Ahh the penny has dropped, i understand what you're trying to do now. I was imagining you requiring two states of polarization, but you only need one because you actually need to remove the polarized light in order to see the specular light.
Exposing pcbs direct from the screen was the idea. If it worked I would have been able to print pcbs with a very tight pitch, trouble was that the leds were peak 400nm so only just in the uv, and glass is good at absorbing uv, so the part of the spectrum I was after was dim to start with and dimmed further by the glass. I'm tempted to revisit the idea at some point with a bit more careful thought. It almost worked by the way, it took about 20mins to partially expose a small area of photoresist but the backlight didn't quite have the umph to do the job properly.
This is exactly why 3D Printers that cure UV resin only work using DLP Projectors
(And as soon as I can afford one I will build such a printer because it has really great accuracy compared to my standard FDM 3D printer :3)