0.5 * (f3 - f2);
	float2 w2 = 1 - w0 - w1 - w3;

	Weight[0] = w0;
	Weight[1] = w1 + w2;
	Weight[2] = w3;

	Sample[0] = tc - 1;
	Sample[1] = tc + w2 / Weight[1];
	Sample[2] = tc + 2;
 
	Sample[0] *= InvSize;
	Sample[1] *= InvSize;
	Sample[2] *= InvSize;
}

MaterialFloat4 Texture2DSampleBicubic( Texture2D Tex, SamplerState Sampler, float2 UV, float2 Size )
{
	float2 Weight[3];
	float2 Sample[3];
	Bicubic2DCatmullRom( UV, Size, Sample, Weight );

	MaterialFloat4 OutColor;
#if 0
	OutColor  = Tex.SampleLevel( Sampler, float2( Sample[0].x, Sample[0].y ), 0 ) * Weight[0].x * Weight[0].y;
	OutColor += Tex.SampleLevel( Sampler, float2( Sample[1].x, Sample[0].y ), 0 ) * Weight[1].x * Weight[0].y;
	OutColor += Tex.SampleLevel( Sampler, float2( Sample[2].x, Sample[0].y ), 0 ) * Weight[2].x * Weight[0].y;

	OutColor += Tex.SampleLevel( Sampler, float2( Sample[0].x, Sample[1].y ), 0 ) * Weight[0].x * Weight[1].y;
	OutColor += Tex.SampleLevel( Sampler, float2( Sample[1].x, Sample[1].y ), 0 ) * Weight[1].x * Weight[1].y;
	OutColor += Tex.SampleLevel( Sampler, float2( Sample[2].x, Sample[1].y ), 0 ) * Weight[2].x * Weight[1].y;

	OutColor += Tex.SampleLevel( Sampler, float2( Sample[0].x, Sample[2].y ), 0 ) * Weight[0].x * Weight[2].y;
	OutColor += Tex.SampleLevel( Sampler, float2( Sample[1].x, Sample[2].y ), 0 ) * Weight[1].x * Weight[2].y;
	OutColor += Tex.SampleLevel( Sampler, float2( Sample[2].x, Sample[2].y ), 0 ) * Weight[2].x * Weight[2].y;
#else
	// Optimized by removing corner samples
	OutColor  = Tex.SampleLevel( Sampler, float2( Sample[1].x, Sample[0].y ), 0 ) * Weight[1].x * Weight[0].y;
	OutColor += Tex.SampleLevel( Sampler, float2( Sample[0].x, Sample[1].y ), 0 ) * Weight[0].x * Weight[1].y;
	OutColor += Tex.SampleLevel( Sampler, float2( Sample[1].x, Sample[1].y ), 0 ) * Weight[1].x * Weight[1].y;
	OutColor += Tex.SampleLevel( Sampler, float2( Sample[2].x, Sample[1].y ), 0 ) * Weight[2].x * Weight[1].y;
	OutColor += Tex.SampleLevel( Sampler, float2( Sample[1].x, Sample[2].y ), 0 ) * Weight[1].x * Weight[2].y;

	// Reweight after removing the corners
	float CornerWeights;
	CornerWeights  = Weight[1].x * Weight[0].y;
	CornerWeights += Weight[0].x * Weight[1].y;
	CornerWeights += Weight[1].x * Weight[1].y;
	CornerWeights += Weight[2].x * Weight[1].y;
	CornerWeights += Weight[1].x * Weight[2].y;
	OutColor /= CornerWeights;
#endif
	return OutColor;
}

	// TANGENTTOWORLD0 is the first row of the tangent to world matrix, w might be needed for padding and is not used yet.
	// TANGENTTOWORLD2 is the last row of the tangent to world matrix, determinant of tangent basis in w

#if FEATURE_LEVEL >= FEATURE_LEVEL_SM5

	#define TANGENTTOWORLD0					TEXCOORD10
	#define TANGENTTOWORLD2					TEXCOORD11

	// _centroid is needed to get better quality with MSAA

	// The D3D shader compiler combines _centroid and non controid. Using float3 would results in a internal
	// shader compiler error. This block is using float4 to prevent that.
	#define TANGENTTOWORLD_INTERPOLATOR_BLOCK	float4 TangentToWorld0 : TEXCOORD10_centroid; float4	TangentToWorld2	: TEXCOORD11_centroid;

#else

	#define TANGENTTOWORLD0					TEXCOORD10
	#define TANGENTTOWORLD2					TEXCOORD11

	// TangentToWorld0 is float4 only to match D3D11
#if METAL_PROFILE || COMPILER_GLSL_ES3_1
		//@todo-rco: FIXME!
		#define TANGENTTOWORLD_INTERPOLATOR_BLOCK	float4 TangentToWorld0 : TANGENTTOWORLD0; float4	TangentToWorld2	: TANGENTTOWORLD2;
#else
	#define TANGENTTOWORLD_INTERPOLATOR_BLOCK	MaterialFloat4 TangentToWorld0 : TANGENTTOWORLD0; MaterialFloat4	TangentToWorld2	: TANGENTTOWORLD2;
#endif
#endif

MaterialFloat Luminance( MaterialFloat3 LinearColor )
{
	return dot( LinearColor, MaterialFloat3( 0.3, 0.59, 0.11 ) );
}

const static MaterialFloat PI = 3.1415926535897932f;

MaterialFloat length2(MaterialFloat2 v)
{
	return dot(v, v);
}
MaterialFloat length2(MaterialFloat3 v)
{
	return dot(v, v);
}
MaterialFloat length2(MaterialFloat4 v)
{
	return dot(v, v);
}

uint Mod(uint a, uint b)
{
#if FEATURE_LEVEL >= FEATURE_LEVEL_ES3_1
	return a % b;
#else
	return a - (b * (uint)((float)a / (float)b));
#endif
}

uint2 Mod(uint2 a, uint2 b)
{
#if FEATURE_LEVEL >= FEATURE_LEVEL_ES3_1
	return a % b;
#else
	return a - (b * (uint2)((float2)a / (float2)b));
#endif
}

uint3 Mod(uint3 a, uint3 b)
{
#if FEATURE_LEVEL >= FEATURE_LEVEL_ES3_1
	return a % b;
#else
	return a - (b * (uint3)((float3)a / (float3)b));
#endif
}

MaterialFloat UnClampedPow(MaterialFloat X, MaterialFloat Y)
{
	return pow(X, Y);
}
MaterialFloat2 UnClampedPow(MaterialFloat2 X, MaterialFloat2 Y)
{
	return pow(X, Y);
}
MaterialFloat3 UnClampedPow(MaterialFloat3 X, MaterialFloat3 Y)
{
	return pow(X, Y);
}
MaterialFloat4 UnClampedPow(MaterialFloat4 X, MaterialFloat4 Y)
{
	return pow(X, Y);
}

// Clamp the base, so it's never <= 0.0f (INF/NaN).
MaterialFloat ClampedPow(MaterialFloat X,MaterialFloat Y)
{
	return pow(max(abs(X),0.000001f),Y);
}
MaterialFloat2 ClampedPow(MaterialFloat2 X,MaterialFloat2 Y)
{
	return pow(max(abs(X),MaterialFloat2(0.000001f,0.000001f)),Y);
}
MaterialFloat3 ClampedPow(MaterialFloat3 X,MaterialFloat3 Y)
{
	return pow(max(abs(X),MaterialFloat3(0.000001f,0.000001f,0.000001f)),Y);
}  
MaterialFloat4 ClampedPow(MaterialFloat4 X,MaterialFloat4 Y)
{
	return pow(max(abs(X),MaterialFloat4(0.000001f,0.000001f,0.000001f,0.000001f)),Y);
} 

