More on Motion Capture…

Why doesn’t full body motion capture or rotoscoping generally work for performance character animation? I mean, sure motion capture works to the extent that it captures the motion of a person and then maps it to the skeleton of a 3d character. Rotoscoping works in that you ‘capture’ the motion of a person and transfer it similarily to a 2d character. It works. But why doesn’t it always ‘work’ and why, when it comes to acting or character performance, does it often fall flat?

The only places I’ve seen motion capture actually work and not fall flat is in realistic game animation, or as background characters in realistic films (like a wide angle shot of the Titanic with virtual people walking around on the deck) or even stunt doubles where the actor is temporarily replaced by a virtual clone. So there are obvious places it works. But why doesn’t it work in films like Final Fantasy, Monster House, or Polar Express?

It seems to me that motion capture only works as long as you don’t have a character trying to portray emotions. In a game when you are an army recruit following your buddies into battle, you don’t see much acting going on – it’s purely motion (get from point A to B and be hunched down). If a soldier gets hit, he goes down realistically. If a soldier is running, he runs realistically. If a soldier is trying to evade and hide, it looks real. But the minute he turns around and tries to tell you he’s scared or to keep your head down… it all falls apart.

Consider the emotional moments in the recent ‘Monster House’ film. There were these intense moments of emotional outburst from the characters and it didn’t feel convincing? Why? It felt… wrong, out of place and toned down – like something was holding the characters back.

I think that there are a couple of factors to consider:

1) Are the characters perceived to be realistic or toony?
2) Is the character’s motion meant to portray simple motion or communicate thought and emotion?

If you try to put motion capture onto a realistic character and the character is doing simple movements (i.e. not trying to communicate they are thinking or emoting), then it seems to work. Any other combination fails. So if you try to put motion capture onto a toony character it doesn’t work. If you try use motion captured acting to convey emotion (or convince us the character is thinking) it will fail.

Here are the permutations:

(a). Basic motion mocap on a realistic character
(b). Basic motion mocap on a toony character
(c). Acting mocap on a realistic character
(d). Acting mocap on a toony character

(b), (c) and (d) all seem to fail. I think the people at Pendulum Studios (the one’s who created the Mark Antony clip and the facial motion capture system) were trying to make (c) work. But as far as I can tell, they didn’t succeed. There is still something missing – be it in the eyes, or facial expression and even the body language. It all seemed… well, mechanical and dead. Let’s pretend that they had a system that captured all the nuances of facial expressions and could map that directly to a special facial/body rig… would it actually work? Well, I guess it would. Wouldn’t this be equivalent to filming a real actor performing their craft?

Seeing an actor in a movie is obviously not the same as seeing a person …err… in person. But visually they are close enough that when we watch a movie we suspend our disbelief and accept the image before us as a real person. I suspect a perfect motion capture system would do the same thing. But a perfect motion capture system already exists… it’s called film. So what’s the point of creating a motion capture system that exactly mimics reality so that you can have a virtual actor? I supose one reason is virtually bringing someone famous back to life. But why do they have to look exactly like the historical figure? We’re more than happy to have an actor be a substitute for a past president in a film. I guess I don’t see the point of pushing motion capture technology to that point. Personally I think people are doing it… well, just because they can. But it’s not doing animation or the film industry any favors.

When any kind of motion capture is placed on a toony character it just does not work at all. Why? I think it is because of our expectations of how a toony character moves. Animation is all about finding the emphasis of a motion. What part of the motion do we want to exaggerate? Then go about doing that. It’s a distilling of motion to the parts we want our audience to focus on. Motion capture muddies the water. It shows us all the nuances when we don’t want to see them. So mocap for toons is out. That’s just talking about regular kinds of motion, not acting. If it doesn’t work for regular motion it certainly won’t work for acting motion.