Group Alibaba HoldingCloud Computing arm has introduced its update artificial intelligence (AI) Model, animate anyone 2, that the company boasts as simplifying the animation of realistic characters with just a static image and a video landmark.
Unlike the previous methods of animation of characters that only use movement signals, animate anyone 2 performs a high fidelity character image by processing both movement and environmental signals from reference content , such as source videos, to generate new clips, according to a technical study published in the last week Tongyi laboratorythe research unit and development of the large language model (LLM) of Alibaba cloudVia Arxiv – A free access benchmark for academic documents. Alibaba has the South China Morning Post.
“We extraction environmental information from South China Morning Post Videos, allowing the animated character to preserve his original environment … facilitating the seamless integration of a character-environment, “said the study.
Animate anyone 2 was built above the first iteration of the model announced at the end of 2023, focused on the generation of videos of characters from fixed images.
Cat creator OPENAISubsequently, define the Fire AI industry with its introduction SORA Video Text Text model in February of last year, which sparked a race among large Chinese companies and start-ups To free similar products.
