232 lines
11 KiB
HLSL
232 lines
11 KiB
HLSL
//--------------------------------------------------------------------------------------------------------------------------------
|
|
// Cartoon FX
|
|
// (c) 2012-2020 Jean Moreno
|
|
//--------------------------------------------------------------------------------------------------------------------------------
|
|
|
|
// Copy of URP specific variables needed for lighting
|
|
|
|
// ================================================================================================================================
|
|
// Input.hlsl:
|
|
// ================================================================================================================================
|
|
|
|
#if defined(SHADER_API_MOBILE) || (defined(SHADER_API_GLCORE) && !defined(SHADER_API_SWITCH)) || defined(SHADER_API_GLES) || defined(SHADER_API_GLES3) // Workaround for bug on Nintendo Switch where SHADER_API_GLCORE is mistakenly defined
|
|
#define MAX_VISIBLE_LIGHTS 32
|
|
#else
|
|
#define MAX_VISIBLE_LIGHTS 256
|
|
#endif
|
|
|
|
// --------------------------------
|
|
|
|
float4 _MainLightPosition;
|
|
half4 _MainLightColor;
|
|
|
|
// --------------------------------
|
|
|
|
half4 _AdditionalLightsCount;
|
|
#if USE_STRUCTURED_BUFFER_FOR_LIGHT_DATA
|
|
StructuredBuffer<LightData> _AdditionalLightsBuffer;
|
|
StructuredBuffer<int> _AdditionalLightsIndices;
|
|
#else
|
|
// GLES3 causes a performance regression in some devices when using CBUFFER.
|
|
#ifndef SHADER_API_GLES3
|
|
CBUFFER_START(AdditionalLights)
|
|
#endif
|
|
float4 _AdditionalLightsPosition[MAX_VISIBLE_LIGHTS];
|
|
half4 _AdditionalLightsColor[MAX_VISIBLE_LIGHTS];
|
|
half4 _AdditionalLightsAttenuation[MAX_VISIBLE_LIGHTS];
|
|
half4 _AdditionalLightsSpotDir[MAX_VISIBLE_LIGHTS];
|
|
half4 _AdditionalLightsOcclusionProbes[MAX_VISIBLE_LIGHTS];
|
|
#ifndef SHADER_API_GLES3
|
|
CBUFFER_END
|
|
#endif
|
|
#endif
|
|
|
|
// ================================================================================================================================
|
|
// UnityInput.hlsl:
|
|
// ================================================================================================================================
|
|
|
|
half4 unity_LightData;
|
|
half4 unity_LightIndices[2];
|
|
|
|
// --------------------------------
|
|
|
|
// ================================================================================================================================
|
|
// Macros.hlsl
|
|
// ================================================================================================================================
|
|
|
|
#define HALF_MIN 6.103515625e-5 // 2^-14, the same value for 10, 11 and 16-bit: https://www.khronos.org/opengl/wiki/Small_Float_Formats
|
|
|
|
// ================================================================================================================================
|
|
// Lighting.hlsl
|
|
// ================================================================================================================================
|
|
|
|
// Abstraction over Light shading data.
|
|
struct Light
|
|
{
|
|
half3 direction;
|
|
half3 color;
|
|
half distanceAttenuation;
|
|
half shadowAttenuation;
|
|
};
|
|
|
|
// Matches Unity Vanila attenuation
|
|
// Attenuation smoothly decreases to light range.
|
|
float DistanceAttenuation(float distanceSqr, half2 distanceAttenuation)
|
|
{
|
|
// We use a shared distance attenuation for additional directional and puctual lights
|
|
// for directional lights attenuation will be 1
|
|
float lightAtten = rcp(distanceSqr);
|
|
|
|
#if SHADER_HINT_NICE_QUALITY
|
|
// Use the smoothing factor also used in the Unity lightmapper.
