Kinect on Stage-- Interference

I recently took some work I have done with the openNi addon for rehearsal in a theatre. Everything was fine until they faded up the lights. I lost all the clarity in the depth camera and user tracking, the user mask had a lot of holes in it (especially where the users hair was- they only had half a head). I am assuming this is because of the infra red frequencies coming out of the lights.

Has anyone used filters to remove these frequencies- I can used LED lights if I have to but would prefer not to as the ones I have access to do not look good at all, and they are a pain to integrate into an existing lighting set-up.


You should be able to use an IR filter gel in front of the lights.

I have used kinect in theatre in different lighting conditions and never saw kinect/light interference, so I assume you work under some very harsh conditions, such as light emmited directly into kinect in the same wave length range. This however leads me to believe that simply moving the lights and kinect around may help.

Eight- I am curious to know your setup. There is a line of sight between the Kinect and the lights but the light is not falling on the kinect, only the person. They are just standard profile spots. The skeleton tracking still works fine but the quality of the user mask is definitely lower with the stage lights on.

Did you use the mask in your setup? Was the user well lit? Was there a colour wash or unfiltered light ont the user?

I have found IR gels and will get them to see what difference it makes.


Fresla: see a picture of one scene here: Other than this depth tracking I used skeleton tracking. The location of kinects and lighting conditions could be seen in some of these rehearsal pictures:


I’ve also heard you can block IR with a combination of red, green and blue gels.
Any IR in the light will definitely interfere with the Kinect, esp sunlight

Just to bring my experience too: I recently had similar problems in 2 different cases:

  1. photographers under the stage (range finders? flash?): the depth image looked consistent, but had some trouble in getting my calibration pose. Solution: I waited for the photographers to stop for 1 sec :slight_smile:

  2. stage lights (the light guy did everything before my arrive, so I don’t know the specs): the depth image looked visibly corrupted, just like fresla said. Needless to say that skeleton tracking was buggy too. Solution: I cursed a lot, asked to have the DMX console on stage and fiddled with it during the performance.

I’m really curious to experiment with gels…