You can cycle through the cameras with an ID, taking a scan from a camera and then making the next camera in the sequence active and taking a scan with that, and so on until you cycle back to the first camera in the sequence and begin again. There is the risk of a certain amount of interference from having multiple cameras in close proximity though.
The newer ZR300 camera though does not suffer from interference though when multiple cameras are used, so that may be an option that you may wish to consider. From the 'Features' list for the ZR300's Intel Click online store entry:
Edit: if you would prefer to use the SR300's that you have, the idea occurs to me that maybe you could put a little Faraday Cage around each camera to reduce its signal emissions.
Thanks for the quick response, we'll give that a try!
With regards to the ZR300 -- our application is around ~10-20cm away from the cameras, is that doable with the ZR300?
See my edit to my first reply please.
In regard to the ZR300, its minimum depth capture range is 0.55 m / 55 cm. Even with the SR300, the minimum is 0.2 m / 20 cm. In my experience, 20 cm is the minimum distance in order to provide satisfactory scan results in terms of objects. I have seen scans of face and skin that look like the camera was closer than 20 cm, though the image tends to have lines and surface break-ups on it.
I was about to ask the same question with SR300. I have two pointing at the same object, and the cameras are at ~45 degree angle with respect to the object. I tried turning on and off the IR projector, but the cameras are too slow, so there is still interference. Any scientific paper you would recommend for this task?
A university reduced interference between multiple Kinect cameras in 2012 and wrote a paper on it.
I was half joking in my earlier suggestion of putting mini Faraday cages around cameras to keep disruptive signals contained but now I see Faraday phone cases on Amazon, I wonder if I was into something after all.