diff --git a/src/image.cpp b/src/image.cpp index fc1b1273..499b54f7 100644 --- a/src/image.cpp +++ b/src/image.cpp @@ -469,28 +469,29 @@ namespace bgfx return; } - float* dst = (float*)_dst; - const float* src = (const float*)_src; + const uint8_t* src = (const uint8_t*)_src; + uint8_t* dst = (uint8_t*)_dst; for (uint32_t yy = 0, ystep = _pitch*2; yy < dstheight; ++yy, src += ystep) { - const float* rgba = src; - for (uint32_t xx = 0; xx < dstwidth; ++xx, rgba += 8, dst += 4) + const float* rgba0 = (const float*)&src[0]; + const float* rgba1 = (const float*)&src[_pitch]; + for (uint32_t xx = 0; xx < dstwidth; ++xx, rgba0 += 8, rgba1 += 8, dst += 16) { float xyz[3]; - xyz[0] = rgba[ 0]; - xyz[1] = rgba[ 1]; - xyz[2] = rgba[ 2]; - xyz[0] += rgba[ 4]; - xyz[1] += rgba[ 5]; - xyz[2] += rgba[ 6]; - xyz[0] += rgba[_pitch+0]; - xyz[1] += rgba[_pitch+1]; - xyz[2] += rgba[_pitch+2]; - xyz[0] += rgba[_pitch+4]; - xyz[1] += rgba[_pitch+5]; - xyz[2] += rgba[_pitch+6]; - bx::vec3Norm(dst, xyz); + xyz[0] = rgba0[0]; + xyz[1] = rgba0[1]; + xyz[2] = rgba0[2]; + xyz[0] += rgba0[4]; + xyz[1] += rgba0[5]; + xyz[2] += rgba0[6]; + xyz[0] += rgba1[0]; + xyz[1] += rgba1[1]; + xyz[2] += rgba1[2]; + xyz[0] += rgba1[4]; + xyz[1] += rgba1[5]; + xyz[2] += rgba1[6]; + bx::vec3Norm( (float*)dst, xyz); } } } @@ -2496,36 +2497,39 @@ namespace bgfx void imageDecodeToRgba32f(bx::AllocatorI* _allocator, void* _dst, const void* _src, uint32_t _width, uint32_t _height, uint32_t _pitch, TextureFormat::Enum _format) { const uint8_t* src = (const uint8_t*)_src; - float* dst = (float*)_dst; - - uint32_t width = _width/4; - uint32_t height = _height/4; + uint8_t* dst = (uint8_t*)_dst; switch (_format) { case TextureFormat::BC5: - for (uint32_t yy = 0; yy < height; ++yy) { - for (uint32_t xx = 0; xx < width; ++xx) + uint32_t width = _width/4; + uint32_t height = _height/4; + + for (uint32_t yy = 0; yy < height; ++yy) { - uint8_t temp[16*4]; - - decodeBlockDxt45A(temp+1, src); - src += 8; - decodeBlockDxt45A(temp+2, src); - src += 8; - - for (uint32_t ii = 0; ii < 16; ++ii) + for (uint32_t xx = 0; xx < width; ++xx) { - float nx = temp[ii*4+2]*2.0f/255.0f - 1.0f; - float ny = temp[ii*4+1]*2.0f/255.0f - 1.0f; - float nz = sqrtf(1.0f - nx*nx - ny*ny); + uint8_t temp[16*4]; - float* block = &dst[( (yy + ii/4)*_pitch+xx*4+ii%4)*16]; - block[0] = nx; - block[1] = ny; - block[2] = nz; - block[3] = 0.0f; + decodeBlockDxt45A(temp+1, src); + src += 8; + decodeBlockDxt45A(temp+2, src); + src += 8; + + for (uint32_t ii = 0; ii < 16; ++ii) + { + float nx = temp[ii*4+2]*2.0f/255.0f - 1.0f; + float ny = temp[ii*4+1]*2.0f/255.0f - 1.0f; + float nz = sqrtf(1.0f - nx*nx - ny*ny); + + const uint32_t offset = (yy + ii/4)*_pitch + (xx*4 + ii%4)*16; + float* block = (float*)&dst[offset]; + block[0] = nx; + block[1] = ny; + block[2] = nz; + block[3] = 