Jiazhi Shu

I am a first-year Ph.D. student at Intelligent Graphics Lab in HKUST (IGL-HKUST), supervised by Prof. Yuan Liu. Previously, I studied at South China University of Technology, majoring in Software Engineering with a minor in Business Administration.

My research interests lie in 3D human reconstruction, human motion modeling, and generative video modeling. I am interested in building realistic, animatable digital humans and understanding how people move and interact with the world.

Google Scholar GitHub Email

Preprints

* Equal contribution · † Corresponding author

Möbius unified motion-language model for motion generation, completion, editing, and understanding

arXiv2026Co-first author

Möbius: A Unified Motion-Language Model Bridging Understanding and Generation

Jiazhi Shu*, Chengfeng Zhao*, Jinyi Wang, Yubo Zhao, Zhiyang Dou, Ronghui Li, Yu Lu, Xinwei Jiang, Zijiao Zeng†, Yuan Liu†

Unifying motion understanding, generation, completion, and instruction-guided editing within a shared multimodal representation space.

Project ↗Paper ↗PDF ↗Code ↗

BibTeX

@article{shu2026mobius,
  title={M{\"o}bius: A Unified Motion-Language Model Bridging Understanding and Generation},
  author={Shu, Jiazhi and Zhao, Chengfeng and Wang, Jinyi and Zhao, Yubo and Dou, Zhiyang and Li, Ronghui and Lu, Yu and Jiang, Xinwei and Zeng, Zijiao and Liu, Yuan},
  journal={arXiv preprint arXiv:2611.05279},
  year={2026}
}

CoMoVi overview of co-generated human motions and realistic videos

arXiv2026

CoMoVi: Co-Generation of 3D Human Motions and Realistic Videos

Chengfeng Zhao, Jiazhi Shu, Yubo Zhao, Tianyu Huang, Jiahao Lu, Zekai Gu, Chengwei Ren, Zhiyang Dou, Qing Shuai, Yuan Liu†

Coupling human motion and video generation within a single diffusion denoising process.

Project ↗Paper ↗Code ↗Dataset ↗

BibTeX

@article{zhao2026comovi,
  title={CoMoVi: Co-Generation of 3D Human Motions and Realistic Videos},
  author={Zhao, Chengfeng and Shu, Jiazhi and Zhao, Yubo and Huang, Tianyu and Lu, Jiahao and Gu, Zekai and Ren, Chengwei and Dou, Zhiyang and Shuai, Qing and Liu, Yuan},
  journal={arXiv preprint arXiv:2601.10632},
  year={2026}
}

Publications

Tavatar animatable human avatars reconstructed from monocular video

CVPR2026Co-first author

Tavatar: Topology-Aware Gaussian Attribute Derivation for Animatable Human Avatars

Hailin Luo*, Yifan Yang*, Jiazhi Shu*, Zixiong Huang, Qi Chen†, Qing Du†, Mingkui Tan†

Deriving Gaussian attributes from mesh topology for robust human avatar animation under unseen poses.

Project ↗Paper ↗Code ↗

BibTeX

@inproceedings{luo2026tavatar,
  title={Tavatar: Topology-Aware Gaussian Attribute Derivation for Animatable Human Avatars},
  author={Luo, Hailin and Yang, Yifan and Shu, Jiazhi and Huang, Zixiong and Chen, Qi and Du, Qing and Tan, Mingkui},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year={2026}
}

WATCH reconstruction of camera trajectories and human motion in world coordinates

Pacific Graphics2026

WATCH: World-aware Allied Trajectory and pose reConstruction for Camera and Human

Zhongyuan Hu*, Qijun Ying*, Jiazhi Shu, Ronghui Li, Yu Lu, Zijiao Zeng†, Xiu Li†

Reconstructing global human motion with explicit camera orientation and learned camera trajectory priors.

Paper ↗PDF ↗Publication ↗

BibTeX

@inproceedings{hu2026watch,
  title={WATCH: World-aware Allied Trajectory and pose reConstruction for Camera and Human},
  author={Hu, Zhongyuan and Ying, Qijun and Shu, Jiazhi and Li, Ronghui and Lu, Yu and Zeng, Zijiao and Li, Xiu},
  booktitle={Pacific Graphics 2026 - Conference Papers and Posters},
  year={2026},
  publisher={The Eurographics Association},
  doi={10.2312/pg.20261036}
}