Optimization of Character Gaze Behavior Animation using an Interactive Genetic Algorithm

Abstract

Character animation is required to appear as natural as human motion. Toward that goal, the approach that is followed is to ensure that the expression appears as natural as human expression by adding gaze behavior to the contextual situation and by adjusting the environment to the general behavior animation. However, a detailed understanding of some gaze control parameters is required for users to create the intended gaze behavior animation that is suitable for situations in which the gaze controller is used. It is difficult for beginners and users who are inexperienced in creating character animation to ensure that the user feels adequate because trial-and-error adjustments are required. In this paper, we propose a method for optimizing character gaze behavior animation by using an interactive genetic algorithm (IGA) based on the viewer’s preference such as “where to gaze” and “how to gaze” without the need to manually set the gaze control parameters. We translate a character’s gaze motion into parameters for the gaze controller. Furthermore, we obtain the optimized gaze target and control parameters such that the viewer experiences natural gaze animation when using the IGA. The experiment compared the viewer’s impression of the method and the result they obtained by using manual creation with that obtained by the proposed method. The results confirmed that the proposed method enables the user to quickly and easily create gaze behavior animation of the same quality possible by using manual creation.