I have found out some interesting things about the economical mask making
process which was discussed here a while back. First I found a paper on this
process which goes into some detail on what works and why. It is:

Suleski, Thomas and Donald O'Shea. "Fidelity of PostScript-generated masks for
diffractive optics fabrication". Applied Optics. February, 1995. p.627-34.

I just ran an experiment using this process of laser printing to film and then
photoreducing by a factor of 10 only to discover exactly what O'Shea has
published. There are indeed limits to this process. Thermal fluctuations, small
dust particles, and lens aberrations are the biggest causes of problems based
upon what I've read and heard from people. The best O'Shea has been able to do
is 8um with his lens system, although he mentions using a better system for a
2um resolution. The people I've talked to have only been able to do 5um stuff
with their ruby-lith process. The best I personally was able to do was about
20um.
Justin Mansell        | "The voyage will be a success only
1626 E. 115th St. #102| if the ships do not collide AND each
Cleveland, OH 44106   | ship is seaworthy."
(216) 754-1691        |             -C.S. Lewis on Morality