0.0f; + } } } } diff --git a/tools/texturec/texturec.cpp b/tools/texturec/texturec.cpp index 25207f5f..3bee4fe7 100644 --- a/tools/texturec/texturec.cpp +++ b/tools/texturec/texturec.cpp @@ -331,6 +331,19 @@ int main(int _argc, const char* _argv[]) , mip.m_format ); + for (uint32_t yy = 0; yy < mip.m_height; ++yy) + { + for (uint32_t xx = 0; xx < mip.m_width; ++xx) + { + const uint32_t offset = (yy*mip.m_width + xx) * 4; + float* inout = &rgba[offset]; + inout[0] = inout[0] * 2.0f/255.0f - 1.0f; + inout[1] = inout[1] * 2.0f/255.0f - 1.0f; + inout[2] = inout[2] * 2.0f/255.0f - 1.0f; + inout[3] = inout[3] * 2.0f/255.0f - 1.0f; + } + } + imageContainer.m_numMips = numMips; imageContainer.m_size = imageGetSize(format, mip.m_width, mip.m_height, 0, false, numMips); imageContainer.m_format = format; @@ -340,14 +353,12 @@ int main(int _argc, const char* _argv[]) for (uint8_t lod = 1; lod < numMips; ++lod) { - ImageMip mip1; - imageGetRawData(imageContainer, 0, lod, output->data, output->size, mip1); - uint8_t* data = const_cast(mip1.m_data); + imageRgba32fDownsample2x2NormalMap(mip.m_width, mip.m_height, mip.m_width*16, rgba, rgba); - uint32_t width = bx::uint32_max(1, mip.m_width >>lod); - uint32_t height = bx::uint32_max(1, mip.m_height>>lod); - imageRgba32fDownsample2x2NormalMap(width, height, width*16, rgba, rgba); - imageEncodeFromRgba32f(data, rgba, width, height, format); + ImageMip dstMip; + imageGetRawData(imageContainer, 0, lod, output->data, output->size, dstMip); + uint8_t* data = const_cast(dstMip.m_data); + imageEncodeFromRgba32f(data, rgba, dstMip.m_width, dstMip.m_height, format); } } else @@ -356,6 +367,14 @@ int main(int _argc, const char* _argv[]) temp = BX_ALLOC(&allocator, size); uint8_t* rgba = (uint8_t*)temp; + imageDecodeToRgba8(rgba + , mip.m_data + , mip.m_width + , mip.m_height + , mip.m_width*mip.m_bpp/8 + , mip.m_format + ); + imageContainer.m_numMips = numMips; imageContainer.m_size = imageGetSize(format, mip.m_width, mip.m_height, 0, false, numMips); imageContainer.m_format = format; @@ -365,14 +384,12 @@ int main(int _argc, const char* _argv[]) for (uint8_t lod = 1; lod < numMips; ++lod) { - ImageMip mip1; - imageGetRawData(imageContainer, 0, lod, output->data, output->size, mip1); - uint8_t* data = const_cast(mip1.m_data); + imageRgba8Downsample2x2(mip.m_width, mip.m_height, mip.m_width*4, rgba, rgba); - uint32_t width = bx::uint32_max(1, mip.m_width >>lod); - uint32_t height = bx::uint32_max(1, mip.m_height>>lod); - imageRgba8Downsample2x2(width, height, width*4, rgba, rgba); - imageEncodeFromRgba8(data, rgba, width, height, format); + ImageMip dstMip; + imageGetRawData(imageContainer, 0, lod, output->data, output->size, dstMip); + uint8_t* data = const_cast(dstMip.m_data); + imageEncodeFromRgba8(data, rgba, dstMip.m_width, dstMip.m_height, format); } }