The existing image-to-video translation methods generally follow a frame-by-frame
generative paradigm, while extracting the temporal information from a reference video or an …