⚡️⚡️ Try it Now with Hunyuan3D-2 for super fast high-quality shape generation within 1 second on 4090.

flashvdm_demo.mp4

What is FlashVDM?

FlashVDM is a general framework for accelerating shape generation Vectset Diffusion Model (VDM), such as Hunyuan3D-2, Michelangelo, CraftsMan3D, CLAY, TripoSG, Dora and etc.

It features two techniques for both VAE and DiT acceleration:

1. Lightning Vectset Decoder that drastically lowers decoding FLOPs without any loss in decoding quality, achieving over 45x speedup.
2. Progressive Flow Distillation that enables flexible diffusion sampling with as few as 5 inference steps and comparable quality.

How to Use?

Visit Hunyuan3D-2 to access the integration of FlashVDM with Hunyuan3D-2.

from hy3dgen.rembg import BackgroundRemover
from hy3dgen.shapegen import Hunyuan3DDiTFlowMatchingPipeline

pipeline = Hunyuan3DDiTFlowMatchingPipeline.from_pretrained(
    'tencent/Hunyuan3D-2',
-   subfolder='hunyuan3d-dit-v2-0',
+   subfolder='hunyuan3d-dit-v2-0-turbo',
    use_safetensors=True,
)
+pipeline.enable_flashvdm()

pipeline(
    image=image,
-   num_inference_steps=50,
+   num_inference_steps=5,
)[0]

Supported Models

Hunyuan3D-2 series

Model	Description	Date	Size	Huggingface
Hunyuan3D-DiT-v2-0-Fast	Guidance Distillation Model	2025-02-03	1.1B	Download
Hunyuan3D-DiT-v2-0-Turbo	Step Distillation Model	2025-03-15	1.1B	Download

Hunyuan3D-2mini series

Model	Description	Date	Size	Huggingface
Hunyuan3D-DiT-v2-mini-Fast	Guidance Distillation Model	2025-02-03	0.6B	Download
Hunyuan3D-DiT-v2-mini-Turbo	Step Distillation Model	2025-03-15	0.6B	Download

Hunyuan3D-2mv series

Model	Description	Date	Size	Huggingface
Hunyuan3D-DiT-v2-mv-Fast	Guidance Distillation Model	2025-03-19	1.1B	Download
Hunyuan3D-DiT-v2-mv-Turbo	Step Distillation Model	2025-03-19	1.1B	Download

Citation

If you found this repository helpful, please cite our report:

@misc{lai2025flashvdm,
      title={Unleashing Vecset Diffusion Model for Fast Shape Generation}, 
      author={Zeqiang Lai and Yunfei Zhao and Zibo Zhao and Haolin Liu and Fuyun Wang and Huiwen Shi and Xianghui Yang and Qinxiang Lin and Jinwei Huang and Yuhong Liu and Jie Jiang and Chunchao Guo and Xiangyu Yue},
      year={2025},
      eprint={2503.16302},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2503.16302}, 
}