MaterialFloat PositiveClampedPow(MaterialFloat X,MaterialFloat Y)
{
	return pow(max(X,0.0f),Y);
}
MaterialFloat2 PositiveClampedPow(MaterialFloat2 X,MaterialFloat2 Y)
{
	return pow(max(X,MaterialFloat2(0.0f,0.0f)),Y);
}
MaterialFloat3 PositiveClampedPow(MaterialFloat3 X,MaterialFloat3 Y)
{
	return pow(max(X,MaterialFloat3(0.0f,0.0f,0.0f)),Y);
}  
MaterialFloat4 PositiveClampedPow(MaterialFloat4 X,MaterialFloat4 Y)
{
	return pow(max(X,MaterialFloat4(0.0f,0.0f,0.0f,0.0f)),Y);
} 

float DDX(float Input)
{
#if COMPUTESHADER
	return 0;
#else
	return ddx(Input);
#endif
}

float2 DDX(float2 Input)
{
#if COMPUTESHADER
	return 0;
#else
	return ddx(Input);
#endif
}

float3 DDX(float3 Input)
{
#if COMPUTESHADER
	return 0;
#else
	return ddx(Input);
#endif
}

float4 DDX(float4 Input)
{
#if COMPUTESHADER
	return 0;
#else
	return ddx(Input);
#endif
}

float DDY(float Input)
{
#if COMPUTESHADER
	return 0;
#else
	return ddy(Input);
#endif
}

float2 DDY(float2 Input)
{
#if COMPUTESHADER
	return 0;
#else
	return ddy(Input);
#endif
}

float3 DDY(float3 Input)
{
#if COMPUTESHADER
	return 0;
#else
	return ddy(Input);
#endif
}

float4 DDY(float4 Input)
{
#if COMPUTESHADER
	return 0;
#else
	return ddy(Input);
#endif
}

#include "FastMath.usf"
#include "Random.usf"	// used by MaterialExpressionNoise
  
/** 
 * Use this function to compute the pow() in the specular computation.
 * This allows to change the implementation depending on platform or it easily can be replaced by some approxmation.
 */
MaterialFloat PhongShadingPow(MaterialFloat X, MaterialFloat Y)
{
	// The following clamping is done to prevent NaN being the result of the specular power computation.
	// Clamping has a minor performance cost.

	// In HLSL pow(a, b) is implemented as exp2(log2(a) * b).

	// For a=0 this becomes exp2(-inf * 0) = exp2(NaN) = NaN.

	// As seen in #TTP 160394 "QA Regression: PS3: Some maps have black pixelated artifacting."
	// this can cause severe image artifacts (problem was caused by specular power of 0, lightshafts propagated this to other pixels).
	// The problem appeared on PlayStation 3 but can also happen on similar PC NVidia hardware.

	// In order to avoid platform differences and rarely occuring image atrifacts we clamp the base.

	// Note: Clamping the exponent seemed to fix the issue mentioned TTP but we decided to fix the root and accept the
	// minor performance cost.

	return ClampedPow(X, Y);
}

#if FEATURE_LEVEL < FEATURE_LEVEL_ES3_1 && !COMPILER_METAL
	// DX11 (feature levels >= 10) feature sets natively supports uints in shaders; we just use floats on other platforms.
	#define uint4	int4
#endif

// Optional VertexID - used by tessellation to uniquely identify control points.
#if USING_TESSELLATION && DISPLACEMENT_ANTICRACK
	#define OPTIONAL_VertexID			uint VertexID : SV_VertexID,
	#define OPTIONAL_VertexID_PARAM		VertexID,
	#define OPTIONAL_VertexID_VS_To_DS	uint VertexID : VS_To_DS_VertexID;
	#define OutputVertexID( Out ) Out.VertexID = VertexID
#else // #if USING_TESSELLATION && DISPLACEMENT_ANTICRACK
	#define OPTIONAL_VertexID
	#define OPTIONAL_VertexID_PARAM
	#define OPTIONAL_VertexID_VS_To_DS
	#define OutputVertexID( Out )
#endif // #if USING_TESSELLATION && DISPLACEMENT_ANTICRACK

// Helper macro used to interpolate the given member
#define TESSELLATION_INTERPOLATE_MEMBER(member) O.member = a.member * aInterp + b.member * bInterp


#if FEATURE_LEVEL >= FEATURE_LEVEL_SM4
	/** 
		* Number of MSAA samples supported by deferred passes in D3D11. 
		* This is hardcoded because it allows deferred passes to optimize for the given value (for example, unrolling a loop).
		*/
	#define NumMSAASamples 4
#endif

// depth in the red channel in DeviceZ
Texture2D		SceneDepthTexture;
SamplerState	SceneDepthTextureSampler;
Texture2D		CustomDepthTexture;
SamplerState	CustomDepthTextureSampler;
// Custom Stencil texture used for mobile platforms
Texture2D		MobileCustomStencilTexture;
SamplerState	MobileCustomStencilTextureSampler;
// scene HDR color
Texture2D		SceneColorTexture;
SamplerState	SceneColorTextureSampler;
// copy of scene alpha for PC ES2 emulation
Texture2D		SceneAlphaCopyTexture;
SamplerState	SceneAlphaCopyTextureSampler;
// shadow and light function
Texture2D		LightAttenuationTexture;
SamplerState	LightAttenuationTextureSampler;

#if FEATURE_LEVEL >= FEATURE_LEVEL_SM4
// Scene depth-stencil buffer used for stencil reads
Texture2D<uint2>	SceneStencilTexture;
#endif

// We don't use an inline function so we can avoid type promotion/ coercion.
#define RETURN_COLOR( Color ) ( Color )

// Tangent space bias
// We don't use a function so we can avoid type promotion/ coercion.
#define TangentBias(X) (X * 2.0f - 1.0f)

float Square( float x )
{
	return x*x;
}

float2 Square( float2 x )
{
	return x*x;
}

float3 Square( float3 x )
{
	return x*x;
}

float4 Square( float4 x )
{
	return x*x;
}

float Pow2( float x )
{
	return x*x;
}

float2 Pow2( float2 x )
{
	return x*x;
}

float3 Pow2( float3 x )
{
	return x*x;
}

float4 Pow2( float4 x )
{
	return x*x;
}

float Pow3( float x )
{
	return x*x*x;
}

float2 Pow3( float2 x )
{
	return x*x*x;
}

float3 Pow3( float3 x )
{
	return x*x*x;
}

float4 Pow3( float4 x )
{
	return x*x*x;
}

float Pow4( float x )
{
	float xx = x*x;
	return xx * xx;
}

float2 Pow4( float2 x )
{
	float2 xx = x*x;
	return xx * xx;
}

float3 Pow4( float3 x )
{
	float3 xx = x*x;
	return xx * xx;
}

float4 Pow4( float4 x )
{
	float4 xx = x*x;
	return xx * xx;
}

float Pow5( float x )
{
	float xx = x*x;
	return xx * xx * x;
}

float2 Pow5( float2 x )
{
	float2 xx = x*x;
	return xx * xx * x;
}

float3 Pow5( float3 x )
{
	float3 xx = x*x;
	return xx * xx * x;
}

float4 Pow5( float4 x )
{
	float4 xx = x*x;
	return xx * xx * x;
}

float Pow6( float x )
{
	float xx = x*x;
	return xx * xx * xx;
}

float2 Pow6( float2 x )
{
	float2 xx = x*x;
	return xx * xx * xx;
}

float3 Pow6( float3 x )
{
	float3 xx = x*x;
	return xx * xx * xx;
}

float4 Pow6( float4 x )
{
	float4 xx = x*x;
	return xx * xx * xx;
}

// Only valid for x >= 0
MaterialFloat AtanFast( MaterialFloat x )
{
	// Minimax 3 approximation
	MaterialFloat3 A = x < 1 ? MaterialFloat3( x, 0, 1 ) : MaterialFloat3( 1/x, 0.5 * PI, -1 );
	return A.y + A.z * ( ( ( -0.130234 * A.x - 0.0954105 ) * A.x + 1.00712 ) * A.x - 0.00001203333 );
}

/**
 * Returns the upper 3x3 portion of the LocalToWorld matrix.
 */
MaterialFloat3x3 GetLocalToWorld3x3()
{
	return (MaterialFloat3x3)Primitive.LocalToWorld;
}

