Weight, Sex, and Facial Expressions : on the Manipulation of Attributes in Generative 3D Face Models

Generative 3D Face Models are expressive models with applicationsin modelling and editing. They are learned from example faces, and offer a compactrepresentation of the continuous space of faces. While they have proven tobe useful as strong priors in face reconstruction they remain to be difficult to usein artistic editing tasks. We describe a way to navigate face space by changingmeaningful parameters learned from the training data. This makes it possible tofix attributes such as height, weight, age, expression or ‘lack of sleep’ while lettingthe infinity of unfixed other attributes vary in a statistically meaningful way.We propose an inverse approach based on learning the distribution of faces inattribute space. Given a set of target attributes we then find the face which has thetarget attributes with high probability, and is as similar as possible to the inputface.