Hands-on, tested tutorials for running MiniMax H3 video generation in ComfyUI: the exact node setup, downloadable workflow files, measured GPU/system-RAM results, and fixes for the errors people actually hit.
Website: https://www.minimaxh3tutorial.com
The JSON files in this repository are the official ComfyUI MiniMax H3 templates (source: Comfy-Org/workflow_templates, commit 7837633a), mirrored here so you can drag them straight into ComfyUI. The tutorials explain the settings that actually matter — resolution, clip length, VRAM headroom, and where the common failures come from.
| Workflow | File | Step-by-step tutorial |
|---|---|---|
| Text-to-Video (T2V) | video_minimax_h3_t2v.json |
T2V tutorial |
| Image-to-Video (I2V) | video_minimax_h3_i2v.json |
I2V tutorial |
| First/Last-frame (R2V) | video_minimax_h3_r2v.json |
R2V tutorial |
- Setup & install — get MiniMax H3 running in ComfyUI from scratch: /comfyui/
- Measured hardware results — real generation-time and VRAM numbers on consumer GPUs (RTX 3060, RTX 4060) and system RAM fallback: /gpus/ · /vram/ · /rtx-3060/ · /system-ram/
- LoRA usage — loading and stacking LoRAs with H3: /loras/
- Troubleshooting — node errors, OOM, bad outputs, and their fixes: /troubleshooting/
No signup, no paywall — everything is free to read.
This is an unofficial community project and is not affiliated with MiniMax or Comfy Org. All product names, logos, and brands are property of their respective owners.
- The workflow JSON files are MIT-licensed by Comfy Org — see
LICENSE-comfy-templates.txt. - All other content in this repository and on the website is © minimaxh3tutorial.com.