The introduction of metallic/interference RGB pigments will probably result in nice screens, but for proving the RGB concept on a more general level I'd prefer a KISS approach. Thinking about the theory, I would think that having gain is not essential for it to work.
Here's a thread that I think is pertinent to this discussion, started by prof55:A Slightly Scientific Study of Colored Light
Now, if we look at the red filter in red light (third picture, left column, second row), we can see that it's almost as efficient at reflecting red light as the white background. And then, when we look at the red filter in ambient light (third picture, left column, first row), the difference to the white is much stronger. What does this mean? I would say it means that if we watched red-and-black movies on a red screen, it would offer improved ambient performance to a white or neutral gray screen. Because it would have the "reds" of a ~0.9 gain screen, but much darker "blacks."
Now, back in reality, where our movies, and our screens, have more color than just red, this advantage will be diminished by the inclusion of greens, blues and whites. The question is, does the advantage go away completely? Remember that all an RGB screen has to do is be more efficient than a "traditional" neutral gray screen at reflecting projector light vs. absorbing ambient to be useful.
Speculation is fun, but how to test this? Again, keeping to the KISS principle, two neutral grays should be mixed into the same flat white base. One should be a "shop color", composed of lamp black and whatever additives will make it neutral. The other mix should be tinted with RGB pigments, and made as neutral as possible. The mixes should be tinted so that in strong ambient light the RGB was very slightly darker when painted and cured. Neither should exhibit any sheen. These samples should then be moved to a light-controlled room, and shot with a PJ. If the RGB panel is brighter, then the concept works, and the RGB will offer improved contrast under ambient conditions. If it's still darker, or too close to call, well...
I've been meaning to do this test myself, but I'm unclear on exactly what pigments of red, green and blue should be used, if one wants to avoid pearlescents and such entirely. Also, using pigment powders instead of pastes might allow one to prevent one paint from developing more sheen than the other.
Originally Posted by benven
And you are correct, in that the thickness of the mirror dsoes impact haloing. I need to repeat with a much thinner mirror. Would a first surface mirror, like a piece of mylar work here?
Yes. In my experiments I've found that mylar gives you gain similar to a glass mirror, with no visible haloing.
A VERY thin second-surface mirror should also give good results, and reduce SDE in the process. It's hard to say how thin you'd have to go, but I'd guess that something in the vicinity of the width of the pixel-gap on digitals would be OK. So we're probably talking rolled goods.