You could direct a video like a real 3D world.

You could direct a video like a real 3D world.

Researchers from Fudan University, HKU, and Tencent introduce VerseCrafter.

It uses a โ€œ4D Geometric Controlโ€ model to give precise, unified control over camera angles and object motion.

It outperforms Yume and Uni3C in generating realistic, controllable videos that closely match desired motion paths.

VerseCrafter: Dynamic Realistic Video World Model with 4D Geometric Control

Paper: https://arxiv.org/pdf/2601.05138ย  Project: https://sixiaozheng.github.io/VerseCrafter_page/ Code: https://github.com/TencentARC/VerseCrafter

Our report: https://mp.weixin.qq.com/s/P2MBsslV2i1Q9v8N7zm_bQ

๐Ÿ“ฌ PapersAccepted by Jiqizhixin

๐Ÿ”— ์›๋ณธ ๋งํฌ

๋ฏธ๋””์–ด

image