Scoping WPT reftest coverage for JPEG XL HDR and wide-gamut color

[Minju-kim-igalia]

web-platform-tests/wpt#58481 added substantial JPEG XL coverage. HDR and wide-gamut color still look like an area where some focused follow-up would be useful.

The Interop 2026 JPEG XL roadmap calls out HDR / wide-gamut testing as an area where a solution or RFC would be useful, so I wanted to open a thread focused specifically on that part of the work.

I’m taking an initial look alongside Chromium issue 483413433 and web-platform-tests/wpt#44320 to better understand what automated coverage here might depend on.

To help structure that discussion, it seems useful to separate the work into:

1. shared screenshot / reference-output questions,
2. wide-gamut SDR cases,
3. HDR-specific cases such as PQ, HLG, and gain maps.

Input would be very helpful on:

• For wide-gamut SDR, should the first strict automated tests focus on the existing Adobe RGB / Display P3 / Rec.2020 coverage?
• If so, how should expected PNG references be generated for those tests: by a browser-independent rule, or by using one decoder implementation as the basis for the expected PNGs?
• For HDR, should the first automated target be PQ / HLG rendering behavior, while keeping gain maps in scope for the same RFC / design work?
• For strict reftest comparison in these cases, should the expectation be exact match, or a tightly-bounded fuzzy policy?
• Do we need small canonical assets for P3 / Rec.2020 / PQ / HLG / gain maps, rather than relying mainly on the current large conformance image and sRGB-only 3x3 fixtures?