Talk:Color temperature

From Wikipedia, the free encyclopedia

This article is within the scope of the following WikiProjects:
This article has an assessment summary page.

Right now this article mainly discusses the practical application of color temperature in cinematography -- we need information on the physics of it...

Agreed. But I don't know anything about the physics of it. --Koyaanis Qatsi

Nice work anyway KQ! I really enjoyed reading it (OK, it does need the techie stuff though) - MMGB

yeh. and the prose needs a little work. but i'm off to bed now.  :-) thanks, though. much appreciated. --Koyaanis Qatsi

Added more physics-oriented discussion, made distinction between colour temperature and white balance. -- The Anome

Hehe - Article title "Color Temperature". First sentence "Colour temperature...". God I love this place. - MMGB

Yeh, it's great. Thanks, Anome, for adding the physics stuff. Sorry to use the term "loosely"; it's very common in the film & documentary industry. --Koyaanis Qatsi

Yep, I know -- The Anome

Cinematographers don't/can't "white balance" in the same way as video camera operators: they can use techniques such as filters, pre-flashing, and after shooting, color grading (both by exposure at the labs, and also digitally, where digital film processes are used). This needs to be incorporated into the article. -- Anon.

Yes, it does need to be integrated into it. I've worked with both film and video, and noticed that, but never noticed the lack of coverage in the article. However, I'm off to bed now, so I'm glad to see you've added it. Thanks.  :-) Koyaanis Qatsi 07:20 17 Jun 2003 (UTC)

Mark Sweep created pure art work from your basic start. I tried to help a little too. I want color temp to link like this, but I'm gettng the wrong words incandescent or incandescance, should link to the section called Incandescent Bulbs but [ "Incandescent" / "Incandescent Bulbs" ]]doesn't give me back my original word.gimma quick lessson please.I want to do this to fluorescence too.[User:Dkroll2|Dkroll2]]--[[User:Dkroll2|Dkroll2]] 04:09, Dec 14, 2004 (UTC)

All you need to do is write this: [[Incandescent light bulb|incandescent]] he pipe character "|" is used to show different text in a wikilink. Put the article name before the pipe, and the desired text of the link after the pipe. It'll show up like this: incandescent. But please remember that we only need to link the first time each term is used. There's no need to go through articles linking every instance of the term "incandescent". Rhobite 05:17, Dec 14, 2004 (UTC)

OOPS!//excuse me whilst I go fix a ton of stuff. DKroll2 OK you Color Temp guys got your physics, well some of it. Enough so It it didn't get too complex. I didnt agree about white balance and photography, since there are so many alternative applications, but at least you HAVE applications sections. AND I learned a bit mre about how white balance relates to color temp. thanks DKroll2

Contents

[edit] spectral power distributions

The plot showing a spectral power distribution looks representative of older "halophosphate" fluorescents, but not of the majority of modern tubes found in offices etc which use a newer "triphosphor" mix. My personal web page has a measured triphosphor spectrum (don't use as a wikipedia source!) http://www.techmind.org/colour/tld32w84hfsml.gif the key point being that the modern triphosphors have a very spikey spectrum with very little continuum at all. You may find emission spectra graphics from Philips/GE-Lighting/...

For spikey emission spectra, even if the overall colour were equivalent to a blackbody, the way it renders reflective (object) colours will always be "off" [colour-rendering issues - reflected colour arise from the product of the lightsource spectral power distribution and the object spectral reflectance]. Consequently, for photography it can still be impossible to properly colour-correct fluorescent light using filters.

I'll try and explain myself better and edit the article itself sometime.

57.66.65.38 12:58, 23 August 2006 (UTC) Andrew Steer ( http://www.techmind.org )

[edit] citations.

Question about citations.

I did 2 with success on those sections I wrote, but:

1: how many sections/paragraphs up from the citation is it valid? problem: I desire to find citations for my contributions, but not to some others, with whom I disagree, etc.

How do I limit my citations to my writings?

A thought: What if my writings are eventually rewritten, or even just editing incorrectly? The citation is invalid and is an insult to the cited author.

