Short Paper: Exploring the Object Relevance of a Gaze Animation Model

Abstract

Models for animating the eyes of virtual characters often focus on making the face appear natural and believable. There has been relatively little work in computer graphics that investigates the relevance of the objects of interest (gaze targets). In this paper, a gaze animation model has been constructed that allocates visual attention to relevant targets from objects that are within the virtual character's field of view in an immersive 3D virtual environment. Relevance is determined by proximity, eccentricity, changes in orientation and velocity of objects in the virtual character's environment. In this paper, two tasks were designed to test the relevance of the objects selected by the gaze animation model during the tasks. Eye tracking data obtained from six human subjects provided benchmark data for measuring the efficiency of the model in picking relevant objects. The gaze animation model largely outperformed a random selection algorithm in predicting the real targets/objects of users' interests within the virtual environment.