Alibaba Cloud AI model Animate Anyone 2 simplifies making of lifelike character animation
Alibaba Cloud has unveiled Animate Anyone 2, an updated AI model that simplifies the animation of lifelike characters using a static image and a video cue. This new model represents advancements in generative AI, allowing for high-fidelity character animation by processing both motion and environmental signals from reference videos.
Unlike previous methods relying solely on motion signals, Animate Anyone 2 extracts environmental information from source videos, enabling seamless integration of the animated character into its environment. This development highlights the progress of Chinese developers in generative AI, despite restrictions on access to advanced semiconductors. The model builds upon the first iteration of Animate Anyone, released in late 2023, and follows OpenAI's Sora, sparking a race among Chinese tech companies to develop similar text-to-video technologies.
via SCMP Full Text Feed
Alibaba Cloud has unveiled Animate Anyone 2, an updated AI model that simplifies the animation of lifelike characters using a static image and a video cue. This new model represents advancements in generative AI, allowing for high-fidelity character animation by processing both motion and environmental signals from reference videos.
Unlike previous methods relying solely on motion signals, Animate Anyone 2 extracts environmental information from source videos, enabling seamless integration of the animated character into its environment. This development highlights the progress of Chinese developers in generative AI, despite restrictions on access to advanced semiconductors. The model builds upon the first iteration of Animate Anyone, released in late 2023, and follows OpenAI's Sora, sparking a race among Chinese tech companies to develop similar text-to-video technologies.
via SCMP Full Text Feed
