MagicPony: Learning Articulated 3D Animals in the Wild

Single-view Images Test Image Training Single-Image Inference Articulated 3D Shape Animated Relit Textured Figure 1 . Learning Articulated 3D Animals in the Wild. Our method trains on a collection of single-view images of an animal category and produces a model that can predict the articulated 3D shape of a new instance from a single test image, which can be animated and relit.