I am currently a Senior Researcher at Tencent Hunyuan, leading a research team focused on AI 3D Pipeline Models. Our team is dedicated to reimagining the traditional 3D art pipeline using 3D generative models. Our research primarily covers: Artistic Mesh Generation, Unified 3D Multimodal Models, Auto UV Unwrapping, Auto Animation, etc. Welcome to explore our AI models at Hunyuan3D HomePage or Hunyuan3D Studio.
🔥🔥🔥We are constantly recruiting both full-time members and interns.🔥🔥🔥 If you are seeking any form of academic cooperation, please feel free to contact me via email
zhuooochen@tencent.com or WeChat
czxytg (To expedite the process, please include basic information, such as your name, school, and research focus.).
🔥 News
- 2026.02: 🎉🎉 Three papers are accepted by CVPR 2026.
- 2026.01: 🎉🎉 Three papers are accepted by ICLR 2026.
- 2025.09: 🎉🎉 Two papers are accepted by NeurIPS 2025.
- 2025.04: 🎉🎉 One paper is accepted by ICML 2025.
- 2025.03: 🎉🎉 One paper is accepted by CVPR 2025.
- 2022.04: 🎉🎉 One paper is accepted by ICML 2022.
- 2020.03: 🎉🎉 One paper is accepted by CVPR 2020.
- 2019.11: 🎉🎉 One paper is accepted by AAAI 2020.
- 2019.03: 🎉🎉 One paper is accepted by CVPR 2019.
📝 Publications
Technical Reports

Hunyuan3D Studio: End-to-End AI Pipeline for Game-Ready 3D Asset Generation
Zhuo Chen is project leader.
- An end-to-end AI-powered content creation platform designed to revolutionize the game production pipeline by automating and streamlining the generation of game-ready 3D assets.

Hunyuan3D 2.0: Scaling Diffusion Models for High Resolution Textured 3D Assets Generation
Zhuo Chen is project leader of the studio part.
Project | Paper | Code | Studio
- An advanced large-scale 3D synthesis system based on VAE & DiT framework for generating high-resolution textured 3D assets.

Hunyuan3D 1.0: A Unified Framework for Text-to-3D and Image-to-3D Generation
Zhuo Chen is project leader of the Multi-view Generation part.
- A two-stage 3D generation approach based on multi-view diffusion and sparse-view reconstruction, that support both text- and image condition.
Papers

Zhuo Chen is project leader.
- A novel 3D-aware RL algorithm named Advantage-guided Ranking Preference Optimization (ARPO).

X-Part: High Fidelity And Structure Coherent Shape Decomposition And Completion
Zhuo Chen is project leader.
Project | Paper | Code | Studio
- A controllable generative model designed to decompose a holistic 3D object into semantically meaningful and structurally coherent parts with high geometric fidelity.

MatPedia: A Universal Generative Foundation for High-Fidelity Material Synthesis
Zhuo Chen is project leader.
- A unified framework handling multiple material tasks—text-to-material, image-tomaterial, and intrinsic decomposition.

Part-X-MLLM: Part-aware 3D Multimodal Large Language Model
Zhuo Chen is project leader.
- A native 3D multimodal large language model that unifies diverse 3D tasks by formulating them as programs in a structured, executable grammar.

QuadGPT: Native Quadrilateral Mesh Generation with Autoregressive Models
Zhuo Chen is project leader.
- The first autoregressive framework for generating quadrilateral meshes in an end-to-end manner.

ArtUV: Artist-style UV Unwrapping
Zhuo Chen is project leader.
- A fully automated, end-to-end method for generating artist-style UV unwrapping.

Auto-Connect: Connectivity-Preserving RigFormer with Direct Preference Optimization
Zhuo Chen is project leader.
- A novel approach for automatic rigging that explicitly preserves skeletal connectivity through a connectivity-preserving tokenization scheme.

Mesh-RFT: Enhancing Mesh Generation via Fine-grained Reinforcement Fine-Tuning
Zhuo Chen is project leader.
- A novel fine-grained reinforcement finetuning framework that employs Masked Direct Preference Optimization (M-DPO) to enable localized refinement 3D mesh generation.

FreeMesh: Boosting Mesh Generation with Coordinates Merging
Zhuo Chen is project leader.
- A mathematical framework, PTME, to evaluate existing mesh tokenizers without any training.

BPT: Scaling Mesh Generation via Compressive Tokenization
- A compressive yet efficient mesh representation called Blocked and Patchified Tokenization.
💻 Experience
Full time
- 2023.08 - Present, Hunyuan 3D, Tencent, Senior Researcher, Leader of AI 3D Pipeline Models.
- 2023.08 - 2024.05, Sensetime, Senior Researcher.
- 2021.08 - 2023.08, Huawei, Postdoctoral Researcher.
Internship
- 2020.06 - 2021.08, Huawei, Intern.
- 2018.04 - 2020.06, Sensetime, Intern.
Educations
- 2015.09 - 2021.06, Institute of Automation, Chinese Academy of Sciences (CASIA), Master and Ph.D.