David Michael Barr

Results 12 issues of David Michael Barr

Runs k-means on logarithms of distortion scale, with k fixed at 3. Needs more comments and AWCY results.

Easier to read and reduces compressed compiled object size by 10%.

For sub-8x8 blocks, uv parameters are not coded on every block but shared between neighbors. The non-coded blocks are not DC_PRED but rather N/A.

Adds doubling of Adiv2, missed while handling odd dimensions.

### subset1 ``` RATE (%) DSNR (dB) PSNR -0.132183 0.00574649 PSNRHVS -0.187846 0.0124809 SSIM -0.0780694 0.00274665 FASTSSIM -0.232905 0.00647426 ``` ### objective-1-fast ``` PSNR YCbCr: -0.06% -1.16% -0.88% PSNRHVS: -0.19%...

Override clusterfuzz's hard-coded maximum unit timeout of 25 seconds. Add a local `run_fuzzers` action that uses our Docker image. Follows the guide in `CONTRIBUTING.md` on fuzzing.

This is an adaptation of some of the changes in https://github.com/KyleSiefring/dump_ciede2000/commit/07bdd82851fbe909556a576f5eefbf2464148f3e with the addition of a `Default` implementation for convenience. The `dump_ciede2000` tool references calibrated values for `KSubArgs` from ["Color...

For `x86_64` builds of rav1e at the time of opening this issue, there are 212 functions included in assembly that are not referenced by rav1e Rust code. These mostly correspond...

enhancement

This should resolve some Windows deployment issues we've been having with [xiph/rav1e](https://github.com/xiph/rav1e). We've been manually preventing `windows-sys` upgrade to `0.48.5` in our `Cargo.lock` as it was causing `x86_64-pc-windows-gnu` builds to...

Inspiration is taken from how git packs achieve high compression rates with good random access performance. The 2 key components are (1) a clustering strategy to store similar objects together...