For those who didn't see it, Dale Eason just published a before and after coating phase contrast set for a 16" mirror - check out this post on the Zambutomirrorgroup (you may have to join the Yahoo group to access the files).
Phase plate images are a good indication.
For those that can't or won't join:
-the roughness features before and after are the same. The coating did not add roughness features of its own.
-the coating doesn't "smooth out" features either; in fact it may make them slightly deeper (that's not a surprise, because that's an effect I know from Al coatings on the simplest integrated circuits, which become "unsmooth" once you add Al layers and etch them away to make the conductive traces).
-(This is an opinion) From what I can see and the sensitivity of the test, the difference is really irrelevant, at least for this mirror - and one that the mirror maker himself rates as being less smooth than e.g. a Zambuto (but certainly smooth enough to make the differences have little relevance).
Contrary to some popular opinion, *very* small amplitude irregularities - too small to induce significant wave front phase shift - don't rob contrast (after all, in a tunnelling microscope, even a "perfectly smooth" mirror has bumps and troughs), and that applies for even some features you can see it in a phase contrast test. It's larger amplitude roughness that it dangerous because it makes some tests unreliable (some tests necessarily undersample the surface, and the smooth surface that fits the test samples may underestimate the phase shift that the real surface introduces).