For the Barbie cinematics, I created a more intuitive control system for the facial animation. Blend shapes still formed the basis of the technique, but they were controlled by a spherical giu that helped solve incorrect combinations of blend shapes creating faulty expressions.
A selection of sounds and expressions formed the basis of reference footage for Barbies facial shapes. (Lotech handheld mirror for simultaneous profile view.)
The Speech GUI is a 3 dimensional layout of phonemes
Shape weights are derived by normalizing the distance of the controller to each of the extremes. ‘Illegal’ blending values that result in ugly geometry are impossible to achieve
The Emotion controller works the same way, and is added to the deformer stack before the speech shape.
A simple facial GUI which you use to point to the expression you want. Shape weights of the various expressions are computed and balanced automatically. Easy to animate with, and the results can’t really do anything illegal.
First animation test with the facial Gui. Pretty good results, and it didn’t take too long to animate.
Animation test of the facial animation system. The facial shapes aren’t that pretty, but the lip-sync is pretty snappy.