The Effect of Using an Auto-Generated Anime Avatar on Player-Avatar Identification

Abstract

Game avatars are a way to represent players within a virtual game world. Player-avatar identification can improve game enjoyment, where avatar customisation and player-avatar similarity are potential factors that contribute to identification. Recent emerging technologies, such as image translation algorithms, offer new alternatives for avatar customisation. However, its effectiveness remains unknown. Image translation algorithms enable users to automatically generate avatars that look similar to themselves from their self images (selfies). This research aims to investigate how player-avatar appearance similarity can affect identification by conducting three comparative experiments with different avatar appearance within an offline single-player game. Our results indicate a positive effect of using an auto-generated anime avatar compared to using a default avatar. No correlation was found between player-avatar appearance similarity and player-avatar identification in general. We therefore posit that avatar customisation acts as a greater contributor than player-avatar appearance similarity to avatar identification in offline single-player games.