"Toys “R” Us has dropped the first ever AI-generated brand video using OpenAI’s Sora to tell the origin story of the company’s mascot Geoffrey the Giraffe."
https://petapixel.com/2024/06/26/toys-r … nais-sora/

One thing I used to like about Unreal Engine demos was how they were industry flagship standards of live rendering and the state of consumer GPUs and tooling at the time.

With generative AI we should start doing annual snapshots to show how far its come year over year as well

two years ago, there was no video generative AI, and at best you could sort of train frame by frame.

one year ago, video generative AI looked like an acid trip, with consistency between frames being impossible but resulting in its own art form.

this year we have these Sora clips, and other video generative AI competitors that simulate physics and everything to do a really good job at understanding consistency and object permanence. some people say "it has no soul" or something, and indeed these are mostly just 10 second panning, but I think its at the place where it wouldn't be judged any differently if you didn't tell people it was "AI" waiting for their reaction.

next year?