Unfortunately without the whole ‘machine learning’ logic to transform read information into blend weights for face capture, you end up with fairly limited result.
The android solution is doing some linear/polynomial blend weight generation, but you really need machine learning for this.
I stopped working on the android plugin and… bought an iphone, literally!
There’s some webcam based solution now as an alternative anyway.
I might add ML logic to it in the future tho, as a training ground for my ML skills!