dmitry, please don't make bug tasks unless someone looked them over. This is also not a critical bug. I am also pretty sure gimp is at fault here.

Hi, @rempt!

Can I ask you not to close the tickets I create until we have at least some agreement about the problem? I am not a dumb bot that just copies all the tickets from bugzilla to phabricator. I do it not so often, you know it.

I can clearly see at least two problems revealed in this report:

1. In your attached images, the image from Krita 2.9 looks much sharper than the one from 3.1. That looks like a regression.
2. Comparing the results of Linear filter in Photoshop and Gimp with Krita, they look really different: Gimp and Photoshop produce the same, nice and sharp result, Krita produces some weird blurry thing. Whatever "bug" could be in Gimp/Photoshop, it solves user's usecase: generates nice and sharp scaled down result. Therefore, to avoid such bugs in the future, we should either fix our linear filter or add some note into the manual, how our linear algorithm differs from Gimp's and PS's algorithm and how to solve the usecase correctly.

The priority of the two listed bugs can be questioned, but they are surely not Invalid.