Here are 3 photos taken of the same specimen. These are crystals of the mineral Calderonite, from the C and B Vanadium Mine in Gila County, Arizona. It would be nice to photograph exactly the same place on the specimen, but given the 3 mm field size, I have a hard time finding the same place twice in a row. They were all photographed using a Mitutoyo 10x M Plan Apo microscope lens in front of a relay lens (in this case a Takumar 200mm f/4) mounted on a Canon 5Dii camera. These are crops from the full frame image, field of view in all cases in 1.8mm.
The process of making this images has gotten me thinking a lot about white balance and light sources. I have seen too many terrible mineral photos with completely screwed up white balance, and I am determined to do this right.
My fiber optic illuminator uses a tungsten/halogen bulb, an EKE 21 volt 150 watt GX5.3 bipin, which is marketed as a microscope illuminator bulb and has a specified color temperature of 3300K. My Canon 5Dii set for "tungsten" expects a 3200K source, so the results I am getting are good, and perhaps just a bit cool if these numbers are accurate. And indeed, the image above that was taken with this light does look a bit cool (bluish).
The inexpensive bedside lamp I bought at Ace Hardware for $21.00 uses a GU10 base 35 Watt, 120 Volt bipin bulb. I have found specifications on these bulbs ranging from 2800K to 3100K and calling them "warm white". Clearly mine is on the warm end of this range, perhaps even warmer than 2800K. (In fact as you will discover if you keep reading, my bulb is more like 2400K -- clearly these bulbs are not carefully controlled). A logical thing to do is to use the custom white balance feature of my camera (where I photograph a white reference object) - which can go as warm as 2000K if needed. A side note: one big plus of this lamp is that the bulb is rated for 2000 service hours, in contrast with 200 service hours for the EKE bulb in the fiber illuminator.
Ikea sells an inexpensive clip on lamp called the Jansjo (for under $20). This has an LED emitter, and the story is that different manufacturing runs of this lamp may use different emitters with different spectral curves. People who use these (and some have great success with them), use custom white balance settings in their cameras).
Another option is to use my tungsten bedside lamp as a "modeling lamp" to focus and orient the subject, then switch to electronic flash for taking the photographs. This could have a multitude of advantages. The oft cited disadvantage (especially with mineral crystals) is that it will be impossible to predict reflections off crystal faces. Another disadvantage is that it adds just that more complexity to my setup. I think that the question of predicting reflections will become almost a non-issue with a diffusion setup that is providing a dome of light from all directions. An electronic flash has a well known color temperature (6000K). I have not done this yet, but stay tuned.
The custom white balance almost looks "too cool" to my eye, but I am inclined to trust the camera. Human beings seem to "like" warm images. I want accurate images, not likeable images -- maybe. The 2500K setting looks "close enough for government work" and I would be happy to use it with this light. It might be a little warmer than the custom setting. The 2600K and 2700K settings are clearly too warm (but likeable). I would almost say that the color temperature of this lamp is something like 2400K. No surprise that with the camera set to "tungsten" (3200K) things looked much too warm.
I am willing to bet that when I replace the bulb, I should check all this again.
Here is a discussion of color temperature and photography. This site looks good and probably has other interesting resources.
Tom's Mineralogy Info / email@example.com