Aura: Consistent Multi-Subject Video Generation via VLM-Grounded Semantic Alignment 文章
ArXiv CS.CV2026-07-07PAPERen作者: Zixiang Zhou, Zhentao Yu, Yifeng Ma, Hongmei Wang, Wenqing Yu, Cong Wang, Zilin Yang, Rui Chen, Jiarong Ou, Yezhou Liu, Yuan Zhou, Qinglin Lu