Everybody Dance Now
This work addresses motion transfer for amateur video creation, offering a practical tool for content generation but is incremental as it builds on existing pose-based methods.
The paper tackles the problem of transferring dance motions from a source video to a target subject using a simple video-to-video translation method with pose as an intermediate representation, producing compelling results as shown in videos. It also introduces a forensics tool for detecting synthetic content and releases an open-source dataset for legal training use.
This paper presents a simple method for "do as I do" motion transfer: given a source video of a person dancing, we can transfer that performance to a novel (amateur) target after only a few minutes of the target subject performing standard moves. We approach this problem as video-to-video translation using pose as an intermediate representation. To transfer the motion, we extract poses from the source subject and apply the learned pose-to-appearance mapping to generate the target subject. We predict two consecutive frames for temporally coherent video results and introduce a separate pipeline for realistic face synthesis. Although our method is quite simple, it produces surprisingly compelling results (see video). This motivates us to also provide a forensics tool for reliable synthetic content detection, which is able to distinguish videos synthesized by our system from real data. In addition, we release a first-of-its-kind open-source dataset of videos that can be legally used for training and motion transfer.