Color Temperature Crash Course: Revised Information Part One

[Ryan Patrick O'Hara]

Just wanted to make a small update on an article I wrote regarding color temperature. I posted it here in the past and have recently added the following information which I thought was interesting and-or had forgotten, and you all might enjoy. The link to the whole article is HERE, but the additions are as written below:


When discussing that natural sunlight varies in color temperature I added this small paragraph:

The Exception

There is however, one natural light source which is always a constant color temperature, no matter the weather conditions or time of day... sunlight in outerspace! If you ever find yourself shooting outside Earth's atmosphere, you should know that our solar system's star (the sun) produces a constant source of closely rated 6500k light. Of course this will change in due time, but do not rush for your space shuttle ticket and color meter just yet, as this will not be for millions or even billions of years from now. There is time.

If wondering why sunlight in space is around 6500k while sunlight on Earth varies around 5600k, it's the exact same reason why a sunset/sunrise is red-orange and why the sky is blue: the atmosphere. To state the process simply, the atmosphere is filled with many different particles. When the sunlight passes through the atmosphere, some wavelengths of light (waves of electromagnetic energy) pass straight through with little disruption while other wavelengths are absorbed by and scattered by the particles within the atmosphere. It is the shorter wavelengths of light (violet and blue) which become absorbed and scattered, creating the atmosphere to appear blue. The absence of the blue wavelength within sunlight, creates the illusion of a yellow like sun. During a sunset or sunrise, the sunlight must pass through a greater amount of atmosphere and hence more of the wavelengths of blue and violet are absorbed or scattered creating very red-orange hues to admire. If one was to look at the sun in space, it would appear almost pure white, as there is no atmosphere to absorb or alter the wavelengths.

And when discussing that no matter how accurate a manufacturer makes it's bulbs, artificial man made lighting inevitably changes color temperature for many reasons. Among the listed reasons I clarified my previous statements about dimming. As most of you know, dimming lowers color temperature. That is what I said, but not as brief. What I forgot to do, was clarify that the statement only pertained to incandescent lighting or that which follows the carbon black body radiating system. When discussing other forms of lighting processes such as HMI or fluorescent, this can change.

The following is my newly written paragraph about dimming:

Never Trust a Light

The color temperature outdoors can vary due to differences in time of day, time of year, geography, geology, and weather conditions. Thankfully man made bulbs and especially motion picture industry globes, are designed and manufactured to operate at a constant and reliable color temperature. Even with these efforts there are many ways for a bulb to stray from it's rated color temperature. A cinematographer must check his globes and be aware of these factors.

Dimmers: Unlike the other methods of reducing a lamps intensity, dimming is simply restraining the amount of electricity from the light emitting process. In incandescent/tungsten lighting, reducing the electric current from the normal recommended amount will reduce its' color temperature. As the filament's electric current drops, so will it's color temperature. This does not apply to lighting instruments which are not radiating (black body like) sources. Non- radiating lighting units behave differently. For instance, some HMI ballasts have a built in dimmer which can be used to change the light's intensity. Unlike incandescent lighting, when a HMI's electrical current is lowered by a dimmer, the color temperature rises. PS: Most non-incandescent lighting (including HMIs and flourescents) cannot have their intensity reduced by way of altering the electrical current unless the ballast is manufactured with a built in dimmer. Do not try to intercept a non-incandescent unit with a hand dimmer or likewise external dimmer.

 

[CovenantPictures]

The funny thing is, even a white balance is relative within reason. It's always been an ironic cunundrum to me. You may see something as 'white' that the camera sees as having a yellow/orange cast, but when the audience views your image, their eyes can adjust to the cast, making your image appear to have nuetral colors! So what exactly IS white light? You only know the temperature of a light source from it's opposite.

 

[CovenantPictures]

I suppose white light looks most nuetral when their is a balanced spectral output, as in equal blue/red intensities

 

[pwyll]

no - the eye does not remove color casts - that is a myth. The brain compensates for them in very subtle ways. But the light at sunset still looks red. Light in the high mountains looks blue. Cheap fluorescents can give things a green tint.

