README.md

metadata

license: apache-2.0
language:
  - en
pipeline_tag: image-to-3d
modalities:
  - image
  - point clouds
  - mesh
arxiv: 2411.14384

[ICCV 2025] DiffusionGS: Baking Gaussian Splatting into Diffusion Denoiser for Fast and Scalable Single-stage Image-to-3D Generation and Reconstruction

Model Description

These three models are trained for image-to-3D generation on object- and scene-level with the spatial resolution of 256x256 and 512x512. For object-level 3D generation, mesh exportation is also supported. Here are some generated examples:

· (a) Object-level Generation

· (b) Mesh Exportation

· (c) Scene-level Generation

· (d) Comparison with Hunyuan3D-v2.5

The first row is the prompt image. The second row is Hunyuan3D-v2.5. The third row is our DiffusionGS.

Our method generates better results while enjoying 7.5x faster inference speed.

Prompt Images at Any Viewpoints
Tencent Hunyuan3D-v2.5 (Inference Time: 180 seconds)
Our DiffusionGS (Inference Time: 24 seconds)

Github Code Link

Please refer to our GitHub repo for more detailed instructions on using our code and models.

https://github.com/caiyuanhao1998/Open-DiffusionGS/

Project Page Link

For more video and interactive generation results, please refer to our project page:

https://caiyuanhao1998.github.io/project/DiffusionGS/

Arxiv Paper Link

For more technical details, please refer to our ICCV 2025 paper:

https://arxiv.org/abs/2411.14384

Citation

If you find our code, data, and models useful, please consider citing our paper:

@inproceedings{diffusiongs,
  title={Baking Gaussian Splatting into Diffusion Denoiser for Fast and Scalable Single-stage Image-to-3D Generation and Reconstruction},
  author={Yuanhao Cai and He Zhang and Kai Zhang and Yixun Liang and Mengwei Ren and Fujun Luan and Qing Liu and Soo Ye Kim and Jianming Zhang and Zhifei Zhang and Yuqian Zhou and Yulun Zhang and Xiaokang Yang and Zhe Lin and Alan Yuille},
  booktitle={ICCV},
  year={2025}
}