good luck on this one, be careful you don't confuse camera matching with motion capture.
camera matching is the process where by you create a camera in your 3d scene and then using information from the video, you update the camera's position and orientation. This is how animators mix live footage with rendered objects.
also, a quick tip that took me ages to work out, when you make your tracker points, make the inner box as small as possible, so it just fits around the tracked object, with maybe a pixels width around the sides, then make the outer box
as big as the greatest distance the point will be between any two frames. So if a ping pong ball moves by say 50 pixels between frame one and two, you'll need the outer box to be big enough to spot it. It's here you have a trade off, the wider the outer box is, the wider the tracker's search area is, and the longer it will take to process.
If you have a camera with the option, film at as many frames per second as you can, in the example i've given above, it would take ages to process a search box that wide, so by having more FPS, it should mean the difference between frames is less.
Ignore what reality forgotten said above, it won't look right. Think about it, you walk very differently if you try to walk in "slow motion" to how you walk in real life. If you simply speed it up, it'll look bizarre, also by doing it slow motion, it'll only mean more frames, and more work for yourself.
Manic
I don't have a sig, live with it.
