Of course with a dance partner they are going to be very close to each other and therefore a great amount of visual occlusion will occur. Visual occlusion in optical motion capture translates to loss of data. The real question is how do you make up for this loss of data? That would probably have to be done by hand. Then the question becomes how much useful information will I be able to obtain by using the motion capture in the first place?

I believe that dance moves and such where there are going to be large amounts of visual occlusion needs/are done with electromagnetic/mechanical motion capture systems.

It would seem that if you had a system that recognized back as well as front for capture then even when the front was occluded the camera would be able capture the back and interpolate relative position. So if you had red makers on the front of you and blue markers in the back the computer would be able to process this data and arrive at a decent relative position.

Hello,

I made a new experiment.
Look at the PDF --> http://www.megaupload.com/?d=X69IVGBP

In some frames the bones are so confused, that I think these are sequence errors.
Or is the problem that the view is from behind?

Best regards

Stephan

I keep getting the message "The file you are trying to access is temporarily unavailable" for this file. Can you please re-upload it or send it to us by email.

Yes, I do see what you mean. The leg bones are crossed and positioned upward.

BTW, after waiting the required time the file downloaded fine from megaupload.