https://bugs.kde.org/show_bug.cgi?id=425190
--- Comment #15 from Claudius Ellsel <claudius.ell...@live.de> --- (In reply to Nate Graham from comment #14) > David, I think Claudius is using the term "scaling" as a user understands > the term, not as a KWin developer understands the term. Looking at the > attachments, clearly there are differences in the different ways that images > can be scaled up. Knowing that fractional scaling downscales from a doubled > pixel-perfect version, it might be interesting to repeat the tests with > downscaled images. Ah, I see my mistake now. Roman Gilg already pointed that out, but I did not understand it then. I just assumed it would be scaled up. Downscaling obviously is the better approach to implementation, since no information has to be interpolated, so that makes sense. This renders the different algorithm comparisons I posted rather pointless. However, I read the the NoHalo algorithms were actually developed for downscaling in particular and that blurriness can also happen when scaling down. So the general idea that there might be better algorithms still seems to be valid. -- You are receiving this mail because: You are watching all bug changes.
