My problem is I absolutely love experimenting.

This is a 'work in process' for an animated vlogging character. I copied the shapes of a 3D model front and side views. Those two views gave me the 'X' and 'Z' values of each point in the face planes.
I learned that using shapes with sharp corners instead of complex curves made it easier to get the result I was after.
There are four vector layers named 'up', 'down', 'left', 'right' and a switch layer for phonemes. I did not use layer order or masking. Unused layers were simply positioned out of the viewable frame.
My next challenge is eyes.