Read the article and I am curious:
We have tons of different individual frames for the characters in e.g. the bar of the Lexington. So its individual images. However, I doubt you are talking about tons of photographs, aka still shots of people, instead I suspect that the actor was recorded doing the motion and the individual frames of that "movie"-shot were encoded as images? If so, that would mean all the blur and motion one gets from a camera is part of the image and in that case, would it not be wiser to actually combine the images back into a "movie" and upscale that one instead of working on the static images?
No critique intended, this is really just a question.