static HRESULT _PremultiplyAlpha( _In_ const Image& srcImage, _In_ const Image& destImage )
{
assert( srcImage.width == destImage.width );
assert( srcImage.height == destImage.height );
ScopedAlignedArrayXMVECTOR scanline( reinterpret_cast<XMVECTOR*>( _aligned_malloc( (sizeof(XMVECTOR)*srcImage.width), 16 ) ) );
if ( !scanline )
return E_OUTOFMEMORY;
const uint8_t *pSrc = srcImage.pixels;
uint8_t *pDest = destImage.pixels;
if ( !pSrc || !pDest )
return E_POINTER;
for( size_t h = 0; h < srcImage.height; ++h )
{
if ( !_LoadScanline( scanline.get(), srcImage.width, pSrc, srcImage.rowPitch, srcImage.format ) )
return E_FAIL;
XMVECTOR* ptr = scanline.get();
for( size_t w = 0; w < srcImage.width; ++w )
{
XMVECTOR v = *ptr;
XMVECTOR alpha = XMVectorSplatW( *ptr );
alpha = XMVectorMultiply( v, alpha );
*(ptr++) = XMVectorSelect( v, alpha, g_XMSelect1110 );
}
if ( !_StoreScanline( pDest, destImage.rowPitch, destImage.format, scanline.get(), srcImage.width ) )
return E_FAIL;
pSrc += srcImage.rowPitch;
pDest += destImage.rowPitch;
}
return S_OK;
}
//-------------------------------------------------------------------------------------
static HRESULT _DecompressBC( _In_ const Image& cImage, _In_ const Image& result )
{
if ( !cImage.pixels || !result.pixels )
return E_POINTER;
assert( cImage.width == result.width );
assert( cImage.height == result.height );
// Image must be a multiple of 4 (degenerate cases of 1x1, 1x2, 2x1, and 2x2 are allowed)
size_t width = cImage.width;
if ( (width % 4) != 0 )
{
if ( width != 1 && width != 2 )
return E_INVALIDARG;
}
size_t height = cImage.height;
if ( (height % 4) != 0 )
{
if ( height != 1 && height != 2 )
return E_INVALIDARG;
}
const DXGI_FORMAT format = result.format;
size_t dbpp = BitsPerPixel( format );
if ( !dbpp )
return E_FAIL;
if ( dbpp < 8 )
{
// We don't support decompressing to monochrome (DXGI_FORMAT_R1_UNORM)
return HRESULT_FROM_WIN32( ERROR_NOT_SUPPORTED );
}
// Round to bytes
dbpp = ( dbpp + 7 ) / 8;
uint8_t *pDest = result.pixels;
if ( !pDest )
return E_POINTER;
// Promote "typeless" BC formats
DXGI_FORMAT cformat;
switch( cImage.format )
{
case DXGI_FORMAT_BC1_TYPELESS: cformat = DXGI_FORMAT_BC1_UNORM; break;
case DXGI_FORMAT_BC2_TYPELESS: cformat = DXGI_FORMAT_BC2_UNORM; break;
case DXGI_FORMAT_BC3_TYPELESS: cformat = DXGI_FORMAT_BC3_UNORM; break;
case DXGI_FORMAT_BC4_TYPELESS: cformat = DXGI_FORMAT_BC4_UNORM; break;
case DXGI_FORMAT_BC5_TYPELESS: cformat = DXGI_FORMAT_BC5_UNORM; break;
case DXGI_FORMAT_BC6H_TYPELESS: cformat = DXGI_FORMAT_BC6H_UF16; break;
case DXGI_FORMAT_BC7_TYPELESS: cformat = DXGI_FORMAT_BC7_UNORM; break;
