

Would that be like on a tripod in the audience area recording the screen?
yes.
Seems lossy somehow
it is! but that’s where Shannon sampling theorem comes in. The sampling only needs to be twice as good as the source, and then you can reconstruct the source perfectly. (with some assumptions, e.g. correct color gamut, focal point, etc.).




I assume the judge told openai to hand it over?