Hmmmm.--[[User:Dkroll2|Dkroll2]] 06:23, Dec 23, 2004 (UTC)

I fixed the link to Planck's Law, but I wondered if that shouldn't be Planck's_law_of_black_body_radiation? --Lemming 21:57, 21 Jun 2005 (UTC)

You're right. I fixed it. PAR 23:02, 21 Jun 2005 (UTC)

[edit] warm and cool colors

hmm I've been thinking (not expert, just a question :) ) how does colour temperature relate to cold/warm colours? I mean, red is a "warm colour", and blue is a "white colour" but according to colour temperature, it's the opposite. Is there anyone who sees the problem (and can clear things)?

The idea of red as warm and blue as a cool color is psychological, it has to do with the psychological effect of colors, as in "hot-blooded" or "cool-headed". Its not a measure of the physical concept of temperature. This article is concerned with the relationship between the temperature of an object and the visible radiation it emits, that is the physiological, observed color of the object. PAR 11:30, 21 August 2005 (UTC)

Our ideas about hot, warm, cool, and cold temperatures are psycological associations based on the common experiences of man since time immemorial. Red is the color of many things hot: fire, embers, lava, molten steel, even sunburned skin. Something "red-hot" is considered very, very hot. Orange and yellow, other "warm" colors, are also often seen in fire. Blue, on the other hand, is associated with water and ice. It's considered very cold. Green is associated with things like leaves and grass. These things are usually somewhat cool to the touch (they conduct heat away from your skin), but not cold like water and ice. These thermal associations were determined before people understood that, as incandesence goes, red is really very cold. Red-hot was just about as hot as anybody could get anything for a long time. PJA 17:30, 19 December 2005 (UTC)

[edit] Color / Colour

Why used color instead colour? The color is the American spelling. And the colour is the more frequent word then the color (Look at Macmillan English Dictionary, 2002).

Wikipedia does not have a preferred version of English, but it does mandate consistency within articles. (See Wikipedia:Manual of Style. Given that the article is called Color Temperature, color and not colour ought to be used throughout. Nareek 20:04, 7 February 2006 (UTC)

There are not frequent of not frequent word. 'Color' is a american term, colour, an english one. As say Nareek, stay constant; if you use one term, keep it all along your article. -- luxorion

[edit] Sky Colors

I believe that there is an error in this page. There where some example Kelvins are noted, I don't think the sky colors count as Black bodies. The blue color of the sky and the yellow color of the sun is actually due to refraction.

From Luxorion : see my answer at the end of this page. For short, you are completely right.

Mostly because of the Oxygen in the air, if I remember correctly. The oxygen has higher refration at the blue spectrum. That's why indirect, refracted light from the sun, all over the sky is blue and the rest of the direct light (image of the sun itself) lies in the yellow to orange (substraction of blue from a white light source gives yellow). Of course, if it's a very bright day, more of the unrefracted light comes from the direction of the sun and it looks whiter. The sun actually has about the same color continuously (except if you gonna count solar flares and solar activity changes). It's just its position in the sky (lower means more air, more refraction) and the composition of the sky (the wheather, height, .. ) that makes it change color (and consequently the color of the sky).

This can also be confirmed by the spectrum of the sky (the indirect light from the sun), it isn't a temperature color on itself! It's rather a refracted part of the spectrum of the sun. Can someone (confirm &) correct ?

R U Bn @ e-builds 13:24, 10 February 2006 (UTC)


This article explains it, its up to you to copy paste :) http://members.shaw.ca/jimht03/light.html -Anon

While it may be true that the blue colour of the sky is due to a specific phenomenon not related to temperature (namely Rayleigh scattering), the fact remains that light colours can be correlated to black-body temperatures, and so it is still useful to speak of 'colour temperature' in relation to the appearance of the sky. The fact that photographers have produced the results desired by using reference to colour temperature in relation to daylight (and have done so for decades) merely reinforces the utility value of the concept, even though what is taking place here is a conceptual mapping from one physical phenomenon to another. Science and technology is, after all, replete with such useful conceptual mappings! Calilasseia 21:36, 16 March 2006 (UTC)


As you say, the underlying physics is different, but pale blue skies could be a reasonable approximation to a black-body colour, and thus be assigned a Correlated Colour Temperature. Deep blue skies however are simply nowhere near the blackbody locus and cannot be assigned a CCT. 57.66.65.38 09:32, 24 August 2006 (UTC) Andrew Steer http://www.techmind.org/


