1 paper · 1 filter
Chenyu Li, Oscar Michel, Xichen Pan +3
Large-scale pre-trained video generation models excel in content creation but are not reliable as physically accurate world simulators out of the box. This work studies the process…