Multiple Kinect skeleton tracking work-around?

[dtr]

Hey,

There's an intriguing post on the OpenNI list about a work-around for
the multiple-Kinect-skeleton-tracking-on-1-machine bug. It states that
an individual process/program for each of the Kinects should work
fine. Just have to send/collect the data to a common point for
merging:

https://groups.google.com/group/openni-dev/browse_thread/thread/bb7b96616c0989dd

Is there a way to test this with SimpleOpenNI? There would have to be
a way to start multiple sketches and identify the Kinect to be used
for each.

I won't have access to my dev machine and Kinects for a week or 2.
Anyone else wanna try this?


Best, Dieter

[Guillermo Marrufo]

Hi there, Dieter.

I will first wait to the new release of OpenNI that it seems it will incorporate a few functions to define your own representation of the data. Nevertheless, it might be worthy to spend some time programming the option it is pointed in the link you send. Inter process communication is well documented and moving around one computer less will be well received for our show. We will see it. I f any, I will try to keep you posted.

Regards,

Guillermo

2012/5/24 dtr <dtr....@gmail.com>

--

[dtr]

Hi,

It's now confirmed that 2 individual apps can run 2 skeleton tracking
processes in parallel: https://groups.google.com/group/openni-dev/browse_thread/thread/bb7b96616c0989dd

I really hope this works with SimpleOpenNI as well! That will make one
less computer to set up on location... I can test after the weekend.

best D.

On May 24, 7:58 pm, Guillermo Marrufo <guillermo.marr...@gmail.com>
wrote:

> Hi there, Dieter.
>
> I will first wait to the new release of OpenNI that it seems it will
> incorporate a few functions to define your own representation of the data.
> Nevertheless, it might be worthy to spend some time programming the option
> it is pointed in the link you send. Inter process communication is well
> documented and moving around one computer less will be well received for
> our show. We will see it. I f any, I will try to keep you posted.
>
> Regards,
>
> Guillermo
>

> 2012/5/24 dtr <dtr.vn...@gmail.com>

>
>
>
>
>
>
>
>
>
> > Hey,
>
> > There's an intriguing post on the OpenNI list about a work-around for
> > the multiple-Kinect-skeleton-tracking-on-1-machine bug. It states that
> > an individual process/program for each of the Kinects should work
> > fine. Just have to send/collect the data to a common point for
> > merging:
>

> >https://groups.google.com/group/openni-dev/browse_thread/thread/bb7b9...

>
> > Is there a way to test this with SimpleOpenNI? There would have to be
> > a way to start multiple sketches and identify the Kinect to be used
> > for each.
>
> > I won't have access to my dev machine and Kinects for a week or 2.
> > Anyone else wanna try this?
>
> > Best, Dieter
>
> --

> <http://guillermomarrufo.com/>

[maxOh]

hi dieter,

maybe you can give it a try.
In the first sketch you call:
cam1 = new SimpleOpenNI(0,this);

in the other sketch you call:
cam1 = new SimpleOpenNI(1,this);

curious if this will work...

max

[dtr]

Hi Max,

Coincidence: I just found a moment to try things out and wanted to
report my findings. Then I see you suggest exactly what I tried.

I had no luck though. It seems selecting a Kinect like that doesn't
work for me, also when I'm not trying to use both at the same time.
The MultiCam sample sketch doesn't work for me either.

OSX 10.7.2, SimpleOpenNI 0.26 (I didn't update yet). My kinects are on
separate busses for sure.

Any ideas?


best, Dieter

On May 31, 3:07 pm, maxOh <m...@paus.ch> wrote:
> hi dieter,
>
> maybe you can give it a try.
> In the first sketch you call:
> cam1 = new SimpleOpenNI(0,this);
>
> in the other sketch you call:
> cam1 = new SimpleOpenNI(1,this);
>
> curious if this will work...
>
> max
>
>
>
>
>
>
>
> > Hey,
>
> > There's an intriguing post on the OpenNI list about a work-around for
> > the multiple-Kinect-skeleton-tracking-on-1-machine bug. It states that
> > an individual process/program for each of the Kinects should work
> > fine. Just have to send/collect the data to a common point for
> > merging:
>

> >https://groups.google.com/group/openni-dev/browse_thread/thread/bb7b9...

[dtr]

The problem seems to be that the second cam is not found. This is
posted by MultiCam:

SimpleOpenNI Version 0.26
0:SensorKinect/PrimeSense
1:SensorKinect/PrimeSense
cam2: Can't open the depthMap, maybe the camera is not connected!

I added "cam2: " in the code to find out which one gives problem.

[dtr]

Ok, just live-blogging as we go here...

I got MultiCam to work with

cam1 = new SimpleOpenNI(0,this);
cam2 = new SimpleOpenNI(0,this);

beats me but it works... no luck with skeletons yet though.

[dtr]

Problem seems to be that 2 sketches with

context1 = new SimpleOpenNI(0,this);
and
context2 = new SimpleOpenNI(1,this); or context2 = new
SimpleOpenNI(0,this);

always connect to the same camera.

I'm working with the User3d sample and changed the kinect
initialization to:

// start OpenNI, loads the library
SimpleOpenNI.start();

// print all the cams
StrVector strList = new StrVector();
SimpleOpenNI.deviceNames(strList);
for(int i=0;i<strList.size();i++)
println(i + ":" + strList.get(i));

context1 = new SimpleOpenNI(0,this);

Starting 1 sketch works fine but when firing up the second it reports:

SimpleOpenNI Version 0.26
0:SensorKinect/PrimeSense
1:SensorKinect/PrimeSense

Can't open the depthMap, maybe the camera is not connected!

SimpleOpenNI not initialised

Both do list the 2 cams though. Grrr... I feel we're so close to it!