Thanks, Calilasseia, you are very right. Though I also agree with Andrew Steer and I would add that the clear midday sun is also not very near BB. But maybe we need to reflect this. Proposal: Let's just put between ( ) which resemble BBs and are not really a (appr.) BB source with for example (approc.) and (no BB source). Wayathink ?R U Bn @ e-builds 13:34, 24 August 2006 (UTC)

Clear midday sun is not far off a blackbody for most practical colour-matching purposes. I took a nice spectrum from Surrey in England a few days ago - hopefully I can get that on the web sometime as it's a nice illustration :-) As to your last point, real BBs vs approx... recall "D65" is NOT the same as 6500K; 6500K is a blackbody colour while "D65" is a "standardised" daylight spectrum with colour close to a BB at 6500K. Similarly other "Dxx" colours - specified spectral power distributions defined by the CIE way back when! Fluorescents and discharge lamps are the ones which are nothing like BBs (and cause real headaches for colour-matching) - and it's the spikiness of the spectrum which prevents smooth filters from being able to balence the light. An arbitrary non-blackbody-colour source, as long as the spectrum is smooth, could be perfectly colour-corrected with a smooth-response filter on a camera. 80.189.152.198 22:01, 24 August 2006 (UTC)Andrew Steer http://www.techmind.org/

From Luxorion (ex-weatherman and pro-photographer among others) : I changed the text about black body and color of the sun in the paragraph were the previous author compared the color of the sun and the sky with the one of a BB brought over 6000 K... It is a non sense ! Only the effetive sun color depends on the BB. The color of the sun seen from the earth surface is only affected by refraction of sunlight and the color of the sky by Rayleigh scattering. See the text for comments. The rainbow graph of the sky from 0-15000 K or so should also be removed of moved elsewhere because it not related to the sky color but only to a black body. More explanations in French on LUXORION website How is it possible to let such errors on wiki !

[edit] Xenon, fluorescent, etc

I suppose we can't say xenon arc has color temperature because it is not produced by a back body.

Fluorescent lamps shouldn't not be referred as having color temperature. They do not have. One can say they have light characteristics that look like (?) another black body light source, but that's all. HMartins

Both do have what is known as correlated color temperature. From the work of MacAdam, we can measure just how close two colors are to each other. For any color then, we can find the black body color that is closest to it. The correlated color temperature of that color is then the color temperature of that closest black body color.

[edit] Daylight

The "common examples" of color temperature needs either citation, cleanup, or further explanation. Right now it lists:

  1. 5000 K: Daylight°
  2. 5500 K: average daylight, electronic flash (can vary between manufacturers)
  3. 5770 K: effective sun temperature
  4. 6500 K: Daylight°

... listing Daylight as *two* different colors, "average daylight" (whatever that means) as a third, and "effective sun temperature" as yet a fourth. That's confusing enough to be utterly unhelpful.

