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
More from Multimodal
- PrunaAI's P-Video-2 Pro adds image/video/audio references from $0.01/s — aziz4ai · 2026-10-16
- ElevenLabs blind tests Eleven v4 voice model, wins all 8 languages on expressiveness — JenniferHli · 2026-10-16
- HyperFrames adds Reference Any Video and auto Brand Kits for AI video agents — alifcoder · 2026-10-16
- 'Alien Metamorphosis': AI short made with Grok Imagine + Topaz — creatoroff · 2026-10-16
- Early local test of distilled FLUX.3 Klein: mediocre t2i, editing hopes remain — Ant_6431 · 2026-10-16
- Krea 3 vs Qwen 2.1 vs ZIT: better facial nuance, but unsettlingly realistic outputs — Crazy-Repeat-2006 · 2026-10-16