The lower 3D model is created along with the upper 3D model, but the range of motion of the mouth is narrower and more constant than the upper model. Is it possible to narrow or widen the range of motion of the mouth on DeepAR?
I think increasing the influence of the mouth bones
should do the trick