97 lines
3.1 KiB
Markdown
97 lines
3.1 KiB
Markdown
|
|
|
|
|
|
-----
|
|
|
|
<div align="center">
|
|
|
|
# daVinci-MagiHuman
|
|
|
|
### Speed by Simplicity: A Single-Stream Architecture for Fast Audio-Video Generative Foundation Model
|
|
|
|
<p align="center">
|
|
<a href="https://plms.ai">SII-GAIR</a> & <a href="https://sand.ai">Sand.ai</a>
|
|
</p>
|
|
|
|
[](https://arxiv.org/abs/2603.21986)
|
|
[](https://huggingface.co/spaces/SII-GAIR/daVinci-MagiHuman)
|
|
[](https://huggingface.co/GAIR/daVinci-MagiHuman)
|
|
[](https://opensource.org/licenses/Apache-2.0)
|
|
[](https://www.python.org/)
|
|
[](https://pytorch.org/)
|
|
|
|
</div>
|
|
|
|
|
|
ComfyUI_MagiHuman
|
|
----
|
|
[DaVinci-MagiHuman](https://github.com/GAIR-NLP/daVinci-MagiHuman):Speed by Simplicity: A Single-Stream Architecture for Fast Audio-Video Generative Foundation Model
|
|
|
|
|
|
Update
|
|
----
|
|
* add layer offload num to fit high vram 大显存 把offload开到你跑得动为止，小显存则从1开始测试，MagiCompiler库不用，但是改了麻烦，还是加了回来
|
|
|
|
|
|
1.Installation
|
|
-----
|
|
In the ./ComfyUI/custom_nodes directory, run the following:
|
|
```
|
|
git clone https://github.com/smthemex/ComfyUI_MagiHuman
|
|
```
|
|
2.requirements
|
|
----
|
|
|
|
```
|
|
pip install -r requirements.txt
|
|
|
|
#python312以下，注释掉MagiCompiler目录下的pyproject.toml的13行的requires-python = ">=3.12" 再安装
|
|
git clone https://github.com/SandAI-org/MagiCompiler.git
|
|
cd MagiCompiler
|
|
pip install -r requirements.txt
|
|
pip install .
|
|
```
|
|
|
|
3.checkpoints
|
|
----
|
|
* dit and TE [links](https://huggingface.co/smthem/daVinci-MagiHuman-custom-comfyUI) or 国内用户 [夸克](https://pan.quark.cn/s/26c7d9d39c87)
|
|
|
|
```
|
|
├── ComfyUI/models/
|
|
| ├── diffusion_models/
|
|
| ├──distill-merger_bf16.safetensors #28G
|
|
| ├──540p_sr_merge_bf16.safetensors #28g For SR ,放大用开源不下
|
|
| ├── vae/
|
|
| ├──sd_audio.safetensors #4.7GM
|
|
| ├──Wan2.2_VAE.pth # 2.7G
|
|
| ├── gguf
|
|
| ├──t5gemma-9b-9b-ul2-Q6_K.gguf # 11G
|
|
|
|
```
|
|
|
|
4.Example
|
|
----
|
|
|
|
|
|
|
|
|
|
|
|
## 🙏 Acknowledgements
|
|
|
|
We thank the open-source community, and in particular [Wan2.2](https://github.com/Wan-Video/Wan2.2) and [Turbo-VAED](https://github.com/hustvl/Turbo-VAED), for their valuable contributions.
|
|
|
|
## 📄 License
|
|
|
|
This project is released under the [Apache License 2.0](https://opensource.org/licenses/Apache-2.0).
|
|
|
|
## 📖 Citation
|
|
|
|
```bibtex
|
|
@misc{davinci-magihuman-2026,
|
|
title = {Speed by Simplicity: A Single-Stream Architecture for Fast Audio-Video Generative Foundation Model},
|
|
author = {SII-GAIR and Sand.ai},
|
|
year = {2026},
|
|
url = {https://github.com/GAIR-NLP/daVinci-MagiHuman}
|
|
}
|
|
```
|