/** Converts a linear input value into a value to be stored in the light attenuation buffer. */
MaterialFloat EncodeLightAttenuation(MaterialFloat InColor)
{
	// Apply a 1/2 power to the input, which allocates more bits for the darks and prevents banding
	// Similar to storing colors in gamma space, except this uses less instructions than a pow(x, 1/2.2)
	return sqrt(InColor);
}

/** Converts a linear input value into a value to be stored in the light attenuation buffer. */
MaterialFloat4 EncodeLightAttenuation(MaterialFloat4 InColor)
{
	return sqrt(InColor);
}

/** return the scene lighting texture */
MaterialFloat3 CalcSceneColor(MaterialFloat2 ScreenUV)
{
#if SCENE_TEXTURES_DISABLED
	return MaterialFloat3(0.0f,0.0f,0.0f);
#else
	return Texture2DSampleLevel(SceneColorTexture, SceneColorTextureSampler, ScreenUV, 0).rgb;
#endif
}

/** return all channels of the scene lighting texture */
MaterialFloat4 CalcFullSceneColor(MaterialFloat2 ScreenUV)
{
#if SCENE_TEXTURES_DISABLED
	return MaterialFloat4(0.0f, 0.0f, 0.0f, 0.0f);
#else
	return Texture2DSample(SceneColorTexture, SceneColorTextureSampler,ScreenUV);
#endif
}

/** Encodes HDR linear scene color for storage in the 8 bit light attenuation texture. */
MaterialFloat3 EncodeSceneColorForMaterialNode(MaterialFloat3 LinearSceneColor)
{
	// Preserving a range from [0, 10]
	// Remap values to get more bits of precision in the darks
	return pow(LinearSceneColor * .1f, .25f);
}


//
// MOBILE WITHOUT FP16 SUPPORT: 8-BIT/CHANNEL MOSAICING FOR LINEAR HDR
//
// This provides linear blending and a {0 to 2} dynamic range.
// This works by splitting the image into a checkerboard of dark and light pixels.
//   DLDL
//   LDLD
//   DLDL
//   LDLD
//
// There is an extra scan line dither pattern to increase precision by 1-bit.
// The HdrMosaic() function computes the exposure level per pixel for the forward render pass.
// The HdrDemosaic() function reconstructs the proper image in the tonemapping pass.
//

// These have been carefully tuned and should probably not be adjusted.
// Dark pixel range is {0 to 1/DRK_MUL}.
#define DRK_MUL 6.0
// Support {0 to 2} in dynamic range.
#define HDR_MUL (1.0/2.0)

#define MOSAIC_ADD   HDR_MUL
// The blend factor causes a smooth blend between the light and dark pixels.
#define MOSAIC_BLEND ((255.0-(DRK_MUL/HDR_MUL))/255.0)
#define MOSAIC_MUL   ((DRK_MUL * MOSAIC_BLEND - MOSAIC_ADD) * 2.0)

// Done during forward shading pass before blending.
MaterialFloat3 HdrMosaic(MaterialFloat3 LinearColor, float2 VPos) 
{
	float2 V;
	V = VPos.xy * 0.5;
	V.y += V.x;
	V = frac(V);
	MaterialFloat2 C;
	C = (V * MaterialFloat2(2.0 * (-0.5/255.0), MOSAIC_MUL)) + MaterialFloat2(-0.5/255.0, MOSAIC_ADD);
	return (LinearColor * C.y) + C.x;
}

#define DEMOSAIC_MUL (((1.0/DRK_MUL) - (1.0/HDR_MUL)) * 2.0)
#define DEMOSAIC_ADD (1.0/HDR_MUL)

// Resolve pass to remove mosaic and restore color.
MaterialFloat3 HdrDemosaic(MaterialFloat3 Pixel, MaterialFloat3 OtherPixel, float2 VPos)
{
	MaterialFloat A = frac(dot(VPos + View.DemosaicVposOffset, float2(0.5, 0.5)));
	MaterialFloat B = 0.5 - A;
	A = A * DEMOSAIC_MUL + DEMOSAIC_ADD;
	B = B * DEMOSAIC_MUL + DEMOSAIC_ADD;

	// On ES2 devices we demosaic during the tonemapping pass which renders upside down, account for that here.
	#if COMPILER_GLSL_ES2 || COMPILER_GLSL_ES3_1
		return max((Pixel * B), (OtherPixel * A)); 
	#else
		return max((Pixel * A), (OtherPixel * B)); 
	#endif
}


// Like RGBM but this can be interpolated.
MaterialFloat4 RGBTEncode(MaterialFloat3 Color)
{
	MaterialFloat4 RGBT;
	MaterialFloat Max = max(max(Color.r, Color.g), max(Color.b, 1e-6));
	MaterialFloat RcpMax = rcp(Max);
	RGBT.rgb = Color.rgb * RcpMax;
	RGBT.a = Max * rcp(1.0 + Max);
	return RGBT;
}

MaterialFloat3 RGBTDecode(MaterialFloat4 RGBT)
{
	RGBT.a = RGBT.a * rcp(1.0 - RGBT.a);
	return RGBT.rgb * RGBT.a;
}



MaterialFloat4 RGBMEncode( MaterialFloat3 Color )
{
	Color *= 1.0 / 64.0;
	
	float4 rgbm;
	rgbm.a = saturate( max( max( Color.r, Color.g ), max( Color.b, 1e-6 ) ) );
	rgbm.a = ceil( rgbm.a * 255.0 ) / 255.0;
	rgbm.rgb = Color / rgbm.a;
	return rgbm;
}

MaterialFloat4 RGBMEncodeFast( MaterialFloat3 Color )
{
	// 0/0 result written to fixed point buffer goes to zero
	MaterialFloat4 rgbm;
	rgbm.a = dot( Color, 255.0 / 64.0 );
	rgbm.a = ceil( rgbm.a );
	rgbm.rgb = Color / rgbm.a;
	rgbm *= MaterialFloat4( 255.0 / 64.0, 255.0 / 64.0, 255.0 / 64.0, 1.0 / 255.0 );
	return rgbm;
}

MaterialFloat3 RGBMDecode( MaterialFloat4 rgbm, MaterialFloat MaxValue )
{
	return rgbm.rgb * (rgbm.a * MaxValue);
}

MaterialFloat3 RGBMDecode( MaterialFloat4 rgbm )
{
	return rgbm.rgb * (rgbm.a * 64.0f);
}

MaterialFloat4 RGBTEncode8BPC(MaterialFloat3 Color, MaterialFloat Range)
{
	MaterialFloat Max = max(max(Color.r, Color.g), max(Color.b, 1e-6));
	Max = min(Max, Range);

	MaterialFloat4 RGBT;
	RGBT.a = (Range + 1) / Range *  Max / (1 + Max);

	// quantise alpha to 8 bit.
	RGBT.a = ceil(RGBT.a*255.0) / 255.0;
	Max = RGBT.a / (1 + 1 / Range - RGBT.a);

	MaterialFloat RcpMax = rcp(Max);
	RGBT.rgb = Color.rgb * RcpMax;
	return RGBT;
}

MaterialFloat3 RGBTDecode8BPC(MaterialFloat4 RGBT, MaterialFloat Range)
{
	RGBT.a = RGBT.a / (1 + 1 / Range - RGBT.a);
	return RGBT.rgb * RGBT.a;
}

#define HDR_ENCODE_NONE 0.0		// 64bpp HDR
#define HDR_ENCODE_MOSAIC 1.0	// 32bpp HDR using Mosaic encoding
#define HDR_ENCODE_RGBA 2.0		// 32bpp HDR using RGBA encoding

half GetHDR32bppEncodeMode()
{
#if ES2_PROFILE
	#if COMPILER_GLSL_ES2 && !IOS && !WEBGL // ANDROID
		return intrinsic_GetHDR32bppEncodeModeES2();
	#endif
	#if MOBILE_EMULATION
		// To enable editor runtime change without recompile, PC always eats the encode cost.
		return View.HDR32bppEncodingMode;
	#endif
#endif
	return HDR_ENCODE_NONE;
}

