Visual Fidelity Effects on Expressive Self-avatar in Virtual Reality: First Impressions Matter

Owning a virtual body inside Virtual Reality (VR) offers a unique experience where, typically, users are able to control their self-avatar’s body via tracked VR controllers. However, controlling a self-avatar’s facial movements is harder due to the HMD being in the way for tracking. In this work we present (1) the technical pipeline of creating and rigging self-alike avatars, whose facial expressions can be then controlled by users wearing the VIVE Pro Eye and VIVE Facial Tracker, and (2) based on this setting, two within-group studies on the psychological impact of the appearance realism of self-avatars, both the level of photorealism and self-likeness. Participants were told to practise their presentation, in front of a mirror, in the body of a realistic looking avatar and a cartoon like one, both animated with body and facial mocap data. In study 1 we made two bespoke self-alike avatars for each participant and we found that although participants found the cartoon-like character more attractive, they reported higher Body Ownership with whichever the avatar they had in the first trial. In study 2 we used generic avatars with higher fidelity facial animation, and found a similar "first trial effect" where they reported the avatar from their first trial being less creepy. Our results also suggested participants found the facial expressions easier to control with the cartoon-like character. Further, our eye-tracking data suggested that although participants were mainly facing their avatar during their presentation, their eye-gaze were focused elsewhere half of the time.