Make-Your-Anchor: A Diffusion-based 2D Avatar Generation Framework

Video for model training. Video for model training. Motion condition Motion condition Generated anchor videos. Generated anchor videos. Figure 1 . We propose Make-Your-Anchor, a diffusion-based 2D avatar generation framework, to map the animatable human 3D mesh sequence into realistic human video. By combining with motion capturing or audio-to-motion methods, our framework could achieve anchor video auto-generation with temporal and lifelike outcomes.