#define DEFAULT_32BPPHDR_ENCODED_RANGE 1024.0

MaterialFloat4 Encode32BPPHDR(MaterialFloat4 Color, float2 SvPosition)
{
	half Mode = GetHDR32bppEncodeMode();
	if (Mode == HDR_ENCODE_MOSAIC)
	{ 
		return MaterialFloat4(HdrMosaic(Color.rgb, SvPosition), Color.a);
	}
	else if (Mode == HDR_ENCODE_RGBA)
	{ 
		return RGBTEncode8BPC(Color.rgb, DEFAULT_32BPPHDR_ENCODED_RANGE);
	}
	else 
	{ 
		return Color;
	}
}

MaterialFloat4 Decode32BPPHDR(MaterialFloat4 Encoded, MaterialFloat3 OtherEncoded = MaterialFloat3(0, 0, 0), float2 SvPosition = float2(0, 0))
{
	half Mode = GetHDR32bppEncodeMode();
	if (Mode == HDR_ENCODE_MOSAIC)
	{
		return MaterialFloat4(HdrDemosaic(Encoded.rgb, OtherEncoded, SvPosition), 0.0f);
	}
	if (Mode == HDR_ENCODE_RGBA)
	{ 
		return MaterialFloat4(RGBTDecode8BPC(Encoded, DEFAULT_32BPPHDR_ENCODED_RANGE), 0.0f);
	}
	else 
	{
		return Encoded; 
	}
}

// Demosaic capable 32bpp hdr decode.
float4 Decode32BPPHDR(float4 Pixel, float2 SvPosition, Texture2D SourceTexture, SamplerState SourceSampler, float2 InTexCoords[4])
{
	half3 PixelColorN = SourceTexture.Sample(SourceSampler, InTexCoords[0].xy).rgb;
	half3 PixelColorE = SourceTexture.Sample(SourceSampler, InTexCoords[1].xy).rgb;
	half3 PixelColorW = SourceTexture.Sample(SourceSampler, InTexCoords[2].xy).rgb;
	half3 PixelColorS = SourceTexture.Sample(SourceSampler, InTexCoords[3].xy).rgb;
	half3 PixelColorV = PixelColorN * 0.5 + PixelColorS * 0.5;
	half3 PixelColorH = PixelColorW * 0.5 + PixelColorE * 0.5;
	if(abs(PixelColorN.g - PixelColorS.g) < abs(PixelColorW.g - PixelColorE.g)) 
	{
		PixelColorH = PixelColorV;
	}
	Pixel.rgb = Decode32BPPHDR(Pixel, PixelColorH, SvPosition.xy).rgb;
	Pixel.a = 0.0;
	return Pixel;
}

/** Get render target write mask value
  * This gets a bit from a write mask texture created with FRTWriteMaskDecodeCS. Only supprted on some platforms.
  */
#if PLATFORM_SUPPORTS_RENDERTARGET_WRITE_MASK
uint DecodeRTWriteMaskTexture(in float2 ScreenPosition, in Texture2D<uint> RTWriteMaskTexture)
{
	int2 IntPosition = int2(ScreenPosition.xy);
	uint RTWriteMaskValue = RTWriteMaskTexture.Load( int3(IntPosition.x/8, IntPosition.y/8, 0) );

	int2 BitCoord = ((IntPosition / int2(4, 4)) % int2(2, 2));
	uint BitIdx = BitCoord.x + (BitCoord.y*2);
	uint RTWriteMaskBit = RTWriteMaskValue & (1<<BitIdx);

	return RTWriteMaskBit;
}
#endif

/** Calculates the ScreenUV given the screen position and an offset fraction. */
float2 CalcScreenUVFromOffsetFraction(float4 ScreenPosition, float2 OffsetFraction)
{
	float2 NDC = ScreenPosition.xy / ScreenPosition.w;
	// Apply the offset in NDC space so that it is consistent regardless of scene color buffer size
	// Clamp to valid area of the screen to avoid reading garbage
	//@todo - soft clamp
	float2 OffsetNDC = clamp(NDC + OffsetFraction * float2(2, -2), -.999f, .999f);
	return float2(OffsetNDC * View.ScreenPositionScaleBias.xy + View.ScreenPositionScaleBias.wz);
}

Texture2D		SceneColorCopyTexture;
SamplerState	SceneColorCopyTextureSampler;

/** Applies an offset to the scene texture lookup and decodes the HDR linear space color. */
float3 DecodeSceneColorForMaterialNode(float2 ScreenUV)
{
#if HIT_PROXY_SHADER || SCENE_TEXTURES_DISABLED
	// Hit proxies rendering pass doesn't have access to valid render buffers
	return float3(0.0f, 0.0f, 0.0f);
#else
	float4 EncodedSceneColor = Texture2DSample(SceneColorCopyTexture, SceneColorCopyTextureSampler, ScreenUV);

	// Undo the function in EncodeSceneColorForMaterialNode
	return pow(EncodedSceneColor.rgb, 4) * 10;
#endif
}

float4 GetPerPixelLightAttenuation(float2 UV)
{
	return Square(Texture2DSampleLevel(LightAttenuationTexture, LightAttenuationTextureSampler, UV, 0));
}

// also see ConvertToDeviceZ()
// @param DeviceZ value that is stored in the depth buffer (Z/W)
// @return SceneDepth (linear in world units, W)
float ConvertFromDeviceZ(float DeviceZ)
{
	// Supports ortho and perspective, see CreateInvDeviceZToWorldZTransform()
	return DeviceZ * View.InvDeviceZToWorldZTransform[0] + View.InvDeviceZToWorldZTransform[1] + 1.0f / (DeviceZ * View.InvDeviceZToWorldZTransform[2] - View.InvDeviceZToWorldZTransform[3]);
}

// inverse operation of ConvertFromDeviceZ()
// @param SceneDepth (linear in world units, W)
// @return DeviceZ (Z/W)
float ConvertToDeviceZ(float SceneDepth)
{
	FLATTEN
	if (View.ViewToClip[3][3] < 1.0f)
	{
		// Perspective
		return 1.0f / ((SceneDepth + View.InvDeviceZToWorldZTransform[3]) * View.InvDeviceZToWorldZTransform[2]);
	}
	else
	{
		// Ortho
		return SceneDepth * View.ViewToClip[2][2] + View.ViewToClip[3][2];
	}
}

/** Returns clip space W, which is world space distance along the View Z axis. Note if you need DeviceZ LookupDeviceZ() is the faster option */
float CalcSceneDepth(float2 ScreenUV)
{
#if SCENE_TEXTURES_DISABLED
	return 0.0f;
#else
	#if FEATURE_LEVEL > FEATURE_LEVEL_ES3_1 || MOBILE_FORCE_DEPTH_TEXTURE_READS
		return ConvertFromDeviceZ(Texture2DSampleLevel(SceneDepthTexture, SceneDepthTextureSampler, ScreenUV, 0).r);
	#else
		#if COMPILER_GLSL_ES2
			#if IOS
				// Only call FramebufferFetch when actually compiling for IOS ES2.
				return FramebufferFetchES2().w;
			#elif WEBGL
				return Texture2DSampleLevel(SceneAlphaCopyTexture, SceneAlphaCopyTextureSampler, ScreenUV, 0).r;
			#else 
				float SceneW = ConvertFromDeviceZ(Texture2DSampleLevel(SceneDepthTexture, SceneDepthTextureSampler, ScreenUV, 0).r);		
				return DepthbufferFetchES2(SceneW, View.InvDeviceZToWorldZTransform[2], View.InvDeviceZToWorldZTransform[3]);
			#endif 
		#elif METAL_PROFILE && !MAC
				return FramebufferFetchES2().w;
		#else
			return ConvertFromDeviceZ(Texture2DSampleLevel(SceneDepthTexture, SceneDepthTextureSampler, ScreenUV, 0).r);
		#endif
	#endif
#endif
}

/** Returns DeviceZ which is the z value stored in the depth buffer. */
float LookupDeviceZ( float2 ScreenUV )
{
#if FEATURE_LEVEL > FEATURE_LEVEL_ES3_1
	// native Depth buffer lookup
	return Texture2DSampleLevel(SceneDepthTexture, SceneDepthTextureSampler, ScreenUV, 0).r;
#else
	#if COMPILER_GLSL_ES2
		// todo: can be optimized
		return ConvertToDeviceZ(CalcSceneDepth(ScreenUV));
	#elif METAL_PROFILE
		// todo: can be optimized
		return ConvertToDeviceZ(CalcSceneDepth(ScreenUV));
	#else
		// native Depth buffer lookup
		return Texture2DSampleLevel(SceneDepthTexture, SceneDepthTextureSampler, ScreenUV, 0).r;
	#endif
#endif
}