 

[Ryan Patrick O'Hara]

Yes our eyes do compensate for color temperature. After long periods of time within an environment, we adapt to see a more correct color rendition.

With that said, the reason sunlight in space should appear white, and not yellow, is because of no atmospheric particles to distort the light.

Sure light may look white on earth when our eyes nearly adapt to it, but then look up. The sun looks yellow, because it has lost it's blue wavelengths within the atmosphere. In space the sunlight will look white because our eyes will adapt to its 6500k temperature, but the difference is, we do not have to look at it through an atmosphere and therefore is not losing part of the blue wavelengths of light, nor has to be compared to that scattered blue color within the atmosphere, which only compounds the effect of looking 'yellow'.

According to the Harvard Spectral classifications, our star is 'G' category and falls within 5,200-6000kelvin... a yellow star... But the chart is not describing the color temperature of the light emitting from the star but rather the blackbody carbon equivelent temperature of the burning surface... because the visible light color temperature is more around 6500kelvin which places it at the surface temperature of a class F star which is between 6000kelvin-7500kelvin; stars described as yellow-white.

The simple fact is, looking at the sun in space is much closer to what the human eye perceives as white, then looking at the sun on earth. Combined with the fact that there lacks atmospheric scattering and consequential comparison to the blue wavelength within that atmosphere helps our eyes adapt and not see any color shift.

 

[pwyll]

I wasn't referring to your article Patrick, rather the comment from CP. I don't shoot in outer space.

 

[Ryan Patrick O'Hara]

no - the eye does not remove color casts - that is a myth. The brain compensates for them in very subtle ways. But the light at sunset still looks red. Light in the high mountains looks blue. Cheap fluorescents can give things a green tint.

Our eyes are very precise and finicky wonders. We are able to adapt and almost correctly see in many different situations, including some drastic changes.

Pwyll is correct with his observations of atmospheric distortion and color spikes, but I have a feeling that if under said lighting conditions for a long time, one might be surprised how much things change to 'normal'.

I would venture to say that if the human eye is given time and the lack of opprotunity to compare, it will adapt to very very off color lighting more surprisingly then many think.

 

[Ryan Patrick O'Hara]

I wasn't referring to your article Patrick, rather the comment from CP. I don't shoot in outer space.

Haha, yes, I figured so. My response after yours was actually started before you submitted. Takes me awhile to proof read. Then I responded again!

Still, I think you are very right, yet would be surprised at the eye's ability to compensate. I'd like to find a research study about just how far we can adapt to seeing "correctly" in very off color situations.

That being said, I am not referring to mixed color temperatures. We, as humans, can compensate that for a nice moderate mix (unlike video) but we are still seeing a difference. This is what I mean by the lack of ability to compare. If the eye is just under one type of lighting it is amazing what we adapt to. Mixed lighting will always give the eye, no matter how it tries to compensate the two, the ability to see color flaws.

 

[CovenantPictures]

I wasn't referring to your article Patrick, rather the comment from CP. I don't shoot in outer space.

no pwyll, I think you misunderstood me. By 'eye' I meant the entire human visual system. The brain DOES adjust for color temperatures within reason. I'll shoot a scene with Mole's. all 3200k, and The light will look white.

But then I'll switch to Kino's and HMIs, and yes, I notice the lights as having an immidiate blue cast. BUT, within a matter of minutes, the lights appear white, just as the tungsten fixtures did just minutes ago. It is certainly not a myth. Blain Brown discusses it in his book Cinematography.

he also mentions that the audience can adjust to color casts in film/video, within reason.

 

[CovenantPictures]

I'd like to find a research study about just how far we can adapt to seeing "correctly" in very off color situations.

I read somewhere that the eye can see 'white' light in situation ranging from 3200k to around 6500k. I'll try to find the article. But no, I am not stating myths.

 

[CovenantPictures]

but I have a feeling that if under said lighting conditions for a long time, one might be surprised how much things change to 'normal'.

