godot/thirdparty/cvtt/ConvectionKernels_BC67.h
K. S. Ernest (iFire) Lee 419b342a9a Faster CVTT by reducing quality.
Make BC6 and BC7 CVTT faster while still having better quality than DXT5.
2022-02-04 15:15:26 -08:00

100 lines
5.8 KiB
C++

#pragma once
#include "ConvectionKernels_ParallelMath.h"
namespace cvtt
{
namespace Tables
{
namespace BC7SC
{
struct Table;
}
}
namespace Internal
{
namespace BC67
{
struct WorkInfo;
}
template<int TVectorSize>
class IndexSelectorHDR;
}
struct PixelBlockU8;
}
namespace cvtt
{
namespace Internal
{
class BC7Computer
{
public:
static void Pack(uint32_t flags, const PixelBlockU8* inputs, uint8_t* packedBlocks, const float channelWeights[4], const BC7EncodingPlan &encodingPlan, int numRefineRounds);
static void UnpackOne(PixelBlockU8 &output, const uint8_t* packedBlock);
private:
static const int MaxTweakRounds = 4;
typedef ParallelMath::SInt16 MSInt16;
typedef ParallelMath::UInt15 MUInt15;
typedef ParallelMath::UInt16 MUInt16;
typedef ParallelMath::SInt32 MSInt32;
typedef ParallelMath::Float MFloat;
static void TweakAlpha(const MUInt15 original[2], int tweak, int range, MUInt15 result[2]);
static void Quantize(MUInt15* color, int bits, int channels);
static void QuantizeP(MUInt15* color, int bits, uint16_t p, int channels);
static void Unquantize(MUInt15* color, int bits, int channels);
static void CompressEndpoints0(MUInt15 ep[2][4], uint16_t p[2]);
static void CompressEndpoints1(MUInt15 ep[2][4], uint16_t p);
static void CompressEndpoints2(MUInt15 ep[2][4]);
static void CompressEndpoints3(MUInt15 ep[2][4], uint16_t p[2]);
static void CompressEndpoints4(MUInt15 epRGB[2][3], MUInt15 epA[2]);
static void CompressEndpoints5(MUInt15 epRGB[2][3], MUInt15 epA[2]);
static void CompressEndpoints6(MUInt15 ep[2][4], uint16_t p[2]);
static void CompressEndpoints7(MUInt15 ep[2][4], uint16_t p[2]);
static void TrySingleColorRGBAMultiTable(uint32_t flags, const MUInt15 pixels[16][4], const MFloat average[4], int numRealChannels, const uint8_t *fragmentStart, int shapeLength, const MFloat &staticAlphaError, const ParallelMath::Int16CompFlag punchThroughInvalid[4], MFloat& shapeBestError, MUInt15 shapeBestEP[2][4], MUInt15 *fragmentBestIndexes, const float *channelWeightsSq, const cvtt::Tables::BC7SC::Table*const* tables, int numTables, const ParallelMath::RoundTowardNearestForScope *rtn);
static void TrySinglePlane(uint32_t flags, const MUInt15 pixels[16][4], const MFloat floatPixels[16][4], const float channelWeights[4], const BC7EncodingPlan &encodingPlan, int numRefineRounds, BC67::WorkInfo& work, const ParallelMath::RoundTowardNearestForScope *rtn);
static void TryDualPlane(uint32_t flags, const MUInt15 pixels[16][4], const MFloat floatPixels[16][4], const float channelWeights[4], const BC7EncodingPlan &encodingPlan, int numRefineRounds, BC67::WorkInfo& work, const ParallelMath::RoundTowardNearestForScope *rtn);
template<class T>
static void Swap(T& a, T& b);
};
class BC6HComputer
{
public:
static void Pack(uint32_t flags, const PixelBlockF16* inputs, uint8_t* packedBlocks, const float channelWeights[4], bool isSigned, int numTweakRounds, int numRefineRounds);
static void UnpackOne(PixelBlockF16 &output, const uint8_t *pBC, bool isSigned);
private:
typedef ParallelMath::Float MFloat;
typedef ParallelMath::SInt16 MSInt16;
typedef ParallelMath::UInt16 MUInt16;
typedef ParallelMath::UInt15 MUInt15;
typedef ParallelMath::AInt16 MAInt16;
typedef ParallelMath::SInt32 MSInt32;
typedef ParallelMath::UInt31 MUInt31;
static const int MaxTweakRounds = 4;
static const int MaxRefineRounds = 3;
static MSInt16 QuantizeSingleEndpointElementSigned(const MSInt16 &elem2CL, int precision, const ParallelMath::RoundUpForScope* ru);
static MUInt15 QuantizeSingleEndpointElementUnsigned(const MUInt15 &elem, int precision, const ParallelMath::RoundUpForScope* ru);
static void UnquantizeSingleEndpointElementSigned(const MSInt16 &comp, int precision, MSInt16 &outUnquantized, MSInt16 &outUnquantizedFinished2CL);
static void UnquantizeSingleEndpointElementUnsigned(const MUInt15 &comp, int precision, MUInt16 &outUnquantized, MUInt16 &outUnquantizedFinished);
static void QuantizeEndpointsSigned(const MSInt16 endPoints[2][3], const MFloat floatPixelsColorSpace[16][3], const MFloat floatPixelsLinearWeighted[16][3], MAInt16 quantizedEndPoints[2][3], MUInt15 indexes[16], IndexSelectorHDR<3> &indexSelector, int fixupIndex, int precision, int indexRange, const float *channelWeights, bool fastIndexing, const ParallelMath::RoundTowardNearestForScope *rtn);
static void QuantizeEndpointsUnsigned(const MSInt16 endPoints[2][3], const MFloat floatPixelsColorSpace[16][3], const MFloat floatPixelsLinearWeighted[16][3], MAInt16 quantizedEndPoints[2][3], MUInt15 indexes[16], IndexSelectorHDR<3> &indexSelector, int fixupIndex, int precision, int indexRange, const float *channelWeights, bool fastIndexing, const ParallelMath::RoundTowardNearestForScope *rtn);
static void EvaluatePartitionedLegality(const MAInt16 ep0[2][3], const MAInt16 ep1[2][3], int aPrec, const int bPrec[3], bool isTransformed, MAInt16 outEncodedEPs[2][2][3], ParallelMath::Int16CompFlag& outIsLegal);
static void EvaluateSingleLegality(const MAInt16 ep[2][3], int aPrec, const int bPrec[3], bool isTransformed, MAInt16 outEncodedEPs[2][3], ParallelMath::Int16CompFlag& outIsLegal);
static void SignExtendSingle(int &v, int bits);
};
}
}