Meta Avat3r – Large Animatable Gaussian Reconstruction Model for High-fidelity 3D Head Avatars

https://tobias-kirschstein.github.io/avat3r

Avat3r takes 4 input images of a person’s face and generates an animatable 3D head avatar in a single forward pass. The resulting 3D head representation can be animated at interactive rates. The entire creation process of the 3D avatar, from taking 4 smartphone pictures to the final result, can be executed within minutes.

https://www.uploadvr.com/meta-researchers-generate-photorealistic-avatars-from-just-four-selfies