Reduce number/mean of pipet coordinates

Hello, I have a question. I’m trying to make kind of game with kinect2 where you can’t touch something. I use pipet to analyze texture and if it find anything yellow, this mean that player touched something. This is easy, but I would like to mark place or places there player fail ;) I have a lot of coordinates in x and y spreads and I tried to reduce/mean this. Best what I made is use Unify (Spread Set) with same spread input and epsilon 0.15 and later Intersect (Spread Set) result with whole spread again. And this is almost working but not in every case. For example in one scenario I have 1 x and 2 y values, so second marker doesn’t fit in place. I would like to ask what is the best way to do that in vvvv 64bits

i would use vvvv 32-bit and use freeframe video analysis plugins. Trautner (Freeframe DShow9) solves the first task (touch someting) and Contour (Freeframe DShow9) the second (get center of blobs)…

Yes, but kinect2 works in 64bits and then I need sharing texture between x64 and x86. I know this is possible, but I hope it’s another way, especially when my try with Spread Set nodes is almost good, but this was more like guess how to do that and doesn’t work in every case ;)
Thanks anyway

I’ve used a histogram for tracking, multiply a grey scale image of your hit points, the ones you need to track, with a pure B/W image from the camera, and if the histogram for that grey value is above a certain pixel count, count it as hit. You can have up to 255 tracked objects this way all in dx11