griefith/test2
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

Vulkan: Add FLOAT to SRGB and Depth24 Conversion

Browse Source 
This PR adds support to convert host/device data from floats
to GPU_SRGB8_A8 and GPU_DEPTH_COMPONENT24.

Pull Request: https://projects.blender.org/blender/blender/pulls/107544
This commit is contained in:
Jeroen Bakker
2023-05-02 15:57:17 +02:00
parent 13fe6d0170
commit 17201752d9
2 changed files with 119 additions and 38 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
source/blender/gpu/tests/texture_test.cc
View File

@@ -262,13 +262,11 @@ static void test_texture_roundtrip__GPU_DATA_FLOAT__GPU_R11F_G11F_B10F()
GPU_TEST(texture_roundtrip__GPU_DATA_FLOAT__GPU_R11F_G11F_B10F);
#endif
#if RUN_SRGB_UNIMPLEMENTED
static void test_texture_roundtrip__GPU_DATA_FLOAT__GPU_SRGB8_A8()
{
texture_create_upload_read_with_bias<GPU_SRGB8_A8, GPU_DATA_FLOAT>(0.0f);
texture_create_upload_read_with_bias<GPU_SRGB8_A8, GPU_DATA_FLOAT>(0.003f);
}
GPU_TEST(texture_roundtrip__GPU_DATA_FLOAT__GPU_SRGB8_A8);
#endif
static void test_texture_roundtrip__GPU_DATA_FLOAT__GPU_RGBA8_SNORM()
{
@@ -405,13 +403,13 @@ static void test_texture_roundtrip__GPU_DATA_FLOAT__GPU_DEPTH_COMPONENT32F()
GPU_TEST(texture_roundtrip__GPU_DATA_FLOAT__GPU_DEPTH_COMPONENT32F);
#endif
#if RUN_COMPONENT_UNIMPLEMENTED
static void test_texture_roundtrip__GPU_DATA_FLOAT__GPU_DEPTH_COMPONENT24()
{
texture_create_upload_read_with_bias<GPU_DEPTH_COMPONENT24, GPU_DATA_FLOAT>(0.0f);
texture_create_upload_read_with_bias<GPU_DEPTH_COMPONENT24, GPU_DATA_FLOAT>(0.0000001f);
}
GPU_TEST(texture_roundtrip__GPU_DATA_FLOAT__GPU_DEPTH_COMPONENT24);
#if RUN_COMPONENT_UNIMPLEMENTED
static void test_texture_roundtrip__GPU_DATA_FLOAT__GPU_DEPTH_COMPONENT16()
{
texture_create_upload_read_with_bias<GPU_DEPTH_COMPONENT16, GPU_DATA_FLOAT>(0.0f);

