FlowVid: Taming Imperfect Optical Flows: Webpage Demo and Quantitative Comparisons

This paper proposes a consistent V2V synthesis framework by jointly leveraging spatial conditions and temporal optical flow clues within the source video.