Intelligent facial animation

Traditional facial animation techniques require either an artist to create and/or clean the animation by hand, which is very time consuming, or rely on motion capture techniques, where fine tuning is usually still required. Generating facial animation becomes even slower when modelling the reactions to external stimuli. Procedural methods are an alternative to these approaches as they rely on an algorithm to automatically generate a short track of animation. Thus, allowing partial automatization of the animation process, however, most research has focused only on procedural body animation, which left the face under-explored. The purpose of this PhD research is to define a novel method for stimuli-based procedural facial animation with believable quality. To be capable of modelling the motions procedurally, a study on facial movements will be done to understand the basic motions and how they vary according to different types of characters. Creating a stimulibased method also requires the definition of the character’s behaviour, unfortunately most research has focused on emotional variation and its expressions, leaving other facial movements to be explored, e.g. tics. Hence, the key contributions of this work are: a study on facial movements; a procedural animation method; a stimuli-reaction model for facial animation. The results will allow reducing the time needed to create facial animation, having great impact in film industry and interactive systems allowing the creation of reactive and highly empathic human-like characters.