Augmented reality, how did he make it?


im starting an augmented reality project.

I have seen this video:

How did he make it?

I guess the first thing he makes is detect the prism and overlay the lines (the first five seconds of the video) but…

The second step is what i dont know exactly:

I dont know how he created a 3d model (a transparent box where is the ball) just from detecting the prism.

Anyone knows it?


that’s not that complicated. If you know the extrinsic and intrinsic from the camera you can simulate this real cam with openGL… you can determine the position of the box in space with computer vision techniques. If you then just render the virtual box into the depth buffer and then the ball normally it seems as if it is rolling in the real box.

but anyway it looks cool!

i have found the web:

As you can see, it speaks about Tracking 3d algorithms to obtain a 3d base (x,y and z) from geometrical objects.

I’m trying to find something TO START with 3d tracking.

Can someone help me?

If you want to start very simple you can use the marker based ARtoolkit. Of course it uses markers but its a very handy toolkit to create nice AR demos…

I dont want markers anymore :slight_smile: thanks,

Anyway, now im finding out what “the extrinsic and intrinsic from the camera” are --> camera calibration.