149
source/blender/gpu/vulkan/vk_data_conversion.cc
View File

@@ -7,6 +7,8 @@
#include "vk_data_conversion.hh"
#include "BLI_color.hh"
#include "Imath/half.h"
namespace blender::gpu {
@@ -47,6 +49,12 @@ enum class ConversionType {
HALF_TO_FLOAT,
FLOAT_TO_HALF,
FLOAT_TO_SRGBA8,
SRGBA8_TO_FLOAT,
FLOAT_TO_DEPTH_COMPONENT24,
DEPTH_COMPONENT24_TO_FLOAT,
/**
* The requested conversion isn't supported.
*/
@@ -90,6 +98,12 @@ static ConversionType type_of_conversion_float(eGPUTextureFormat device_format)
case GPU_R16_SNORM:
return ConversionType::FLOAT_TO_SNORM16;
case GPU_SRGB8_A8:
return ConversionType::FLOAT_TO_SRGBA8;
case GPU_DEPTH_COMPONENT24:
return ConversionType::FLOAT_TO_DEPTH_COMPONENT24;
case GPU_RGB32F: /* GPU_RGB32F Not supported by vendors. */
case GPU_RGBA8UI:
case GPU_RGBA8I:
@@ -114,7 +128,6 @@ static ConversionType type_of_conversion_float(eGPUTextureFormat device_format)
case GPU_R11F_G11F_B10F:
case GPU_DEPTH32F_STENCIL8:
case GPU_DEPTH24_STENCIL8:
case GPU_SRGB8_A8:
case GPU_RGB8UI:
case GPU_RGB8I:
case GPU_RGB8:
@@ -131,7 +144,6 @@ static ConversionType type_of_conversion_float(eGPUTextureFormat device_format)
case GPU_RGBA8_DXT5:
case GPU_SRGB8:
case GPU_RGB9_E5:
case GPU_DEPTH_COMPONENT24:
case GPU_DEPTH_COMPONENT16:
return ConversionType::UNSUPPORTED;
}
@@ -512,6 +524,8 @@ static ConversionType reversed(ConversionType type)
CASE_PAIR(UI32, UI8)
CASE_PAIR(I32, I8)
CASE_PAIR(FLOAT, HALF)
CASE_PAIR(FLOAT, SRGBA8)
CASE_PAIR(FLOAT, DEPTH_COMPONENT24)
case ConversionType::UNSUPPORTED:
return ConversionType::UNSUPPORTED;
@@ -529,6 +543,44 @@ static ConversionType reversed(ConversionType type)
/** \name Data Conversion
* \{ */
template<typename InnerType> struct ComponentValue {
InnerType value;
};
template<typename InnerType> struct PixelValue {
InnerType value;
};
using UI8 = ComponentValue<uint8_t>;
using UI16 = ComponentValue<uint16_t>;
using UI32 = ComponentValue<uint32_t>;
using I8 = ComponentValue<int8_t>;
using I16 = ComponentValue<int16_t>;
using I32 = ComponentValue<int32_t>;
using F32 = ComponentValue<float>;
using F16 = ComponentValue<uint16_t>;
using SRGBA8 = PixelValue<ColorSceneLinearByteEncoded4b<eAlpha::Premultiplied>>;
using FLOAT4 = PixelValue<ColorSceneLinear4f<eAlpha::Premultiplied>>;
class DepthComponent24 : public ComponentValue<uint32_t> {
public:
operator uint32_t() const
{
return value;
}
DepthComponent24 &operator=(uint32_t new_value)
{
value = new_value;
return *this;
}
/* Depth component24 are 4 bytes, but 1 isn't used. */
static constexpr size_t used_byte_size()
{
return 3;
}
};
template<typename InnerType> struct SignedNormalized {
static_assert(std::is_same<InnerType, uint8_t>() || std::is_same<InnerType, uint16_t>());
InnerType value;
@@ -550,35 +602,33 @@ template<typename InnerType> struct SignedNormalized {
};
template<typename InnerType> struct UnsignedNormalized {
static_assert(std::is_same<InnerType, uint8_t>() || std::is_same<InnerType, uint16_t>());
static_assert(std::is_same<InnerType, uint8_t>() || std::is_same<InnerType, uint16_t>() ||
std::is_same<InnerType, DepthComponent24>());
InnerType value;
static constexpr size_t used_byte_size()
{
if constexpr (std::is_same<InnerType, DepthComponent24>()) {
return InnerType::used_byte_size();
}
else {
return sizeof(InnerType);
}
}
static constexpr int32_t scalar()
{
return (1 << (sizeof(InnerType) * 8)) - 1;
return (1 << (used_byte_size() * 8)) - 1;
}
static constexpr int32_t max()
{
return ((1 << (sizeof(InnerType) * 8)) - 1);
return ((1 << (used_byte_size() * 8)) - 1);
}
};
template<typename InnerType> struct ComponentValue {
InnerType value;
};
using UI8 = ComponentValue<uint8_t>;
using UI16 = ComponentValue<uint16_t>;
using UI32 = ComponentValue<uint32_t>;
using I8 = ComponentValue<int8_t>;
using I16 = ComponentValue<int16_t>;
using I32 = ComponentValue<int32_t>;
using F32 = ComponentValue<float>;
using F16 = ComponentValue<uint16_t>;
template<typename StorageType>
void convert_component(SignedNormalized<StorageType> &dst, const F32 &src)
template<typename StorageType> void convert(SignedNormalized<StorageType> &dst, const F32 &src)
{
static constexpr int32_t scalar = SignedNormalized<StorageType>::scalar();
static constexpr int32_t delta = SignedNormalized<StorageType>::delta();
