The face is the most complex part of the human body to set up. We spend our entire lives looking at human faces, and we're pretty good at detecting the subtlest changes and what they mean emotionally. It's a tough job trying to replicate this amount of complexity in CG: from the carefully built mesh topology to the detailed shading, texturing, and—of course—the animation controllers and movements.

The more you try to build a realistic human face in CG, the more details you have to add in order to avoid the uncanny valley effect. This terms refers to a hypothesis in the field of robotics which holds that the human observers tend to be repulsed by something that looks and moves barely like a human being: if it's not exactly like a human nor a clear abstraction of it (like a cartoon character), the observers notice that there's something "wrong" or "strange" to the character—often referred to as "zombie like".

There are books dedicated to the subject of facial rigging, and it wouldn't be feasible to try and cover all the details in this cookbook. We're going to learn some general guidelines and techniques which are applied to a cartoon character that would help you get acceptable results, though. Grosso modo, it's the amount of detail and refinement used with these techniques that will lead you to better results.

A good thing to do is using a mixed approach when rigging faces: you should use Shape Keys controllers for specific expressions (such as a smile), a free deformation tool (such as a Lattice or a MeshDeform modifier), and Armatures (as drivers or to accomplish pivotal deformations, such as the jaw movement) to deform the character's mesh as you wish. In this chapter, we'll see how to create these kinds of controllers.