Jon Sneyers

> Let me clarify my position: I am against this change, unless we know that the fraction of such JPEGs actually in the wild is high enough. That's fair enough....

The main difference between MAANS and MANIAC is the entropy coding 'backend' — both use MA trees as a way to determine the context, but in MAANS the entropy coding...

I don't think there's anything wrong with cjxl/djxl here, and it's just the viewers that are not properly doing color management. The original image does use the Adobe RGB (1998)...

For now, I made a simple variant of the Thorium logo that besides the black ring also has "Th" in the middle: https://jpegxl.info/icons/sw_thorium.svg Feel free to use this one if...

The icon doesn't matter at all in my daily use of Thorium, but it is practical to have a somewhat clearer distinction between Chromium and Thorium, e.g. when you are...

It would be good to know if JPEG XL support will remain in Thorium after M109. I'm certainly hoping @Alex313031 can confirm this intention, since Thorium is currently the only...

What can we do to reduce development pressure? Would it help if we (libjxl devs) maintain an up-to-date chromium patch for jxl somewhere (e.g. on the chromium gerrit itself)?

Some examples of the workarounds people have resorted to 'in the wild': Just about any image on https://www.pixilart.com/ is actually a nearest-neighbor upsampled image. As a random example: the image...

Currently cjxl silently strips exif/xmp from a PNG, and only encodes the pixel data and the ICC profile. I didn't know Preview actually looks at DPI information to do "Actual...

The spec says decoders should assume there is an implicit first frame that is all zeroes in all channels (so transparent black). Looks like we don't handle this case properly.