Inspiration
I'm a aspiring videographer and so I play around with many video generation models.
What it does
The project runs a video generation model on a single H100 gpu.
How we built it
I had codex read some recent papers. I first started with the Mobile Wan paper.
Challenges we ran into
I'm not an ML Researcher by trade so it took me a bit to understand.
Accomplishments that we're proud of
It works! Codex got a video model running in real time!
What we learned
Combining next token prediction and diffusion is the future. Using Codex @goal is powerful!
What's next for Real-time Video Generation
I want to make it into a mobile app.
Log in or sign up for Devpost to join the conversation.