Digital Light & Color

This may seem like an obtuse problem, but it's a real life issue.

Monitor Curves are great if you could compare a test print with a monitor at the location where the print is to be displayed and then 'pre-correct' it. But what if one can't physically go to the display location, but the illumination level and light temperature (and more or less its spectrum) are known?

For instance: I know that the prints at the gallery will be illuminated with a broad-spectrum light source with a colour temperature of some 4100K and the print illumination level of some 320Lux. How do I guess the Monitor Curves to pre-lighten and 'pre-bluish' the prints so that they look somewhat better there?

Trying to correct the colour shift, I toyed with the idea of using Colour Balance, picking highlight, and using the 'colour to remove' with a colour picker set to Colour Temperature, moving the temperature from 6500K to 4100K, but I'm not sure if this is the way to go. With prelighting, I have no idea how to relate the lower light level (in the daylight the print is probably illuminated with some 10,000 Lux) to any prelighting curves.

I guess experimenting would be the way to go, but this time I have only one shot at it and it's quite dangerous to play with lightness and colour shifts when I'm not entirely sure what I'm doing and won't be able to see how it looks like at the location. What to do?

I don't have a good suggestion here. Given that the eye does a good job of correcting for color temperature differences I would focus on getting the gamma right. After all you can't change the color of the paper white.

But getting the gamma right, how do I start knowing the numerical value of the target illumination level at the gallery? Is there an approximate model that would relate the decrease in illumination with what the pre-lighting gamma should be to compensate for it? Cheers.

Not that I know of - I would make several test prints and see which ones look best.

I haven't studied concepts of "appearance models" in color science, but apparently those are the basis for automating color adjustments to account for different illumination levels and surroundings. More to read at http://www.handprint.com/LS/CVS/color.html

LittleCMS, the color management system included in PWP, does have some functions that can be used for such adjustments, see the "CIECAM02" section of http://www.littlecms.com/TUTORIAL.TXT But none of this is incorporated in the design of PWP color processing, and it all looks like (mathematical) greek to me. Maybe JS or KS have some comments on how PWP might make those color transformations approachable and usable by non-g(r)eeks. I think it would be a first in a photo editor.

There is a freely available Photoshop plug-in(which works in any Photoshop plug-in host, some of which are freeware as well) that does these kind of color transformations at http://cliff.rames.googlepages.com/ciecam02plugin It all looks pretty complicated, but there are more literature references on that page as well.

Many thanks! As I'm reading the references, I wonder if someone could offer a three-sentences-or-less, layman, physiological/psychological explanation as to why pre-lighting prints (forget colour shifts) to be viewed in environments dimmer than daylight ambient light (but not that overly dim, with no ambient brightness levels way higher than the print luminance range and its background, and allowing time for eye adaptation) will make them look better?

The amount of shadow detail that you can see in a dense black depends on illumination level. So if you anticipate lower illumination levels you must make your blacks less dense or your shadows will appear blocked.

Kiril

Could anyone point me out to the actual data of black/white reflectivities of typical photographic papers (or their densities Dmax/Dmin)? Cheers.

The black density varies with the printer and ink. Coated papers get a much denser black.

For more information on a closely related issue, see http://cool-palimpsest.stanford.edu/waa ... 1-308.html and a former discussion, at http://www.dl-c.com/discus/messages/2/1 ... 1211306971.