1 paper
Xiaojie Xu, Zhengyuan Lin, Runyi Li +3
Learning to generate or reconstruct explorable worlds requires video paired with more than RGB: camera motion, scene geometry, temporal correspondence and, for interactive models,…