Yan Zhang
| homepage at ETH Zurich | homepage at MPI Tuebingen |
About Me
I am a senior/lead research engineer at Epic Games, working in the Digital Human Research Department led by Prof. Michael J. Black. My research experiences have been focusing on generative human foundation models, covering human motion and behavior synthesis, 3D human perception, generative models, and their applications in embodied AI, interactive characters, and beyond.
Short Bio
- Research Lead at Meshcapade(acquired by Epic Games.), generative human foundation models, 2024-2026.
- Postdoc researcher at VLG, ETH Zurich, supervised by Prof. Siyu Tang, 2020-2023.
- PhD defense in June 2020 from Ulm University, supervised by Prof. Heiko Neumann.
- Research intern at MPI for Intelligent Systems, working with Prof. Michael J. Black, Prof. Siyu Tang, and PS folks, 2018-2020.
- More details in my CV.
News
- [May. 2026] EgoGen wins the EgoVis 2024/2025 Distinguished Paper Award.
- [May. 2026] Awarded as outstanding area chair at CVPR’26.
- [Aug. 2025] Serve as area chair for CVPR’26.
- [Aug. 2025] serve on the Senior Program Committee, AAAI’26
- [Jun. 2025] PRIMAL is accepted by ICCV’25.
- [Apr. 2025] Guest lecturer of Artificial Intelligence for Digital Characters, ETH Zurich.
- [Apr. 2025] SmallGS for camera motion estimation in Tiktok-like videos is accepted as a CVPR’25 workshop paper.
Services
Chair & Review
- serve on the Senior Program Committee, AAAI'26
- Area Chair: 3DV'24, CVPR'24, CVPR'25, CVPR'26
- Reviewers of Siggraph Asia'23, CVPR, ICCV, ECCV, 3DV regularly
- TPAMI, EuroGraphics
Activities
- Organizer of the workshop New Challenges in 3D Human Understanding at CVPR'25.
- Organizer of the workshop Foundation Models for 3D Humans at ECCV'24.
- Invited guest lecture of RL in Behavior Modeling in Artificial Intelligence for Digital Characters - SS 24/25 Computer Graphics Lab, ETH Zurich.
- Organizer of the workshop Human Body, Hands, and Activities from Egocentric and Multi-view Cameras at ECCV'22.
Other Affiliations
- Max-Planck ETH Center for Learning Systems (CLS), 2020-2023
- ETH AI Center, 2020-2023
Teaching
Co-supervised Student Projects at ETH Zurich
- Pascal Troxler & Onat Vuran, Multiview hand motion capture with diffusion models in Immsersive Design Lab, 2023
- Chuqiao Li, 3D human pose estimation from egocentric images, 2023
- Haoliang Shang, Retargeting SMPL-X bodies to Metahumans, 2023
- Yelan Tao, Virtual Humans meet Real Drones: Drone flight simulation in populated scenes, 2023
- Lukas Bösiger, Navigating digital humans in mixed reality, 2022
- Kaifeng Zhao, Semantically Controllable Human-Scene Interaction Synthesis, 2022
- Jonathan Lehner, Digital human navigation in the verticle city, 2022
- Yan Wu & Jiahao Wang, Stochastic Whole-Body Grasping with Contact, 2021
- Dexin Yang, 4D Human Body Capture from Egocentric Video via 3D Scene Grounding, 2020
- Siwei Zhang, Proximity learning of articulation and contact in 3D environments, 2020
Lectures at ETH Zurich
- Reinforcement Learning in Behavior Modeling, Artificial Intelligence for Digital Characters, 2024
- Human motion modeling I and II, Virtual Humans, 2022
- The wanderings of Odysseus in 3D scenes, Mixed Reality, 2022
Seminars at ETH Zurich
- Seminar in Advanced Topics in Vision, 2020,2021,2022,2023
- Seminar on Digital Humans, 2022,2023
Selected Publications
-
Please see all at Google Scholar
-
arxiv
arXiv, 2026.An agentic approach to generate 3D HSI in the open world with open-vocabulary command -
ICCV
International Conference on Computer Vision (ICCV), 2025.A diffusion model produces avatar's motions and reactions in real time. -
CVPR
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024.A compact implicit motion field with spatiotemporal regularity. -
CVPR
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024. (ORAL)Human motion synthesis with egocentric visual perception; Rendered synthetic data boosts several egocentric vision tasks. -
CVPR
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024.A two-stage coarse-to-fine diffusion architecture for extremely long dance generation; characteristic dance primitives. -
ICCV
IEEE/CVF International Conference on Computer Vision (ICCV), 2023.Generative motion pritimives + RL to synthesize lifelike behavior in 3D scenes. -
ICCV
IEEE/CVF International Conference on Computer Vision (ICCV), 2023. (ORAL)A scene-conditioned diffusion model regresses 3D bodies from single images. The denoising process guided by COAP solves interpenetration. -
CVPR
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022.Generative motion pritimives + RL to synthesize locomotion behavior of diverse human identities. Featured at the homepage of ETH Zurich. -
CVPR
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021.A novel marker-based body representation for stochastic motion prediction. -
CVPR
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020 (ORAL) .generative model to populate scenes + test-time optimization -
ECCV
European Conference on Computer Vision (ECCV), 2022.A dataset of people interactions in 3D scenes, captured by Kinects and Hololens. -
ICCV
IEEE/CVF International Conference on Computer Vision (ICCV), 2021 (ORAL) .Marker motion priors to capture body motions in 3D scenes.
Powered by Jekyll and Minimal Light theme.