Exploring Non-verbal Behavior Models for Believable Characters

Believable characters constitute an important component of interactive stories. It is, therefore, not surprising to see much research focusing on developing algorithms that enhance character believability within interactive experiences, such as games, interactive narrative, and training environments. These efforts target a variety of problems, including portraying and synchronizing gestures with speech, developing animation tools that allow artists to manipulate and blend motions, or embed emotions within virtual character models. There has been very little research, however, devoted to the study of non-verbal behaviors, specifically mannerisms, patterns of movement including postures, gaze, and timing, and how they vary as a function of character attributes. This paper presents a work in progress of a study conducted to (1) identify key character characteristics recognized by animators using an acting model, and (2) formalize non-verbal behaviors patterns that animators use to express these character characteristics.