forums.naturalpoint.com

anther problem I want to say.
we want to use 13 mark in human body for motion capture, but because the capture volume was not enough, we have to reduce the mark to 6.
the scene will see 6-mark in forepart, but several mark were disappear when the body motion. and the data which gained from software is mixed each other. I can not use that data at all.
would you have any suggestion for me ?
thank you very much!

Hello:

Point Cloud software does not track the individual points, it just reports them, it is up to you to track the points over time. Creating rigid bodies, or trajectorizing points, or using Arena are the best ways to track points over time.

[I am new to OptiTrack and MoCap, so forgive my asking what maybe everyone knows. If that's the case please point me in the right direction, and I will RTFM]

What do you mean by "trajectorize" points? I assume you may mean to look for some form of regression functions suitably restricted to yield reasonable properties about continuity and derivatives, etc.

Is there any support for doing this in the Rigid Body Toolkit or SDK?

On how to distinguish a point from others nearby, I've thought of looking for direction of movement and speed, but don't know if that is an option in real-time. Another might be to have a model of 2-marker distances that are expected to remain constant and help discriminate the points. Any experience you might share?

shameless but relevant link follows:

http://fie.us/blog/mocap_series/

Thanks, Brad. Very relevant. I'll be experimenting with these ideas for some time.

BTW, found your blog very interesting, and will visit often.