|
I am a Ph.D. student in the School of Data Science, The Chinese University of Hong Kong, Shenzhen (CUHK-SZ), supervised by Prof. Li Jiang. Before joining CUHK-SZ, I received the B.E. degree from Tianjin University (TJU) in 2025. My research interests lie in computer vision and artificial intelligence, with a focus on world models, generative models (video & 3D), and multimodal learning. Feel free to contact me for discussion or collaboration! Google Scholar / Github / 📧 |
|
|
|
|
Guian Fang, Shengju Qian, Dongyue Lu, Junchao Huang, Lutao Jiang, Xin Wang, Yiren Song, Li Jiang, Mike Zheng Shou Technical Report, 2026 pdf / project page / code SolarRenderer turns structural references into finished video in four sampling steps with a lightweight control adapter for fast, streaming-friendly rendering. |
|
Junchao Huang, Guian Fang, Shengju Qian, Xianghao Kong, Zhuoran Zhao, Wei Huang, Yihua Du, Zixin Zhang, Justin Cui, Yuchao Gu, Yukang Chen, Xinting Hu, Tianyu He, Shaoshuai Shi, Zhuotao Tian, Xin Wang, Mike Zheng Shou, Li Jiang Technical Report, 2026 pdf / project page / code / data We introduce SolarWM, a fully open foundation for interactive video world models that unifies 1.43 million clips, scalable backbone-native training, and real-time long-horizon interaction within one reproducible framework. |
|
|
|
|
Junchao Huang, Ziyang Ye, Xinting Hu, Tianyu He, Guiyu Zhang, Shaoshuai Shi, Jiang Bian, Li Jiang International Conference on Machine Learning (ICML), 2026 pdf / project page / code We introduce LIVE, a novel training framework that enforces bounded error accumulation for long-horizon video world modeling. Through a cycle-consistency constraint, LIVE performs forward rollout followed by reverse generation to reconstruct the initial state, enabling efficient autoregressive diffusion training. |
|
|
Junchao Huang, Xinting Hu, Bohan Han, Shaoshuai Shi, Zhuotao Tian, Tianyu He, Li Jiang Conference on Neural Information Processing Systems (NeurIPS), 2026 pdf / project page / code (coming soon) We introduce Memory Forcing, a training framework that enables video diffusion models to balance exploration and revisit consistency for autoregressive video generation in Minecraft using spatio-temporal memory mechanisms. |
|
|
Junchao Huang, Xinting Hu, Shaoshuai Shi, Zhuotao Tian, Li Jiang IEEE International Conference on Computer Vision (ICCV), 2025 Highlight (2.3% acceptance rate) pdf / project page / code A tuning-free framework that extends 2D modifications to multi-view consistent 3D editing, enabling user-specific editing from arbitrary viewpoints while ensuring structural coherence across all views. |
|
|
|
Dongyue Lu, Guian Fang, Hur Lim, Yonggang Wu, Yanrong Wang, Hongshan Chen, Shengju Qian, Yue Liao, Wei Chow, Lingdong Kong, Tianxin Huang, Yu Yang, Hongji Yang, Junchao Huang, Ao Liang, Yihao Wang, Zihan Wang, Rong Li, Hanlin Chen, Xin Wang, Wei Tsang Ooi, Mike Zheng Shou, Shuicheng Yan Technical Report, 2026 pdf / project page / code / data Orbia benchmarks whether generated videos form a consistent 3D world, using six revisit protocols to evaluate input preservation, generated-content persistence, and 3D self-consistency. |
|
Ziyang Ye, Junchao Huang, Evelyn Zhang, Zhihao Xie, Ruicheng Zhang, Boyao Han, Litao Ban, Ziye Wang, Xinting Hu, Shaoshuai Shi, Zhuotao Tian, Li Jiang Preprint, 2026 pdf / project page WorldCast builds a distributed multiplayer world model in which each player generates a local view while exchanging player and scene states to maintain a shared world. |
|
Zhuoran Zhao, Shengju Qian, Tongtong Liang, Xianghao Kong, Songchun Zhang, Junchao Huang, Guian Fang, Xin Wang, Pan Hui, Anyi Rao Preprint, 2026 pdf / project page / code Mask Forcing uses dual-noise masking during autoregressive video diffusion distillation to broaden mode coverage and improve generation quality. |
|
Zhihao Xie, Junfeng Wu, Xinting Hu, Junchao Huang, Li Jiang Preprint, 2026 pdf / project page / code A representation autoencoder that turns frozen video foundation model features into compact continuous and discrete latents, enabling strong reconstruction and approximately 5x faster convergence for downstream autoregressive and diffusion video generators. |
|
Guiyu Zhang, Yabo Chen, Xunzhi Xiang, Junchao Huang, Zhongyu Wang, Li Jiang IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026 pdf / project page / code A unified motion-control framework that jointly models camera trajectories and object dynamics for coherent video generation, supported by geometry-aware control and the new RealCOD-25K dataset. |
|
Yilin Ye, Junchao Huang, Xingchen Zeng, Jiazhi Xia, Wei Zeng International Conference on Machine Learning (ICML), 2025 pdf / code A novel dimensionality reduction technique designed to visualize cross-modal embeddings with enhanced accuracy by learning kernel regression of the metric landscape in the projection space. |
|
|
|
Bohong Chen, Yajing Chen, Yihua Du, Junchao Huang, Guian Fang, Shengju Qian SolarX-Team Blog, October 2026 blog post / code A training-free approach to decoding H3 video VAE latents as they arrive, without waiting for future latents. |
|
|
|
The Chinese University of Hong Kong (Shenzhen)
Ph.D. in School of Data Science September 2025 - Present Merit Fellowship (Ph.D. Scholarships at CUHK-SZ) Advisor: Prof. Li Jiang |
|
Tianjin University
Bachelor of Science: Mathematics and Applied Mathematics September 2021 - June 2025 Elite Program: "Qiushi" Class of the Mathematics Department GPA: 3.83/4.00 (91.17/100), Ranking: 3/55 |
|
|
|
|
Programming: Python, C++, PyTorch, R, LaTeX, Visual Basic Software: PyCharm, Visual Studio Code, Mathematica, MATLAB, MS Office Suite Languages: Chinese (Native), English (IELTS 6.5/9, CET-4: 584) Interests: Piano (since age 4), Table Tennis, Cycling |