Wan-Animate is a setup for animating or swapping characters in video. You give it a character image and a video and it can make that character move just like the person in the video. It matches the same face details and body motion to make a clear new video. You can also drop the animated character into the original scene and it’ll try to match the lighting and colors so it blends right in.
The system runs on a version of the Wan model. They changed how it takes input so it can handle different parts of the job under the same process. It uses aligned skeleton motion for the body and pulls face details straight from the original image to match emotion. That helps make videos that feel real and easy to control.
To help characters blend better into scenes when swapping them in, it uses something called Relighting LoRA. That bit adjusts the light and tone but keeps the character looking the same.
Both the model weights and inference code are open sourced.
If you'd like to access this model, you can explore the following possibilities:
Weights