1 paper
Jiaqi Wang, Weijia Wu, Yi Zhan +6
With AI-generated videos increasingly indistinguishable from reality, current benchmarks primarily focus on broad semantic alignment and basic physical consistency, offering limite…