Closures

SGX is based on a muscle-dynamic model of speech. To achieve high-quality lip sync in any language, the main requirement is to define the character’s muscle poses correctly, and SGX does the rest. However, certain speech sounds require a little more precision beyond basic muscle setup; namely, closure consonants.

What are closures?

In phonetics, some consonants are produced by closing off or constricting the flow of air through the mouth. For example, to produce the ‘b' in ‘boy’, the lips form a complete seal, temporarily blocking the flow of air. To produce the ‘f' in ‘feel’, the lower lip presses into the upper teeth, which constricts the airflow. In each case, two parts of the oral anatomy come into direct contact. These types of consonants are technically called ‘stops’ and 'fricatives’. We refer to them collectively as closures.

Not all closures are equally important in facial animation. For example, the ‘k' in ‘kin’ is produced with the back of the tongue pressed against the back of the palate. But because this closure happens back and high in the mouth, it is not very visible from the outside; if the tongue is in the right area but not exactly touching the palate, the viewer won’t perceive any inaccuracies.

However, closures involving frontal parts of the oral anatomy, such as the lips or teeth, are highly visible from the outside. If these visible closures are inaccurate, either by undershooting (showing a gap) or overshooting (interpenetrating), the viewer will sense the lip sync is off. Below is an animation of a dental closure (tongue to upper teeth), in three states: undershooting, in good contact, and overshooting.

To ensure a character always produces high-quality closures, the character setup process includes a step of refining a set of sample closure poses.

Closure refinement

As part of character setup, the user can refine a set of sample closure poses. The refined samples then instruct the animation system how to correctly execute closures generally. Closure poses are refined on the character by moving a single slider, which scales the muscle activations producing closure – to the point where there is neither undershoot or overshoot. The goal is to achieve firm contact: no gap between the two vocal organs involved, but also not too much interpenetration between them.

SGX Studio exposes the following three closures. For each closure, a number of sample poses are provided for refinement.

For instructions on using the closure refinement tool in SGX Studio, see Editing Closures (Maya) or Editing Closures (UE).

Note: Closure editing should take place only after muscles have been defined. If any changes are to be made to the muscle poses or the rig, closures may need to be re-edited.