I use a dash cam to gather source material for OSM editing. Sometimes the video quality isn’t good enough to read the text on a sign, e.g. a street name. I wonder if there is some software tool that could take a couple of seconds long clip and I would assign an area of interest and the software would gather all data from the frames for that area and produce a photo where the street sign would hopefully be readable?
Are you thinking along the lines of “enhance… zoom… enhance” type of interpolation from the Bladerunner movie?
Something which takes multiple frames of video and picks the “clearest” pixels from each frame before stitching them all together to produced higher definition output than the input.
Give this a whirl - it apparently runs in wine without any issues.
Alternatively, VapourSynth can do what you want. Linux native binary too, so no messing around in wine if that’s not your cup of tea.