Alibaba open sources video generation model Wan2.2-S2V, which can generate movie-quality digital human videos with just one image.

On the evening of August 26th, Ali released a new open-source multimodal video generation model called WanXiang Wan2.2-S2V. With just one static image and a piece of audio, this model can generate movie-quality digital human videos with natural facial expressions, consistent lip movements, and smooth body movements. The length of the video generated by this model can reach minutes, greatly improving the efficiency of video creation in industries such as digital human live streaming, film and television production, and AI education.