I just did a very non scientific test. I paused a scene from the movie Lost World Jurassic Park (it's on my Tivo) that contained both blue light and orange light (probably 3200 k and 7000k under daylight balance I'm estimating). I used my thumb to cover the section of the frame containing blue light, and I stared at the 'orange' light for a minute. I was suprised by how white, and nuetral, it started to appear. Then I removed my thumb and the blue light looked extremely blue! So yes, it's all relative in the end, within reason. The eye knows things by their opposite.

 

[CovenantPictures]

no - the eye does not remove color casts - that is a myth. The brain compensates for them in very subtle ways. But the light at sunset still looks red. Light in the high mountains looks blue. Cheap fluorescents can give things a green tint.

If the eye does not remove color casts whatsoever (which is false), than what is the color balance of the eye? 3200k? 5600k? There is none.

 

[j1clark@ucsd.edu]

If the eye does not remove color casts whatsoever (which is false), than what is the color balance of the eye? 3200k? 5600k? There is none.

I don't think 'the eye', that is the lens, cones, and rods, etc. 'remove color cast'. What happens is that neurons associated with color processing, along with some forms of 'pattern' recognition involving some other group of neurons, form a feedback loop, and the eye sensor data, is 'adjusted' to compensate for the illumination's color cast.

In essence it is a reflex response, and as such, does not involve human 'conscious' processing.

 

[CovenantPictures]

I don't think 'the eye', that is the lens, cones, and rods, etc. 'remove color cast'. What happens is that neurons associated with color processing, along with some forms of 'pattern' recognition involving some other group of neurons, form a feedback loop, and the eye sensor data, is 'adjusted' to compensate for the illumination's color cast.

In essence it is a reflex response, and as such, does not involve human 'conscious' processing.

Exactly, forgive my miscommunication, I meant the nuerons in the brain most literally, not the eye itself. But yes, your brain does have a way of removing temperature casts within reason.

My question is, how do we qualify "white" light.. is it light containing a equal spectral distribution of the visual light spectrum? i.e. equal ratio across the red-blue axis? Does that light look most noticeably 'white' to the eye at first? It's all very interesting.:smile:
Again, sorry if I caused a sense of confusion or misinterpretation.

 

[j1clark@ucsd.edu]

Exactly, forgive my miscommunication, I meant the nuerons in the brain most literally, not the eye itself. But yes, your brain does have a way of removing temperature casts within reason.

My question is, how do we qualify "white" light.. is it light containing a equal spectral distribution of the visual light spectrum? i.e. equal ratio across the red-blue axis? Does that light look most noticeably 'white' to the eye at first? It's all very interesting.:smile:
Again, sorry if I caused a sense of confusion or misinterpretation.

Well... naming 'color' is a big problem... So it is difficult to test for such questions as 'is this blue, green, whatever...'.

However, in some cases, such as 'white', one can define that as some material which reflects the three color bands, R, G, and B, equally, and which given various experiments and tests, humans usually preceive, and name as 'white'...

Of course in addition to any other point of confusion... there are people who do not process R, G, B, in various combinations, 'equally', and so, one has color blindness... that is the image processing function maps two 'dissimilar' colors to the 'same' color as preceived by the observer...

Based on the last 150 years of experiments in human color preception, I think one can state that given the RGB response of the human vision system, that 'equal' amounts of R, G, B, are preceived as color neutral, and at the near 'brightest' value is named 'white', and ad the darkest, dark grey...

 

[pwyll]

Yes - as far as we know the human visual system is trichromatic (RGB) much like a digital camera. But our visual system processes luminance before chrominance - in a totally different part of the brain. More importantly the human brain can apply "gain" and "color correction" on a "pixel by pixel" basis, where you can assume an infinite number of pixels.

So if I put on some red tinted wrap around sunglasses the snow will still look white. But if I hold the red lens six inches in front of my eyes then the snow I see through the lens looks red. The eye does indeed "correct" for the lens when the entire frame is (lightly) tinted, but not when the tint is only applied to a portion of the field of view. The correction applied by the human visual system is not applied uniformly across the chroma spectrum. It is not a single correction but rather a series/constellation of thousands of calculations and adjustments.