|
|
half factor = distanceSqr * distanceAttenuation.x;
|
|
half smoothFactor = saturate(1.0h - factor * factor);
|
|
smoothFactor = smoothFactor * smoothFactor;
|
|
#else
|
|
// We need to smoothly fade attenuation to light range. We start fading linearly at 80% of light range
|
|
// Therefore:
|
|
// fadeDistance = (0.8 * 0.8 * lightRangeSq)
|
|
// smoothFactor = (lightRangeSqr - distanceSqr) / (lightRangeSqr - fadeDistance)
|
|
// We can rewrite that to fit a MAD by doing
|
|
// distanceSqr * (1.0 / (fadeDistanceSqr - lightRangeSqr)) + (-lightRangeSqr / (fadeDistanceSqr - lightRangeSqr)
|
|
// distanceSqr * distanceAttenuation.y + distanceAttenuation.z
|
|
half smoothFactor = saturate(distanceSqr * distanceAttenuation.x + distanceAttenuation.y);
|
|
#endif
|
|
|
|
return lightAtten * smoothFactor;
|
|
}
|
|
|
|
half AngleAttenuation(half3 spotDirection, half3 lightDirection, half2 spotAttenuation)
|
|
{
|
|
// Spot Attenuation with a linear falloff can be defined as
|
|
// (SdotL - cosOuterAngle) / (cosInnerAngle - cosOuterAngle)
|
|
// This can be rewritten as
|
|
// invAngleRange = 1.0 / (cosInnerAngle - cosOuterAngle)
|
|
// SdotL * invAngleRange + (-cosOuterAngle * invAngleRange)
|
|
// SdotL * spotAttenuation.x + spotAttenuation.y
|
|
|
|
// If we precompute the terms in a MAD instruction
|
|
half SdotL = dot(spotDirection, lightDirection);
|
|
half atten = saturate(SdotL * spotAttenuation.x + spotAttenuation.y);
|
|
return atten * atten;
|
|
}
|
|
|
|
// Fills a light struct given a perObjectLightIndex
|
|
Light GetAdditionalPerObjectLight(int perObjectLightIndex, float3 positionWS)
|
|
{
|
|
// Abstraction over Light input constants
|
|
#if USE_STRUCTURED_BUFFER_FOR_LIGHT_DATA
|
|
float4 lightPositionWS = _AdditionalLightsBuffer[perObjectLightIndex].position;
|
|
half3 color = _AdditionalLightsBuffer[perObjectLightIndex].color.rgb;
|
|
half4 distanceAndSpotAttenuation = _AdditionalLightsBuffer[perObjectLightIndex].attenuation;
|
|
half4 spotDirection = _AdditionalLightsBuffer[perObjectLightIndex].spotDirection;
|
|
half4 lightOcclusionProbeInfo = _AdditionalLightsBuffer[perObjectLightIndex].occlusionProbeChannels;
|
|
#else
|
|
float4 lightPositionWS = _AdditionalLightsPosition[perObjectLightIndex];
|
|
half3 color = _AdditionalLightsColor[perObjectLightIndex].rgb;
|
|
half4 distanceAndSpotAttenuation = _AdditionalLightsAttenuation[perObjectLightIndex];
|
|
half4 spotDirection = _AdditionalLightsSpotDir[perObjectLightIndex];
|
|
half4 lightOcclusionProbeInfo = _AdditionalLightsOcclusionProbes[perObjectLightIndex];
|
|
#endif
|
|
|
|
// Directional lights store direction in lightPosition.xyz and have .w set to 0.0.
|
|
// This way the following code will work for both directional and punctual lights.
|
|
float3 lightVector = lightPositionWS.xyz - positionWS * lightPositionWS.w;
|
|
float distanceSqr = max(dot(lightVector, lightVector), HALF_MIN);
|
|
|
|
half3 lightDirection = half3(lightVector * rsqrt(distanceSqr));
|
|
half attenuation = DistanceAttenuation(distanceSqr, distanceAndSpotAttenuation.xy) * AngleAttenuation(spotDirection.xyz, lightDirection, distanceAndSpotAttenuation.zw);
|
|
|
|
Light light;
|
|
light.direction = lightDirection;
|
|
light.distanceAttenuation = attenuation;
|
|
/// light.shadowAttenuation = AdditionalLightRealtimeShadow(perObjectLightIndex, positionWS);
|
|
light.shadowAttenuation = 1;
|
|
light.color = color;
|
|
|
|
// In case we're using light probes, we can sample the attenuation from the `unity_ProbesOcclusion`
|
|
#if defined(LIGHTMAP_ON) || defined(_MIXED_LIGHTING_SUBTRACTIVE)
|
|
// First find the probe channel from the light.
|
|
// Then sample `unity_ProbesOcclusion` for the baked occlusion.
|
|
// If the light is not baked, the channel is -1, and we need to apply no occlusion.
