08:02 <dnkl> emersion: from https://codeberg.org/dnkl/foot/issues/2035, it appears kwin (and I suspect hyprland too) uses a plain 2.2 gamma function when given a linear surface. This causes more color differences than necessary (i.e. more than loss of color precision due to foot using 8-bit surfaces), since foot currently uses the split-function sRGB transfer function to decode to linear

08:02 <dnkl> thoughts on that? I believe your wlroots patch also uses the split function?

08:03 <dnkl> I have a vague memory reading an article (by pq?) on this topic, where it is suggested to not bother with the split-function variant, and only use the 2.2 gamma function

08:09 <dnkl> ah, yes, I was correct: https://zamundaaa.github.io/wayland/2023/12/18/update-on-hdr-and-colormanagement-in-plasma.html

08:09 <dnkl> "KWin now also uses gamma 2.2 instead of the sRGB piece-wise transfer function for sRGB applications, as that more closely matches what displays actually do in SDR mode. This means that in the dark regions of sRGB content things will now look like they do with HDR disabled, instead of things being a little bit brighter and looking kind of washed out"