The Junkyard Jumbotron

This is great.  I’ve been thinking for a while about how to use the silly little Augmented Reality markers (that I hate) in a different way, and this new video out of MIT looks like they’re heading down the same path.

Since the purpose of the AR tag is to provide a physical basis for the calculation of relative distance, position, and orientation between the camera and the tag, I’ve always wanted to do a project that uses them to make physical calculations instead of visual overlays.

If you have two AR tags in a scene, you should be able to calculate the relative distance and orientation between them, independent of the position of the camera.  Same thing for having two users snap  photos of a single AR tag — a common system should then be able to calculate the position of each photographer.

This could be helpful for 3D scene-building, because if you have a series of photographs that each have a different angle on the same AR tag, any basic Augmented Reality engine should be able to assume the tag is fixed and spit out exactly where each photo was taken in relation to it.

I’d be super interested to see a big AR tag displayed in a stadium at some sporting event, and then to pass a bunch of unmarked flickr photos through it and come back with a plot of where each photographer was sitting.