CineDance: Towards Next-Generation Multi-Shot Long-Form Cinematic Audio-Video Generation

Yuheng Chen, Teng Hu, Yuji Wang +7

The fidelity and structural diversity of training datasets fundamentally determine the capabilities of video generation models. While commercial systems showremarkableabilitytogene…