Uncertainty about gamma correction context for custom shader

[nick87720z]

While writing shader for subpixel image filtering, I only just discovered, that it's one of those things, sensitive on gamma. Before I added gamma correction to my shader, it's result produced color fringes in many places. Previously I did not understand, how could these artifacts appear. So now I can tell why - because it must be done in the linear color space. Because after I applied the correction, artifacts disappeared and text started to look almost like from freetype (may be like in some older qt version, e.g. qt4).

However, it seems, that gamma correction affects more graphical operations, including image scaling and filtering at all.

For now I ended up in the confusion about what color space where to be used. At least in my setup - I have only laptop LCD without hardware knobs. There are two ways for display-wide correction, which seem to stack each on other if set both (thus should be never used simultaneously for same thing):

• xgamma way, also used by monica (xgamma fltk gui) and (in exagerated way) Section "Monitor" / Gamma option in xorg.conf.
• xrandr way, used by redshift, sct and xcalib (judging by how previous two override xcalib settings).

It's unclear, if entire desktop is expected to be in processing (linear) color range or only software with special support. Monica help says, that what it helps for is making sure it's in gamma 2.2 color space. But I can't understand:

• whether it means, that linear color values must need additional correction in order to have expected look in properly calibrated system
• Or this correction it rather to ensure, that those color values are automatically converted to perception color space without need for additional correction.

In case if first case takes place, this means, that picom has to expect all window content to be surely uncorrected, as well as make sure, that its composite output is uncorrected as well.