That indeed is rather cool. Though another method that would have the ability to increase the size of an image without necessarily blurring it is to convert the image to a fractal or vector-based representation, and then resizing it. Those technologies do in fact exist - though I am not sure if they would do better or worse than what is presented in that video. Also, I suppose one could simply do a bicubic resize of the image and then apply a sharpening (or unsharp mask) filter, but that's cheating, and will likely not provide good results unless it's a very modest increase in size. (this can be done in nearly any image editing program such as The GIMP)
This is about as close as you can get with current technology.