Of course no one knows the future, but I think it's very plausible that the future of films/games (especially games) tech resembles something like this:

1. Generation something that looks good in 2D latent space

2. Generation 3D representation from 2D

3. Next time the same scene is shown on the screen, reuse information from step 2 to guide step 1

That's an interesting idea! I'm thinking step 2 might be inserting 3d foreground/hero objects in front of a 2d background/inside a 2d worldbox