Taming the Virtual Mane: Enhancing Photorealistic VR Avatars with Interactive, Strand Based Hair Models
Description
This thesis aims to improve the appearance and behaviour of photogrammetrically generated Virtual Reality (VR) avatars through the implementation of interactive, strand-based hair. Previous research has shown how discrepancies between realistic appearance and behaviour of VR avatars can lead to an uncanny valley effect, thus motivating the improvement of behaviour realism. Photogrammetrically created avatars will be processed with Blender and hair models will be created using Maya.
A comparative user study will investigate how well the avatars are accepted and if the Illusion of Virtual Body Ownership (IVBO) changes depending on the hair solution used. In the final phase of the thesis, adaptations to the photogrammetry pipeline for the purpose of automating the process of creating avatars with the strand-based hair system will be evaluated. This includes the post-processing of the created avatars as well as the customisation within Unity.
