I must be missing something here. I rather thought the point to using a smallish number of polygons per unit area was that the computation required got thoroughly out of hand when you tried to draw too many. Splitting the same area into a much larger number of slightly spaced apart polygons, even if calling them atoms, doesn't obviously improve matters.
Why would rendering a spectacular number of tiny dots be easier to do than a large number of polygons? Tragically there doesn't seem to be any reasoning behind the idea in the youtube video, just lots of pictures.