Home // International Journal On Advances in Systems and Measurements, volume 17, numbers 3 and 4, 2024 // View article


Evaluating Digital Avatars - A Systematic Approach to Quantify the Uncanny Valley Effect by Using Real Life Samples

Authors:
Hakan Arda
Andreas Henneberger
Karsten Huffstadt
Nicholas Müller

Keywords: Uncanny valley; human-like avatars; human-likeness; database; virtual reality.

Abstract:
Virtual reality has seen significant advancements in recent years, particularly in rendering nearly perfect replicas of real-world objects. These improvements in environmental realism enhance user immersion. However, increasing the realism of human-like avatars seems to have the opposite effect, often leading to negative emotional reactions and breaking immersion. This is explained by the uncanny valley curve, where subtle deviations in lifelike avatars can evoke discomfort. In this study, we developed a method to evaluate human-like avatars based on the uncanny valley curve, aiming to pinpoint where this discomfort originates. A database of over 200 avatar images was created, and studies were used to identify key characteristics that make these avatars resemble humans. Additionally, we conducted a follow-up experiment with 30 participants, where real avatars were developed based on our findings and tested in live, direct conversations. This allowed us to validate our approach and offer a more precise method for future research on avatar evaluation and uncanny valley effects. Our results open up new avenues for refining avatar design, potentially mitigating negative reactions and increasing overall immersion in virtual reality experiences.

Pages: 146 to 155

Copyright: Copyright (c) to authors, 2024. Used with permission.

Publication date: December 30, 2024

Published in: journal

ISSN: 1942-261x