## Version 0.5.2 - Re-remove harmonic mean. Unfortunately it cannot be used on sets that have negative numbers, which includes ssimulacra2 scores. Geometric mean was also tested, and in _theory_ it may be computed on sets with negative values, in practice it cannot be due to computer algorithms using logarithms for the computation. ## Version 0.5.1 - Add harmonic mean output - Add ttf feature to `plotters` library - Bump all dependencies ## Version 0.5.0 - Bump minimum MSRV to 1.74.1 - Add rolling average to progress bar - Add increment parameter for skipping video frames - Bump all dependencies ## Version 0.4.1 - Add support for piping y4m input from stdin as one of the sources ## Version 0.4.0 - Update to [version 2.1 of the metric](https://github.com/cloudinary/ssimulacra2/compare/v2.0...v2.1) ## Version 0.3.5 - Minor performance improvements ## Version 0.3.4 - Add `--frame-threads` argument to support frame parallel multithreading - This is set to 1 by default because it does linearly increase memory usage - Various performance improvements ## Version 0.3.3 - Fix graphing ## Version 0.3.2 - Use crossterm instead of termion for Windows compatibility ## Version 0.3.1 - Add a progress bar - Bump dependencies for a pretty healthy speed increase of about 20% ## Version 0.3.0 - Implement upstream changes from https://github.com/libjxl/libjxl/pull/1848 ## Version 0.2.2 - Fix decoding of AV1 files with overlays when using `ffmpeg_build` feature - Or at least attempt to. There still seems to be some flakiness with linking libdav1d. - Add support for YUVJ files ## Version 0.2.1 - Fix compilation with no default features ## Version 0.2.0 - Add support for video comparison - This change splits ssimulacra2_rs into two subcommands: - `ssimulacra2_rs image input.png output.png` for comparing still images. All popular input formats are supported. - `ssimulacra2_rs video input.y4m output.y4m` for comparing videos. All popular input formats are supported. This feature requires the `ffmpeg` feature be enabled, which is on by default.