I'd love to see a Core ML-powered deblur feature in Pixelmator Pro.
I saw one demoed years ago in Photoshop, and there are a few open source projects out there (convolution deblur). Old photos that are blurred because of motion (slow shutter, blurred because of the shutter being pressed too hard or motion from an unsteady hand) can be fixed up pretty well. These photos are often deleted, because most people don't know that they can be improved or even fixed completely.
I used SmartDeblur (https://github.com/Y-Vladimir/SmartDeblur) to fix a late 19th century glass negative that was blurred and the results were pretty astounding, but the software was slow and tricky to use. The negative was about to be thrown away because it was 'useless'.
This could work really nicely in tandem with ML Super Resolution.
What features would you like to see in Pixelmator Pro?
We've actually been doing research in this area recently but weren't quite satisfied with the results we were getting. We've currently put this project aside, but if you have any blurry images you'd be able to share with us for testing purposes, that would be very much appreciated! Feel free to send them through to me at firstname.lastname@example.org.