1. Introduction
With the ability of graphical engines to display more and more complex scenes, the creation of visually plausible digital content has become a major problem: asking designers to manually shape each element of a scene is labour intensive, while automatic generation methods can restrict the variety of generated shapes and only offer indirect control. With these methods, the user has to set high level parameters from which the global shape emerges, often a trial and error approach if a specific design is to be reached. This is particularly true in the case of hair generation, on which this paper focuses.