He came up with the 80m figure by multiplying the number of dots, or pixels, in a high-resolution image (about 1m-2m) by the number of surfaces that an imaginary ray of light corresponding to each pixel would encounter as it bounced around a complex scene (say, 50-100).