Indexing the archive…
Your Universe of Digital Possibilities
A smooth ramp from a dark plateau to a bright one. The light has no stripe in it — the plot underneath proves it — and yet you will see a bright one at the right knee and a dark one at the left. Cancel each with a real counter-line until you cannot tell, and the instrument reads off how strongly your retina subtracts its neighbours: the weight of your own receptive field’s surround.
A narrow excitatory centre minus a broad inhibitory surround of weight w, normalised so a uniform field comes through unchanged. On a ramp it is silent in the middle and overshoots at the knees — a bright band where the ramp meets the bright plateau, a dark one where it meets the dark.
Each unit’s response r is its own excitation e minus its neighbours’ responses above a threshold r⁰, weighted by k — measured in the horseshoe crab’s eye. A network of such units exaggerates every edge it sees.
A pixel value V is not a luminance. Every Iris stimulus is specified in linear light and encoded through this curve (IEC 61966-2-1), and ramps are dithered in linear light — so a ramp on the screen is a ramp of light, and two equal codes are two equal luminances.
The second percept turns from what a surface is to what an edge is. A centre–surround unit answers a uniform field with silence and a change with a shout, so a ramp’s smooth middle is quiet and its knees ring — Mach saw the ringing in 1865, eighty years before anyone recorded a receptive field. Your nulls weigh your surround; your fitted filter then fails, honestly, on the curvy grid, where straightness — not inhibition alone — decides. What the same visual system does with a shadow is The Checker; what a bank of such Mexican-hat filters, one per scale, makes of a signal is The Loom; what noise does to a threshold is The Noise.