Henghaofan Zhang

About me. I am an undergraduate student in Software Engineering at the University of Electronic Science and Technology of China (UESTC), where I have studied since September 2023. Since January 2026, I have also been a research intern at the Institute for AI Industry Research (AIR), Tsinghua University, working on 3D/4D vision and generative models.

My research interests. I am interested in 3D/4D reconstruction and generation, physically grounded world models and spatial intelligence, and structured multimodal generation with LLMs. A common thread is how explicit structure can make generated worlds easier to edit, control, and simulate. My work explores this through semantic-part-aware 3D editing (PartFlow), simulation-ready 4D reconstruction from video (OVOW), editable reconstruction of objects and lighting (Lumera), and mesh-conditioned video generation (DAR). These projects connect geometric and physical representations with generative models, informing my broader interest in controllable multimodal systems.

Recently. I have been exploring how executable code and discrete tokens can represent images, video, audio, and scenes, and how LLMs could use these representations for unified multimodal generation and editing. I am particularly interested in composing such content into coherent films, interactive scenes, and games. I am seeking MPhil / PhD opportunities for Fall 2027 and am also open to visiting student and research internship opportunities. If our research interests align, please get in touch.