Instead of going into it blind, I decided to make a mold of a small piece and try the technique first. I built the mold tonight from some XPS scraps, using some Goop adhesive and filleting the interior with RTV silicone. Tonight it will sit to cure, and tomorrow I will pour a few layers to test mah theory.
And you of course want pictures.
XPS (pinkboard) and RTV silicone caulk, smoothed with naphtha. The mold will likely be destroyed in the demolding process, but the filleted corners may make a pop-out with enough mold release first. If I can re-use it, I can do a couple of tests, but it's not that difficult to make up another one.
I want to see how well this stuff releases from an XPS-faced mold. Mold release is my friend, but this run will determine whether I want to create a positive from the XPS-and-glue method and do a neg-mold with the silicone-and-fiber method. I'm tending toward the second because it's a method I need to try for the paver and stone molds as well; it's using fiber to reinforce the spaces between the pieces and counter some of the need for the far more expensive silicone. It's a little tricky to embed the fiber in the silicone, but I have a theory: it could be bunnies.
Naw, I'm just kidding, it's not really bunnies. But there is a trick to it.
Tomorrow morning I will sleep in. Forward to this looking I am.