Color Temperature Crash Course: My attempt to explain.

[Ryan Patrick O'Hara]

Hey everyone! I just finished writing another article on Cinematography, and I wanted to share it here. Anyone who dares read the entire thing, I would be very grateful to. (Do you remember my camera reports article.... phew)

I literally just finished writing the damn thing and am about to hibernate for the next 12 hours and not think about color temperature!

When I get back up I'll do some light editing and spice up the language. Otherwise, please tell me if I forgot anything important, am wrong about something, and etc. I count on folks like you guys to proof my writings.

I hope this is helpful to someone here, and I hope someone can help me improve this article. It fizzes out at the end, but I think that is just because I got tired of writing.

Read my article on Color Temperature here:
http://ryanpatrickohara.blogspot.com/2009/01/color-temperature-basics.html

Enjoy!!!

 

[j1clark@ucsd.edu]

Hey everyone! I just finished writing another article on Cinematography, and I wanted to share it here. Anyone who dares read the entire thing, I would be very grateful to. (Do you remember my camera reports article.... phew)
Enjoy!!!

I just quickly read through it. So here's one typo...

... digital camera sensors will me manufactured ... (should be ... will be... )

And on the topic of Kelvin...

----------------------
Color temperature is measured in increment units referred to as degrees Kelvin. Kelvin is the scale of which a black body source is heated to the point it emits light. From the moment it begins to emit light, it begins on the bottom of the Kelvin scale. As the temperature of the black body increases, it will emit different color wavelengths of light until reaching the top of the Kelvin scale. By referring to a lights color temperature in a specific number of degrees Kelvin, we have a system to which we can always reference the exact same color.
-----------------------

Perhaps a little review on temperature measurement... but the Kelvin Scale measures temperature from Absolute Zero, which is currently estimated to be -273.15 Celseus.

When a 'black body' is heated, at approximately 1500 K, the black body begins to emit human visible radiation (lower than this point the BB radiates InfraRed radation... etc...). (I know some people are horrified to thing of 'radiation'... but anyway...).

At about 20,000 K the 'black body' radiation will pass into UltraViolet, and be invisible to the human eye.

Film material is more sensitive to blue and upwards to UV than the human eye, and digital sensors are more sensitive to IR... etc....

----------------------

Color Temperature is the industry standard term, so there's no 'real' problem there... its just confusing because we think of 'higher temperature', means 'warm/heat', when for example 'shade' has a higher color temperature than being in the broad sun light...

Since I've not needed to get 'mired' into correcting as yet, I don't have to much to say on that portion of your write up. However, I think perhaps a web based calculator would be beneficial... if one exists... as well as seeing if someone has a more interactive tutorial that shows the effects of filtration on a scene... I don't know if it has to be color wise absolutely correct, but at least give the suggestion of how things would be 'seen' by the film or digital sensor...

 

[Ryan Patrick O'Hara]

Thanks!

Yes you are right about the kelvin scale. It does begin and end around visible light, due to infrared/UV light.

I'll have to reword that! Thanks!

I'll see what I can do about mired tutorial or a calculator.

Thanks! :beer:

 

[Ryan Patrick O'Hara]

Is this better? I guess I didn't pay too much attention to the non-visible part of the kelvin scale because it has no meaning for film and television. In my haste I did completely mess up the true description of how kelvin works!

Color temperature is measured in increment units referred to as degrees Kelvin. Kelvin is the scale of which a black body source is heated. From the moment it begins to heat, the black body will produce light radiation starting with infrared. Somewhere around 1500k the radiation enters the human eye's visual spectrum. As the black body continues rise in temperature, the color of emitting light radiation increases. Because the color of light is dependent on the temperature of the black body, this scale can be used to make references to a lights exact color.... it's color temperature. Eventually, the radiation will surpass the human eye's visual spectrum and will enter the realm of Ultra Violet radiation. By referring to a lights color temperature in a specific number of degrees Kelvin, we have a system to which we can always reference the exact same color.

haha, btw, I couldn't find where the light becomes visible so I took your number of 1500k! I hope you are right, teach'.

Btw, it is my understanding that it is light radiation being radiated, correct?

 

[j1clark@ucsd.edu]

Is this better? I guess I didn't pay too much attention to the non-visible part of the kelvin scale because it has no meaning for film and television. In my haste I did completely mess up the true description of how kelvin works!

haha, btw, I couldn't find where the light becomes visible so I took your number of 1500k! I hope you are right, teach'.

Btw, it is my understanding that it is light radiation being radiated, correct?

Here's the Wiki article on Black Body radiation...

http://en.wikipedia.org/wiki/Black_body

Probably more than you'd ever want to know... But if you look at the pictures... the lava flow is listed at about 1000 K... probably not good for reading, and there is mention in other articles of a 'candle flame' being at about 1700-1800 K... unless of course, you're thinking about a remake of some volcanic film... then perhaps the color temperature of hot lava may be useful...

