Lighting temperature considerations for print evaluation?

Question

I am setting up a color management workflow for fine art photos that includes monitor and printer/paper calibration, ambient light control and a print viewing area.
I'm having trouble deciding which temperature lighting to use in a print viewing area for evaluating the color of final prints. Some experts say 5000K (kelvin) lighting is ideal, others tout 4700K (D50) as ideal, while others like John Paul Caponigro say prints are best evaluated at the temperature lighting they are to be displayed at -- like 3500K.
That makes sense to me, but how do you get photos to print exactly how they appear on the monitor when the monitor is calibrated at the recommended 6500K, and the prints are evaluated under 3500K lighting? Or am I comparing apples & oranges here? Could anyone help clarify this for me? Thanks.

Answer 1

RichPate wrote:
Thanks Larry. Very interesting article. Andrew Rodney and others over at the Photo.net Digital Darkroom Forum have weighed in on the article and my original question.
Its all quite simple really, and the long posts over on PhotoNet seem to be making this all far more complex than it needs to be.
5000K (or for that matter any value provided in Kelvin) is a range of colors. This is why its useful when expressing a kelvin value to use Correlated Color Temperature (CCT) prior to any value. Many different colors correlate to the same numeric value. One light source that is said to be 5000K could produce a different color than another light source said to be 5000K. A Standard Illuminant, like D50 is at least a definition of a fixed, single color based on its Spectral Power Distribution(SPD). That doesn’t mean that someone specifying a light source or calibrating a display to D50 results in the product producing D50. In a prefect world this should be the case. Another issue is one can take the same instrument and two different calibration products and ask for D50 or worse (because its somewhat ambiguous, 5000K), calibrate the display and get differing results.
Figure out what light source you will use to view the prints next to the display. Many are recommending CCT 3500K Solux bulbs. This can be used to view prints away from the display environment but that isn’t a rigid requirement. Your eyes will adapt to the conditions under which you view your prints either next to the display or 10 miles away. The goal is WYSIWYG in terms of the print and the display. That means you have to have this display in the environment where you view the prints. It means you have to calibrate the display and control the viewing environment of the print next to the display. If you select CCT 3500K Solux (or a Fluorescent GTI booth etc), you now need to set the calibration target values of your software for a display calibration and the resulting display profile. What are the right settings? One would assume that you would ask for a CCT 3500K white point in the software. That’s not likely going to work, again due to the CCT values here, the differences between the Fluorescent light (or maybe LED) display backlight and so forth. The right values are those that produce a visual match, period! 
The value may be D50 on an NEC SpectraView using their software. It might be CCT 6000K using EyeOne Match. It might be D55 using another type of display. Your Mileage May Vary. You have to enter various values until you result in a visual match. This is a reason why the better calibration products for displays allow one to insert an X/Y chromaticity value instead of relying just on a CCT or standard illuminant value.
To the same degree, the backlight intensity, specified in cd/m2 needs to be adjusted to taste. Again, YMMV. How bright is the light illuminating the print? Can you control this either using a dimmer (not useful for Solux, it alters its color) or by moving the light closer or farther from the print. The dreaded “my prints are too dark” issue is 98 out of 100 times, a severe disconnect between the intensity of the viewing booth and the target calibration of the display backlight. These people generally need to dial down the display luminance and/or raise the viewing booth intensity to result in a visual match.
Getting a CCT 3400K Solux and a decent display, driven with decent software to produce a match is totally possible and many people do this all the time. The same could be said of a good GTI or Just Normlitch (or similar) Fluorescent booth (although there are issues with Fluorescent lighting and interaction with OBAs in some papers).
Why CCT 34000K Solux? Simple, they look better. While a CCT 4700K bulb (and its new cousin the 5000K bulb) in theory should be a “better match” because they are numerically closer to the so called 5000K standard (its really D50), due to these values being not very reliable descriptors of the color, the cooler bulbs don’t produce as pleasing effect. If you go into galleries with Solux bulbs, especially those who have tested several different flavors of CCT values, almost all select the 3500K Solux. No reason not to use that for your custom made viewing booth. And as I said, if you then get your visual match between booth and display, take the print into the kitchen, outside, to another gallery etc, your eyes will adapt to the new white point, assuming its not some weird-*** illuminant (metal halide mixed with daylight etc). You will like the appearance of the prints when you move them away from this display to viewing booth environment.
Of course, all this begs for good displays, using good instrumenting and software. You may wish to produce multiple calibration targets and resulting profiles for differing papers (contrast ratio settings). Or you may be working with Solux and Fluorescent booths because customers are using the later and you need to collaborate with them. Mucking around with the display OSD’s to produce the ideal cd/m2 and white point is not the best approach, better are “smart monitors” that do all this internally in high bit (the NEC SpectraView II line comes to mind). Effective soft proofing is only as strong as the weakest link in the chain. But it is possible and achieved every day by many.
Could you please explain "lux" to me and the 250 lux figure you came up with? Is this something I can use the spot meter on my DSLR to measure? Is it similar to cd/m2? I'm an artist/photographer who is rather challenged at the technical aspects of color management.
Lux like cd/m2 is just a unit of measurement of light intensity. Lower for the ambient conditions is always better. It can’t be too low (it can be too high). Any ambient light in the environment will have an effect on the blackest black the display can produce so again, lower is better. You want the brightest and darkest object you view to be the display. When you are editing, the viewing booth can be off but at some point, you’ll want to view print and display. Here you need to correctly setup the soft proof (unfortunately that today means Photoshop) with the simulate check boxes on, with the image in full screen mode. No palettes, no UI elements which can’t undergo the white simulation. Black bkgnd for the image (hit tab key, F key until all you see is the image on screen, filling the screen or as close to the size of the print as possible).