Here is his photo of quartz from Clara (from mindat) Mindat photo of the day on Nov 4, 2006.

Here is a link to a proustite photo of his, and here is the photo:

This specimen is labelled as from the Neue Hoffnung Gottes Mine, Bräunsdorf, Freiberg District, Erzgebirge, Saxony, Germany Included in Quartz. Picture width 3 mm. Collection and photo Stephan Wolfsried

Photo added: 30th Oct 2005 The quartz photo was added Oct 30, 2005, and the proustite photo was added 12-23-2006, so both were taken with the Zeiss V12 and the Nikon Coolpix 8400 based on the information given below.

Stephan gives a lot of information about his methods (and history of the development of his methods) on his mindat user page.

He began taking photographs through the ocular of his first microscope (which he does not specify), but says it was "not very satisfying".

His next microscope was an Olympus SZ60, which had a phototube. He says be built his own adapter, and used an SLR camera (with film) on this setup for quite a long time. He says though that his exposure times onto tungsten balanced film were from 30-60 seconds, and that he never knew for sure how things had turned out till the film came back from the lab, and typically only 2-3 exposures on a 36 exposure roll were satisfactory.

Digital photography brought a big breakthrough and in late 2004, he bought his third microscope, a Carl Zeiss Discovery V12. He used a Nikon coolpix 8400 camera. Attempts with a Canon D5 were not successful, due to dust problems. (He notes that the 8400 had some hot pixels that he needed to deal with)

He says that lighting is at least as important as the microscope, and makes the following comments.

All the time I use halogen light sources (250W from Schott). I refuse LEDs because of the high light density which spoils almost every picture, at least without diffusing. In addition, the colours aren't really true, see also Tony Petersons remarks on the micromounting message board.

In 2008 he upgrades his microscope to the V.20 (FOV 20...1 mm) and obtained an adapter for the Canon G9 camera (12 Mipxels instead of the previous 8). This microscope is capable of 225x magnification (the V12 allowed 150x). His pictures dated 6/2008 or later are 12 Mp fotos compressed to 1000x750 pixels. I have been told that the V20 is a $40,000 microscope (more or less).

In December 2008 he switched from the Canon G9 to a brand new Nikon Coolpix P6000, with slightly better noise and resolution preformance. The big reason for the upgrade though is the infrared remote shutter control.
He says:

This gives the possibility to take lots of layer pictures without having a hand on the camera or the microscope either, which improves quality at high magnifications. Before, every touch on the camera in order to release the shutter (with 2 sec. delay of course) gave a micro impact to the whole system with vibrations and the risk of a movement of the object.

He also notes:

The Canon G9 lenses were not fully sealed to atmosphere and suffered from dust in the lens I had to work with dust maps then. With the P6000 there is actually no need for cleaning artifacts or anything else. Pictures on mindat from January 2009 on are compressed from 13.5 Mp pictures to 1024x768 pixels.

In December of 2010, he switched to using a bellows with Zeiss CZ Luminar lenses and a Panasonic Lumix GH2 camera. The Panasonic GH2 is a micro 4/3 camera which is an interesting concept. There is no mirror to flip up and cause vibration, and the camera are quite thin and compact, but we are drifting off the topic.

I corresponded with him in February of 2011. I asked about whether he was having dust problems with the GH2 as with the Canon 5D. He said that indeed there was dust to deal with, but the big difference is that the dust does not move laterally with focus changes (as it does when shooting on a microscope using a phototube), so it is easy to deal with (lateral motion produces trails and streaks on stacked images).

His own words abut his current setup:

In December 2010 I finished assembling a second stand with motor driven stative from Carl Zeiss like the Discovery V.20, but with a bellows and CZ Luminar lenses in front instead of a microscope. See thread at the Photography topic. I choose the Panasonic Lumix GH2 for that purpose: 16 MPixel sensor with 5,5 MPixel per square centimeter imager, no mirror, tilt and swivel display, radio controlled wireless shutter release. With no practice at all I could achieve pretty good pictures at the very first attempts. Their resolution is remarkably better than taken with the Discovery V.20, but only down to FOV of appr. 1,5 mm. The whole setup seems to be more robust against reflexions and bright light spots. Maybe the PlanApo 1,5 of the V.20 gets too much diffuse light. The 25 mm Luminar seems to become the most appropriate lens with a FOV from 3,5 to 2,5 mm with the bellows only and down to 1,5 mm with a bellows extension. 11 to 21 mm working distance allows enough external illumination with gooseneck fiber lighting. Some first photos are now uploaded, more will come. There is obviously a limitation to FOVs larger than 1,5 mm for the bellows. For FOVs smaller than 1,5 mm the Discovery V.20 remains for me benchmark. Pictures taken with the Luminar 16mm or the Nikon M Plan 40x have a good resolution but suffer from very small depth of view, which lets the transition from sharp to unsharp areas appear very prompt and artificial. The Luminar 25 and 40 mm lenses are my favorites now and I take a lot of photos from my personal collection once again. Some of them I could improve over years step by step. I cannot see my very first attempts now any longer but don't erase them to see and show the amazing progress.