> A computer first would have to be able to recognize objects, then use that > information to analyze how they are moving an where their borders are and > then it still must make up the parts that are not visible in either frame There are TVs out there which do it fairly convincingly, if such a thing can ever be convincing. It's also a technique common to special effects tools like ReVision FX's Twixtor, which is designed to produce slow motion effects from footage filmed at normal rates. If you think about it, it's pretty much the same technique used by motion compensated video codecs, so code to do the critical bits of it is available. Yes, of course, it does fail in certain circumstances, and special effects software designed to do this sort of thing usually includes tools for manual intervention on problem material. It's still a pretty horrible idea from the perspective of watching a movie, though. P
