Color Temperature and White Balance
Color temperature gives a way to describe the composition of light, specifically the spectrum of colors in light used to illuminate a subject. Light from 3200K tungsten photo flood bulbs contains more orange and red than blue while daylight, which typically has a color temperature between 5500K and 6500K, contains a more equal amount of the various colors of the spectrum. So, what does knowing this get us?
Visualize, if you will, daylight striking a flag. It can be a United States flag or, maybe you'd prefer a Union Jack, a French flag, or even one from the Netherlands. It doesn't really matter, as long as the flag you visualize has red, white, and blue in it. Since daylight has a fairly equal spectrum, it follows that the intensity of the light reflecting from the red and blue parts of the flag and toward us (and our camera) is roughly equal. And the light from the white part of the flag is reflecting all the components of the daylight equally.
Now, lets take that flag indoors and light it up with tungsten light. Because tungsten light is rich in reds and poor in blues, as illustrated by the graph to the right, the light reflecting from the red part of the flag is going have a higher intensity than the light from the blue part. In addition, the white part is going to reflect all the spectrum of the illuminating light, so the white in the flag is actually going to be composed of more red and orange than blue as well.
Because of the differences in the light between these two situations, the data captured by our camera is going to be different. And, without doing something about this difference, the image taken under tungsten light might not be recognized as a flag from whatever country it originally came from.
I'm not sure I follow. I've seen these flags both outdoors and inside. I don't remember seeing any difference between them.
You, my friend, have a highly evolved visual cortex which automatically compensates for the kind of light you're standing in. But, the sensors in our digital cameras don't compensate. Instead, they literally record the absolute amounts of red, green, and blue light that is reflected by the subject. In fact, if you take a neutral surface and photograph it using different sources of illumination, you'll end up with very different amounts of red, green, and blue data being recorded.
Here's an illustration that shows the relative amounts of red, green, and blue that is captured using different illumination sources:

In order to make a neutral surface look neutral under different kinds of light, the software in either the camera or in your RAW processing tool uses a color temperature value to derive a compensation to apply to the data recorded by the sensor. The result, if performed correctly, is a neutral rendering of your subject. It's the visual equivalent of using an equalizer to balance the various frequencies in music.

Of course, not all the surfaces in your scene will be neutral, but neutral surfaces are easy to talk about and use as illustrations. Also, because a neutral surface should reflect all colors equally, the compensation to make a neutral surface appear neutral will usually make all the colors in your image render correctly.
So, that's it? That's all there is to it?
Well, almost. The one thing I've been glossing over is that color temperature is defined by a theoretical black body. Some light sources are modeled by a black body fairly well. Other sources emit light that aren't modeled as well by black body radiation. And sometimes, our subjects are illuminated by multiple kinds of light sources, each adding its own special spectrum to the mix. In many cases, a color temperature value can get us pretty close, but other times there will be differences between theory and practice.
These differences will usually show up in your images as a green or magenta cast after the color temperature correction has been applied. To help you tame this difference between theory and real life, RAW software processing tools provide you with a tint control to dial in a compensation for this green or magenta shift. You can use it to perfectly tune the colors in your photograph.
OK, smarty-pants, how come we never had to worry about this with film?
Back in the days where we had to hike to school uphill both ways in the snow, this whole discussion of color temperature did matter, at least to professionals and advanced amateurs. When you bought film, you had your choice between daylight film for shooting in 5500-6500K light, and tungsten film for shooting in 3200K light. If your light was different than this, you had to break out the filters for your lens and start correcting for it the old fashioned way. Or, hope that your lab was good enough to correct everything up when it printed your color negatives.
In any case, it was a concern, but there was less flexibility to deal with it. So most people didn't even think about it. And, most print labs did a good enough job to where a lot of people were able to ignore the problem. Some would say that ignorance is bliss, but I for one am actually glad to be able to adjust to different lighting conditions so easily. And I'm glad to have the ability to tweak white balance so easily after the fact. If I want a warmer tone for a photo, I can just tweak the color temperature a tad in Aperture, Lightroom, or Camera Raw and get the results I want after the fact instead of being bitter that I didn't use a warming filter when I took the picture.
What about other light sources, like fluorescent and sodium lights?
Fluorescent and sodium lights are much harder to talk about when it comes to white balance as they don't have the same kind of continuous spectrum output that the sun and incandescent lights do. The better, newer fluorescent bulbs are getting there, but old school fluorescent bulbs have a huge green spike in their spectrum. But really, a discussion about how to deal with color correction with fluorescent lights is one for another day. As far as sodium lights, their spectrum is so spiky as to require beer or even more suitable beverages to really talk about.
Related Posts:
- Measuring the Color of Light, which gives some background on why the color composition of light is measured using the Kelvin unit of temperature.
Related Links:

2 Comments
One of the better explanations of this issue I've ever read.
And a fun read to boot.
~~ chaos pilot ~~
Reply to this comment
This is indeed one of the better explanations I've read. I had sort of worked a lot of things out... most of it comes down to making sure that your whitebalance is correctly set (which is why I always have a white paper with me).
I just can't believe you actually forgot to mention Luxembourg for the blue-white-red example ;-)
Reply to this comment
Leave a comment