|
This really depends on the performance you want to capture. If it's dance, for example, you're better off capturing each dancer in his own pass. Actually, this is a good rule for almost any multi-character performance, from conversations to fighting. You can sync the performances by having the performers act to a music track, audio or video. This has been working very well for our animated music video project.
To get the mocap to 'interact' with precision, you can edit and re-time the motions for multiple characters simultaneously using Motion Builder, Animeeple, or other 3D application.
If you absolutely must get multiple performances in one take, to get the best results the entire body of each performer really should be fully visible to all the cameras. This becomes very impractical in a small space unless the actors are basically just standing in place, in which case you may as well shoot them separately anyway.
Another option is to use the Kinect, and have the actors perform side by side in clear view of the camera. Be advised that the Kinect's range of what it can capture is somewhat limited when compared to a multi-camera setup. For example, no walking in circles. In quick tests I've done, walking towards camera seems to work okay though. (To be honest, I haven't spent a lot of time with the Kinect setup but that's been my impression so far.)
Hope this helps.
G.
|
