1 paper
Shanwen Tan, Hao Li, Jingtao Zhang +4
Streaming long-video generation faces a central challenge in continuous semantic switching, requiring adaptive memory to preserve coherent visual evolution. Current approaches rely…