You would probably have better luck examining the code for mocapped animations, particularly for 3-D fighting games and movies. I think mocap has a lot of pote

You would probably have better luck examining the code for mocapped animations, particularly for 3-D fighting games and movies. I think mocap has a lot of potential (though I’m biased here[1]). For example, there have been some interesting experiments capturing performances by sportspeople of different levels and comparing them. These might be looking for patterns in what the better performers do differently, which might then give some ideas for how to train less experienced performers. Some elite sports teams and athletes also use mocap technology to analyse their form and look for details that are difficult to see with human eyes in real time. The challenge is that even state-of-the-art mocap technology is still a very long way from modelling the full complexity of the human musculoskeletal system. Remember that a lot of what you see in games, and particularly in big budget movies, was originally based on mocap data but then manually refined by skilled people for a considerable period of time. The mocap is certainly a useful starting point and it’s getting better all the time, but there’s a long way to go before it will get the fine details right. That said, even with today’s systems, the amount of data you get from a decent mocap set-up means there is a lot you can investigate, as long as you know what you want to look for. As you say, there are a lot of degrees of freedom in a human body, and coming up with the right abstractions to capture the important details is a challenge in itself. [1] I help to run www.ballroomgenie.com, which is a training aid featuring animations of ballroom moves using mocap-driven CGI instead of traditional live video footage. This lets us show various useful effects that live footage can’t, typically because of physics or because in our little niche market no-one can afford movie-level camera set-ups and editing to record the same action from many different perspectives simultaneously. On the other hand, even working with high calibre performers and studio as we do, it still takes a lot of work to get acceptable quality, and we’re still a long way from being able to reproduce truly lifelike footage. This is partly because we don’t have a Hollywood budget and a whole team of people to get each animation just right, but it’s also because as far as I know the technology to model a human body accurately enough that things like contact and pressure are spot on simply doesn’t exist yet.