ajoshi

I have trying to connect two kinects to the freenect object, but that seems to make the patch crash. Is there a particular method that you need to follow? Can you only toggle between the two cameras or is it possible to get them running simultaneously? Can you attach more than two kinects to a mac?

Also, what kind of data can you extract from this object besides the depth matrix. Can it, for instance, give you the depth of a person in terms of distance from the camera, or give the number of people in the kinect's field of view?

Hope you can answer some or all of my questions.

jit-freenect-grab-help

The object only outputs a depth matrix (integers between 0 and 2048) and a standard 4-plane char matrix of RGB data.

2 kinects: By using 2 freenect objects you can read 2 kinects BUT they need to be on 2 separate USB busses. This is something else than 2 USB ports. Several ports can be on the same bus. Turns out my older Macbook Pro's USB ports are both internally connected to the same bus so I'm screwed. Using a 2nd MBP for the 2nd Kinect at the moment. To check the number of busses you have you can connect a USB2.0 device to each of your USB ports and use System Profiler to see if they're on 1 or more internal USB busses.

Other functions besides depth and RGB: The freenect object was hacked together before the release of the official OpenNI and NITE libraries and doesn't do things like user identification, skeleton tracking, etc. I've been using OSCeleton (

 sending OSC to Max for that. I'm now looking into SimpleOpenNI for Processing which seems more robust but also much more CPU intensive (

http://code.google.com/p/simple-openni/),

Other functions besides depth and RGB: The freenect object was hacked together before the release of the official OpenNI and NITE libraries and doesn't do things like user identification, skeleton tracking, etc. I've been using OSCeleton (https://github.com/Sensebloom/OSCeleton) sending OSC to Max for that. I'm now looking into SimpleOpenNI for Processing which seems more robust but also much more CPU intensive (http://code.google.com/p/simple-openni/), also sending OSC to Max.


jit.freenect.grab help