CIE defines a series of "daylight" (D-class) illuminants, among which are D50 (5003 K CCT), D55 (5503), D65 (6504) and D75 (7504). The color temperature of the Sun simply equals the temperature of its surface (5778 K), because the Sun itself is true black body. There is no confusion with multiplicity of values, because perceived color of daylight is obviously varying due to atmosphere conditions: from reddish sunrise to bluish evening and silver-white clouds. 217.172.21.161 18:32, 31 July 2007 (UTC)
It doesn't make much sense to talk about the color temp of the sun; only sunlight as received on earth is relevant, unless you want to talk about photography in outer space. Dicklyon 19:00, 31 July 2007 (UTC)
Practically, you're right. But this value helps to understand the enormous influence of the atmosphere, as actual CCT on Earth's surface can be much lower and much higher than original, greatly varying across the day. I think it's good to know the true unbiased value, at least for theoretic meditations. :) 217.172.21.161 06:41, 10 August 2007 (UTC)
OK, but then what is the source for it being a "true black body"? The ref'd source for the temp calls it "Effective temperature: 5778 K" and shows different higher and lower temps at different levels of the sun's atmosphere. Doesn't sound much like a black body to me. Does "effective" here mean the CCT? Or something different? Dicklyon 06:54, 10 August 2007 (UTC)
No, the "effective" 5778 K temperature is real temperature of the solar surface, and no chemical reactions go on this surface, so it is the "true" (purely planckian) color temperature of the Sun itself. But all other "apparent temperatures" are CCTs, because they refer to atmosphere-modified spectrum of black body. As solar surface doesn't heat up or cool down, they just can't be "true" color temperatures.
No, that's wrong on several counts. The temp does change. And a lot of the light is from hot gasses, so has narrowband spectral lines. It's overall not too far from a blackbody, but certainly not near an ideal one. Dicklyon 15:29, 11 September 2007 (UTC)
By the way, while we discussed this, the article almost completely lost its connection to physics, including the mentioned reference to "effective" (true) Sun temperature. This ain't good, I think. 213.234.235.82 15:16, 11 September 2007 (UTC)
Color temperature is more of a color topic than a physics topic. Of course, the connection to an ideal blackbody radiator should be made clear, too. Dicklyon 15:29, 11 September 2007 (UTC)
I have reverted your edit because it introduced a lot of misinformation. The characterization of blackbody in the lead was not accurate, not required. The word "perceived" detracts from the objective definition of what we're talking about. The introduction of a human body as a blackbody is neither correct nor relevant to color temperature. etc. There may have been a few good links or something reverted, too. Please try again, being more careful to get it right. Dicklyon 16:45, 11 September 2007 (UTC)
Please be consistent in your judgements. Are we talking about color perception or pure physics?
  • The main thing to know about black-body radiator in context of color theory, I think, is the fact that emission is caused solely by thermal radiation. The other fact, that is usually considered primary in scholarship, — that no transmission and reflection is present — is not of great concern here, as we are talking of self-luminous objects and discounting any light actually reflected by them. This is exactly why I put human body in the examples: no matter that it can reflect (and even transmit) light, the point is that in a dark room it emits some thermal radiation, just as any black body would. This radiation is only defined by temperature of skin, not the chemical and biological processes under it. If our eyes could register such a weak emission at least in visible wavelength range, we would perceive 310 K as highly saturated red (according to plot of planckian locus on xy cromaticity diagram).
  • Of course, there is no perfect black-body radiator in the real world. So, for the purpose of this article, any object whose emission is near-planckian, can be named to have "true" color temperature. Some small spikes on the spectral plot don't make great deal, as long as much more power is carried by wideband radiation. Yes, maybe I was wrong by having put no disclaimer about this approximation, but the sense change little in this context.
  • The Sun. In the corresponding article I found no reference to some kind of temperature fluctuation. Even if it really goes on, it has no connection with terrestrial atmospherical modification (absorption) of solar light. Temperature of solar surface is really about 6000 K. As I understand it, the emission of light (well, of bigger part of it) is not done by chemical reaction — no gas burns there, almost no oxygen for it — but is defined by thermal state of photosphere.
  • The word "perceived" is inevitable in context of color perception by living organisms, especially by human being with his sophisticated perception mechanisms, especially when we talk about correlated color temperature. Chromaticity is about perception and nothing else, as long as neither objects nor emissions are "red" or "bluish white" or "saturated green". All those color models (CIE RGB, XYZ) are built upon experimental data with human observers trying to match two colors. Other animals have very different chromaticity perception, because their eyes consist of different number of receptors with different spectral sensivity and stimulus adaptation, and their minds (which are the main color recognition units) work completely differently.
I just don't like that this article loses connection with science. I think that the main thing to know about color temperature is its nature, for the people to understand the difference between black-body and artificial illuminants. Whithout this understanding, the term CRI makes no sense, along with some other publishing-related issues. And as to the physics, it is good to understand that the black body abstraction is used because it gives the unique ability to describe emitted spectrum (and apparent color) with a single number — the temperature. It is good to understand that the temperature of, let's say, match flame is not 1700 K, but much lower. That the Moon seems invisible without reflected light, because our eyes don't sense its own "red" ("infra-red") emission.
For the moment, this article is very "shallow", doesn't give deep view of the problem; unfortunately, most of color theory articles in Wikipedia are too. I think, simplifying the content further is a wrong way. —Preceding unsigned comment added by 62.118.220.182 (talk) 09:31, 12 September 2007 (UTC)
Oh, by the way. Early you asked about the meaning of "effective" temperature. There is an article in Wikipedia of the same name: Effective temperature. According to it, 5780 K is of course CCT, not theoretically "true" temperature. But having in mind the aforementioned approximation of "near black-body" objects, and that the actual temperature of surface is almost equal, what's the difference? In context of color topic, there is none, I suppose. The goal is to show that Sun is not like a fluorescent lamp or display screen. 213.234.235.82 13:11, 12 September 2007 (UTC)

Unindenting.

