Yukai Shi (石瑜恺)

I am a Ph.D. student (2022-present) at Tsinghua University, supervised by Prof. Heung-Yeung Shum (former Executive Vice President of Microsoft). I am also advised by Prof. Lei Zhang at IDEA Research, working closely with Prof. Ping Tan.

Currently, I am an intern at Adobe Research, working on agentic long video generation advised by Dr. Yuqian Zhou and Dr. Zhe Lin. Previously, I was a K-star intern at Kling Group, focused on data balancing and dynamic physics for video generation pretraining, advised by Dr. Xin Tao (Multimedia Pretraining Manager).

I am actively seeking full-time opportunities for 2027. Any opportunities are welcome, please feel free to reach out!

Email: shiyk22@mails.tsinghua.edu.cn / shiyukai22@gmail.com

Email / Google Scholar / Github /

Research

	Imbalance in Balance: Online Concept Balancing in Generation Models Yukai Shi, Jiarong Ou, Rui Chen, Haotian Yang, Jiahao Wang, Xin Tao, Pengfei Wan, Di Zhang, Kun Gai. IEEE/CVF International Conference on Computer Vision (ICCV 2025) · A self-equalization training loss for visual generation models to balance data distribution. · Employed in the mid-training and domain-specific sft stage of Kling video foundation model. Paper / Code
	Koala-36M: A Large-scale Video Dataset Improving Consistency between Fine-grained Conditions and Video Content Qiuheng Wang, Yukai Shi, Jiarong Ou, Rui Chen, Ke Lin, Jiahao Wang, Boyuan Jiang, Haotian Yang, Mingwu Zheng, Xin Tao, Fei Yang, Pengfei Wan, Di Zhang IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR 2025) · A novel video generation data curation pipeline considering data heterogeneity. · The first datasets technical report of Kling video team. Project page / Paper / Code
	SceneMaker: Open-set 3D Scene Generation with Decoupled De-occlusion and Pose Estimation Model Yukai Shi, Weiyu Li, Zihao Wang, Hongyang Li, Xingyu Chen, Ping Tan, Lei Zhang. IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR 2026) · Further extend 3D scene generation from indoor to open-world domains. Project page / Paper / Datasets / Code
	TOSS: High-quality Text-guided Novel View Synthesis from a Single Image Yukai Shi, Jianan Wang, He Cao, Boshi Tang, Xianbiao Qi, Tianyu Yang, Yukun Huang, Shilong Liu, Lei Zhang, Heung-Yeung Shum International Conference on Learning Representations (ICLR 2024)* · Utilize text as semantic guidance to further constrain the solution space of Novel View Synthesis. Project page / Paper / Code
	DreamTime: An Improved Optimization Strategy for Diffusion-Guided 3D Generation Yukun Huang, Jianan Wang, Yukai Shi, Boshi Tang, Xianbiao Qi, Lei Zhang International Conference on Learning Representations (ICLR 2024) Paper
	DreamWaltz: Make a Scene with Complex 3D Animatable Avatars Yukun Huang, Jianan Wang, Ailing Zeng, He Cao, Xianbiao Qi, Yukai Shi, Zheng-Jun Zha, Lei Zhang Conference on Neural Information Processing Systems (NeurIPS 2023) Project page / Paper / Code
	OccScene: Semantic Occupancy-based Cross-task Mutual Learning for 3D Scene Generation Bohan Li, Xin Jin, Jianan Wang, Yukai Shi, Yasheng Sun, Xiaofeng Wang, Zhuang Ma, Baao Xie, Chao Ma, Xiaokang Yang, Wenjun Zeng. IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI) Paper
	Cococo: Improving text-guided video inpainting for better consistency, controllability and compatibility Bojia Zi, Shihao Zhao, Xianbiao Qi, Jianan Wang, Yukai Shi, Qianyu Chen, Bin Liang, Rong Xiao, Kam-Fai Wong, Lei Zhang. AAAI Conference on Artificial Intelligence (AAAI 2025) Project page / Paper / Code
	LipsFormer: Introducing Lipschitz Continuity to Vision Transformers Xianbiao Qi, Jianan Wang, Yihao Chen, Yukai Shi, Lei Zhang, International Conference on Learning Representations (ICLR 2023) Paper/ Code

Academic Service

Conference reviewer of CVPR, NeurIPS, ICLR, ICML, ICCV, ECCV, etc..

This cool template is stolen from Jon Barron!