Another sort of 'bizarre' twist is that Black Body radiation only holds for light producers which have a nearlly continuous spectrum. So, such light sources as florescent lights, which have spikes or holes depending on the particular type of light are not 'modelled' by BB radiation... doesn't mean they don't have color... in fact, part of the problem is the spikes and holes of their visible spectra...

However, like most things, the use of Black Body radiation, and a simple number like K, use more beneficial to most, to quickly select filters or camera settings...

If you have Java (enabled) on your machine, here's a widget that may help. Put the physics aside, and just select the K temperature. The widget attempts to give an indication of the 'color' that would be created at that degree K. So for 15000 K the result is definitely 'bluish', at 2500 K, definitely orangeish... etc.

http://webphysics.davidson.edu/Applets/java11_Archive.html

 

[Ryan Patrick O'Hara]

Fluorescents dont matter anyhow. They do not emit light by 'black body like' radiating means, therefore have correlated color temperatures instead. A color meter can offer the appropriate amount of color compensation and color temperature information.

I'll try to get that link working but it seems to be errored for my computer.


Thanks again.

 

[Postmaster]

Nice read.

But in a time where almost every single picture is blown thru postproduction/colorgrading it does not mean that much anymore.

Look, I use only the daylight preset on my HVX. Why?
Cause setting it to tungsten means cranking up the db in the blue sensor, that leads to noise in the blue chanel.

But I can set the mood I want easily in post without getting noise.

Temperature may be intresting if you shoot on film, but for video, it´s obsolete in my book when you go to the post anyway.

Lot´s a guys try to skip post and try to get the look in the camera.
I think that´s a mistake. The sensors need to be saturated as much as possible and one needs to shoot as flat as possible to employ the whole dynamic range the camera is capable of.

That usualy looks like crap. but it gives you sooooo much room in post to get any look you want without sacrifying sharpness, cleanness and overall quality.

Frank

 

[j1clark@ucsd.edu]

Temperature may be intresting if you shoot on film, but for video, it´s obsolete in my book when you go to the post anyway.
Frank

Depends of course on the situation. It is true that because of the limitations on mid/low end video cameras 'balancing' causes some problems. But perhaps as resolution increases in both the spatial and intensity domains, white balancing may become more significant.

An example in the case of still digital, last The Wife got permission to take still digital images of Gustav Klimts 'Der Kuß'. The Belvedere had 'recreated' some set of rooms that approximately matched the dimensions of the Kunstschau 1908, when several of Klimts paintings were first premiered. In 1908, none of the buildings were lit with artificial light... speed forward 100 years... the lighting was only artificial... and tungsten artificial...

Selecting the tungsten balance on the camera eliminated the yellow cast. As it was we opted for shooting 'raw' and will be correcting in post. The point being, at the moment, there is no 'raw' mode available to the typical digitial video user... Unless of course everyone here but me has a Red, or Blue, or whatever high end camera...

 

[Postmaster]

Don´t know if that counts for still cameras too - probably not, but take a HVX and shoot under tungsten with daylight setting and the with tungsten setting - if you bored try a manual balance with a card too.

Than look at your blue chanels and you know what I mean.

Frankl

 

[Ryan Patrick O'Hara]

Nice read.

But in a time where almost every single picture is blown thru postproduction/colorgrading it does not mean that much anymore.

Look, I use only the daylight preset on my HVX. Why?
Cause setting it to tungsten means cranking up the db in the blue sensor, that leads to noise in the blue chanel.

But I can set the mood I want easily in post without getting noise.

Temperature may be intresting if you shoot on film, but for video, it´s obsolete in my book when you go to the post anyway.

Lot´s a guys try to skip post and try to get the look in the camera.
I think that´s a mistake. The sensors need to be saturated as much as possible and one needs to shoot as flat as possible to employ the whole dynamic range the camera is capable of.

That usualy looks like crap. but it gives you sooooo much room in post to get any look you want without sacrifying sharpness, cleanness and overall quality.

Frank

An opinion worthy of a 'Postmaster'. Haha.

Do you get perfectly balanced images by shooting the wrong color temp and correcting in post? Is that what you are saying. Because I was under the impression that is bad news bears. I've always thought proper color temperature is best attained in the sensor before it is compressed/written into one flat video file.

Nice read.

But in a time where almost every single picture is blown thru postproduction/colorgrading it does not mean that much anymore.

This may be true for small shifts of the entire picture, but matching the lighting units to remove color temp difference is still very important... unless you like incredible hard roto work in color. Especially when the off lighting starts to blend, etc.

I can partly see doing post work on what traditionally has been lens filtration for stylized effect. But even then, there is something different by actually absorbing the actual light waves versus shifting them in post.

