For a potential new project I was re-invigorating an old project to produce a proof-of-concept using Google's cross-plattform ZXing QRcode reader library in a client/server setup much like reacTIVision or tbeta operate. For now I've created a bridge to wrap Processing's PImage (and camera capture) objects and so allow them to be used with ZXing. The server side is running the video capture & marker recognition and then broadcasts the decoded QR code along with its centroid's XY coordinates to connected clients. At this stage it's just using a proprietary message format over standard TCP, however it would make sense to integrate this into a TUIO architecture in the future. It'd even be interesting to do more with the actual corner point coordinates of the marker since in theory this should provide the same perspective information as traditional AR markers. Though QR codes as AR markers would allow for all sorts of new use cases also in this field (e.g. the code itself could point to a URL from which the 3D model to be displayed is to be downloaded from - no more locally known markers only...)
Comments & further discussion very much appreciated, please get in touch!
Btw. The super low framerate of this video is purely down to the screencasting software used. The software currently runs at 32fps using a PS3 Eye cam @ 640x480 video stream...
Source code for the above demo available from here: