Project Lyra: Open Generative 3D World Models

NVIDIA Spatial Intelligence Lab

https://github.com/user-attachments/assets/f47c24c0-453e-4134-84f1-80c56613f4af

🫨 News

• 2026-07-20: 👋 We released the Lyra-2.0 GUI and training code — see the GUI instructions and training instructions.
• 2026-04-15: 👋 We released Lyra-2.0. Explorable generative 3D worlds with long-horizon, 3D-consistent generation.
• 2025-09-23: 👋 We released Lyra-1.0. Feed-forward 3D and 4D scene generation from a single image/video via video diffusion model self-distillation.

📝 Overview

Project Lyra is a series of open generative 3D world models developed at NVIDIA.

This repository provides the official implementations of Lyra 1.0 and Lyra 2.0.

License

Lyra source code is released under the Apache 2.0 License. Please refer to Lyra-1 and Lyra-2 for their respective model licenses.

Citation

If you find our series of models or the report helpful to your research or applications, please consider citing our paper.

@inproceedings{bahmani2026lyra,
  title={Lyra: Generative 3D Scene Reconstruction via Video Diffusion Model Self-Distillation},
  author={Bahmani, Sherwin and Shen, Tianchang and Ren, Jiawei and Huang, Jiahui and Jiang, Yifeng and 
          Turki, Haithem and Tagliasacchi, Andrea and Lindell, David B. and Gojcic, Zan and Fidler, Sanja and 
          Ling, Huan and Gao, Jun and Ren, Xuanchi},
  booktitle={International Conference on Learning Representations (ICLR)},
  year={2026}
}

@article{shen2026lyra2,
    title={Lyra 2.0: Explorable Generative 3D Worlds},
    author={Shen, Tianchang and Bahmani, Sherwin and He, Kai and Srinivasan, Sangeetha Grama and Cao, Tianshi and Ren, Jiawei and Li, Ruilong and Wang, Zian and Sharp, Nicholas and Gojcic, Zan and Fidler, Sanja and Huang, Jiahui and Ling, Huan and Gao, Jun and Ren, Xuanchi},
    journal={arXiv preprint arXiv:2604.13036},
    year={2026}
}