Hanxiao Wang
Portrait
PhD Candidate
Institute of Automation, Chinese Academy of Sciences (CASIA)

I am a PhD candidate at CASIA. My research focuses on 3D representation, mesh generation, and world models.

I am interested in building generative systems that learn structured 3D understanding and produce high-quality geometry efficiently.

I am always open to collaboration and would be glad to discussā€”please feel free to reach out.

Google Scholar GitHub ORCID
Education
  • University of Chinese Academy of Sciences
    B.S. in Computer Science and Technology
    Sept 2018 - June 2022
  • University of Chinese Academy of Sciences, Institute of Automation (CASIA)
    M.S. in Computer Application Technology (Supervised by Prof. Dong-Ming Yan)
    Sept 2022 - Aug 2024
  • University of Chinese Academy of Sciences, Institute of Automation (CASIA)
    Ph.D. in Computer Application Technology (Supervised by Prof. Dong-Ming Yan)
    Sept 2024 - June 2027 (Exp.)
Experience
  • VAST, Beijing, China
    June 2025 - Present
    Research Intern, Vstar (Mentors: Dr. Yan-Pei Cao, Dr. Yuan-Chen Guo, Dr. Ying-Tian Liu, and Dr. Peng Wang)
    • AR Mesh Generation
      June 2025 - Dec. 2025
    • Nexus
      Dec. 2025 - Mar. 2026
    • World Model
      Mar. 2026 - Present
  • King Abdullah University of Science and Technology (KAUST)
    May 2024 - Apr 2025
    Remote Research Intern (Supervised by Prof. Peter Wonka and Dr. Biao Zhang)
News
2026
Selected (Apr 2026): FACE: A Face-Based Autoregressive Representation for High-Fidelity and Efficient Mesh Generation has been selected as a CVPR 2026 Highlight.
Apr
Accepted (Apr 2026): Nexus has been accepted to the SIGGRAPH 2026 Journal Track.
Apr
Launched (Mar 5): šŸ”„ TripoP1.0 is officially online! The fastest and best next-generation polygon mesh algorithm, generating low-poly meshes in just 2 seconds! Try it here
Mar
Accepted (Feb 2026): FACE: A Face-Based Autoregressive Representation for High-Fidelity and Efficient Mesh Generation. IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2026.
Feb
2025
Accepted (Dec 2025): Autoregressive Generation of Static and Growing Trees. Proceedings of the SIGGRAPH Asia 2025 Conference Papers, 1-12.
Dec
Accepted (Oct 2025): E(3)-Net: Efficient E(3)-Equivariant Normal Estimation Network. IEEE Transactions on Visualization and Computer Graphics (TVCG).
Oct
iFlame (arXiv 2025) released: Interleaving full and linear attention for efficient mesh generation. Google Scholar
Mar
New ICLR 2025 paper: On the Performance Analysis of Momentum Method: A Frequency Domain Perspective. Google Scholar
Jan
2024
VQ-CAD (Computer Aided Geometric Design 2024) paper page. Paper page
Nov
2023
ICCV 2023 paper: Structure-aware surface reconstruction via primitive assembly. Paper page
Nov
Selected Publications (view all )
Nexus: Native Mesh Generation with Diffusion
Nexus: Native Mesh Generation with Diffusion

H Wang*, YT Liu*, YC Guo, QY Feng, ZX Zou, D Liang, B Zhang#, YP Cao# (* equal contribution, # corresponding author)

ACM Transactions on Graphics (SIGGRAPH 2026) 2026

Diffusion-based mesh generation with decoupled vertex and topology modeling via Spacetime Interval embeddings.

[DOI]

Nexus: Native Mesh Generation with Diffusion

H Wang*, YT Liu*, YC Guo, QY Feng, ZX Zou, D Liang, B Zhang#, YP Cao# (* equal contribution, # corresponding author)

ACM Transactions on Graphics (SIGGRAPH 2026) 2026

Diffusion-based mesh generation with decoupled vertex and topology modeling via Spacetime Interval embeddings.

[DOI]

FACE: A Face-based Autoregressive Representation for High-Fidelity and Efficient Mesh Generation
FACE: A Face-based Autoregressive Representation for High-Fidelity and Efficient Mesh Generation

H Wang, YC Guo, YT Liu, ZX Zou, B Zhang, W Quan, D Liang, YP Cao, DM Yan

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2026

Face-based autoregressive representation for high-fidelity and efficient mesh generation.

[arXiv] [PDF]

FACE: A Face-based Autoregressive Representation for High-Fidelity and Efficient Mesh Generation

H Wang, YC Guo, YT Liu, ZX Zou, B Zhang, W Quan, D Liang, YP Cao, DM Yan

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2026

Face-based autoregressive representation for high-fidelity and efficient mesh generation.

[arXiv] [PDF]

Autoregressive Generation of Static and Growing Trees
Autoregressive Generation of Static and Growing Trees

H Wang, B Zhang, J Klein, DL Michels, DM Yan, P Wonka

Proceedings of the SIGGRAPH Asia 2025 Conference Papers 2025

Autoregressive framework for generating both static and growing tree structures.

Autoregressive Generation of Static and Growing Trees

H Wang, B Zhang, J Klein, DL Michels, DM Yan, P Wonka

Proceedings of the SIGGRAPH Asia 2025 Conference Papers 2025

Autoregressive framework for generating both static and growing tree structures.

iFlame: Interleaving Full and Linear Attention for Efficient Mesh Generation
iFlame: Interleaving Full and Linear Attention for Efficient Mesh Generation

H Wang, B Zhang, W Quan, DM Yan, P Wonka

arXiv preprint arXiv:2503.16653 2025

Efficient mesh generation with interleaved full and linear attention.

[Project] [Code] [arXiv]

iFlame: Interleaving Full and Linear Attention for Efficient Mesh Generation

H Wang, B Zhang, W Quan, DM Yan, P Wonka

arXiv preprint arXiv:2503.16653 2025

Efficient mesh generation with interleaved full and linear attention.

[Project] [Code] [arXiv]

E(3)-Net: Efficient E(3)-Equivariant Normal Estimation Network
E(3)-Net: Efficient E(3)-Equivariant Normal Estimation Network

H Wang, M Zhao, W Quan, Z Chen, DM Yan, P Wonka

IEEE Transactions on Visualization and Computer Graphics 2025

Efficient E(3)-equivariant neural network for normal estimation.

E(3)-Net: Efficient E(3)-Equivariant Normal Estimation Network

H Wang, M Zhao, W Quan, Z Chen, DM Yan, P Wonka

IEEE Transactions on Visualization and Computer Graphics 2025

Efficient E(3)-equivariant neural network for normal estimation.

All publications