iTryOn: Mastering Interactive Video Virtual Try-On with Spatial-Semantic Guidance
Interactive video virtual try-on with spatial and semantic control.
Algorithm Engineer · Alibaba Group
Jun Zheng
My research focuses on virtual try-on.
I am an Algorithm Engineer in the AI Native Applications department at Alibaba, where I work on virtual try-on. My broader research interests include image and video generation.
I received my master’s degree in Control Science and Engineering from Sun Yat-sen University in June 2026, advised by Prof. Xiaodan Liang. I also received my bachelor’s degree from Sun Yat-sen University in 2023.
News
Recent updates- New work on compact image generation models, capability-centric data design, and real-world image editing evaluation.
- I joined Alibaba’s AI Native Applications department as an Algorithm Engineer, working on virtual try-on.
- I graduated from Sun Yat-sen University with a master’s degree in Control Science and Engineering.
- iTryOn was accepted to ICML 2026.
- We released Tstars-Tryon 1.0 and the Tstars-VTON benchmark.
Earlier updates
- WISA was accepted to NeurIPS 2025 as a Spotlight.
- Dynamic Try-On was accepted to BMVC 2025; our monocular-to-3D virtual try-on work was accepted to IEEE TPAMI.
- We released WISA for physics-aware text-to-video generation.
- We released Dynamic Try-On for video virtual try-on.
- We released VITON-DiT for in-the-wild video virtual try-on.
Selected Research
Google Scholar ↗Virtual try-on and visual generation. * Equal contribution.
Dynamic Try-On: Taming Video Virtual Try-on with Dynamic Attention Mechanism
Dynamic attention for temporally consistent video virtual try-on.
VITON-DiT: Learning In-the-Wild Video Try-On from Human Dance Videos via Diffusion Transformers
Learning in-the-wild video try-on from human dance videos.
Tstars-Tryon 1.0: Robust and Realistic Virtual Try-On for Diverse Fashion Items
Authors listed alphabetically in the report.
Robust virtual try-on across diverse fashion items and real-world conditions.
Exploring the Performance Frontier of Compact Unified Image Generation Models
Swift-Image: compact models for unified image generation and editing.
From Corpora to Co-Evolving Capabilities: Capability-Centric Data Design for Generalist Image Generation
Capability-aligned data construction and training curricula for image generation.
Other Publications
Experience
Alibaba Group
Jul. 2026 – PresentAlgorithm Engineer
AI Native Applications · Virtual try-on
Previously: Algorithm Intern, from Feb. 2025
ByteDance
Mar. 2024 – Dec. 2024Algorithm Intern
Education
Sun Yat-sen University
Graduated Jun. 2026Master’s degree in Control Science and Engineering
Advisor: Prof. Xiaodan Liang
Sun Yat-sen University
Sep. 2019 – Jun. 2023Bachelor’s degree in Intelligence Science and Technology