Yep, unindenting has dramatic power. It helps to loose the line of sight. :)

Colorimetry is related to perception, but translated to objective physical measurements and calculations. Perception is variable and subjective; chromaticity and color temperature are not. But it's not really about physics per se, either, just a representation of a 3D subspace of spectrum space.

You mess different types of "perception". General everyday perception under arbitrary conditions is of course variable (but not too "subjective", as it's widely considered). Perception within fixed "colorimetric conditions" is very uniform between individuals, and that makes possible the creation of colorimetic models such as RGB and XYZ. Then measurements are formalized and formulae are strict, but they still reflect human vision model under certain conditions. Thus, correlation of apparent color to a reference (black-body or wharever) stimulus is also a colorimetic issue, not just abstract mathematics.

I think it's great that you're interested in working on improving the article. But improving it with information from reliable sources will be a better contribution; sourced material is usually correct, and usually lasts better, than stuff made up from what's in you head.

As to the "purely thermal" explanation of blackbodies, I think you made that up. It's not enough. Emissions from a low-pressure gas will be in discrete lines, nothing like a blackbody spectrum. That's why the sun has so many well known solar spectrum lines (actually the Fraunhofer lines are mostly absorption lines from cooler gases higher up in the solar atmosphere). Also see Solar variation.

I wasn't aware of the atronomical term effective temperature, but I see it's about stellar luminosity, not color. I don't think it has any place in this article, but if you find a source that connects it, then please do cite it.

As for importance of spectrum to Color-rendering index, you are correct. But that's not really very relevant to this article. Dicklyon 14:52, 12 September 2007 (UTC)

Unfortunately (or fortunately?) it is. Otherwise, what is this all about? Be sincere, do you really need 10'000 gradations between "match flame" and "cloudly sky" for practical applications, such as "visual arts, filmmaking and stagecraft"? Differences less than 500 K are virtually undistinguishable, not only by eye, but even by a camera. So a maximum of 20 steps is required, and for general usage some 5–10 would be enough. Why just don't call it "bluish" and "reddish" or "warm" and "cool"?
Talking about color temperature (and CCT) without mentioning CRI and underlying metamerism with perfect radiator is a dead end. The problem of "CCT vs CT" is not just that small difference in xy coordinates. Someone who wants "fixed and well-known white point" will not succeed with simply achieving cherished 5000 K reported by XYZ-colorimeter. Readers should know that applying a "spiky" light of generic daylight lamp won't render them correct colors. They should clearly understand, why they need to approach as closely as possible to some smooth, uniform-like spectum of black body or its real approximations. Without it, such things as "D50 test" patch pairs (which demonstrate metamerism only under D50 and look different with any other illuminant) simply make no sense.
Do you catch my point? For the moment, this article has little practical value, neither it is scientific enough. 62.118.220.182 08:56, 14 September 2007 (UTC)
So fine, add some mention of metamerism, and a discussion of why "spiky" light sources won't render correct colors. That seems to me like a perfectly relevant thing to add to this article. I doubt Dicklyon would be opposed to including such background information, as long as it stays concise and links out to the main metamerism article. --jacobolus (t) 05:42, 15 October 2007 (UTC)
Agreed. A brief section titled illuminant metamerism, with a main link to metamerism (color), to point out that color temperature isn't all you need to know about a light source if you care about how things are going to look, would be a good idea. As to 500K being fine enough, that's not true at the red end of things; 2000 vs. 2500 are hugely different. That's why reciprocal color temperature is sometimes used; see mired, where numbers 0 to 500 or so are plenty (somewhat finer than you need for good white balance correction, but that's OK). Dicklyon 06:41, 15 October 2007 (UTC)

