Animate DIff

AnimateDiff is an artificial intelligence python code that can animate personalized text-to-image models without specific tuning.

It can append a motion modeling module to any existing text-to-image model based on Stable Diffusion, and train it on video clips to learn a motion prior.

Once trained, the motion modeling module can inject motion dynamics to the generated images according to the text input. AnimateDiff can produce diverse and personalized animated images for various styles and domains, such as cartoons, portraits, landscapes, and more.

You can find the official implementation of AnimateDiff on GitHub or visit the project page for more details and examples.