 

[Postmaster]

Well it depends much of the look you want, shooting style and workflow.

I record everything uncompressed (when ever possible) direct into the computer wia component. So I skip the compression part in the first place - wich is a good thing for grading.

I think it´s safe to say, that we are d`acord that raising db leveels cause noise - especialy on the HVX 200 - right?

Colorbalancing from daylight to tungsten means raising db level on the blue chip - thus rising noise in the blue cheanel.

So I prefer, having less noise in a unbalanced picture than vice versa.
Getting rid of the noise (or live with it) is worse than color grading - wich I do anyway - at least in my book.

But to each his own.

Here is an example of what I mean.
Shoot flat, unbalanced, saturate the sensors and colorgrade later.

Frank

 

[AccurateFilms]

Well it depends much of the look you want, shooting style and workflow.

I record everything uncompressed (when ever possible) direct into the computer wia component. So I skip the compression part in the first place - wich is a good thing for grading.

I think it´s safe to say, that we are d`acord that raising db leveels cause noise - especialy on the HVX 200 - right?

Colorbalancing from daylight to tungsten means raising db level on the blue chip - thus rising noise in the blue cheanel.

So I prefer, having less noise in a unbalanced picture than vice versa.
Getting rid of the noise (or live with it) is worse than color grading - wich I do anyway - at least in my book.

But to each his own.

Here is an example of what I mean.
Shoot flat, unbalanced, saturate the sensors and colorgrade later.

Frank

Why doesn't everyone shoot with auto whte balance (which has fairly neutral colours) and do a little colour grading in post? Also, don't you lose some clarity when capturing in component? I mean, sensors are analog, processors are digital, component connections are analog, and your computer is digital. Why don't you capture with Firewire?

 

[j1clark@ucsd.edu]

Why doesn't everyone shoot with auto whte balance (which has fairly neutral colours) and do a little colour grading in post? Also, don't you lose some clarity when capturing in component? I mean, sensors are analog, processors are digital, component connections are analog, and your computer is digital. Why don't you capture with Firewire?

Was that a rhetorical question... 'autobalance' is about the worst you can do... that is, each shot may be 'different' in color, and somewhat randomly so, since what is used for the 'autobalance' reference is anyone's guess, especially the engineers who designed the 'autobalance' mechanism.

As for the analog signal... well every method of recording adds distortion... by digitizing, and for those who can afford it, digitizing with as little or no compression at all, will at least fix the 'distortion' introduced by the digitization process.

And of course, not to mention that to keep the process all analog would put 'video' back to 2 in. mag tape and a couple million dollars worth of editing equipment... either that, or crappy Beta and U-Matics...

 

[Ryan Patrick O'Hara]

Autobalance? Why don't you just turn on autofocus, auto-iris, and run around with your eyes shut. You can fix it in post.

PS: I think you are on to something. Those Rosco, Lee, GAM, and Formatt salesmen must be some god damn good ones, to make us filmmakers think (for over eight decades) that color temperature is worth controlling or thinking about. Those guys could sell sand to the Saudi's!

That concludes my weekly release of pent up mis guided aggression. Goodnight! I'll finish it off with a passive aggressive smiley.

 

[pwyll]

Auto functions work very well in modern DSLRs where the auto settings are locked down milliseconds before the image is recorded.

You need to do the same - lock down your settings before you start recording. You can't have the camera deciding to change exposure /wb /focus as the talent moves back and forth in front of a window, or somesuch occurence.

Shooting run and gun most pros use the presets. The idea of leaving the cam on 56K might work ok in a highly planned shoot with external monitors etc but it is not recommended if you rely on a color evf/lcd - you will miss all sorts of color opportunities/problems.

 

[DSWMEDIA]

Color temperature has been made a little less of an issue with RAW shooting. You can just adjust your white balance in post. Of course, you still need to provide some sort of prevailing, matching temperature on set so all light can appear nuetral when you balance. THEN you can take the image red, blue, or anyway you want.

 

[thome]

That article is really good..and personally i feel you need to get the right color when shooting no matter what your gonna do in post.

 

[DSWMEDIA]

You have to add some gain to the red channel when daylight balancing...and not every camera's sensor is biased towards dayight. That being said, if youre shooting on a camera like the RED which is clearly biased towards daylight (the published color balance of the sensor ias 5000k) then you should filter optically and accept the subsequent F stop loss.

 

[PaPa]

So are you guys saying that generally it is better to shoot under daylight preset? If so i have noticed in my footage when im balanced for daylight and correct in post that i have less noise than if i go tungsten and adjust in post. Is this why?

 

[Ryan Patrick O'Hara]

It's always best to get the color temp you want on set, but the reason there might be more noise is because many sensors are manufactured with a daylight balance bias.