Performing a custom white balance will not reliably allow the camera to reproduce what the human eye sees. That is a Myth. Yes it is perpetuated by almost every video text book ever written - so I do understand your reluctance to accept the more accurate paradigm - that's why I call it a Myth rather than a simple falsehood.

Next time you are sitting beside the Indian ocean enjoying a tropical sunset take a few seconds to glance over a white piece of paper - it will look white, even though the entire scene glows orange/red. Pull out your camera if you must - do your custom white balance and then shoot a few seconds of footage. When you review the footage later you will wonder why you bothered - the magic of the golden hour is gone.

Alll the best

 

[j1clark@ucsd.edu]

Performing a custom white balance will not reliably allow the camera to reproduce what the human eye sees. That is a Myth.

I don't know if I'd agree with the use of the word 'myth' so much as a misstatement...

The 'white' balance, given an appropriate target which has equal values of RGB, will eliminate any(*) shift in the ambient light from 'equal values' as received by the camera's sensor.

Since this processing for many low end or prosumer cameras is 'before' compression, either bit depth compression, or spatial compression, (at least I believe that to be true....), one will get the best 'adjustment' for removing that color shift from the resulting footage.

If we had cheap cameras that shot all 'raw' frames, then, we could adjust for a color shift in post production, perhaps even more accurately than on set.

The idea of obtaining a 'neutral' or 'standard' color balance, is so that later processing can occur from a known state. Or, that footage shot at different times, has the same 'color' balance, at least initially.

Back in the good old days of film, this was done with a combination of filters on the camera during shooting, or 'timing', that is by changing the film chemical processing procedure, while developing the negative (or positive... as the case may be...). Also back in the good old days, some types of processing were pretty much 'do it and hope it comes out ok'... others had some amount of technical repeatability, such as preflashing negative stock, or so called 'bleach by pass', etc.

Fortunately in these modern times, given footage that starts from some 'standard' place, one can add digital processing with far greater control than in those olden times...

I'm using 'standard' to really mean something on the order of having a target that has been calibrated to have a given amount of RGB, and that the footage has been captured in such a way as to as accurately represent that calibration standard when the RGB components are analyzed. No appeal to 'what a human eye would do'... other than the sensor is capturing imagery in the 'visible' range...

 

[CovenantPictures]

yes, I agree, white balancing is a feature designed to 'restore' a scene to it's equal RGB values; your eye sees nuetral white light within a certain range. It just gets interesting, when in the end, it's all relative. I can view a film scene and look around the frame. Soon enough, my visual system will adjust to it's color temperature and i'll see white light.

My question is, at what point does the human visual system NOT remove the color cast? How can you see tungsten light as white, yet a sunset at red/orange? Does it have something to do with mired values?

 

[Ryan Patrick O'Hara]

yes, I agree, white balancing is a feature designed to 'restore' a scene to it's equal RGB values; your eye sees nuetral white light within a certain range. It just gets interesting, when in the end, it's all relative. I can view a film scene and look around the frame. Soon enough, my visual system will adjust to it's color temperature and i'll see white light.

My question is, at what point does the human visual system NOT remove the color cast? How can you see tungsten light as white, yet a sunset at red/orange? Does it have something to do with mired values?

No. Mired values are simply a system of quantifying color values in a non linear fashion, so that one can identify a difference between color temperature no matter where on the scale.

My assumption on why sunsets and other off balance lighting stays off balance to our eyes, is because

1) It's a very strong color cast. Possibly bordering our eyes ability to conform.
2) The color change could happen rather quickly allowing us the initial opportunity to see crazy off color balance lighting.
3) The off color balance lighting is still happening in the balanced color environment, otherwise, we can compare the two color temperatures and easily recognize the new temperature as being very different.

Depending on the event, these will vary.

Talking about a sunset in particular I would venture to say it is mainly because the atmospheric dust makes the sunlight so low on the kelvin scale, our eyes cannot easily and quickly adjust to such a change.... AND especially because the rest of the sky is still relatively normal in scope, meaning our eyes won't immediately try to adjust to the sunset and we can see the comparative difference between the sky and sunset, which makes the extra low kelvin colors very obvious.