Whoops! I thought I was replying to a different thread. Sorry. Anyway - I’m not sure exactly what that would look like, but I’ll try. Take for example CuddleMocap’s “sex on the beach” scene. If you change the default model with one that has different bone lengths, the alignment can diverge and wacky things happen. The model you’re trying to swap in could be the same height but have long legs and a short torso, etc, and the leg mocap will diverge as it plays out. I’m thinking that we could make something that trains the mocap data to fit a different model, and maybe use the feet, ankles, elbows, and hands as anchor points? I’m just speaking hypothetically and probably not making a lot of sense.