Jimeng AI launches action imitation function: picture rotation dynamic video

Jimeng AI has released a new "action imitation" feature aimed at providing users with a more vivid and realistic AI video creation experience.

This function is technically supported by ByteDance's intelligent creative digital people team, allowing users to generate dynamic videos by simply uploading a character picture and a reference video through the digital person entrance of Jimeng AI, so that the characters in the picture can imitate the actions in the reference video and accurately restore emotions.

According to the official introduction of Zhimeng AI, the "action imitation" function supports a variety of pattern types such as portraits, half-body and full-body, and can synchronize the body movements and facial expressions under different patterns. In terms of the control of facial expressions, Dream AI adopts the independently developed face motion tokenizer technology, which can accurately extract the expression details in the driver video, thereby greatly improving the vividness and reality of the generated video.

In order to further enrich users' creative methods, Dream AI also provides three action templates for users to choose from. In addition, users can upload local files by themselves, and the generated video can last up to 30 seconds. The launch of this function not only broadens the application scenarios of AI technology in the field of content creation, but also brings more creative inspiration and possibilities to users.

In order to ensure the security and compliance of platform content, Dream AI promises to conduct strict security audits on video content uploaded by users. At the same time, a "AI Generation" watermark will be added to the output video to ensure that all generated content complies with platform regulations.