#pragma warning(disable : 3568)
#pragma exclude_renderers gles gles3 d3d11_9x

#include "../StdLib.hlsl"
#include "../Colors.hlsl"

RWStructuredBuffer<uint4> _WaveformBuffer;
Texture2D<float4> _Source;
SamplerState sampler_Source;

CBUFFER_START(Params)
    float4 _Params; // x: source width, y: source height, z: linear, w: unused
CBUFFER_END

#define GROUP_SIZE 256
#define GROUP_SIZE_X 16
#define GROUP_SIZE_Y 16

#ifdef DISABLE_COMPUTE_SHADERS

TRIVIAL_COMPUTE_KERNEL(KWaveformGather)
TRIVIAL_COMPUTE_KERNEL(KWaveformClear)

#else

#pragma kernel KWaveformGather
[numthreads(1, GROUP_SIZE, 1)]
void KWaveformGather(uint2 dispatchThreadId : SV_DispatchThreadID)
{
    // Gather local group histogram
    if (dispatchThreadId.x < uint(_Params.x) && dispatchThreadId.y < uint(_Params.y))
    {
        float3 color = _Source[dispatchThreadId].rgb;
        color = saturate(color);

        // We want a gamma-corrected histogram (like Photoshop & all)
        if (_Params.z > 0)
            color = LinearToSRGB(color);

        // Convert channel values to histogram bins
        uint3 idx = (uint3)(round(color * (_Params.y - 1)));
        idx += dispatchThreadId.x * _Params.y;

        if (idx.x > 0u) InterlockedAdd(_WaveformBuffer[idx.x].x, 1u); // Red
        if (idx.y > 0u) InterlockedAdd(_WaveformBuffer[idx.y].y, 1u); // Green
        if (idx.z > 0u) InterlockedAdd(_WaveformBuffer[idx.z].z, 1u); // Blue
    }
}

#pragma kernel KWaveformClear
[numthreads(GROUP_SIZE_X, GROUP_SIZE_Y, 1)]
void KWaveformClear(uint2 dispatchThreadId : SV_DispatchThreadID)
{
    if (dispatchThreadId.x < uint(_Params.x) && dispatchThreadId.y < uint(_Params.y))
        _WaveformBuffer[dispatchThreadId.y * uint(_Params.x) + dispatchThreadId.x] = uint4(0u, 0u, 0u, 0u);
}

#endif // DISABLE_COMPUTE_SHADERS