Why OpenAI’s Sora is big
Slides
Sora is able to create physically accurate animation using Diffusion technology, instead of a computationally expensive physics engine.
If combined with technologies like Turbo XL from Stability, which can generate near-instant txt2img, it can potentially mean near real-time video animation with temporal coherence.
That’s physically accurate animation and visualization in real time that don’t require a physics engine — think realistic fluid simulation e.g. RealFlow, but done without all the computationally expensive math — because it uses Diffusion tech instead.
Do you know which industries need this?
Gaming, movie production, CG — all of these industries are now semi-bottlenecked by the tremendous computational power required to create productions in a workable timeline.
An additional point of interest is that nVidia has long been a player in the GPU market that enabled these techs, but since they have also invested heavily in technologies related to AI, they’ll keep on being an important player.
I know that all of these companies are currently unrelated to each other, but there are so many potential collaboration opportunities, that if they’re not always collaborating now, I suspect that they will do it very soon.