@whensdeyi says Lisa Simpson copies their face in real time, using 2-step MiniMax-H3-Image and Jev. Every frame is a fresh generation from MiniMax-H3-Image with the 2-step pDMD LoRA, about 3 seconds each, and the facial expression is detected by Jev from @typesafeai.
- 7.0k views
- 51 likes
- 40 bookmarks
- 2 reposts
About this case
Product. Dawna Yi (@whensdeyi) says this was made in like 30 minutes on the way home, and that a lot is going to happen with these super fast models. Video attached. The post quotes AK (@_akhaliq) on MiniMax-H3-Image in 2 steps, realtime: a pDMD LoRA on MiniMax's 33B image model, realtime editing on ZeroGPU, a LoRA plus one-file ComfyUI checkpoint at https://huggingface.co/akhaliq/minimax-h3-image-pdmd-2step, and an app at https://huggingface.co/spaces/akhaliq/minimax-h3-image-2step.
Original posthttps://x.com/whensdeyi/status/2120527444692308171