default: cformat = cImage.format; break;
}
// Determine BC format decoder
BC_DECODE pfDecode;
size_t sbpp;
switch(cformat)
{
case DXGI_FORMAT_BC1_UNORM:
case DXGI_FORMAT_BC1_UNORM_SRGB: pfDecode = D3DXDecodeBC1; sbpp = 8; break;
case DXGI_FORMAT_BC2_UNORM:
case DXGI_FORMAT_BC2_UNORM_SRGB: pfDecode = D3DXDecodeBC2; sbpp = 16; break;
case DXGI_FORMAT_BC3_UNORM:
case DXGI_FORMAT_BC3_UNORM_SRGB: pfDecode = D3DXDecodeBC3; sbpp = 16; break;
case DXGI_FORMAT_BC4_UNORM: pfDecode = D3DXDecodeBC4U; sbpp = 8; break;
case DXGI_FORMAT_BC4_SNORM: pfDecode = D3DXDecodeBC4S; sbpp = 8; break;
case DXGI_FORMAT_BC5_UNORM: pfDecode = D3DXDecodeBC5U; sbpp = 16; break;
case DXGI_FORMAT_BC5_SNORM: pfDecode = D3DXDecodeBC5S; sbpp = 16; break;
case DXGI_FORMAT_BC6H_UF16: pfDecode = D3DXDecodeBC6HU; sbpp = 16; break;
case DXGI_FORMAT_BC6H_SF16: pfDecode = D3DXDecodeBC6HS; sbpp = 16; break;
case DXGI_FORMAT_BC7_UNORM:
case DXGI_FORMAT_BC7_UNORM_SRGB: pfDecode = D3DXDecodeBC7; sbpp = 16; break;
default:
return HRESULT_FROM_WIN32( ERROR_NOT_SUPPORTED );
}
XMVECTOR temp[16];
const uint8_t *pSrc = cImage.pixels;
const size_t rowPitch = result.rowPitch;
for( size_t h=0; h < cImage.height; h += 4 )
{
const uint8_t *sptr = pSrc;
uint8_t* dptr = pDest;
for( size_t count = 0; count < cImage.rowPitch; count += sbpp )
{
pfDecode( temp, sptr );
_ConvertScanline( temp, 16, format, cformat, 0 );
if ( !_StoreScanline( dptr, rowPitch, format, &temp[0], 4 ) )
return E_FAIL;
if ( result.height > 1 )
{
if ( !_StoreScanline( dptr + rowPitch, rowPitch, format, &temp[4], 4 ) )
return E_FAIL;
if ( result.height > 2 )
{
if ( !_StoreScanline( dptr + rowPitch*2, rowPitch, format, &temp[8], 4 ) )
return E_FAIL;
if ( !_StoreScanline( dptr + rowPitch*3, rowPitch, format, &temp[12], 4 ) )
return E_FAIL;
}
}
sptr += sbpp;
dptr += dbpp*4;
}
pSrc += cImage.rowPitch;
pDest += rowPitch*4;
}
return S_OK;
}
//-------------------------------------------------------------------------------------
static HRESULT _DecompressBC( _In_ const Image& cImage, _In_ const Image& result )
{
if ( !cImage.pixels || !result.pixels )
return E_POINTER;
assert( cImage.width == result.width );
assert( cImage.height == result.height );
const DXGI_FORMAT format = result.format;
size_t dbpp = BitsPerPixel( format );
if ( !dbpp )
return E_FAIL;
if ( dbpp < 8 )
{
// We don't support decompressing to monochrome (DXGI_FORMAT_R1_UNORM)
return HRESULT_FROM_WIN32( ERROR_NOT_SUPPORTED );
}
// Round to bytes
dbpp = ( dbpp + 7 ) / 8;
uint8_t *pDest = result.pixels;
if ( !pDest )
return E_POINTER;
// Promote "typeless" BC formats
