JPEG XL is a technically proficient image codec developed by the JPEG Committee, and it was rejected from Chrome in 2023 despite being royalty-free and flexible. Recently, a JPEG XL decoder in Rust has appeared in Firefox and Chrome, prompting questions about whether the format deserves a place on the Web.
According to compression engineer Gianni Rosato, JPEG XL’s practical advantages are limited for most Web use cases. The format’s primary strength is lossless compression, where it achieves roughly 11.9% smaller file sizes than lossless WebP on test datasets. However, Rosato argues this advantage is insufficient to justify introducing a new codec to browsers, since most Web content relies on versatile lossy compression rather than lossless encoding.
For lossy compression—where the real Web demand exists—JPEG XL faces mounting competition. Modern encoders for competing formats like AVIF have received specialized perceptual tuning based on controlled human trials, and performance metrics show AVIF and other codecs performing significantly better than JPEG XL’s reference encoder on both speed and fidelity-per-bit measures.
Rosato identifies several technical limitations contributing to JPEG XL’s lossy performance gap. The codec lacks directional prediction modes, which other block-based codecs use to preserve edges more effectively. Its proposed solution—using splines—is more complex to implement and lacks proof-of-concept demonstrations for edge preservation. Additionally, JPEG XL lacks a proper deblocking loop filter, instead relying on alternative tools that cannot fully replace this functionality.
Other factors include JPEG XL’s XYB colorspace, which relies on aggressive quantization that sometimes yields subpar color preservation, and weaker performance on non-photographic images compared to alternatives like AVIF’s Intra Block Copy feature. Patches, JPEG XL’s proposed solution for such content, carry higher encoding overhead and currently require explicit enabling below effort level 7 due to performance issues.
Rosato, who previously endorsed JPEG XL and has worked on advancing AVIF, decided not to build his own encoder around JPEG XL after analyzing the compression landscape. He emphasizes these observations reflect the current state of image compression in 2026 rather than critiques of the format’s authors or their technical conduct.
Key facts
- JPEG XL was rejected from Chrome in 2023 despite being royalty-free and developed by the JPEG Committee
- JPEG XL’s lossless advantage over WebP is approximately 11.9% on test datasets, but applies to a small volume of Web content
- Modern AVIF encoders, particularly with perceptual tuning, outperform JPEG XL on both speed and fidelity-per-bit metrics
- JPEG XL lacks directional prediction modes and proper deblocking loop filtering, limiting edge preservation capabilities
- The format performs poorly on non-photographic images, with patches offering a less efficient alternative than competing solutions