How to Build Digital Humans?
From Priors to Photorealistic Avatars

Wojciech Zielonka1†, Tobias Kirschstein2†, Timo Bolkart3†, Simon Giebenhain2, Vanessa Sklyarova5,8, Xiang Deng4, Donglai Xiang6, Shunsuke Saito1, Yebin Liu4, Matthias Niessner2, Justus Thies7†
1Meta    2Technical University of Munich    3Google    4Tsinghua University
5Max Planck Institute for Intelligent Systems    6NVIDIA    7Technical University of Darmstadt    8ETH Zurich
† These authors contributed equally.
Code arXiv Paper Slides
Teaser figure showing digital human avatar components
Figure 1: Building digital avatars requires considering many components, such as the human head, hair, hands, and garments, and ultimately full-body avatars. Much like da Vinci's Vitruvian Man symbolizes the ideal proportions and unity of the human form, the construction of digital avatars demands a holistic approach in which each part integrates into a coherent whole. Each of these elements presents distinct challenges and numerous potential design choices across different stages. In this report, we introduce and examine the details of each creation stage, develop a practical taxonomy of the current literature, and discuss future directions in the field of digital avatars. The background image was generated using Google Gemini.

Abstract

This state-of-the-art report provides an overview of controllable 3D human avatar creation. We describe current 3D avatar systems, which typically consist of three stages: (i) learning priors of human appearance and motion, (ii) creating a personalized avatar, and (iii) animating the avatar. To limit the scope, we focus on the prior learning and avatar creation stages. We define current avatar representations and introduce a taxonomy that categorizes existing work along multiple axes, including body regions and employed priors. We review methods for full-body and head avatars, as well as layered representations that decompose the body into components such as hands, hair, and garments. Finally, we outline common underlying principles, reference key literature for newcomers, and discuss open challenges and future research directions.

Loading publications…