Aurelia Molzer

This was fixed at some point, at least to the limits of lossy 8-bit jpeg compression. (We don't provide an API with more guarantees than that right now). Reproduction code...

The nature of our filter implementation makes performance quite dependent on auto-vectorization. If the issue is specific to a rustc version, there is very likely an llvm regression that I...

Revive https://github.com/image-rs/image-core with this, sure. Developing the `Color` type here allowed me a sort of reality-check, is it sufficiently well-defined to describe all parameters necessary for a full transformation of...

No I won't 😆 (well, the next part is consideration). The maths are completely pointless from there given that they are not vectorized and monomorphizing everything over color representations as...

It's way too many conceptual changes at once for what it's trying to add (also causing that a need to rebase again and again). We can take individual PRs for...

Yes that would be great :tada: We have an `Entry`/`Directory` structure that is supposed to be shared between different styles of directories and thus unrelated ID lists so it would...

Correct: - [x] Converting sRGB to Oklab is fine albeit newer versions round differently. Known issues: - [x] ~~Converting between sRGB and ITU.BT2020 is inaccurate.~~ ```python // theirs: [255, 255,...

So far so good. The biggest issue found after fixing coefficient bugs in transfer functions is that the conversion to XYZ and back is using multiplication in 32-bit floats and...

Would this interface work? ``` struct SbitChunk { // copied from the IHDR default_depth: u8, color_type: ColorType, // Filled from the chunk in order. parsed_depths: [u8; 4], } impl SbitChunk...

Using impls is inherently a bad fit for this purpose. As @kornelski sketched, the most approachable way to generalize is by listing the available and supported representations at some point...