Same Image, Different Score? | Halide Compression
halide.cx · 1,489 words · saved by 1 readers
Metrics don't just measure encoder performance; decoding matters too, and even after decoding, not all post-processing is created equal.
In developing our proprietary encoder Iris for WebP, our aim with public and private testing is to properly demonstrate the value of the encoder compared to alternatives. Cheating benchmarks, overfitting for metrics, or unfairly testing other encoders does not help sell our product, which is meant to provide quality-of-experience improvements for human users above all else. Investigating WebP's decoding performance led us to begin evaluating different means of presenting the decoded images to metrics. Even within the dwebp reference decoder, there are a number of different options that…
saved by
related reading
- Guillaume Técherguillaumetech.github.io
- Shared Dictionaries: compression that keeps up with the agentic webblog.cloudflare.com
- Image formats: Pixel data from encoders to decodersdeveloper.mozilla.org
- H.264 is magic: a technical walkthrough of a remarkable technology.sidbala.com
- What's In a JPEG?blog.bithole.dev
- How to Optimize a CUDA Matmul Kernel for cuBLAS-like Performance: a Worklogsiboehm.com
- Aaron Leslie — Software Engineer & Co-founderaaronleslie.dev
- Unraveling The JPEGparametric.press
- Color: From Hexcodes to Eyeballsjamie-wong.com
- [2309.10668] Language Modeling Is Compressionarxiv.org
- Mediumpub.towardsai.net
- Compression and Intelligencegreene.sh