Having tried the stuff from the scratch now a few times I'd like to point out some implicit knowledge which this video does not point out.

First: think about tracking when shooting the footage. The most important thing is to move with the camera base to get some depth information into the markers. It helps if you know the focal length while shooting too.

Then there is the camera correction: the K1-3 parameter set. Basically you will want to correct the lens abberration if you have problems with it: some test-shots of square things reveal it well. You can do without but in case you start to wonder why the image and composite don't line up: the most likely reason is the aberration.

Also Sebastian is quite aware of how he places the additional markers: you need some to solve the scene well and you need some to get a nice 3d representation of the points. The latter points he adds are nicely aligned with the wall and help orienting the cube and bottom-plate nicely.

So don't forget to move the camera base while shooting a bit (don't forget that you can cut it afterwards), remember the lens aberration part (there are compositing nodes to remove/add it) and think about the markers.

I have a your compositing tutorial and I was wondering what the specs of your camera was so I could take a swing at tracking.

Why wouldn't you add some dialogue describing what you're doing instead of useless music ?? :(