I’m searching for a way to connect the standalone application “reacTIVision” with vvvv to use the comfortable multi-touch recognition and undistortion (and maybe to do the tracking on another PC).
Has anyone done something like a TUIODecoder?

Even after reading this
which is dealing with problems about the TUIO-messages from reacTIVision, i can’t set up a working TUIODecoder-Patch.


I’m considering doing the Multi-Touch thing with a contour-node. But that would only be a workaround. Iwould still miss the undistortion. Plus: i had the feeling, that the tracking in reacTIVision was a bit faster. (could be the output too)

Tried to receive message from Touchlib. Same thing as reacTIVision: Nonsense in my OscDecoder.

Can anyone help me?
I’m quite new in here.

ai gekko,

as mentioned above the tuio protocol cannot be decoded within a vvvv-patch. therefore i wrote the FiducialTracker, which is now still lacking the latest multitouch-features of reacTiVision.

i hear there is a new release (1.4) of reacTiVision upcomming with even better fingertracking support…one could then take the code and update FiducialTracker easily, probably.

what does touchlib send? is there a protocol specification somewhere online?

hoi gekko,

was messing around with reactivision a year ago. the problem with the tuio protocol is that is is not completely conform with the osc-standard (it sends more than one message with the same address), thats why it you receive nonsense. (at least it was like this at that time)

there’s an old posting somewhere in the vvvvorums.

a work around is to receive the messages with another application, repacking it and sending it to vvvv

i hope, this info is still correct with the lastest version


ah, and for undistortion you may be interested in the Undistort (DShow9 OpenCV) node.

Hi everyone,

thanks a lot for your replies!

joreg, how much time/skill did it take to transfer the code from reactivision to the fiducialtracker-plugin?

one could then take the code and update FiducialTracker easily
Ah, why do i ask? I won’t manage it, anyway. Just for my interest.

Investigated the messages from reactivision:
only two addresses: /tuio/2Dobj and /tuio/2Dcur
they differ in TypeTag and
especially in the “Data1” field: set, source, alive and fseq
wouldn’t be too difficult to extract the needed data.
But as mentioned many times befre in this forum, only the last message from the bundle (fseq) is extracted by the OSCDecoder.


We use Community Core Vision in order to tracking data from a camera.
There isn’t fiducial tracking support but if you need blob coordinates works fine with TUIODecoder.