Mike,
Thanks for the additional info about the MP-E-65.
None of this is any surprise, except that I think you're being pretty optimistic about that 1.5mm DOF at
5:1.
Let's assume you image at 5:1 onto a 22mm sensor, and from that, make a final image that's 110mm wide. Then your total magnification is Mtot = 25X, with enlarging magnification Menl = 5X.
As summarized in
Ted Clarke's article, diffraction limits you to 330/Mtot^2 =
0.5 mm DOF, even if you're tolerating a pretty fuzzy result (3 lines/mm). Notice that 110mm x 3 lines/mm x 2 pixels/line = 660 pixels. This is strictly web-resolution stuff.
To get this resolution, you'll be running at an f/number of 330/(Mtot+Menl) = f/11.
Stopping down farther doesn't help -- it just makes the whole image even fuzzier. There's a reason that Canon limits you to f/16 on that lens -- they don't want people complaining that the glass is crappy, when the real problem is diffraction.
If you want a sharper image or more magnification, you have to open up even farther. Make that final image be 220mm at 3 lines/mm (1320 pixels) and you've got Mtot=50, Menl=10. You're only going to get 0.13mm DOF, and you'll need around f/5.6 to get it. Make the aperture any smaller and you get useless magnification -- diffraction blur bigger than your pixels.
Next time you have your MP-E-65 set up, run some tests and check it out for yourself. One of my friends who ran exactly that experiment with his MP-E-65 was rather surprised to see that f/16 was not his friend at 5:1.
All of the above explains why I stacked this cluster of frost spikes. I needed about 5mm total DOF for a subject that's only about 6 mm wide. But that's Mtot=18, and the best I can get in an individual frame is only 1 mm DOF at web resolution. Solution: step focus and stack 'em.
Yeah, high-magnification stuff is a pain. Those dang light waves are just too big!
--Rik