Blendshape

If you don't have the blendshapes your character needs then your character cannot act except with body language.

Similarly, avatar facial animation through vision-based methods has been explored following a similar approach in which facial features on the user's face are tracked via a Kinect [15] or single image acquisition methods [3, 4, 10] to generate animation via detailed face rigs or pre-defined blendshapes.

Second, and more widely used for real-time purposes, facial blendshapes (predefined morph targets) can be defined as morph properties of the avatar's face and deformed according to tracking software inputs [Orvalho et al., 2012].