Using multiple kinects to track a skeleton?

kizaph

Hey! :)

I’m currently working on project where i have to track a skeleton with 3 Kinects that are standing in a row to capture a larger area. I did some research and found this tutorial where they describe how they did it. But i can’t find the pde-file they mention in the video.

Even worse, when opening the SimpleOpneNI example »MultiCamUser.pde« the comment says that this example isn’t working?

So i’m wondering how it’s done with the current SimpleOpenNi version? Is there a way? And if not, how would you suggest to move on?

I’m really thankful for your help!

Greetings and a awesome day :) Kizaph

kizaph

Oha, just found the the UserCoordsys.pde. I will try it out soon! Somehow i was blind :D

mschi

any results already? :)

kizaph

We will try it this weekend! But i think we should get this work! :)

kizaph

Mhh is it possible that the setUserCoordsys() function doesn't work? At least i can't see any difference if i use it? In both cases my skeleton is drawn with an offset between the two Kinects and i have to align the points manualy. Else they make a big jump when the confidence from the second kinect is higher.

Is this normal? And if yes, waht's the best way to calculate the offset? At the moment i just println the converted points and calculate the offset by hand...not very elegant. O:-)

Have a nice day :)

brigdonia

Hello, Were you able to get the .setUserCoordsys() function to work? I have tried as the video explained but I am unable to get it to work. Is there an example sketch that I would be able to look at with it working?

tlecoz

Hello Kizaph,

The idea of using 2 kinects is interesting but in the context of your project, if all what you want is squeleton data ; it should be much more easier to get the squeleton one by one and then, with code, putting them together.

You could then focus on the core of your project which is probably "how to render skeleton data in an artistic way" (but maybe not)