In the course of the projects we have been working on (our first movie, about our SocializAR app) should become available any day now), I have compiled a little wishlist for the fantastic piece of work that is QCAR. Maybe you guys are keeping a tally on requested features, so here are a couple of +1's :)
- I wanted to create multi-marker tracking using frame-markers because these are trackable over a larger distance than image targets. For this (and other purposes too) it would be convenient to have access to the matrix-form of the trackable position in camera-space. Using matrices, it's easy to find the relative pose of one coordinate system to that of another. With quaterinions and position and scale vectors, not so easy. (Unless I'm missing something of course!)
- We are using frame markers in our SocializAR app to 'track' and connect people on trade shows, networking events, etc. So we would really like to have more ID's in the set of frame-markers, preferably in the range of millions but if that's a problem, 10 to 20 thousand would help al lot already. Would it be an option to look at the color in the camera-image once a square is found? Or add a second row of dots?
- For the same app it would be good to have access to the 2D locations of the center/corners of the marker, without having to do the reprojection of the 3D point ourselves.
- (even) better detection of frame-markers at a distance. Maybe by making the dots around the edge larger?
- For testing, it would be very nice to be able to run a preview in Unity's editor using a webcam. (I realize that a calibration procedure would have to be added for this to work properly..)
- We have an application at hand where it would be great to have a pose estimation of all markers at a very high precision. If we could snap some photos using the full resolution of the device's camera, and then feed these to QCAR and get some matrices/TRS-data out if it, that would be great.
- We would really like to be able to create a resources package with the tracking data for a multimarker, for more than 6 markers. Imagine a scenario where you are inside a room with 18 well-trackable paintings on the walls... Right now we would have to create 3 multimarkers and combine the resulting camera pose, only because we can only get a resources package for up to 6 markers. It's no problem to define the location of the markers in the config.xml file.
- And of course, a SLAM tracker with loop-closing and matching with known locations :-)
Hope this helps! And thanks,