Color Temperature
Oct 2nd, 2007 by Leo A. Geis
If you are a professional photographer it is absolutely necessary that you understand Color Temperature in order to manipulate or compensate for it. Photoshop practitioners have access to R, G, and B Curves (but remember…in Photoshop they are adulterated), and on some cameras you even have Color Temperature bracketing, but like a chainsaw these tools can be incredibly destructive when misapplied-and their application is not intuitive.
Back to Max Planck we go. This page on techmind.org has an exceptional, concise review of Color Temperature fundamentals. The graphics are superb-you should be able to digest the entire matter in less than 5 minutes.
Please take special note of the definition of Color Rendering Index, which will be a critical matter in our upcoming exploration of Metamerism.
Outdoors, Color Temperature is situationally varied:
- Since the sun (which effectively functions as an “Absolute Black Body”) has a surface Color Temperature of 5800K, that metric serves as our basal Color Temperature.
- In the early morning or late evening the sun’s light rays must travel more obliquely through the earth’s atmosphere than when the sun is overhead. This Color Temperature “situation” will obviously vary because the sun’s position relative to the observer is continually changing. The atmosphere tends to absorb and otherwise affect the Red end of the spectrum more aggressively than the Blue-thus, remnant light is biased toward Red. The more atmosphere the sunlight encounters, the more pronounced the effect. Technically, there is some absorption by O2/Oxygen and O3/Ozone, but diffusion and reflection (not completely unrelated to Rayleigh Scattering) are the primary machinery of the effect.
- Color Temperature increases in shade. This can play havoc with automatic Color Temperature sensing and calculations performed in camera or an improperly used incidence meter.
In digital photography (as opposed to the science, Colorimetry) Color Temperature is calculated only as a shift between Red and Blue with no Green abstraction. “Green Compensation,” to the best of my knowledge, is included in some software simply to counterbalance composite Luminance changes produced by R and B manipulations. Green adjustments in Camera-based Color Temperature bracketing are designed to speed workflow by facilitating off-axis grid-based adjustments and/or to specifically allow for the addition or reduction of Green for a desired subjective effect.
Considering that outdoor Color Temperature is constantly changing due to atmospherics and the sun’s procession, and that Color Temperature varies from direct sunlight to shade, you can sense the potential for variable results when relying upon in camera processing to either compensate for or homogenize Color Temperature during a (typical aerial…) shoot. Fortunately, RAW images have no Color Temperature character (I’m not saying that the sensor or algorithms are unable to differentiate R and B registration within composite light…) and the application of a standardized Color Temperature character during demosaicing (conversion to .tif) is easily automated. It’s yet another in a very long line of reasons to shoot RAW, even if you are using a midgrade or “Prosumer” camera.
For our purposes of developing an understanding of Metamerism, it is important to know that various light sources such as flourescent bulbs and incandescent bulbs also have distinct Color Temperatures. General Electric has a very comprehensive library on that matter here.
For those who wish to expand their knowledge a bit beyond these basics, I’d recommend familiarizing yourself with the concept of a Mired and the Color Temperature Chart and Normograph contained on this page. The best resource on Planckian Locus I could find is here (see pp. 4-5).
Finally, I strongly recommend that aerial photographers play with a few images shot in color temperature extremes to familiarize themselves with the unique Histogram characteristics of such images and the limitations of common Photoshop techniques for bias resolution. I have found that the Hue, Saturation, Color and Luminosity Blending Modes, combined with Layer Masking, are much more potent (if tedious) tools for extremely problematic bias situations (including mixed lighting) than Curves.
Incidentally…I just realized that I have neglected to include the Luminosity Blending Mode as a member of the Luminance-only Histogram adjustments available in Photoshop. My apologies.
L


