A narrower test for a web codec
A compression engineer has published a detailed case against adopting JPEG XL as a general web image codec, arguing that the format’s broad feature set does not translate into a compelling advantage for typical browser use. The analysis acknowledges that JPEG XL is royalty-free, technically capable and a clear upgrade over legacy JPEG, but says the relevant comparison in 2026 is with modern alternatives, particularly AVIF.
The author, Gianni Rosato, says most images delivered on the web need versatile lossy compression rather than archival-grade lossless performance. In his assessment, JPEG XL’s lossless advantage is modest and drawn from test material that is not representative of ordinary web pages. He cites an approximately 11.9 percent size improvement over lossless WebP in one dataset, then argues that this saving applies to too small a share of traffic to justify another browser codec by itself.
Rosato’s broader claim is that encoder quality matters as much as the format specification. He says modern AV1 and AVIF encoders have received perceptual tuning and currently perform better on speed and fidelity per bit in the comparisons he examined. He also cautions that no benchmark can establish a codec’s permanent ceiling: measurements compare particular encoders and settings, while future software could produce different results. His conclusion is therefore about present implementations and the likely effort needed to close the gap.
Features weighed against deployment cost
JPEG XL supports capabilities including high channel counts, flexible color depth, progressive decoding and lossless recompression of existing JPEG files. The analysis questions how often web developers need the most expansive options. For mainstream pages, Rosato argues, RGB or YUV plus alpha, practical high-dynamic-range depth and fast loading cover the dominant requirements.
Progressive rendering is another contested point. The post says AVIF can now present a usable early preview after only a small fraction of an image has arrived, weakening an advantage once associated with JPEG XL. It also examines JPEG recompression, which can reduce the size of an existing JPEG without changing its pixels. Rosato cites a commonly discussed saving of about 20 percent but says decoding such recompressed files takes longer, making the trade-off less than free.
The article further reports size-matched decoding tests in which WebP remained more than ten times faster than the tested Rust JPEG XL decoder despite producing a somewhat larger file. Those figures reflect the author’s selected images, software and settings, not a universal performance guarantee. Independent testing across devices and content types would be needed before generalizing them.
The post arrives as Rust-based JPEG XL decoding work appears in Firefox and Chrome projects, renewing interest after Chrome dropped experimental support in 2023. Rosato does not dispute that safer implementations could reduce security concerns. Instead, he argues that browser support also carries engineering, maintenance and compatibility costs, so a new decoder must deliver clear benefits for common workloads. The analysis sharpens the technical debate, but it remains one engineer’s evidence-based position rather than a browser-vendor decision.
