Translated data: The VideoSwap model, developed in collaboration between the National University of Singapore and Meta, enables the replacement of a video's main subject with a single photo using a few semantic points, showcasing advanced technology and smooth effects. By encoding the source video and extracting semantic points, VideoSwap has achieved state-of-the-art (SOTA) results, making it the top video transformation model. This breakthrough AI video editing technology not only significantly alters shapes but also ensures temporal consistency, ranking it among the most advanced video editing tools currently available.