#if FEATURE_LEVEL >= FEATURE_LEVEL_SM4 || MOBILE_EMULATION

	// depth in in DeviceZ
	Texture2D<float> SceneDepthTextureNonMS;

	/** Returns clip space W, which is world space distance along the View Z axis. */
	float CalcSceneDepth(uint2 PixelPos)
	{
#if SCENE_TEXTURES_DISABLED
		return 0.0f;
#else
		float DeviceZ = SceneDepthTextureNonMS.Load(int3(PixelPos, 0));

		// Fetch the depth buffer Z / W value, solve for W
		return ConvertFromDeviceZ(DeviceZ);
#endif
	}
#endif

/**
* Returns scene color in rgb, depth in a
*/
float4 CalcSceneColorAndDepth( float2 ScreenUV )
{
	return float4(CalcSceneColor(ScreenUV), CalcSceneDepth(ScreenUV));
}

#if FEATURE_LEVEL >= FEATURE_LEVEL_SM4
#if FEATURE_LEVEL == FEATURE_LEVEL_SM4
	// SM4 requires texture size to be explicitly stated and expressed as the number of samples.
	Texture2DMS<float4, NumMSAASamples> SceneColorSurface;
	Texture2DMS<float, NumMSAASamples> SceneDepthSurface;
#else
	Texture2DMS<float4> SceneColorSurface;
	Texture2DMS<float> SceneDepthSurface;
#endif // FEATURE_LEVEL
	float CalcSceneDepthMSAA(float2 ScreenUV,uint SampleIndex)
	{
		int2 IntUV = int2(trunc(ScreenUV * View.BufferSizeAndInvSize.xy));
		float DeviceZ = SceneDepthSurface.Load(IntUV,SampleIndex);

		return ConvertFromDeviceZ(DeviceZ);
	}
#endif
	 
// ----------------------------

float2 ScreenPositionToBufferUV(float4 ScreenPosition)
{
	return float2(ScreenPosition.xy / ScreenPosition.w * View.ScreenPositionScaleBias.xy + View.ScreenPositionScaleBias.wz);
}

float2 SvPositionToBufferUV(float4 SvPosition)
{
	return SvPosition.xy * View.BufferSizeAndInvSize.zw;
}

// Used for post process shaders which don't need to resolve the view	
float3 SvPositionToTranslatedWorld(float4 SvPosition)
{
	float4 HomWorldPos = mul(float4(SvPosition.xyz, 1), View.SVPositionToTranslatedWorld);

	return HomWorldPos.xyz / HomWorldPos.w;
}

// Used for vertex factory shaders which need to use the resolved view
float3 SvPositionToResolvedTranslatedWorld(float4 SvPosition)
{
	float4 HomWorldPos = mul(float4(SvPosition.xyz, 1), ResolvedView.SVPositionToTranslatedWorld);

	return HomWorldPos.xyz / HomWorldPos.w;
}

// prefer to use SvPositionToTranslatedWorld() for better quality
float3 SvPositionToWorld(float4 SvPosition)
{
	return SvPositionToTranslatedWorld(SvPosition) - View.PreViewTranslation;
}

// investigate: doesn't work for usage with View.ScreenToWorld, see SvPositionToScreenPosition2()
float4 SvPositionToScreenPosition(float4 SvPosition)
{
	// todo: is already in .w or needs to be reconstructed like this:
//	SvPosition.w = ConvertFromDeviceZ(SvPosition.z);

	float2 PixelPos = SvPosition.xy - View.ViewRectMin.xy;	

	// NDC (NormalizedDeviceCoordinates, after the perspective divide)
	float3 NDCPos = float3( (PixelPos * View.ViewSizeAndInvSize.zw - 0.5f) * float2(2, -2), SvPosition.z);

	// SvPosition.w: so .w has the SceneDepth, some mobile code and the DepthFade material expression wants that
	return float4(NDCPos.xyz, 1) * SvPosition.w;
}

// Used for vertex factory shaders which need to use the resolved view
float4 SvPositionToResolvedScreenPosition(float4 SvPosition)
{
	float2 PixelPos = SvPosition.xy - ResolvedView.ViewRectMin.xy;	

	// NDC (NormalizedDeviceCoordinates, after the perspective divide)
	float3 NDCPos = float3( (PixelPos * ResolvedView.ViewSizeAndInvSize.zw - 0.5f) * float2(2, -2), SvPosition.z);

	// SvPosition.w: so .w has the SceneDepth, some mobile code and the DepthFade material expression wants that
	return float4(NDCPos.xyz, 1) * SvPosition.w;
}

float2 SvPositionToViewportUV(float4 SvPosition)
{
	// can be optimized from 2SUB+2MUL to 2MAD
	float2 PixelPos = SvPosition.xy - View.ViewRectMin.xy;	

	return PixelPos.xy * View.ViewSizeAndInvSize.zw;
}

float2 BufferUVToViewportUV(float2 BufferUV)
{
	float2 PixelPos = BufferUV.xy * View.BufferSizeAndInvSize.xy - View.ViewRectMin.xy;
	return PixelPos.xy * View.ViewSizeAndInvSize.zw;
}
// ----------------------------

/** 
 * aligns the clip space position so that it can be used as a texture coordinate
 * to properly align in screen space
 */
MaterialFloat2 ScreenAlignedPosition( float4 ScreenPosition )
{
	return MaterialFloat2(ScreenPositionToBufferUV(ScreenPosition));
}

/** 
 * Aligns the [0,1] UV to match the view within the backbuffer
 */
MaterialFloat2 ScreenAlignedUV( MaterialFloat2 UV )
{
	return (UV*MaterialFloat2(2,-2) + MaterialFloat2(-1,1))*View.ScreenPositionScaleBias.xy + View.ScreenPositionScaleBias.wz;
}

/**
 * Compute viewport coordinates from the given fragment coordinates.
 */
MaterialFloat2 GetViewportCoordinates(MaterialFloat2 InFragmentCoordinates)
{
	return InFragmentCoordinates;
}

/**
 * Unpack a normal stored in a normal map. The X and Y components are rescaled from [0,1] to [-1,1] and Z is reconstructed.
 */
MaterialFloat4 UnpackNormalMap( MaterialFloat4 TextureSample )
{
#if COMPILER_GLSL_ES2 && IOS
	return MaterialFloat4(TextureSample.rgb * 2 - 1, 1);
#else
	#if DXT5_NORMALMAPS
		MaterialFloat2 NormalXY = TextureSample.ag;
	#else
		MaterialFloat2 NormalXY = TextureSample.rg;
	#endif

	NormalXY = NormalXY * MaterialFloat2(2.0f,2.0f) - MaterialFloat2(1.0f,1.0f);
	MaterialFloat NormalZ = sqrt( saturate( 1.0f - dot( NormalXY, NormalXY ) ) );
	return MaterialFloat4( NormalXY.xy, NormalZ, 1.0f );
#endif
}

// Antialiased version of a binary comparison between ThresholdConst and a texture channel.
float AntialiasedTextureMask( Texture2D Tex, SamplerState Sampler, float2 UV, float ThresholdConst, int Channel )
{
	// By setting MaskConst to 0001, 0010, 0100 or 1000 individual channels can be chosen (the compiler should be able to optimize that).
	MaterialFloat4 MaskConst = MaterialFloat4(Channel == 0, Channel == 1, Channel == 2, Channel == 3);

	// border width in pixels, for antialiasing 1 .. 1.5 is good but 1.0 is good for optimizations 
	const float WidthConst = 1.0f;			
	float InvWidthConst = 1 / WidthConst;

	// Problem:

	// A simple texture lookup with a comparison against some thresold value allows to get a mask useful
	// for many purposes (e.g. text rendering, signs, oil/water/paint). Antialiased masks look much better
	// and mip mapping provides that but only for minification. So when the texture resolution is lower than
	// the rendering size results get blurry.