@@ -586,32 +636,29 @@ void convert_component(SignedNormalized<StorageType> &dst, const F32 &src)
dst.value = (clamp_i((src.value * scalar + delta), 0, max));
}
template<typename StorageType>
void convert_component(F32 &dst, const SignedNormalized<StorageType> &src)
template<typename StorageType> void convert(F32 &dst, const SignedNormalized<StorageType> &src)
{
static constexpr int32_t scalar = SignedNormalized<StorageType>::scalar();
static constexpr int32_t delta = SignedNormalized<StorageType>::delta();
dst.value = float(int32_t(src.value) - delta) / scalar;
}
template<typename StorageType>
void convert_component(UnsignedNormalized<StorageType> &dst, const F32 &src)
template<typename StorageType> void convert(UnsignedNormalized<StorageType> &dst, const F32 &src)
{
static constexpr int32_t scalar = UnsignedNormalized<StorageType>::scalar();
static constexpr int32_t max = scalar;
dst.value = (clamp_i((src.value * scalar), 0, max));
}
template<typename StorageType>
void convert_component(F32 &dst, const UnsignedNormalized<StorageType> &src)
template<typename StorageType> void convert(F32 &dst, const UnsignedNormalized<StorageType> &src)
{
static constexpr int32_t scalar = UnsignedNormalized<StorageType>::scalar();
dst.value = float(src.value) / scalar;
dst.value = float(int32_t(src.value)) / scalar;
}
/* Copy the contents of src to dst with out performing any actual conversion. */
template<typename DestinationType, typename SourceType>
void convert_component(DestinationType &dst, const SourceType &src)
void convert(DestinationType &dst, const SourceType &src)
{
static_assert(std::is_same<DestinationType, UI8>() || std::is_same<DestinationType, UI16>() ||
std::is_same<DestinationType, UI32>() || std::is_same<DestinationType, I8>() ||
@@ -623,24 +670,34 @@ void convert_component(DestinationType &dst, const SourceType &src)
dst.value = src.value;
}
static void convert_component(F16 &dst, const F32 &src)
static void convert(F16 &dst, const F32 &src)
{
dst.value = imath_float_to_half(src.value);
}
static void convert_component(F32 &dst, const F16 &src)
static void convert(F32 &dst, const F16 &src)
{
dst.value = imath_half_to_float(src.value);
}
static void convert(SRGBA8 &dst, const FLOAT4 &src)
{
dst.value = src.value.encode();
}
static void convert(FLOAT4 &dst, const SRGBA8 &src)
{
dst.value = src.value.decode();
}
/* \} */
template<typename DestinationType, typename SourceType>
void convert_per_component(MutableSpan<DestinationType> dst, Span<SourceType> src)
void convert(MutableSpan<DestinationType> dst, Span<SourceType> src)
{
BLI_assert(src.size() == dst.size());
for (int64_t index : IndexRange(src.size())) {
convert_component(dst[index], src[index]);
convert(dst[index], src[index]);
}
}
@@ -655,7 +712,17 @@ void convert_per_component(void *dst_memory,
total_components);
MutableSpan<DestinationType> dst = MutableSpan<DestinationType>(
static_cast<DestinationType *>(dst_memory), total_components);
convert_per_component<DestinationType, SourceType>(dst, src);
convert<DestinationType, SourceType>(dst, src);
}
template<typename DestinationType, typename SourceType>
void convert_per_pixel(void *dst_memory, const void *src_memory, size_t buffer_size)
{
Span<SourceType> src = Span<SourceType>(static_cast<const SourceType *>(src_memory),
buffer_size);
MutableSpan<DestinationType> dst = MutableSpan<DestinationType>(
static_cast<DestinationType *>(dst_memory), buffer_size);
convert<DestinationType, SourceType>(dst, src);
}
static void convert_buffer(void *dst_memory,
@@ -746,6 +813,22 @@ static void convert_buffer(void *dst_memory,
case ConversionType::HALF_TO_FLOAT:
convert_per_component<F32, F16>(dst_memory, src_memory, buffer_size, device_format);
break;
case ConversionType::FLOAT_TO_SRGBA8:
convert_per_pixel<SRGBA8, FLOAT4>(dst_memory, src_memory, buffer_size);
break;
case ConversionType::SRGBA8_TO_FLOAT:
convert_per_pixel<FLOAT4, SRGBA8>(dst_memory, src_memory, buffer_size);
break;
case ConversionType::FLOAT_TO_DEPTH_COMPONENT24:
convert_per_component<UnsignedNormalized<DepthComponent24>, F32>(
dst_memory, src_memory, buffer_size, device_format);
break;
case ConversionType::DEPTH_COMPONENT24_TO_FLOAT:
convert_per_component<F32, UnsignedNormalized<DepthComponent24>>(
dst_memory, src_memory, buffer_size, device_format);
break;
}
}