Did you know you can play Doom on a diffusion model now? It’s true, Google just announced it! Just don’t read the paper too closely.
But it’s glitchy, the numbers don’t work, and you’ll notice the player never looks behind them.
I watched the video and there are two scenes where the player turns to look directly back where they just came from.
The demo was neat, but it was hilariously overhyped, even in the abstract paper. It was pretty damn obvious that the researchers were just trying to continue funding their research with a PR push.
The more interesting aspect was the potential for better AI video processing, not creating a game engine. You can’t create a game engine without a series of defined rules, and you can’t define those rules without documenting it in programming language.
