As someone who's worked on very cutting edge digital humans before, muscle simulation has a very poor cost to benefit ratio for facial animation. It's almost a

As someone who's worked on very cutting edge digital humans before, muscle simulation has a very poor cost to benefit ratio for facial animation. It's almost always better to learn external deformations that emulate the muscle triggers than to simulate it. The reasons are usually: - setting up muscles takes a long time and isn't very scalable to many characters - animators or even actors will do a motion that is needed for the shot that the muscle rig will fail at. Especially for things like high velocity impacts. - muscles are really hard to art direct. Sometimes you want to end up sculpting a shape into something just slightly different. - simulations need to run linearly in time and are slow. So for the situations where you need to be able to skip around in time or for realtime, you want an emulation of muscles rather than a simulation Doing more superficial muscle emulations like https://zivadynamics.com/zrt-face-trainer https://zivadynamics.com/zrt-face-trainer give a better mix of control and realistic behaviour. Of course, for bodies it becomes a lot easier to use muscle simulations ( and Ziva's own tetrahedral muscle solver is amazing for that )