Select one or more identity photos.
Select the video whose movement/camera you want to use.
The system will analyze the reference video, detect useful parts, generate each part, and combine the final result.