Generating Virtual Characters from Personality Traits via Reverse Correlation and Linear Programming

The usefulness of a virtual character depends on its ability to fulfill the user’s expectations: among other aspects, the character’s appearance should match her/his personality as well as her/his behavior [11]. It has indeed been shown that the better a character looks the part, the more believable and effective she/he will be in the narrative [13, 8]. This paper presents a system which generates a virtual character defined along three personality traits: Dominance, Trustworthiness, and Agreeableness. From these three traits, 14 surface physical attributes of the target character are automatically inferred: chin bones, chin height, eyebrow angle, eye size, head ovality, height, muscularity, weight, mouth and lip width, mouth and lip height, neck width, nose width, stomach tone, and torso V-shape. The system is based on a character model where each physical attribute is modulated