Yawen Luo | 罗亚文

Ph.D. Student

The Chinese University of Hong Kong

luoyw0207@gmail.com

     

Your Photo

Biography

I am a first year Ph.D. student at MMLab, The Chinese University of Hong Kong, supervised by Prof. Tianfan Xue. My research interests lie in the fields of computer vision and deep learning, and I am currently focusing on controllable video generation, long video generation, and unified model.

Previously, I received my Bachelor's degree from Huazhong University of Science and Technology.

Publications

ShotStream: Streaming Multi-Shot Video Generation for Interactive Storytelling
Yawen Luo, Xiaoyu Shi†, Junhao Zhuang, Yutian Chen, Quande Liu, Xintao Wang, Pengfei Wan, Tianfan Xue†
ECCV, 2026.

[Project Page] [paper] [Code] [Models] [机器之心]

ShotStream is a causal multi-shot video generation framework for interactive storytelling, enabling efficient on-the-fly frame generation at 16 FPS on a single NVIDIA GPU.
AnyRecon: Arbitrary-View 3D Reconstruction with Video Diffusion Model
Yutian Chen, Renbiao Jin, Tianshuo Yang, Xin Cai, Yawen Luo, Mingxin Yang, Mulin Yu, Linning Xu, Shi Guo†, Tianfan Xue
ACM (SIGGRAPH Asia), 2026.

[Project Page] [paper] [Code]

AnyRecon reconstructs 3D scenes from arbitrary unordered sparse views by combining video diffusion with persistent global scene memory and geometry-aware conditioning.
CamCloneMaster: Enabling Reference-based Camera Control for Video Generation
Yawen Luo, Xiaoyu Shi†, Jianhong Bai, Menghan Xia, Xintao Wang, Pengfei Wan, Di Zhang, Kun Gai, Tianfan Xue†
ACM (SIGGRAPH Asia), 2025

[Project Page] [paper] [Code] [Dataset] [机器之心]

CamCloneMaster enables reference-based camera control by replicating camera movements from videos without camera parameters or test-time fine-tuning, supporting both I2V and V2V.
CineMaster: A 3D-Aware and Controllable Framework for Cinematic Text-to-Video Generation
Qinghe Wang*,Yawen Luo*(co-first author), Xiaoyu Shi†, Xu Jia†, Huchuan Lu, Tianfan Xue†, Xintao Wang, Pengfei Wan, Di Zhang, Kun Gai
ACM (SIGGRAPH), 2025.

[Project Page] [paper] [机器之心]

CineMaster is a 3D-aware controllable text-to-video framework that lets users jointly manipulate objects and cameras in 3D space for cinematic video generation.
Video Bokeh Rendering Paper Thumbnail
Video Bokeh Rendering: Make Casual Videography Cinematic
Yawen Luo, Min Shi, Liao Shen, Yachuan Huang, Zixuan Ye, Juewen Peng, Zhiguo Cao†
ACM Multimedia (MM), oral, Best Paper Candidate, 2024.

[paper]

Video Bokeh Renderer brings cinematic shallow-depth-of-field effects to casual videos with temporal consistency and adjacent-frame information.

Preprints

OmniDirector: General Multi-Shot Camera Cloning without Cross-Paired Data
Jiwen Liu, Shujuan Li, Zhixue Fang, Xiaohan Li, Yan Zhou, Zijie Meng, Zhimin Zhang, Yawen Luo, Guoxin Zhang, Yu-Shen Liu, Pengfei Wan
Under Review, 2026.

[Project Page] [paper] [Code]

OmniDirector clones diverse multi-shot camera motions to animate source images using a camera grid representation and a hierarchical prompt expansion agent for multimodal control.

Honors

Internship

Academic Service

© Yawen Luo | Last update: June 2026