Zeqiang Lai — Multimodal & 3D Generation

About

I build scalable generative models for high-fidelity multimodal content creation, with a current focus on 3D. My recent work contributes to the Hunyuan3D series.

Hunyuan3D Series

Hunyuan3D 3.0

High-fidelity 3D generation at up to 1536³ resolution.

demo ↗

Hunyuan3D 2.5

A major upgrade in geometric fidelity and details.

report ↗

Hunyuan3D 2.0

Large-scale shape and texture generation.

report ↗ code ↗

Selected Publications See more →

CVPR 2026

LATTICE: Democratize High-Fidelity 3D Generation at Scale

Z. Lai*, Y. Zhao*, Z. Zhao, H. Liu, Q. Lin, J. Huang, C. Guo, and X. Yue

CVPR 2026 highlight

NaTex: Seamless Texture Generation as Latent Color Diffusion

Z. Lai*, Y. Zhao*, Z. Zhao, X. Yang, X. Huang, J. Huang, X. Yue, and C. Guo

ICCV 2025 highlight

Unleashing Vecset Diffusion Model for Fast Shape Generation

Z. Lai*, Y. Zhao*, Z. Zhao, H. Liu, F. Wang, H. Shi, X. Yang, Q. Lin, J. Huang, Y. Liu, J. Jiang, C. Guo, and X. Yue

NeurIPS 2024

VisionLLM v2: An End-to-End Generalist Multimodal Large Language Model for Hundreds of Vision-Language Tasks

J. Wu*, M. Zhong*, S. Xing*, Z. Lai*, Z. Liu*, W. Wang*, Z. Chen, X. Zhu, L. Lu, T. Lu, et al.

SIGGRAPH 2023

∇-Prox: Differentiable Proximal Algorithm Modeling for Large-Scale Optimization

Z. Lai*, K. Wei*, Y. Fu, P. Härtel, and F. Heide

Awards

Meshy Fellowship Grand Prize2026
CIE–Tencent Doctoral Research Fellowship2025
National Scholarship, China2022

Misc

Rubik’s Cube · photography · Dota 2 · currently playing Black Myth: Wukong