RealWonder: Real-Time Physical Action-Conditioned Video Generation

作者：Wei Liu, Ziyu Chen, Zizhang Li, Yue Wang, Hong-Xing Yu, Jiajun Wu · 单位：Stanford University, University of Southern California · 会议/期刊：arXiv · 日期：2026-03-05 · 来源：High-Level Structured Planning / Video-based Planning