mirror of
https://github.com/modelscope/DiffSynth-Studio.git
synced 2026-03-19 06:48:12 +00:00
1.2 KiB
1.2 KiB
HunyuanVideo
HunyuanVideo is a video generation model trained by Tencent. We provide advanced VRAM management for this model, including three stages:
| VRAM required | Example script | Frames | Resolution | Note |
|---|---|---|---|---|
| 80G | hunyuanvideo_80G.py | 129 | 720*1280 | No VRAM management. |
| 24G | hunyuanvideo_24G.py | 129 | 720*1280 | The video is consistent with the original implementation, but it requires 5%~10% more time than hunyuanvideo_80G.py |
| 6G | hunyuanvideo_6G.py | 129 | 512*384 | The base model doesn't support low resolutions. We recommend users to use some LoRA (example) trained using low resolutions. |
Gallery
Video generated by hunyuanvideo_80G.py and hunyuanvideo_24G.py:
https://github.com/user-attachments/assets/48dd24bb-0cc6-40d2-88c3-10feed3267e9
Video generated by hunyuanvideo_6G.py using this LoRA:
https://github.com/user-attachments/assets/2997f107-d02d-4ecb-89bb-5ce1a7f93817