DirectXTex

• HDR (RGBE Radiance) file format reader and writer
• Evaluate and Transform functions for computing user-defined functions on images
• Fix BC6H GPU shaders on WARP device
• Fix for alignment issues on ARM devices in software compression codec
• Added TEX_THRESHOLD_DEFAULT (0.5f) constant default alpha threshold value for Convert & Compress
• Minor CaptureTexture optimization
• texconv/texassemble: Support for .hdr file format
• texconv: added -gpu switch to specify adapter to use for GPU-based compression codecs
• texconv: added -badtails switch to enable loading of legacy DXTn DDS files with incomplete mipchain tails
• texconv: added -c switch for old-school colorkey/chromakey transparency to alpha conversion
• texconv: added -alpha switch for reverse premultiply along with TEX_PMALPHA_REVERSE flag
• texconv: added wildcard support for input filename and optional -r switch for recursive search

See this blog post for more information about .hdr.

• CompileShader script updated to build external pdbs
• Regenerated shaders using Windows 10 Anniversary Update SDK (14393)

• Updated for VS 2015 Update 3 and Windows 10 SDK (14393)

• Workaround for bug in XMStoreFloat3SE (impacts conversions to DXGI_FORMAT_R9G9B9E5_SHAREDEXP)
• DDSTextureLoader12, WICTextureLoader12, and ScreenGrab12 for Direct3D 12 support
• Minor code cleanup

• texconv command-line tool -wicq and -wiclossless switches
• Code cleanup

• Optional callback from WIC reader functions to query additional metadata
• Retired obsolete adapter code
• Minor code cleanup

• Fix to clean up partial or zero-length image files on failed write
• Retired VS 2012 projects

• texconv command-line tool -fl switch now supports 12.0 and 12.1 feature levels
• Updated for VS 2015 Update 1 and Windows 10 SDK (10586)

This is the last release to support VS 2012

• DDS support for legacy bumpmap formats (V8U8, Q8W8V8U8, V16U16)
• Fix for buffer overread in BC CPU compressor
• Minor code cleanup

• Added GetWICFactory and SetWICFactory
• Updates for new DXGI 1.3 types
• Xbox One platform updates

• Fixed rounding problem with 32-bit RGBA/BGRA format conversions
• texconv: use CPU parallel compression for BC1-BC5 (-singleproc disables)
• Updated for VS 2015 and Windows 10 SDK RTM
• Retired VS 2010 and Windows 8.0 Store projects

• New BC_FLAGS_USE_3SUBSETS option for BC7 compressors; now defaults to skipping 3 subset blocks
• Fixed bug with MakeTypeless and A8_UNORM
• Fixed file length validation problem in LoadDDSFromFile

• Updates for Visual Studio 2015 Technical Preview
• Minor code cleanup

• Format conversion improvements and bug fixes (depth/stencil, alpha-only, float16, RGB -> 1 channel)
• Fixed issue when BC decompressing non-standard compressed rowPitch images
• Explicit calling-convention annotation for all 'public' functions
• Code cleanup
• Xbox One platform updates

• texconv command-line tool fixes
• Fixed problem with 'wide' images with CPU Compress
• Updates to Xbox One platform support

Windows phone 8.1 platform support

• Direct3D 11 video and Xbox One extended format support
• New APIs: IsPlanar, IsPalettized, IsDepthStencil, ConvertToSinglePlane
• Added 'alphaWeight' parameter to GPU Compress breaking change
• texconv '-aw' switch to control the alpha weighting for the BC7 GPU compressor
• Fixed bug with ordered dithering in non-WIC conversion codepaths
• Fixed SaveToDDS* functions when using arbitrary row pitch values

• Added sRGB flags for Compress (TEX_COMPRESS_SRGB*)
• Added 'compress' flag parameter to GPU versions of Compress breaking change
• Minor fix for potential rounding problem in GPU Compress
• Code cleanup (removed DXGI_1_2_FORMATS control define; ScopedObject typedef removed)
• Dropped VS 2010 support without the Windows 8.1 SDK (removed USE_XNAMATH control define)

• texconv updated with -fl and -pow2 command-line switches
• Fixed bug in Resize when doing custom filtering which occurred when exactly doubling the image size
• Added move operators to ScratchImage and Blob classes
• Xbox One platform support

• Updated for Visual Studio 2013 and Windows 8.1 SDK RTM
• PremultiplyAlpha updated with new 'flags' parameter and to use sRGB correct blending
• Fixed colorspace conversion issue with DirectCompute compressor when compressing for BC7 SRGB