[need testing] New opacity handling in the brushes needs testing by users

[dkazakov]

Hi, all!

I’ve been recently working on an ancient opacity calculation bug from Krita 2.9.x era, and I have finally fixed it! The problem is that it changes the way how we handle opacity on a very low level, so I would really like someone to test the actual code before I merge it into master!

• the actual MR: Draft: Switch KisPainter to use floating point opacity instead of 0...255 integer range (!2268) · Merge requests · Graphics / Krita · GitLab

• testing packages: KDE - Experience Freedom!

Test plan:

1. Paint with different brushes with different non-100% opacity and flow levels. They should behave as usual
   • with normal Pixel bursh (non-standard opacity and flow)
     • Build-up mode
     • Incremental mode
   • with Color Smudge brush (non-standard opacity):
     • new algorithm
     • old algorithm
     • lightness-mode
2. Fill Tool with custom opacity value
3. Enclosed fill tool with custom opacity value (@Deif_Lou, ping )

Basically, everything should work “as usual” The main worry for me is the colorsmudge brush. Theoretically, the patch changes its algorithm a little bit. And again, theoretically, it should become a little bit smooth and precise. But there might be some issues that I don’t see while testing.

Could you please briefly check the packages and see if there any obvious regressions in the opacity handling?

[Reinold]

I’m not sure about incremental but if that is “wash” then that and build up mode with pixel brushes behave the same as the old algorithm.

I also used the wet bristles brush on normal and lightness and those also behave the same as old.

Will try to get to the other test cases later.

[cgidesign]

I tried with a 16 bit sRGB document, Windows, Krita 5.2.6, 5.3 alpha from above linked source.

with normal Pixel bursh (non-standard opacity and flow)

• Build-up mode
• Incremental mode (I assume this is meant to be wash mode?)

with Color Smudge brush (Blender Blur from Krita3 resources):

• lightness-mode (from HSL blending modes)
• normal-mode
  (I don’t know about new / old algorithm)

Fill Tool Bucket.

I did not notice a relevant difference in how drawing, blending, wash / build up, fill behave.

some patches are made 5.3 some in 5.2.6

[Zoldyako]

Tested here and everything looks alright

[dkazakov]

Hi, @Reinold, @cgidesign and @Zoldyako!

Thank you very much for testing! I will merge the patch then!