has anyone seen an AI video model that uses video as a reference input, but not for video-to-video? more like "take this clip as inspiration" and generate something new from scratch?
5,58K