Quote - and wants to dismiss the variance in colour output to limitations of the preview pane. Which I don't think is correct at all, and I intend to use Bagginsbill recipe to prove it.

 

I wouldn't discard the "preview is hopeless" hypothesis just yet. From my tedious 30 minutes looking at one preview and the other, I could detect no real pattern as to the changes taking place in the red and blue components (which is where the real puzzle lies). If it were something mathematically precise going on in the brick arithmetic, I'd expect some kind of predictable pattern to emerge. But as far as I could tell, there wasn't anything but seemingly random fluctuations. Which suggests it might be a preview-display issue. Of course, further tests will add some extra info to work on. I'm even tempted to download a trial of DSA and install it, now!