I have tried the three photo programs and Photo AI on some photos that have a lot of flaws. The method I found with the best success was to create multiple duplicates of the single photo, import that into Video AI 2.64 I then use the various models to see what looks best. Artemis MQ and LQ and Dehalo give me the best results at 100% deblock. I then have to turn it into a video then capture a single frame with some other software then apply Photo AI to the captured frame which is now a photo. I would like to streamline this so that Video AI can accept a single photo which I can then apply the various Video models too and then have it output as a single image.
Can’t you just set the output to an image sequence?
I have used both VEAI and TVAI for enhancing single images, like you said, by making about 30 copies and processing them as a video. Then I pick one image from the output folder as my final image.
I’m not going to vote for this feature because the AI models are made to work on movies. They need image sequences to work how they were trained. This use case is more of an outlier that can happen, rather than what it was ever meant to do.
