A video generation model for character animation and replacement. It can animate a single character image according to the motions and expressions in a reference video, or seamlessly replace the character into the target video scene. Two modes:
- Animation mode: single character image + reference video → generate a new video where the character follows the reference video’s motions and expressions, typically preserving the original image’s appearance and proportions.
- Replacement mode: integrate the generated character into the reference video to replace the original character, matching the target scene’s lighting and color for seamless compositing.