	// Idea:

	// We compute the distance to the threshold line in pixels (with subpixel precision). We can visualize
	// the problem as a heightmap that intersects a axis aligned plane at the threshold height. Only surface
	// above the threshold plane contributes to the mask. Looking at one pixel the heightmap can be approximated
	// by a plane. We can easily get the plane center value form a texture lookup and get the plane equation from
	// ddx and ddy of that value (only one value per 2x2 block) or some other more precise method. We can reduce the
	// 3d problem to 2d (looking at the steepest angle only) and the resulting value tells us how much the texture value
	// changes for one pixel. This allows us to scale and bias (threshold) the texture value the so it maps to the
	// distance function. We rescaling the distance to 0.5 coverage at the line, >1 MaterialFloat a pixel inside and <0 MaterialFloat
	// a pixel outside. Clamping this value in the range from 0..1 gives us a good approximation of the pixel coverage.

	// We tried multiple possible implementations - this is the cheapest and looks ok is most cases.
	// If quality improvements are needed we can add an option to the node later on.
	float Result;
	{
		// optimized, ddx/ddy only for every 2x2 block (bad for distant stuff)
		float Sample1 = dot(MaskConst, Texture2DSample(Tex, Sampler, UV));

		// compute the derivatives of the texture content
		float2 TexDD = float2(DDX(Sample1), DDY(Sample1));

		float TexDDLength = max(abs(TexDD.x), abs(TexDD.y)); 
		float Top = InvWidthConst * (Sample1 - ThresholdConst);
		Result = Top / TexDDLength + ThresholdConst;
	}

	Result = saturate(Result);	// no always needed (e.g. DX9 framebuffer blending)

	return Result;
}

float Noise3D_Multiplexer(int Function, float3 Position, int Quality, bool bTiling, uint RepeatSize)
{
	// verified, HLSL compiled out the switch if Function is a constant
	switch(Function)
	{
		case 0:
			return SimplexNoise3D_TEX(Position);
		case 1:
			return GradientNoise3D_TEX(Position, bTiling, RepeatSize);
		case 2:
			return FastGradientPerlinNoise3D_TEX(Position);
		case 3:
			return GradientNoise3D_ALU(Position, bTiling, RepeatSize);
		case 4:
			return ValueNoise3D_ALU(Position, bTiling, RepeatSize);
		default:
			return VoronoiNoise3D_ALU(Position, Quality, bTiling, RepeatSize, true).w * 2. - 1.;
	}
	return 0;
}

// @param LevelScale usually 2 but higher values allow efficient use of few levels
// @return in user defined range (OutputMin..OutputMax)
MaterialFloat MaterialExpressionNoise(float3 Position, float Scale, int Quality, int Function, bool bTurbulence, uint Levels, float OutputMin, float OutputMax, float LevelScale, float FilterWidth, bool bTiling, float RepeatSize)
{
	Position *= Scale;
	FilterWidth *= Scale;

	float Out = 0.0f;
	float OutScale = 1.0f;
	float InvLevelScale = 1.0f / LevelScale;
	
	LOOP for(uint i = 0; i < Levels; ++i)
	{
		// fade out noise level that are too high frequent (not done through dynamic branching as it usually requires gradient instructions)
		OutScale *= saturate(1.0 - FilterWidth);

		if(bTurbulence)
		{
			Out += abs(Noise3D_Multiplexer(Function, Position, Quality, bTiling, RepeatSize)) * OutScale;
		}
		else
		{
			Out += Noise3D_Multiplexer(Function, Position, Quality, bTiling, RepeatSize) * OutScale;
		}

		Position *= LevelScale;
		RepeatSize *= LevelScale;
		OutScale *= InvLevelScale;
		FilterWidth *= LevelScale;
	}

	if(!bTurbulence)
	{
		// bring -1..1 to 0..1 range
		Out = Out * 0.5f + 0.5f;
	}

	// Out is in 0..1 range
	return lerp(OutputMin, OutputMax, Out);
}


// Material node for noise functions returning a vector value
// @param LevelScale usually 2 but higher values allow efficient use of few levels
// @return in user defined range (OutputMin..OutputMax)
MaterialFloat4 MaterialExpressionVectorNoise(MaterialFloat3 Position, int Quality, int Function, bool bTiling, float TileSize)
{
	float4 result = float4(0,0,0,1);
	float3x4 Jacobian = JacobianSimplex_ALU(Position, bTiling, TileSize);	// compiled out if not used

	// verified, HLSL compiled out the switch if Function is a constant
	switch (Function)
	{
	case 0:	// Cellnoise
		result.xyz = float3(Rand3DPCG16(int3(floor(NoiseTileWrap(Position, bTiling, TileSize))))) / 0xffff;
		break;
	case 1: // Color noise
		result.xyz = float3(Jacobian[0].w, Jacobian[1].w, Jacobian[2].w);
		break;
	case 2: // Gradient
		result = Jacobian[0];
		break;
	case 3: // Curl
		result.xyz = float3(Jacobian[2][1] - Jacobian[1][2], Jacobian[0][2] - Jacobian[2][0], Jacobian[1][0] - Jacobian[0][1]);
		break;
	default: // Voronoi
		result = VoronoiNoise3D_ALU(Position, Quality, bTiling, TileSize, false);
		break;
	}
	return result;
}


/*
* Clips a ray to an AABB.  Does not handle rays parallel to any of the planes.
*
* @param RayOrigin - The origin of the ray in world space.
* @param RayEnd - The end of the ray in world space.  
* @param BoxMin - The minimum extrema of the box.
* @param BoxMax - The maximum extrema of the box.
* @return - Returns the closest intersection along the ray in x, and furthest in y.  
*			If the ray did not intersect the box, then the furthest intersection <= the closest intersection.
*			The intersections will always be in the range [0,1], which corresponds to [RayOrigin, RayEnd] in worldspace.
*			To find the world space position of either intersection, simply plug it back into the ray equation:
*			WorldPos = RayOrigin + (RayEnd - RayOrigin) * Intersection;
*/
float2 LineBoxIntersect(float3 RayOrigin, float3 RayEnd, float3 BoxMin, float3 BoxMax)
{
	float3 InvRayDir = 1.0f / (RayEnd - RayOrigin);
	
	//find the ray intersection with each of the 3 planes defined by the minimum extrema.
	float3 FirstPlaneIntersections = (BoxMin - RayOrigin) * InvRayDir;
	//find the ray intersection with each of the 3 planes defined by the maximum extrema.
	float3 SecondPlaneIntersections = (BoxMax - RayOrigin) * InvRayDir;
	//get the closest of these intersections along the ray
	float3 ClosestPlaneIntersections = min(FirstPlaneIntersections, SecondPlaneIntersections);
	//get the furthest of these intersections along the ray
	float3 FurthestPlaneIntersections = max(FirstPlaneIntersections, SecondPlaneIntersections);

	float2 BoxIntersections;
	//find the furthest near intersection
	BoxIntersections.x = max(ClosestPlaneIntersections.x, max(ClosestPlaneIntersections.y, ClosestPlaneIntersections.z));
	//find the closest far intersection
	BoxIntersections.y = min(FurthestPlaneIntersections.x, min(FurthestPlaneIntersections.y, FurthestPlaneIntersections.z));
	//clamp the intersections to be between RayOrigin and RayEnd on the ray
	return saturate(BoxIntersections);
}

/** Computes distance from an AABB to a point in space. */
MaterialFloat ComputeDistanceFromBoxToPoint(MaterialFloat3 Mins, MaterialFloat3 Maxs, MaterialFloat3 InPoint)
{
	MaterialFloat3 DistancesToMin = InPoint < Mins ? abs(InPoint - Mins) : 0;
	MaterialFloat3 DistancesToMax = InPoint > Maxs ? abs(InPoint - Maxs) : 0;

	//@todo - this is actually incorrect, it gives manhattan distance
	MaterialFloat Distance = dot(DistancesToMin, 1);
	Distance += dot(DistancesToMax, 1);
	return Distance;
}

