This guide helps you use ComfyUI to run HunyuanVideo-1.5 model.
Install the latest version of ComfyUI from the official repository:
git clone https://github.com/comfyanonymous/ComfyUI.git
cd ComfyUI
For detailed installation instructions, please refer to the ComfyUI official repository.
Download the ComfyUI checkpoint from Hugging Face:
Model: https://huggingface.co/Comfy-Org/HunyuanVideo_1.5_repackaged
Find workflow templates at:
- 720p Image-to-Video: https://github.com/Comfy-Org/workflow_templates/blob/main/templates/video_hunyuan_video_1.5_720p_i2v.json
- 720p Text-to-Video: https://github.com/Comfy-Org/workflow_templates/blob/main/templates/video_hunyuan_video_1.5_720p_t2v.json
Important: If you enable the Tiled VAE decoding node, please set the 'temporal_size' to the maximum value (4096) to avoid generating artifacts.
- Important: Workflows are still being updated. Please ensure you are using the latest version of the workflow templates to guarantee the best generation results.
- Important: The distilled models include both CFG-distilled models and step-distilled models. CFG-distilled models should run with 50 inference steps for best generation quality. Step-distilled models (480p I2V) can run with 8 or 12 steps (recommended) for faster generation, or 4 steps for even faster speed with slightly reduced quality. The recommended configurations are provided in the following table.
- Important: Please strictly follow the configuration below to guarantee the best generation results.
Important! The following table is for the best generation quality. If you want to generate faster, you can use the following configurations:
| Model | CFG Scale | Embedded CFG Scale | Flow Shift | Inference Steps |
|---|---|---|---|---|
| 480p T2V | 6 | None | 5 | 50 |
| 480p I2V | 6 | None | 5 | 50 |
| 720p T2V | 6 | None | 9 | 50 |
| 720p I2V | 6 | None | 7 | 50 |
| 480p T2V CFG Distilled | 1 | None | 5 | 50 |
| 480p I2V CFG Distilled | 1 | None | 5 | 50 |
| 480p I2V Step Distilled | 1 | None | 7 | 8 or 12 (recommended) |
| 720p T2V CFG Distilled | 1 | None | 9 | 50 |
| 720p I2V CFG Distilled | 1 | None | 7 | 50 |
| 720p T2V CFG Distilled Sparse | 1 | None | 9 | 50 |
| 720p I2V CFG Distilled Sparse | 1 | None | 7 | 50 |
| 480→720 SR Step Distilled | 1 | None | 2 | 6 |
| 720→1080 SR Step Distilled | 1 | None | 2 | 8 |
- Use detailed, descriptive prompts for better results
- Refer to the HunyuanVideo-1.5 Prompt Handbook for guidance
- Include details about camera movement, style, lighting, etc.
If you develop/use HunyuanVideo-1.5 in your ComfyUI projects, welcome to let us know.
- ComfyUI-MagCache - ComfyUI-MagCache: MagCache is a training-free caching approach that accelerates video generation by estimating fluctuating differences among model outputs across timesteps. It achieves 1.7x speedup for HunyuanVideo-1.5 with 20 inference steps.
- ComfyUI Repository: https://github.com/comfyanonymous/ComfyUI
- HunyuanVideo-1.5 Repository: https://github.com/Tencent-Hunyuan/HunyuanVideo-1.5
- Model Checkpoint: https://huggingface.co/Comfy-Org/HunyuanVideo_1.5_repackaged
- Workflow Templates: https://github.com/Comfy-Org/workflow_templates/tree/main/templates
- Prompt Handbook: https://github.com/Tencent-Hunyuan/HunyuanVideo-1.5/blob/main/assets/HunyuanVideo_1_5_Prompt_Handbook_EN.md
For issues or questions, please visit the HunyuanVideo-1.5 repository to submit issues or check documentation.
