GitHub radar
Fix for distorted faces in MiniMax-H3 video model for ComfyUI
This is a tuned version of the MiniMax-H3 video model that removes facial distortion and unnatural skin shine. Install it in place of the regular model in ComfyUI and get cleaner video without extra steps.
It's a fine-tuned (reworked) version of the MiniMax-H3 video model, made by an independent creator, not by MiniMax.
When it helps
Useful if you already generate video in ComfyUI with MiniMax-H3 and have seen distorted faces in medium and wide shots. Skip it if you don't use ComfyUI or lack a powerful GPU.
Pros
- Fixes facial distortion in medium and wide shots
- Removes unnatural skin shine, colors look more realistic
- Drops in for the original — no workflow changes needed
- Dynamic scenes and effects look sharper
Cons
- Community fix, not an official MiniMax release — no guarantees
- Needs a GPU with 16+ GB VRAM, won't run on a laptop
How to set it up — step by step
- 1Open your AI agent (Claude Code, Codex) in your ComfyUI project folder.
- 2Send the agent the text from the block below — it will download the model and swap it in for the base checkpoint.
- 3Check that your GPU has 16+ GB VRAM — otherwise the model won't run.
- 4Open your current video generation workflow and run a test generation.
- 5Compare the result with the original MiniMax-H3 — especially faces in medium and wide shots.
Text for your agent
Copy this and send it to your agent — Claude Code, Codex, any of them:
Install this MiniMax-H3 variant in ComfyUI in place of the base checkpoint: https://huggingface.co/WarmBloodAban/Minimax-h3_Singularity — load the model and open my existing video generation workflow so I can compare the output quality against the original.
Requires a GPU with at least 16 GB of VRAM — the same as standard MiniMax-H3. A regular laptop will not run it.
Open on Hugging Face▌ More finds