DXGI_FORMAT cformat;
switch( cImage.format )
{
case DXGI_FORMAT_BC1_TYPELESS: cformat = DXGI_FORMAT_BC1_UNORM; break;
case DXGI_FORMAT_BC2_TYPELESS: cformat = DXGI_FORMAT_BC2_UNORM; break;
case DXGI_FORMAT_BC3_TYPELESS: cformat = DXGI_FORMAT_BC3_UNORM; break;
case DXGI_FORMAT_BC4_TYPELESS: cformat = DXGI_FORMAT_BC4_UNORM; break;
case DXGI_FORMAT_BC5_TYPELESS: cformat = DXGI_FORMAT_BC5_UNORM; break;
case DXGI_FORMAT_BC6H_TYPELESS: cformat = DXGI_FORMAT_BC6H_UF16; break;
case DXGI_FORMAT_BC7_TYPELESS: cformat = DXGI_FORMAT_BC7_UNORM; break;
default: cformat = cImage.format; break;
}
// Determine BC format decoder
BC_DECODE pfDecode;
size_t sbpp;
switch(cformat)
{
case DXGI_FORMAT_BC1_UNORM:
case DXGI_FORMAT_BC1_UNORM_SRGB: pfDecode = D3DXDecodeBC1; sbpp = 8; break;
case DXGI_FORMAT_BC2_UNORM:
case DXGI_FORMAT_BC2_UNORM_SRGB: pfDecode = D3DXDecodeBC2; sbpp = 16; break;
case DXGI_FORMAT_BC3_UNORM:
case DXGI_FORMAT_BC3_UNORM_SRGB: pfDecode = D3DXDecodeBC3; sbpp = 16; break;
case DXGI_FORMAT_BC4_UNORM: pfDecode = D3DXDecodeBC4U; sbpp = 8; break;
case DXGI_FORMAT_BC4_SNORM: pfDecode = D3DXDecodeBC4S; sbpp = 8; break;
case DXGI_FORMAT_BC5_UNORM: pfDecode = D3DXDecodeBC5U; sbpp = 16; break;
case DXGI_FORMAT_BC5_SNORM: pfDecode = D3DXDecodeBC5S; sbpp = 16; break;
case DXGI_FORMAT_BC6H_UF16: pfDecode = D3DXDecodeBC6HU; sbpp = 16; break;
case DXGI_FORMAT_BC6H_SF16: pfDecode = D3DXDecodeBC6HS; sbpp = 16; break;
case DXGI_FORMAT_BC7_UNORM:
case DXGI_FORMAT_BC7_UNORM_SRGB: pfDecode = D3DXDecodeBC7; sbpp = 16; break;
default:
return HRESULT_FROM_WIN32( ERROR_NOT_SUPPORTED );
}
XMVECTOR temp[16];
const uint8_t *pSrc = cImage.pixels;
const size_t rowPitch = result.rowPitch;
for( size_t h=0; h < cImage.height; h += 4 )
{
const uint8_t *sptr = pSrc;
uint8_t* dptr = pDest;
size_t ph = std::min<size_t>( 4, cImage.height - h );
size_t w = 0;
for( size_t count = 0; count < cImage.rowPitch; count += sbpp, w += 4 )
{
pfDecode( temp, sptr );
_ConvertScanline( temp, 16, format, cformat, 0 );
size_t pw = std::min<size_t>( 4, cImage.width - w );
assert( pw > 0 && ph > 0 );
if ( !_StoreScanline( dptr, rowPitch, format, &temp[0], pw ) )
return E_FAIL;
if ( ph > 1 )
{
if ( !_StoreScanline( dptr + rowPitch, rowPitch, format, &temp[4], pw ) )
return E_FAIL;
if ( ph > 2 )
{
if ( !_StoreScanline( dptr + rowPitch*2, rowPitch, format, &temp[8], pw ) )
return E_FAIL;
if ( ph > 3 )
{
if ( !_StoreScanline( dptr + rowPitch*3, rowPitch, format, &temp[12], pw ) )
return E_FAIL;
}
}
}
sptr += sbpp;
dptr += dbpp*4;
}
pSrc += cImage.rowPitch;
pDest += rowPitch*4;
}
return S_OK;
}
