I was able to make the mask for this pretty soft and flexible, so for the short term I didn't need to implement amplification in the mouth and cheek area to get motion to transfer through.  It's still muted of course, but 'good enough' for Halloween.

However, the eyebrow area could still use some help.

So the plan is this.  I've said as much in comments, but time to make it a log.

Use a conductive foam variable resistor or actual pressure sensor over the target movement area, with a matching servo moving the outer skin over the same area.  They are coupled in that the outer skin movement translates a bit to the inner.  The programming would simply aim to maintain the same resistance by moving the servo until that resistance is met.  Anything off of the target resistance would result in a correction of the outer skin in either direction to match.

So you raise your eyebrow, which puts pressure on the sensor and decreases the resistance.  The servo reacts and pulls up on the outer eyebrow to raise the resistance back to what it was, relieving the pressure on the inner eyebrow sensor.  

You lower your eyebrow and the resistance shoots up.  The servo reacts accordingly by letting go of the eyebrow and the resistance returns to the original.

The baseline resistance reading is read on startup and movement in either direction is then tracked from there so it can be auto-calibrating.

AKA.. Servo amplification of small facial movement.  Sounds good in theory..

Problem is I have zero conductive foam left and I'm too cheap to spring for something I normally got for free back in the day when buying MOS/CMOS components.  :)  That and the conductive foam decays with age, and would be sensitive to moisture (aka, sweating would screw it up).

So I'm currently figuring out what around the house I can use to make a suitable pressure sensitive capacitor instead and adapting my Teensy code for reading multiple channels of home-made capacitors.

The real issue here is the target capacitance for an area the size of an eyebrow is about 5x lower than the designed capacitor size for that project so my Mylar dielectric won't cut it.  The pressure differential applied is also about 10x lower.  

So I would need a dielectric at least 50x more effective to get a similar response, or I suppose I could just live with the noise of a 50x weaker signal...  Not yet.

I have some pure white colorant for paint from back when I wanted to color urethane foam myself, otherwise known as titanium dioxide, in the garage.  Titanium dioxide as a dielectric is actually right in the range I would need... I have zero idea how or if it will be possible to make capacitors from paint colorant so... experimentation ensuing.