Version 4.2 of Compressonator was released in July 2021. Here are the new features introduced with this version:
BC1 improvements by up to 38% in performance and 0.6 dB in quality
We added new refine steps to improve quality of images with mixed low and high-frequency content.
- We boosted max quality encoding performance by 38% and +0.6 dB for smooth texture-mapped surfaces.
The BC1 codec now has four discrete step settings; prior releases for CPU encoding only used Step 4. Any value can be set in the ranges to enable the specific algorithm used. Changing values within the range will not change the resulting image quality.
The first three steps will run on CPU and GPU, while the last step will run on CPU. On GPU, Step 4 achieves the same results as Step 3.
|1||0.0 to 0.24||Uses a fast min-max algorithm which generates poor quality images.|
|2||0.25 to 0.49||Uses the prior step and adds a principal component-based algorithm to improve image quality.|
|3||0.50 to 0.74||Uses the prior step with a course ramp search on indexed end points.|
|4||0.75 to 1.0||On CPU use above and multiple ramp searches on indexed end points. |
On GPU no channels are adjusted so quality will be less than the CPU version (same as Step 3).
Users should adjust quality settings used in prior releases to new values to match or exceed prior performance and quality results.
- BC7 HLSL bug fixes and improved overall quality of the shader.
API interface updates for BCn codecs
The interfaces and include dependencies are been updated to support switching from CPU/GPU API and switching from CMP codecs and any developer-based codecs. Currently BC1 and BC7 have been modified, and additional changes are pending for these and all other BCn codecs.
- Added “Refine Steps” settings for BC1 image processing.
- Removed “Use Alpha” and “No Alpha” setting for BC1. Just set a threshold above 0 to enable BC1 punch through ranges.
framework.hfile, replacing it with
- Added new kernel setting options for BCn codecs in CMP_Framework. See the
Readme.mdor SDK example code for details.
- New single header shader files added for CPU-GPU common APIs and type definitions.
- Added FidelityFX types to
common_defs.hto support programming FidelityFX shaders with CMP type definitions. These new definitions will allow developers using FidelityFX code to run and debug shader code on CPU, and then compile them on GPU for production.
- Improved CompressonatorLib BC1 performance and quality.
CMake build updates
- Added options to select build of external libs, SDK libs, and applications.
- Build updated for CLI on macOS.
- OpenCV 4 supported.
- Improved compiler standard revision detection and extensions.
- Visual Studio® 2019 support.
- Qt 5.15 support.
Compressonator is a set of tools to allow artists and developers to more easily work with compressed assets and easily visualize the quality impact of various compression technologies.
The Windows version of Compressonator 4.0 supports GPU encoding with DirectX® Compute (DXC) or OpenCL™ (OCL) shaders.
Compressonator v4.1 adds SNORM support for BC4 and BC5, MIP map filter support using either GPU or CPU, GPU-based compression and MIP map generation, and much more.