This isn't augmented reality, this is a CG trick by a film student.
Please don't submit comments complaining that a submission is inappropriate for the site. If you think something is spam or offtopic, flag it by going to its page and clicking on the "flag" link.
(Although if votes on my past submissions are any indication of belonging, I would have to observe that SMB seems more appropriate to this forum than an essay by V.I. Arnold on teaching math, or an article linking databases to category theory. This is also mystifying.)
First, you need footage of the game being played through, obviously. Then, take the static elements (the playfield) and create a giant animated map, including coin animations and enemies walking. Then, splice in the capture of Mario running through the levels, removing any collected coins or defeated enemies from the "static" playfield.
Once your giant resolution animation is finished, you then need to shoot some video for the backdrop. In his video, he obviously timed his walking to where the playfield would be (possibly by playing back the video on a handheld device while filming?).
Now that you have the video game animation and backdrop, it's simply a matter of layering the animation onto the video using camera matching. Most video post editors have this capability.
Most of the work looks like it resides in the actual animation. If you notice in many underground levels, they aren't arranged the way you see in the video; he did alter some portions to form a continuous map.