Developing a multitouch specific camera

[reml]

Seth,

Thanks! I'm hoping egosgame/Justin has experimented with this and that it added to the effort to use these cameras for multitouch with CCV.

Rob

[reml]

All,

Has anyone at Natural Point or in this user community used the TrackIR or Optitrack cameras and/or software for multi-touch applications? I am getting closer to my deadline and would appreciate knowing about any successes or failures to use this hardware. I have 3 SLIM v100 cameras from old projects and would like to put them to use for this.

The CCV/NUI Group community has robust software, but I am not committed to using it.

My needs are not even multi-touch at the moment. I am trying to use touch as input on a large rear projected fabric screen. The fabric is partially elastic and creates a clear peak when �pressed�.

I need a simple �point and left click� input for my limited application.

I would really appreciate any feedback.

Rob

[Birch]

We saw the following post over at the NUI Group multi-touch forums announcing a support layer for OptiTrack cameras. Maybe it will be of use to you.

http://nuigroup.com/forums/viewthread/2178/P30/#23640

[jmrb2002]

One interesting idea (not sure if it has been developed elsewhere) could be to been able to substract a "flat field" before recognition, but I'm not sure if this could be possible with actual Optitrack programation.

A little better explained :


Let it be a multitouch system mounted in a rear projection setup (see the "draw")

................................................
..................................|.............
Projector ------------>..|.................
..................................|.............
..................................|......|.............
(and flash)...................|.............
............................. screen...........
...............................................

(well, not the best draw, I know!)

One of multitouch problems is the big amount of external ilumination in daylight ambients. The idea is to take two photos with the optitrack camera: one with flash and another one without flash. If we substract the images, there is only the reflected light in the fingers of the user detected (the flash of the camera is suposed to reflect only there).

This reduce to a half the bandwith of data, but could be much robust to external ilumination. The same could be usefull for 3D setup, if there is some daylight interference.

Even it could be possible to take by example only one or two reference images (without flash) per second, because external light is not suposed to vary very fast. In this way, bandwith could be almost maintained.



The problem is that there is necesary to modify (very little) the cameras microcontroller/microprocesor code.


What do you think about this? could be possible?


Thanks in advance

JM

[VincentG]

This is an interesting feature, however it would not be possible to modify the cameras to perform this behavior. Perhaps it can be considered for a future camera. This could be accomplished with the cameras operating in grayscale image mode, but you would lose the (low system processing
load) benefit of the in-camera processing.