Interactive Avatar Creation System from Learned Attributes for Virtual Reality

Tomohiro Hibino, Haoran Xie, Kazunori Miyata

Abstract: Demands for 3D models are increasing rapidly as metaverse culture rises. Communication in virtual spaces with avatars is supposed to play an important role in the near future. While photo-realistic 3D avatars are already available to be made by many methods, anime-like avatars are seldom considered due to their difficulties in analysis. Existing creation support systems like for character modeling need to operate too many parameters, so a user may lose what they want to make and what to do. Our main goal of this work is to help users to create it interactively with no work stress and low time cost. To solve this issue, we developed Subdivison Shrink method and implemented interactive VR (Virtual Reality) anime-like avatar creation system. In the proposed system, a user can create an anime-like avatar model in VR space and modify it interactively.