|
|
|
|
// probeChannel is the index in 'unity_ProbesOcclusion' that holds the proper occlusion value.
|
|
int probeChannel = lightOcclusionProbeInfo.x;
|
|
|
|
// lightProbeContribution is set to 0 if we are indeed using a probe, otherwise set to 1.
|
|
half lightProbeContribution = lightOcclusionProbeInfo.y;
|
|
|
|
half probeOcclusionValue = unity_ProbesOcclusion[probeChannel];
|
|
light.distanceAttenuation *= max(probeOcclusionValue, lightProbeContribution);
|
|
#endif
|
|
|
|
return light;
|
|
}
|
|
|
|
uint GetPerObjectLightIndexOffset()
|
|
{
|
|
#if USE_STRUCTURED_BUFFER_FOR_LIGHT_DATA
|
|
return unity_LightData.x;
|
|
#else
|
|
return 0;
|
|
#endif
|
|
}
|
|
|
|
// Returns a per-object index given a loop index.
|
|
// This abstract the underlying data implementation for storing lights/light indices
|
|
int GetPerObjectLightIndex(uint index)
|
|
{
|
|
/////////////////////////////////////////////////////////////////////////////////////////////
|
|
// Structured Buffer Path /
|
|
// /
|
|
// Lights and light indices are stored in StructuredBuffer. We can just index them. /
|
|
// Currently all non-mobile platforms take this path :( /
|
|
// There are limitation in mobile GPUs to use SSBO (performance / no vertex shader support) /
|
|
/////////////////////////////////////////////////////////////////////////////////////////////
|
|
#if USE_STRUCTURED_BUFFER_FOR_LIGHT_DATA
|
|
uint offset = unity_LightData.x;
|
|
return _AdditionalLightsIndices[offset + index];
|
|
|
|
/////////////////////////////////////////////////////////////////////////////////////////////
|
|
// UBO path /
|
|
// /
|
|
// We store 8 light indices in float4 unity_LightIndices[2]; /
|
|
// Due to memory alignment unity doesn't support int[] or float[] /
|
|
// Even trying to reinterpret cast the unity_LightIndices to float[] won't work /
|
|
// it will cast to float4[] and create extra register pressure. :( /
|
|
/////////////////////////////////////////////////////////////////////////////////////////////
|
|
#elif !defined(SHADER_API_GLES)
|
|
// since index is uint shader compiler will implement
|
|
// div & mod as bitfield ops (shift and mask).
|
|
|
|
// TODO: Can we index a float4? Currently compiler is
|
|
// replacing unity_LightIndicesX[i] with a dp4 with identity matrix.
|
|
// u_xlat16_40 = dot(unity_LightIndices[int(u_xlatu13)], ImmCB_0_0_0[u_xlati1]);
|
|
// This increases both arithmetic and register pressure.
|
|
return unity_LightIndices[index / 4][index % 4];
|
|
#else
|
|
// Fallback to GLES2. No bitfield magic here :(.
|
|
// We limit to 4 indices per object and only sample unity_4LightIndices0.
|
|
// Conditional moves are branch free even on mali-400
|
|
// small arithmetic cost but no extra register pressure from ImmCB_0_0_0 matrix.
|
|
half2 lightIndex2 = (index < 2.0h) ? unity_LightIndices[0].xy : unity_LightIndices[0].zw;
|
|
half i_rem = (index < 2.0h) ? index : index - 2.0h;
|
|
return (i_rem < 1.0h) ? lightIndex2.x : lightIndex2.y;
|
|
#endif
|
|
}
|
|
|
|
// Fills a light struct given a loop i index. This will convert the i
|
|
// index to a perObjectLightIndex
|
|
Light GetAdditionalLight(uint i, float3 positionWS)
|
|
{
|
|
int perObjectLightIndex = GetPerObjectLightIndex(i);
|
|
return GetAdditionalPerObjectLight(perObjectLightIndex, positionWS);
|
|
}
|
|
|
|
int GetAdditionalLightsCount()
|
|
{
|
|
// TODO: we need to expose in SRP api an ability for the pipeline cap the amount of lights
|
|
// in the culling. This way we could do the loop branch with an uniform
|
|
// This would be helpful to support baking exceeding lights in SH as well
|
|
return min(_AdditionalLightsCount.x, unity_LightData.y);
|
|
} |