MOCAP - any opensource software available?

[IrrWHAT?!]

Me and another good friend want to look into motion capture, to create amazingly realistic 3D models. We want this to help with future projects, for the animation. But mainly we want it just for the fun of it, and the experience. Essentially it is for HOBBY purposes.

The problem is just that. We aren't professionals, we don't have a company and we don't have access to much money at all. But we are enthusiastic about experiencing this unique technology. Also, we don't care if what we use isn't of a professional standard.

Sensor-based motion capture is out of the question. Optical motion capture (setting up cameras, and white balls on the actor's joints, then filming them in motion, and getting a 3d model from that) is the way to go. We can obtain the cameras, but the software is more difficult. The cheapest software we've found so far is $2000, which is out of our price range. You might suggest we give up, 'if you aren't even willing to pay that much', and you might be right. But our hope is that somebody has made homebrewn software that can take a series of video files and find the balls/points on them and generate a 3d model, given the appropriate camera positions.

My question is, does anyone know of such open-source mocap software? As I said, I don't expect professional standard. Even something dodgy would be fine.

Naturally, with all the commercial products out there, I'm not likely to find such software with a search engine. Asking in a forum is the only way I know how to 'search' for it.

[killthesand]

I don't like seeing people with really cool ideas getting shot down immediately, but I'm going to grasp from my limited knowledge of the subject and let you know why I think this would be a difficult thing for you to set up.

First, to clear one thing up.

to create amazingly realistic 3D models.

Motion Capture does not 'create' models. You need to have a very well rigged model already created. Motion capture will then animate the bones/joints for that model.

I had some friends taking classes dealing with motion capture. It's not as easy as it seems; the raw animation data that is generated isn't all that good. They were writting their own custom software to fix and clean up that data alongside weeks worth of manually adjusting keyframes.

You might suggest we give up, 'if you aren't even willing to pay that much', and you might be right.

Even with free software, you're going to need at least a dozen (more would be better) high quality digital cameras at 10 megapixels or more, and those will need to be very precisely aligned. On top of that you'll need studio lights to properly light and expose the scene. Then a computer (network) powerful enough to process the high volume data stream. The equipment seems to me like a more daunting financial battle.

If mocap is something you wish to pursue, go research. Find a bunch of books or papers about it and figure out a way you can do it. Understand that the further you get from a 'proper' set up, the less likely that you will achieve good results.

[Halifax]

Well I know I'm going to sound crazy for this, but you could use digital motion synthesis as a replacement. All you would have to do is program it, or use existing software. And by the way, a user is implementing it, as a Google Summer of Code project, for Ogre.

[IrrWHAT?!]

You're right, killthesand. The more I research this technology, the more I realize it is too impractical for hobby/personal experience use, and even more impractical for actual realistic 'modelling'. Thanks for responding . It is disappointing to have to abandon the idea. Oh well, atleast I now know how difficult it is and can truly appreciate the work of those that do use it .

On the other hand, Halifax's response about the DMS idea sounds very interesting. It sounds very powerful and practical for use in my project. I was thinking along the lines of designing synthesised digital movement before I heard about it (came up with the idea myself). It's interesting hearing that 'my' idea is actually being used in the computer games industry already. Maybe I won't have to make it myself then (I don't think I have the expertise to do it anyway; To do it properly is damn hard). I would have to understand the physics behind real movement, be able to make it and it would involve practically programming a physics engine from scratch. I don't think I would be willing to spend many months doing that and only that.

Does anyone have any suggestions about the easiest possible way I can incorporate it into my game? Naturally, I use Irrlicht C++, and currently no physics engine. Obviously I will need to find one that maybe supports DMS. Is there such a thing available free? Would it be simple enough to actually learn to use, given that I don't have alot of free time spare?

As you may have inferred from me not having a physics engine, I am new to making games with Irrlicht, so 'the simpler, the better', is my current policy, even if the physics engine quality is not so good.

Sorry for changing the post topic .