1 paper
Zijian Dong, Longteng Duan, Jie Song +2
We present MoGA, a novel method to reconstruct high-fidelity 3D Gaussian avatars from a single-view image. The main challenge lies in inferring unseen appearance and geometric deta…