I once made a topic here about a software that analyzed videos recorded by iirc 2-4 cameras / webcams and translates that into bone movements. It was combined with some physics engine and so the results were pretty good. I don't know it's name anymore - you should be able to find a thread about it here though. It had different costs in terms of how many cameras your capture should have and they did sound reasonable (something around 200$ for the 2-Cam-method afaik)...

Real MoCap equipment with like IR cameras and tracking balls etc. never was cheap...

Enjoy your meal
Toast