1 paper
Deyin Liu, Yisheng Ding, Zhe Jin +3
Existing zero-shot video editing methods rely on pre-trained diffusion models, successfully achieving spatial control and basic temporal consistency but fundamentally fail to prese…