Correlated color temperature/color temperature

T

TheJackyl

Active member
These are all quotes out of the American Cinematographer Manual, I'm trying to understand color temperature.

"Color temperature describes the 'true' temperature of a 'blackbody radiator' and thereby completely defines the spectral energy distribution (SED) of the object. "

What exactly is a 'backbody radiator' and is that also saying that color temperature equals the SED of the object?

"One of the most important features of incandescent radiators is that they have a continuous spectrum. This means that energy is being radiated at all the wavelengths in its spectrum. The term 'color temperature' can only be properly applied to radiating sources that meet this requirement. When the term "color temperature" is applied to fluorescent lamps (or other sources that do not meet the criteria for incandescence), it really refers to "correlated color temperature.""

"The Term correlated color temperature is used to indicate a visual match where the source being described is not a blackbody radiator. The term is often abused, an example being its application to light sources such as mercury-vapor lamps. From a photographic standpoint, the correlated color temperature can be extremely misleading. It is important to keep in mind that its connotations are visual. It is a number to be approached with extreme caution by the cinematographer."

I guess basically I don't understand at all what they mean by correlated color temperature and was hoping someone could explain it. Things like why a fluorescent light isnt a blackbody radiator, and what they mean by its connotations are visual, possibly with examples to compare with other light temperatures? and lastly, why are they misleading to the cinematographer? thanks a lot!
 
That's a lot of technical jargon that I think boils down simply to the fact that these "correlated" sources are a bit unpredictable if you're relying solely on numbers to figure out their color temperature. An incandescent bulb has a very smooth spectrum and is always reasonably close to 3200K unless you dim it, but sodium vapor, fluoros, metal halides, etc. have what they call "fragmented" specturms with spikes in them that affect the consistency of the color temp.
 
A black body radiator is a theoretical thing being heated up.

Think of a branding iron in a fire. At about 2,000 degrees it starts to glow red.
At about 3,200 degrees it is orangish hot, and at about 5,600 degrees it glows blue-white hot. Sound familiar?

The outer surface of the sun burns at 5,600k.

Tungsten lights give off light by warming up to ~3,200 degrees. Any hotter and they burn out too quickly to be useful.

These light sources hava bell shaped curve that peaks at the top (say 3200k for tungsten) and falls off gently to either side. Some on the blue side, and some on the red side.

Other lights like fluorescents, sodium, mercury vapor, HMI, kinoflo, LEDs, and television sets create light by electronics or electric discharge. Their spectral fingerprint has all kinds of spikes and gaps. Our eye tries to read it, but we fill in the blanks. Film does not, that's why a fluorescent light might look white but read as green on film. That's what the ASC manual was referencing.
 
Thanks a lot for the replies and information it is very helpful, I think I am beginning to understand this.

i - when you say "These light sources hava bell shaped curve that peaks at the top (say 3200k for tungsten) and falls off gently to either side. Some on the blue side, and some on the red side." What would determine what side, blue or red, the color would fall off too, would this be the color temp, 3200K+ would be blue? and 3200K- would go towards red?

Also could could someone touch on what is the difference between a regular fluorescent light and a kinoflo would be, since they both have spectral spikes and gaps. Does digital treat them the same way as film does, where a fluo could look white to the eye but on digital look green?

thanks again for helping me out
 
I found the following link very good:
http://images.google.ca/imgres?imgu...refox-a&rls=org.mozilla:en-US:official_s&sa=G

A "regular" fluorescent tube flickers at 50 or 60Hz but these:
http://www.theodoregray.com/PeriodicTable/Samples/063.5/s7s.JPG
as well as Kinoflo work at 20KHz providing "flicker free" lights for film/video cameras.
More about the fluorescent spike spectrum here:
http://en.wikipedia.org/wiki/Color_temperature
There are also "full spectrum" tubes as well as "daylight" in hardware stores.
 
If the kinflo are 'flicker free' because they run at 20KHz then why are they classified with the rest of the lights with spectral spikes and gaps (non-incandescent lights correct?)?
 
flicker refers to the consistency of light output over time. Spikes and gaps refer to the spectrum of the light that is output when the lamp is emitting. You cannot think of light as being a single temperature or color. Light contains a spectrum of colors, as you would observe if you looked at what comes out of a prism when a light beam is shot into it. That spectrun may have gaps or spikes (less or more of a specific frequency than the average of its output). Your eye has very sophisticated ways of ignoring all this. Your camcorder does not.
 
Thanks for the clarity between flicker and spikes Chris. I guess I am just unsure then on how different a fluorescent light would look compared to a tungsten on the DVX. If I light with all tungsten and white balanced as opposed to lighting with all fluorescent and white balanacing, what would be the advantages and disadvatages I'd see from each, relating to their flicker and spectrum?
 
If you white balance under the right conditions, you shouldn't see much/any difference. The real issues come into play when you're shooting film and can't preview the results on a color monitor, or when you mix sources, whether accidentally or on purpose, or when you use the presets (I almost never white balance, prefer presets and gels).
 
In most cases you can ignore the flicker problem if you are in the USA (NTSC and 60hz power). A few old fluor fixtures could present a problem, so test first. The real potential problem is white balance. Cheap fluors can present a spectrum to your cam that fools the cam. Step one is to do a manual white balalnce under the lights you will be shooting under. That handles most conditions. Auto white balance will frequently do real damage to your footage, especially if there is ANY other light source - even a different brand of fluor bulb in some fixtures. Most problems (if you use manual white balance) can be fixed in post.
 
When using scene file presets with controlled lighting, should I manually white balance, or will that change some of the color that was set up in the custom preset?
 
that is intersting also, if im corrected on what aking is asking, Id like to know the same thing, If you change/adjust the color setting for the scene files, then use that scene file, then manually white balanace, do you lose the color setup in your custom preset by going to white? or is the color preset part of the white balanced image?

ookay after typing all that i realized it comes down to, do you manually white balance BEFORE or AFTER tweaking the color controls on the Camera(DVX)?
 
You must log in or register to reply here.
Back
Top