Google's Super-Resolution Algorithm SR3?

Any chance that we might get something like this for video in the future?

This would be incredible. Even if it takes a long time to render, I’d love for something like this to make it into VEAI. VEAI doesn’t really add details that well. If you took a pixelated video like this and tried to upscale it, it would not look great in VEAI. Being able to add detail like this to old videos would be a game changer.


I just worry that once put to images “in the wild”, this AI wouldn’t do so amazing. All the example pixelated images they show are pretty clean compared to many of the background faces I’ve been seeing in the videos I try to enhance.
But if not, it would be a great improvement.

There are ESRGAN models that can do this. It’s not as effective, and much, much slower. Would be great to see an AI or a model for this on VEAI.