/** Computes squared distance from a point in space to an AABB. */
MaterialFloat ComputeSquaredDistanceFromBoxToPoint(MaterialFloat3 BoxCenter, MaterialFloat3 BoxExtent, MaterialFloat3 InPoint)
{
	MaterialFloat3 AxisDistances = max(abs(InPoint - BoxCenter) - BoxExtent, 0);
	return dot(AxisDistances, AxisDistances);
}

/** Computes distance from point inside an AABB to the AABB's surface. */
float ComputeDistanceFromBoxToPointInside(float3 BoxCenter, float3 BoxExtent, float3 InPoint)
{
	float3 DistancesToMin = max(InPoint - BoxCenter + BoxExtent, 0);
	float3 DistancesToMax = max(BoxCenter + BoxExtent - InPoint, 0);
	float3 ClosestDistances = min(DistancesToMin, DistancesToMax);
	return min(ClosestDistances.x, min(ClosestDistances.y, ClosestDistances.z));
}

bool RayHitSphere(float3 RayOrigin, float3 UnitRayDirection, float3 SphereCenter, float SphereRadius)
{
	float3 ClosestPointOnRay = max(0, dot(SphereCenter - RayOrigin, UnitRayDirection)) * UnitRayDirection;
	float3 CenterToRay = RayOrigin + ClosestPointOnRay - SphereCenter;
	return dot(CenterToRay, CenterToRay) <= Square(SphereRadius);
}

bool RaySegmentHitSphere(float3 RayOrigin, float3 UnitRayDirection, float RayLength, float3 SphereCenter, float SphereRadius)
{
	float DistanceAlongRay = dot(SphereCenter - RayOrigin, UnitRayDirection);
	float3 ClosestPointOnRay = DistanceAlongRay * UnitRayDirection;
	float3 CenterToRay = RayOrigin + ClosestPointOnRay - SphereCenter;
	return dot(CenterToRay, CenterToRay) <= Square(SphereRadius) && DistanceAlongRay > -SphereRadius && DistanceAlongRay - SphereRadius < RayLength;
}

/** Transforms a vector from tangent space to world space */
MaterialFloat3 TransformTangentVectorToWorld(MaterialFloat3x3 TangentToWorld, MaterialFloat3 InTangentVector)
{
	// Transform directly to world space
	// The vector transform is optimized for this case, only one vector-matrix multiply is needed
	return mul(InTangentVector, TangentToWorld);
}

/** Transforms a vector from world space to tangent space */
MaterialFloat3 TransformWorldVectorToTangent(MaterialFloat3x3 TangentToWorld, MaterialFloat3 InWorldVector)
{
	// Transform from world to tangent space with the transpose of TangentToWorld (achieved by swapping vector / matrix multiply order)
	// Note that the transpose is only equal to the inverse for orthonormal matrices - aka only uniform scaling
	return mul(TangentToWorld, InWorldVector);
}

float3 TransformWorldVectorToView(float3 InTangentVector)
{
	// Transform from world to view space
	return mul(InTangentVector, (float3x3)ResolvedView.TranslatedWorldToView);
}

/** Computes the distance from the center to the edge of an AABB with the given extents in the given direction. */
MaterialFloat GetBoxPushout(MaterialFloat3 Normal,MaterialFloat3 Extent)
{
	return dot(abs(Normal * Extent), MaterialFloat3(1.0f, 1.0f, 1.0f));
}

/** Generates arbitrary but valid perpendicular unit vectors to ZAxis.  ZAxis should be unit length. */
void GenerateCoordinateSystem(float3 ZAxis, out float3 XAxis, out float3 YAxis)
{
	if (abs(ZAxis.x) > abs(ZAxis.y))
	{
		float InverseLength = 1.0f / sqrt(dot(ZAxis.xz, ZAxis.xz));
		XAxis = float3(-ZAxis.z * InverseLength, 0.0f, ZAxis.x * InverseLength);
	}
	else
	{
		float InverseLength = 1.0f / sqrt(dot(ZAxis.yz, ZAxis.yz));
		XAxis = float3(0.0f, ZAxis.z * InverseLength, -ZAxis.y * InverseLength);
	}

	YAxis = cross(ZAxis, XAxis);
}

// Define passthrough implementations of EvaluateAttributeAtSample for non-D3D11 platforms.
#if !SM5_PROFILE
	float EvaluateAttributeAtSample(float Attribute,uint SampleIndex) { return Attribute; }
	float2 EvaluateAttributeAtSample(float2 Attribute,uint SampleIndex) { return Attribute; }
	float3 EvaluateAttributeAtSample(float3 Attribute,uint SampleIndex) { return Attribute; }
	float4 EvaluateAttributeAtSample(float4 Attribute,uint SampleIndex) { return Attribute; }
#endif

/** Output of the screen vertex shader. */
struct FScreenVertexOutput
{
#if METAL_PROFILE || COMPILER_GLSL_ES3_1
	//@todo-rco: FIXME!
	noperspective float2 UV : TEXCOORD0;
#else
	noperspective MaterialFloat2 UV : TEXCOORD0;
#endif
	float4 Position : SV_POSITION;
};


// for velocity rendering, motionblur and temporal AA
// velocity needs to support -2..2 screen space range for x and y
// texture is 16bit 0..1 range per channel
float2 EncodeVelocityToTexture(float2 In)
{
	// 0.499f is a value smaller than 0.5f to avoid using the full range to use the clear color (0,0) as special value
	// 0.5f to allow for a range of -2..2 instead of -1..1 for really fast motions for temporal AA
	return In * (0.499f * 0.5f) + 32767.0f / 65535.0f;
}
// see EncodeVelocityToTexture()
float2 DecodeVelocityFromTexture(float2 In)
{
	const float InvDiv = 1.0f / (0.499f * 0.5f);
	// reference
//	return (In - 32767.0f / 65535.0f ) / (0.499f * 0.5f);
	// MAD layout to help compiler
	return In * InvDiv - 32767.0f / 65535.0f * InvDiv;
}

// Used for the Global Illumination in the GIReplace material expression
bool GetGIReplaceState()
{
#if REFLECTIVE_SHADOW_MAP
	return true;
#else
	return false;
#endif
}


#if FEATURE_LEVEL >= FEATURE_LEVEL_SM4
struct FWriteToSliceGeometryOutput
{
	FScreenVertexOutput Vertex;
	uint LayerIndex : SV_RenderTargetArrayIndex;
};
#endif

// Helper macro to globally ignore requests for non-offset world positions in materials when lower than shader model 4. We do this
// because we are using an extra interpolator for this second world position, and in < ES31 there may not be enough
#define USE_WORLD_POSITION_EXCLUDING_SHADER_OFFSETS	(NEEDS_WORLD_POSITION_EXCLUDING_SHADER_OFFSETS && FEATURE_LEVEL >= FEATURE_LEVEL_ES3_1)

/** Used for calculating vertex positions and UVs when drawing with DrawRectangle */
void DrawRectangle( in float4 InPosition, in float2 InTexCoord, out float4 OutPosition, out float2 OutTexCoord)
{
	OutPosition = InPosition;
	OutPosition.xy = -1.0f + 2.0f * (DrawRectangleParameters.PosScaleBias.zw + (InPosition.xy * DrawRectangleParameters.PosScaleBias.xy)) * DrawRectangleParameters.InvTargetSizeAndTextureSize.xy;
	OutPosition.xy *= float2( 1, -1 );
	OutTexCoord.xy = (DrawRectangleParameters.UVScaleBias.zw + (InTexCoord.xy * DrawRectangleParameters.UVScaleBias.xy)) * DrawRectangleParameters.InvTargetSizeAndTextureSize.zw;
}

/** Used for calculating vertex positions when drawing with DrawRectangle */
void DrawRectangle( in float4 InPosition, out float4 OutPosition)
{
	OutPosition = InPosition;
	OutPosition.xy = -1.0f + 2.0f * (DrawRectangleParameters.PosScaleBias.zw + (InPosition.xy * DrawRectangleParameters.PosScaleBias.xy)) * DrawRectangleParameters.InvTargetSizeAndTextureSize.xy;
	OutPosition.xy *= float2( 1, -1 );
}

