I've been experimenting, sometimes with pretty good results. My ultimate goal is to be able to take amazing pictures for the "single image science lessons" being developed for a kids' program. Too much else going on to adequately document progress so far, but I can make some comments.
1) If it's just reflected DIC you want, it's pretty easy and relatively affordable to buy used gear. And the prisms in an epi unit can provide a basis for further experimentation.
2) My take is that there are three limitations to achieving somewhat decent and affordable transmitted DIC results. First, is finding affordable DIC prisms. Even used, a complete set of prisms top (above the objective) and bottom (in the condenser) runs near $2000 for older gear (e.g. PZO, 1st generation Nikon, older Zeiss), maybe $3000 for slightly more popular gear (e.g. Olympus BH but not BH2), and $4000 up for more modern scopes (BH2, BX series, etc.). The cost is prohibitive for most hobbyists. So, I suspect what most DIY hobbyists have done is buy most any prism that shows up cheap, hoping it can be matched up in a system. Even at this level, you're looking at around $100-150 per prism and might want as many as eight (four top and bottom) prisms.
2) Second problem is that practically no prisms are marked with shear angles or even orientations and few objectives have an indication of where the prism (with some specified shear angle) should be located. So, barring optical testing equipment and greater knowledge of optics than I have, it's a trial and error process. If anyone can point me to a somewhat decipherable optical tract that could help with this, I'd be grateful.
3) Third problem is more mechanical - setting up rigs where one could fairly easily experiment with prism locations. They ideally want three degrees of freedom (rotational, lateral, up and down). That one is a bit easier for me, having been cursed/blessed with the "tool gene."
I've had decent results with these combinations:
1) PZO DIC condenser, re-oriented 45 degrees on a Leica DM series scope, fitted with Nikon 60mm infinity objectives, with Olympus epi prisms fitted to a DIC topside slider. Probably easier, though, to just look for a complete PZO system (though limited a bit by the older PZO objectives). There has been a glut of very cheap 20x infinity Nikon Apo objectives on Ebay -- and DIC results with this objective are very satisfying (at least to me).
2) Various Olympus stands with a proper top slider, with epi prisms removed and re-located as close as possible to the condenser diaphragm. If anyone wants to sell an Olympus condenser populated with Olympus prisms -- for just an arm OR a leg -- please let me know.
3) An actual and complete Nikon v2 DIC condenser and slider (but not holder) fitted to a makeshift holder. This works well with both Nikon top end finite and Olympus top end finite objectives. Olympus SPlan objectives are what they recommend for DIC and they seem to work pretty well. The Nikon and Olympus finite Apos I have also work well.
4) Olympus epi DIC prisms (sometimes affordable) fitted to Olympus SPlan objectives and a makeshift 1.25x polarizer holder up top, and those same type prisms removed and fitted to condenser apertures as close as possible to the iris. I modify condensers to fit a slider holding these prisms.
Sometimes there is banding evident. Sometimes the effect is best with a combination of oblique (prism shifted) and shear angle illumination. And sometimes it's very good. When time permits, I've got a batch of older Leica/Reichert epi prisms to try as well.
In general, my experience is that the cost of experimentation is about the same as biting the bullet and spending the money just once on a proper and complete DIC system. Still, I've enjoyed the experiment and in the end I expect to have half a dozen pretty decent (image-wise) DIC systems.
Someone wanting DIC for a clinical application would likely be crazy to head down this path -- you'd have even less idea of what you were seeing in terms of the false relief. But, someone wanting cool images might actually find greater control with DIC prisms that can be combined with oblique movements.
I realize folks would like photos and better descriptions, but I simply don't have time to make that a priority now. What I can say, is that it's possible -- if time consuming -- to cobble together a pretty decent (image wise) DIC system with just the prisms costing under $1K. But recognize it's also a bit like diving into a black hole