[edit] Outstanding Article

I just want to say this is an outstanding article. I've been trying to find the right color temperature to request LED lighting, and wanted to make sure I understand color temperature. It would be interesting, if someone has the information, to describe the differences in LED lighting (relating to color temperature, or CCT) as has been done with fluorescent. 206.124.31.24 17:39, 31 January 2007 (UTC)

[edit] Class differences

Can someone explaine the difference in ratings between wikiprojects? -JWGreen 04:00, 15 March 2007 (UTC)

Yes, this article has a long way to go, but the filmmaking wikiproject apparently has low(er) standards. Though it was my impression that "good article" meant an article had actually been through WP:GAC. It's possible that the "good article" assessment from the filmmaking wikiproject is simply a mistake. --jacobolus (t) 05:13, 28 April 2007 (UTC)

[edit] Incandescent

I have a question about the following statement made in this article.

"The power of a lightbulb (20 or 100 W) seems to change its color but in reality it only affects its luminosity (luminance) to which our eyes are very sensitive."

Is there a reference for this? This doesn't seem to agree with a statement in the Incandescent Light Bulb entry (see quote below), which infers that color temperature of tungsten lighting is variable.

"The current heats the filament to an extremely high temperature (typically 2000K to 3300K depending on the filament type, shape, size, and amount of current passed through). Heated atoms within the filament intensely vibrate. The electrons, which are charged particles now strongly oscilating, radiate excess energy in the form of black body radiation, according to Maxwell's equations."

I'm not saying that a 20W bulb and a 100W bulb couldn't have the same color temperature, it just means that they aren't necessarily the same since there is a lot of variability in filament temperatures for tungsten bulbs.

Furthermore, I have seen many photography websites which list color temperatures for tungsten lighting as a range, rather than a single value.

On a different (but related) subject, I would infer from the Incandescent Light Bulb entry that a dimmer switch will vary not only the luminosity of the bulb, but also the color temperature. This might be worth mentioning in this article. —The preceding unsigned comment was added by 202.33.240.34 (talk) 02:00, 14 May 2007 (UTC).

Since incandescent lamp is near a black-body radiator, its (correlated) color termerature must be almost equal to the actual temperature of filament. As almost all drawed power is spent on heating, the rated nominal (20W or 100W) defines the color temperature too. Of course, some considerations and approximations apply here, but roughly this is it: if the lamp manages to heat the filament to 3300 K, it will look chromatically different thant 2000 K. Of couse, luminance level is also essential in saturation and even hue perception, but the xy chromaticity coordinates are really not the same. 62.118.220.182 11:06, 14 September 2007 (UTC)
As a matter of convenience in filament sizing, the higher-wattage standard bulbs tend to run a little hotter than the lower-wattage bulbs. It doesn't have to be that way, and it doesn't apply to all classes of bulbs, but within standard indoor bulbs there is a moderate increase in color temperature as you go up in wattage. Search up some sources and fix it... Dicklyon 15:01, 14 September 2007 (UTC)

[edit] full spectral power plot

Please get a real, full spectral power plot for incandescent lamp - don't cut it off at 700 nm!-69.87.203.221 01:44, 26 May 2007 (UTC)

Why? Only the visible part of the spectrum affects color temperature. Do you have a plot you'd like to contribute? Dicklyon 01:56, 26 May 2007 (UTC)

[edit] leds?

If anyone knows about LED lights and their color temperature, that is what i wondered next when reading this article. good job y'all; thx.

Colored LEDs are nearly monochromatic, so talking about a color temperature makes no sense. But white LEDs do have a correlated color temperature, usually in the range of 5000 K to 8000 K. Dicklyon 03:00, 6 August 2007 (UTC)
I know that even white LEDs are not full range, and even with a pure green, red, and blue LED colormix you cannot create every color. I don't know if that is relevant to color tempurature, but thats what I know from my theatre lighting experience. -JWGreen 03:14, 6 August 2007 (UTC)
No, it's not relevant to color temperature. The links I cited above, by the way, are for phosphor-type LEDs, I think, not RGB LEDs. Dicklyon 03:16, 6 August 2007 (UTC)