OSCeleton options

[Ste]

Hi all, 

I am working on a project using osceleton and maxmsp. I have to take head coordinates of one person and track his movements.

The problem is that, when there is more than one person in kinect range my max animation bugs. Is it possible to give some option to osceleton to take data from only one person, and can anyone give me a list of osceleton options?

To have calibrated data in max I start osceleton with this comand   ./osceleton -p 7110 -mx 2 -my -2 -mz -1 -ox -1 -oy 1 -oz 1.15 -a 127.0.0.1  witch I took from max forum. 

Any help will be appreciated! 

[Jon Paul]

Ste,

  While it's been awhile since I've looked at the OSCeleton options, you can always route and separate user data inside Max. For how to route individual users to non-conflicting packets, see my Max interface for OSCeleton.  http://jpbellona.com/kinect  The Max source patches are there so you can grab out the guts of what is relevant to use. If you do use, let me know how the project turns out!  Best of luck!

[Ste]

Yes, I tried this, but I have the same problem. In more details the problem is that I take head coordinates in max and I look when people jump (difference between head position now end before if the difference is bigger than.. => jump ) which makes an animation. It looks like when kinect sees two peoples with different height it measures separately the sizes of the two,meaning first the smaller one and then the taller one,despite that in max they are separated. And my animation jumps every time when there two or more people in front of it,despite the fact that they are not jumping.

[Jon Paul]

Ste,

  Not quite sure how to pinpoint without seeing the actual data streams. One issue is that incoming Kinect data can be jittery.  With spikes in data and depending on your thresholds, you may see inconsistent triggers. Smoothing out this data inside Max with zl_ objects might be one direction to check. (don't have the patcher on me to include)  I've seen a nice work around for reducing data jitter by placing a roto-motor on the Kinect device itself. http://hackaday.com/2012/08/13/building-a-better-kinect-with-a-pager-motor/

  For software, another app that I've seen that only tracks a single user skeleton is Synapse.  http://synapsekinect.tumblr.com/

  Unless there is a single-user tracking mode, I don't think OSCeleton options will provide the fix that you require. Multiple users crowding in a shared space may shift the skeleton between users and also incur the same types of spikes in data.

[Ste]

:) this motor thing is very interesting!

"Multiple users crowding in a shared space may shift the skeleton between users and also incur the same types of spikes in data."    ...hm maybe this is my problem. I exclude another variants because I don't have any problems if there is only one person in front of kinect.   

I am sending you my patch, and tomorrow will try with synapse.

Thanks a lot for your replay!     

[Ste]

Hi again, 

Today I tried with synapse, it offers a vey good tracking, but requires a figure with two hands in the air to start. http://synapsekinect.tumblr.com/post/6610177302/synapse witch isn't relevant in this setup, because we are dealing with children.

So after many tries with software, person witch I work with offer to put two peaces of ducktape on the eye of the kinect. This simple hack is very suitable in this situation because the program don't need to track walking figures but only jumps.

This is a solution for now, but I will try to find another way for the future.