One Fused Turbo model, three ways to start a clip: from text, from your own picture, or from a picture used as a reference for a new shot.
Four tested graphs (plus an 8-step version for comparison), a sample picture and a quick start guide. No model files inside.
Watch the video: https://www.youtube.com/watch?v=a6hvehUq1nY
What you need
- ComfyUI with the core MiniMax H3 nodes (tested on 0.32)
- MiniMax H3 Fused Turbo in models/diffusion_models
- The H3 text encoder and the video and audio VAEs from Comfy-Org/MiniMax-H3
- QUICKSTART.md in the zip lists every file name and folder
How to use it
- Unzip and drag one of the .json graphs into ComfyUI (the .api.json files are for API use).
- Text: describe a shot. Image: start the clip from your picture. Reference: use a picture as a guide for a different shot.
- Pick the model files in the loaders, keep the CLIP loader type on minimax.
- Run one graph at a time while you test.
The step-by-step guide, the video and more free workflows are on my portal.


