Alibaba PAI open-sources unified ControlNet for MiniMax-H3: one 7GB checkpoint, five control modes
linoy_tsaban · x · 2026-08-24
Alibaba's PAI team released MiniMax-H3-Fun-Controlnet-Union on Hugging Face, trained with the VideoX-Fun pipeline to give the MiniMax-H3 video generator all-in-one control:
- A single 6.8GB checkpoint supports Canny, Depth, HED, MLSD and Pose control conditions plus video inpainting, no per-condition checkpoint switching
- The control branch attaches to only 5 of 50 transformer blocks (layers 0, 10, 20, 30, 40), with control skips added to the main branch via zero-gated projections
- Guidance-distilled: runs with guidancescale = 1.0, one forward pass per step, no CFG needed
- controlcontextscale tunes control strength; 0 disables the control branch entirely
Related event: Alibaba PAI Open-Sources Unified ControlNet for MiniMax-H3(2 posts)→
More from Multimodal
- Seedance 2.5 Video So Realistic You Forget It's AI Generated — Aiden_Tech_Ai · 2026-08-24
- Making a fake 80s-style movie trailer with Minimax H3 and Krea — NathanTheSnake · 2026-08-24
- Cohere's Tiny Aya Vision: sub-4B multilingual VLM covering 70+ languages — Cohere_Labs · 2026-08-24
- Simulating 90s Handheld-Cam MV Style with Minimax + ComfyUI — jordek · 2026-08-24
- France's first fully AI-generated talent show: no actors, no physical set — Smooth_School1283 · 2026-08-24
- Hands-on: Wan 3.0 generates 30-second clips with sound on Magnific — iamfakhrealam · 2026-08-24