Effects of Personalized Avatar Texture Fidelity on Identity Recognition in Virtual Reality

Abstract

Recent advances in 3D scanning, reconstruction, and animation techniques have made it possible to rapidly create photorealistic avatars based on real people. While it is now possible to create personalized avatars automatically with consumer-level technology, their visual fidelity still falls far short of 3D avatars created with professional cameras and manual artist effort. To evaluate the importance of investing resources in the creation of high-quality personalized avatars, we conducted an experiment to investigate the effects of varying their visual texture fidelity, specifically focusing on identity recognition of specific individuals. We designed two virtual reality experimental scenarios: (1) selecting a specific avatar from a virtual lineup and (2) searching for an avatar in a virtual crowd. Our results showed that visual fidelity had a significant impact on participants' abilities to identify specific avatars from a lineup wearing a head-mounted display. We also investigated gender effects for both the participants and the confederates from which the avatars were created.