For 3D rotoscoping it is best to have two video views 90d apart (front and side, for instance). Even better if you can get three perpendicular views (one on each axis). It can be done from a single video, but you are going to be guessing on many body part angles due to foreshortening. It is very difficult to precisely line up figures, especially when you have more than one and they are moving around and the filming conditions aren't optimal. You need to get the figures placed rather precisely on the first frame (in all respectes - position, rotation, distance, pose) and continuously check the figures against the video as you go. In some cases, I've had to start over from scratch because my assumptions were not correct and led to an impossibility to continue further. Use Wireframe mode to do figure/person comparisons.

C makes it easy to shoot yourself in the foot. C++ makes it harder, but when you do, you blow your whole leg off.

 -- Bjarne Stroustrup

Contact Me | Kuroyume's DevelopmentZone