OneWorld: Taming Scene Generation with 3D Unified Representation Autoencoder

Sensen Gao*, Zhaoqing Wang*, Qihang Cao, Dongdong Yu, Changhu Wang, Tongliang Liu📧, Mingming Gong📧, Jiawang Bian📧

TL;DR: **We present OneWorld, the first diffusion framework for 3D scene generation that operates directly in the feature space of pretrained 3D foundation models**, enabling unified modeling of geometry, appearance, and semantics for superior cross-view consistency.

✅ TODO

• Release inference code and checkpoints
• Release training code & dataset preparation

🎓 Citation

Please cite our paper if you find this repository useful:

@misc{gao2026oneworldtamingscenegeneration,
      title={OneWorld: Taming Scene Generation with 3D Unified Representation Autoencoder}, 
      author={Sensen Gao and Zhaoqing Wang and Qihang Cao and Dongdong Yu and Changhu Wang and Tongliang Liu and Mingming Gong and Jiawang Bian},
      year={2026},
      eprint={2603.16099},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2603.16099}, 
}