I want to make data-driven facial animation based on FACS (Facial Action Coding System). I know that you can set a Shape Key to for example create a smile, however, as far as I understand, the model changes in a linear fashion from base model to the morph target. The deformation of a human face however happens in a non-linear fashion. Some wrinkles of a smile only appear when you stretch your mouth to the maximum. It would be weird that wrinkles already slightly appear when the Shape Key value is
One way to create this effect is by setting key frames, but then it becomes an animation. My data, however, comes from facial motion capture. I still want to be able to export the model to e.g. Unity 3D, where I assign a value between 0 and 1 to every shape key for the desired facial expression.
Is there a way to have 1 Shape Key to have 2 or more target morphs? E.g. between values
0.2 it targets morph 1, between
0.7 morph 2, and between
1.0 morph 3. Or another way to achieve the above described effect without animation (that can be exported)?