I've seen musculoskeleton simulation (although passive and not active) used throughout VFX quite a lot, its basically what Weta Digital has been doing for years

I've seen musculoskeleton simulation (although passive and not active) used throughout VFX quite a lot, its basically what Weta Digital has been doing for years. (For example from the Hobbit movie 8 years ago: https://www.youtube.com/watch?v=r45e5Xky35k https://www.youtube.com/watch?v=r45e5Xky35k). Hell, Maya even has its own muscle system (Maya Muscle) that animators can already use out of the box. So the technology to simulate these kinds of things (at adequate realism for animations and movies) is already there. The problem is that it takes a painstakingly long time to do the modelling for these kinds of stuff. Anatomy is hard to get right, high quality commercial human template models cost you $10000+, and even then you're going to take a lot of time tweaking mesh geometry to make the simulation work. And the musculoskeletal models you're creating are all going to have different body shapes and might not even be human, so you need to create a lot of things from scratch (there was a recent paper on a musculoskeleton design tool for precisely this: https://www.dgp.toronto.edu/projects/modeling-musculoskeletal-anatomy https://www.dgp.toronto.edu/projects/modeling-musculoskeleta...) So you need a lot of work to do even just passive simulation (where the articulated skeleton just kinematically mimics the animation and the muscle effects are secondary) Though the bullet points you've listed are definitely all good research topics in itself. DeepMimic-style active musculoskeleton simulations (where models physically mimic given animation data using only muscle activations) are currently possible for line-segment muscles (the paper: https://mrl.snu.ac.kr/research/ProjectScalable/Page.htm https://mrl.snu.ac.kr/research/ProjectScalable/Page.htm) It uses Deep Reinforcement Learning with lots of training and fine-tuning though, it's not plug-and-play in its current state. Maybe such a similar thing can be also done with volumetric muscles too, but that's definitely going to be a hard paper to write since FEM simulation is incredibly costly and finicky to run stably, and you need a lot of simulation frames to use DRL. But I think the fundamental issue is: does this all have to do with actually making a good film/animation, outside of the realm of biomechanics? I really don't think so. The latest remastered Disney films all lack the fluidity of 2D animation of the past; the remade Lion King has one of the best realistically simulated animal characters but is utterly void of any artistic merit. And these graphical improvements don't seem to make the artist's life easier and is making animation even more labor-intensive. Nowadays it feels like artists in those big industries have less room to experiment because of all these "realistic" technical requirements. From this viewpoint, it makes sense why Disney/Pixar has still focused more on superficial skin deformation instead of a more realistic biomechanical approach: when making cartoonish 3D animations you need a lot of squash-and-stretch, and that is hard to do with rigid muscle systems. Better to model your character as a blob of mass, it's easier to animate, keyframe, and employ all kinds of effects to express artistic intent.