//Since some platforms don't remove Nans in saturate calls, 
//SafeSaturate function will remove nan/inf.    
//Can be expensive, only call when there's a good reason to expect Nans.
//D3D saturate actually turns Nans -> 1  since it does the min(x, 1) first, and D3D nan rules specify the non-nand operand wins in such a case.  
//See: http://msdn.microsoft.com/en-us/library/windows/desktop/jj218760(v=vs.85).aspx  
#define SafeSaturate_Def(type)\
type SafeSaturate(type In) \
{\
	return saturate(In);\
}

SafeSaturate_Def(float)
SafeSaturate_Def(float2)
SafeSaturate_Def(float3)
SafeSaturate_Def(float4)

// Experimental way to allow adjusting the OpacityMask for shadow map rendering of masked materials.
// Can be accessed with a Custom material node. If this turns out to be very useful we can expose as MaterialFunction
// and potentially expose other queries as well (e.g. SkeletalMesh, HitProxy, ).
// @return 0:no, 1:yes
float IsShadowDepthShader()
{
#ifdef SHADOW_DEPTH_SHADER
	return 1;
#else
	return 0;
#endif
}
#define TERRAIN_ZSCALE (1.0f/128.0f)

// Decodes a value which was packed into two 8 bit channels
float DecodePackedTwoChannelValue(float2 PackedHeight)
{
	return PackedHeight.x * 255.0 * 256.0 + PackedHeight.y * 255.0;
}

float DecodeHeightValue(float InValue)
{
	return (InValue - 32768.0) * TERRAIN_ZSCALE;
}

float DecodePackedHeight(float2 PackedHeight)
{
	return DecodeHeightValue(DecodePackedTwoChannelValue(PackedHeight));
}

uint ReverseBits32( uint bits )
{
#if SM5_PROFILE && !METAL_SM5_PROFILE
	return reversebits( bits );
#else
	bits = ( bits << 16) | ( bits >> 16);
	bits = ( (bits & 0x00ff00ff) << 8 ) | ( (bits & 0xff00ff00) >> 8 );
	bits = ( (bits & 0x0f0f0f0f) << 4 ) | ( (bits & 0xf0f0f0f0) >> 4 );
	bits = ( (bits & 0x33333333) << 2 ) | ( (bits & 0xcccccccc) >> 2 );
	bits = ( (bits & 0x55555555) << 1 ) | ( (bits & 0xaaaaaaaa) >> 1 );
	return bits;
#endif
}

SamplerState GetBilinearWrappedSampler()
{
#if PS4_PROFILE
	sce::Gnm::Sampler BuiltinSampler = GetDefaultBuiltinSampler();
	BuiltinSampler.setWrapMode(sce::Gnm::kWrapModeWrap, sce::Gnm::kWrapModeWrap, sce::Gnm::kWrapModeWrap);
	BuiltinSampler.setXyFilterMode(sce::Gnm::kFilterModeBilinear, sce::Gnm::kFilterModeBilinear);
	BuiltinSampler.setZFilterMode(sce::Gnm::kZFilterModeLinear);
	BuiltinSampler.setMipFilterMode(sce::Gnm::kMipFilterModePoint);
	return SamplerState(BuiltinSampler);
#else
	return BuiltinSamplers.Bilinear;
#endif
}

SamplerState GetBilinearClampedSampler()
{
#if PS4_PROFILE
	sce::Gnm::Sampler BuiltinSampler = GetDefaultBuiltinSampler();
	BuiltinSampler.setXyFilterMode(sce::Gnm::kFilterModeBilinear, sce::Gnm::kFilterModeBilinear);
	BuiltinSampler.setMipFilterMode(sce::Gnm::kMipFilterModePoint);
	BuiltinSampler.setZFilterMode(sce::Gnm::kZFilterModeLinear);
	BuiltinSampler.setWrapMode(sce::Gnm::kWrapModeClampLastTexel, sce::Gnm::kWrapModeClampLastTexel, sce::Gnm::kWrapModeClampLastTexel);
	return SamplerState(BuiltinSampler);
#else
	return BuiltinSamplers.BilinearClamped;
#endif
}

SamplerState GetPointWrappedSampler()
{
#if PS4_PROFILE
	sce::Gnm::Sampler BuiltinSampler = GetDefaultBuiltinSampler();
	BuiltinSampler.setXyFilterMode(sce::Gnm::kFilterModePoint, sce::Gnm::kFilterModePoint);
	BuiltinSampler.setMipFilterMode(sce::Gnm::kMipFilterModePoint);
	BuiltinSampler.setZFilterMode(sce::Gnm::kZFilterModePoint);
	BuiltinSampler.setWrapMode(sce::Gnm::kWrapModeWrap, sce::Gnm::kWrapModeWrap, sce::Gnm::kWrapModeWrap);
	return SamplerState(BuiltinSampler);
#else
	return BuiltinSamplers.Point;
#endif
}

SamplerState GetPointClampedSampler()
{
#if PS4_PROFILE
	sce::Gnm::Sampler BuiltinSampler = GetDefaultBuiltinSampler();
	BuiltinSampler.setXyFilterMode(sce::Gnm::kFilterModePoint, sce::Gnm::kFilterModePoint);
	BuiltinSampler.setMipFilterMode(sce::Gnm::kMipFilterModePoint);
	BuiltinSampler.setZFilterMode(sce::Gnm::kZFilterModePoint);
	BuiltinSampler.setWrapMode(sce::Gnm::kWrapModeClampLastTexel, sce::Gnm::kWrapModeClampLastTexel, sce::Gnm::kWrapModeClampLastTexel);
	return SamplerState(BuiltinSampler);
#else
	return BuiltinSamplers.PointClamped;
#endif
}

SamplerState GetTrilinearWrappedSampler()
{
#if PS4_PROFILE
	sce::Gnm::Sampler BuiltinSampler = GetDefaultBuiltinSampler();
	BuiltinSampler.setXyFilterMode(sce::Gnm::kFilterModeBilinear, sce::Gnm::kFilterModeBilinear);
	BuiltinSampler.setMipFilterMode(sce::Gnm::kMipFilterModeLinear);
	BuiltinSampler.setZFilterMode(sce::Gnm::kZFilterModeLinear);
	BuiltinSampler.setWrapMode(sce::Gnm::kWrapModeWrap, sce::Gnm::kWrapModeWrap, sce::Gnm::kWrapModeWrap);
	return SamplerState(BuiltinSampler);
#else
	return BuiltinSamplers.Trilinear;
#endif
}

SamplerState GetTrilinearClampedSampler()
{
#if PS4_PROFILE
	sce::Gnm::Sampler BuiltinSampler = GetDefaultBuiltinSampler();
	BuiltinSampler.setXyFilterMode(sce::Gnm::kFilterModeBilinear, sce::Gnm::kFilterModeBilinear);
	BuiltinSampler.setMipFilterMode(sce::Gnm::kMipFilterModeLinear);
	BuiltinSampler.setZFilterMode(sce::Gnm::kZFilterModeLinear);
	BuiltinSampler.setWrapMode(sce::Gnm::kWrapModeClampLastTexel, sce::Gnm::kWrapModeClampLastTexel, sce::Gnm::kWrapModeClampLastTexel);
	return SamplerState(BuiltinSampler);
#else
	return BuiltinSamplers.TrilinearClamped;
#endif
}

// see PixelShaderOutputCommon
struct FPixelShaderIn
{
	// read only
	float4 SvPosition;

	// Pixel Shader InCoverage, only usable if PIXELSHADEROUTPUT_COVERAGE is 1
	uint Coverage;

	//
	bool bIsFrontFace;
};
// see PixelShaderOutputCommon
struct FPixelShaderOut
{
	// [0..7], only usable if PIXELSHADEROUTPUT_MRT0, PIXELSHADEROUTPUT_MRT1, ... is 1
	float4 MRT[8];

	// Pixel Shader OutCoverage, only usable if PIXELSHADEROUTPUT_COVERAGE is 1
	uint Coverage;

	// Pixel Shader OutDepth
	float Depth;
};