I too noticed this when doing pixel art. Krita doesn’t seem to get it right when the mirror position is not exactly on the seam of the pixels.
what other software should i use? Krita clearly does a poor job, but the art still needs to be done, i am on linux.
I am gonna make a bug report, it doesn’t look to be intentional
Perhaps you’ll try this? →
At least in my eyes, $5 that every pixel art artist should consider spending. But you should ask the author if it solves your issue!
Michelist
The problem i am trying to fix looks like a bug to be honest, while the tool you linked looks cool, the behavior of the mirror tool should be fixed.
Then you should create a bug report on the KDE bug tracker at https://bugs.kde.org/, for this you need an e-mail that can be viewed publicly, because it can after you have submitted your report. But there doesn’t seem to be a flood of spam to worry about, at least I haven’t received a single mail from an addressee other than KDE on my mail account set up just for the bug tracker in the years I’ve been registered with KDE.
After sending in the report, please add a link to your report here in this topic, so there is a chance someone will see and confirm it if the user has the same issue you describe.
Please do not forget to name your OS, and the version of Krita, and add a link to this topic in the bug tracker, so the devs can find this topic for more information on the topic.
Michelist
100% a bug
I believe it is a bug or at least an oversight. A floating point precision/rounding issue would be my guess. Since the mirror tool allows to set the origin on half a pixel but Krita doesn’t have half pixels.
My workaround was to only put the origin on the pixel seams not on pixels itself. But that obviously only works for even pixel counts.
I believe this is the same problem?
But still it seems different
It’s definitively related. A rounding issue is suspected there too.
BTW thanks for linking this, i got it and it works like a charm…
For pixel artists using Krita, these are the best invested 5 bucks ever.
Michelist
I reproduced this specific bug and could confirm it also on Windows. I also found out that its only happening if you move the cursor while placing the pixel. I tried to show this on a video I attached to the bug report: https://bugs.kde.org/attachment.cgi?id=181332
I hope that helps finding the cause. It might be a rounding error, but somehow the input itself seems to be involved since it does not happen if the cursor is stationary.
Confirmed and added a comment to this other bug too. Its a different failure mode to the one you reported, but it could have the same root cause. It looks like if you place a mirror in the center of a pixel, something happens to the subpixel-coordinate which causes them to be shifted/misinterpreted.
I dont have much coding experience and am still learning how to use GitHub, but I would love to take a look at this part of the code. Any advice on I could approach this efficiently?