Alibaba to Open Source Its Video Generation Model Wanxiang 2.1

Sources reveal that #Alibaba will fully open source its video generation model Wanxiang 2.1 tonight. The upgraded version, launched in January, enhanced both video and image generation capabilities.

 

Wanxiang 2.1 leverages an efficient self-developed VAE and DiT architecture to improve spatiotemporal context modeling, enabling efficient encoding and decoding of unlimited-length 1080p videos. It also introduced Chinese text-to-video generation for the first time, securing the top spot on the VBench leaderboard.

#ArtificialIntelligence