| Summary: | brush hardness and opacity blending bug | ||
|---|---|---|---|
| Product: | [Applications] krita | Reporter: | Jessie <Youcanthahalol> |
| Component: | Brush Engine/Bristle | Assignee: | sh_zam <shzam> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | CC: | dimula73, halla, shzam |
| Priority: | NOR | Keywords: | triaged |
| Version First Reported In: | 5.1.5 | ||
| Target Milestone: | --- | ||
| Platform: | Android | ||
| OS: | Linux | ||
| Latest Commit: | https://invent.kde.org/graphics/krita/commit/2a2d308626560db60fd12d81b956321c579f7bf3 | Version Fixed/Implemented In: | |
| Sentry Crash Report: | |||
| Attachments: |
Video of bug
Better look at the bug |
||
|
Description
Jessie
2023-03-28 20:25:08 UTC
https://youtu.be/2QZobGEOJgo I couldn't attach the video file so I uploaded it to YouTube Created attachment 157671 [details] Video of bug https://youtu.be/2QZobGEOJgo longer version Created attachment 157673 [details]
Better look at the bug
*** Bug 467896 has been marked as a duplicate of this bug. *** I haven't been able to reproduce. Sharaf, can you take a look? Confirming, this seems to happen on 32-bit float colorspace. A possibly relevant merge request was started @ https://invent.kde.org/graphics/krita/-/merge_requests/1782 Git commit 2a2d308626560db60fd12d81b956321c579f7bf3 by Sharaf Zaman. Committed on 13/04/2023 at 05:17. Pushed by dkazakov into branch 'master'. Fix colorsmudge brush creating NaN values This happened because xsimd:min, which calls vminq_f32 on ARM return NaN if any of there parameter contains a NaN, which is different from behavior of the instruction on x86. Now, we create a mask out of nan values and for those indices set the value to unitValue. This should work cross-platform. M +8 -1 libs/pigment/compositeops/KoOptimizedCompositeOpCopy128.h https://invent.kde.org/graphics/krita/commit/2a2d308626560db60fd12d81b956321c579f7bf3 |