yitongx / Gen4P

git clone https://github.com/yitongx/Gen4P.git

Gen4P: Video Generators are 4D Perceivers

—

stars, 24h

+26

stars, 7 days · #1,455

+26

stars, 30 days · #9,693

★ total · +26% this week

forks · 1 open issues

Star history starts 2026-10-20; the chart appears after the second daily snapshot.

License—

Created2026-10-13 (7d ago)

Last push4d ago

Watchers5

OwnerUser

Links

GitHub: github.com/yitongx/Gen4P
Homepage: https://gen4p.github.io/
Releases · Issues · Hacker News mentions

README excerpt

# Gen4P: Video Generators are 4D Perceivers
Yitong Xia 1 ,
Zeren Jiang 2 ,
Chuanxia Zheng 1 
 1 Physical Vision Group (PVG), Nanyang Technological University (NTU) 
 2 Visual Geometry Group (VGG), Oxford University
 Faithful structure preservation and motion tracking on challenging scenes like flamingo. 
> **TL;DR:** Gen4P repurposes a video generator for feed-forward 4D perception, with single-pass encoding, on-demand decoding for spatial flexibility, continuous-time embeddings for temporal flexibility, and a frozen depth LoRA as a plug-in geometry prior.
## 🛠️ Installation
WIP.
## 🚀 Usage
WIP.
## 🙏 Acknowledgment
This research is supported by the NTU SUG-NAP and National Research Foundation, Singapore, under its NRF Fellowship Award NRF-NRFF17-2025-0009. We would like to acknowledge the High Performance Computing Centre of NTU Singapore for providing the computing resources, facilities, and services that have contributed significantly to this work. Zeren Jiang is supported by the Clarendon Scholarship.
## 📖 BibTeX
 [code]