Colour temperature basics

[David Owen James]

I'm trying to wrap my head around how colour temperatures work with film. I will be shooting some footage with the new Wittner Chrome 200D colour reversal, which is balanced for 5500 K. If I understand correctly, this is daylight temperature, and if shot on a sunny day the colours would look natural. I'd like to know how to manage colour temperatures to get the best image possible. Can anyone suggest a book or website which gives some practical explanations? Any tips or advice would also be great.

 

[David Mullen ASC]

Daylight varies quite a bit in color temperature (hence why many digital still cameras have higher color temp settings to take out the extra blue in shade or overcast.) 5600K or 5500K is considered "photographic daylight" in that many film stocks are balanced for that (in theory, this will give you the correct rendition of colors at Noon in full sunlight, etc.). See:

 

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

 

The thing is that for creative and visual reasons, you don't always want to perfectly match the color temperature of the light because you want that light to render with its natural color bias, or have a bias for creative effect. For example, as it gets later in the day, the sun gets lower in the sky and gets warmer but the shadows get cooler, especially if the sunlight gets dimmer. So if you matched the color temperature of the sunlight so that it renders white, the overall image might feel very cool because the shadows will go even more blue. And if you balanced for the shade, the sun would look very orange. So there is no right or wrong color temperature setting, it just depends on the look you want to achieve.

 

Also, at some point, small adjustments in warmth or coldness can be done in color-correction later rather than carrying a lot of warming and cooling filters. Though people shooting reversal film will tend to use camera filters since slide film was designed for direct projection more than it was for post color-correction.

 

If you really want to know what filters shift color temperature around by specific amounts of Kelvin, you will have to learn the MIRED shift system. Once you figure out the amount of MIRED units you need to move towards orange or blue, you can pick the right filter (or lighting gel) -- the manufacturers list the MIRED shift that the filter causes.

 

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

 

http://www.leefilters.com/lighting/mired-shift-calculator.html

[David Owen James]

Is the MIRED system useful for practical application - do you use it on set? It seems difficult, but I want to really understand how to control colour and get the best image possible. I was wondering if you use a colour temperature meter? I was thinking about buying one until I saw the price they are going for. Is it possible to understand how to use gels and filters correctly without such devices?

 

I've ordered a few books on cinematography and lighting; I hope they will give me a basic understanding about colour temperatures.

 

By the way, saw your Northfolk reel. Impressive, great work!

[David Mullen ASC]

Most warming and cooling of lights with gels is done by eye (and monitor, with digital). A color temp meter can make things precise, like for gelling two HMI's to match each other if one looks a bit cooler or warmer than the other -- I don't own one, but my gaffer generally carries one. Useful when going into some location with odd compact fluorescents, etc. and figuring out what combination of gels are needed to correct them or to match our lights to them.

[David Owen James]

I think I will have to learn it by eye too as a meter is not in the budget. How precise must the DOP in regards to colour temperature? How much can be corrected after without losing quality?

[David Mullen ASC]

It's not hard to shift an overall image in post to be a bit cooler or warmer.

 

It matters more in matching sources within a frame -- let's say someone walks from a patch of real sunlight in a room to an artificial patch created by an HMI. Since both lights are in the same frame, it's not easy to correct one to match the other in post.

 

And some light sources with color spikes and missing parts of the spectrum can look different to the eye than how they render on film, so it's harder to judge correction by eye.

 

But I'm not sure what your shooting situation is in regards to needing to match or correct color temperature. If you are shooting on daylight balanced film in the daytime, then why are you worried about color temperature? Are you planning on shooting in the shade a lot and don't want it to have a cool cast? You can probably fix that in post but you could also use a light warming filter if you are concerned. It's just that with 200 ISO film, you will definitely need ND filters outside so I'm not sure you also want to add another filter as well to fix small variations in daylight color.

[David Owen James]

A lot of the film will take place outdoors in direct sunlight (in India). I'm mostly concerned with the interior shots as I will probably have to use tungsten lighting for cost reasons. If I am shooting in a room with several tungsten light sources as well as a window with daylight coming in, then I imagine the best solution would be to correct the tungsten light to match daylight with blue gels, and leave the lens unfiltered. If I am shooting a subject in the shade during peak sunlight hours, my subject might require a warming filter?

[David Mullen ASC]

Do you realize that the correction for 3200K to 5500K loses you 2-stops, whether it's full blue gel on tungsten lights or a full blue filter on the camera? At some point, it may be worth the money to rent some HMI's and daylight Kinos and LED's.

 

Yes, to match daylight, your tungsten lamps will need Full CTB gel. However, you can only partially correct the tungstens if you want a warm light afternoon sunlight effect, or leave them uncorrected for an sunset orange look.

 

You don't need a warming filter if shooting in the shade, you can fix that in post.

 

Why are you shooting in color reversal film? It doesn't give you much latitude for correction. Is this for a fantasy or flashback effect?

[David Owen James]

I may be able to rent some HMI's, but will likely have to use tungsten lights for this project.

 

The film will be shot in India and will use a lot of religious imagery which I feel will be best captured on colour reversal film. In general, I prefer the look of colour reversal over colour negative, and prefer the look of older stocks to the newer ones.

[David Owen James]

I just picked up the book 'Cinematography: 3rd edition', which I was happy to see has your name on it! I assume I'm speaking with the same David Mullen. The lighting section looks extremely useful. I'm going to devour the book over the next week. Thanks for the help.

Edited June 8, 2013 by David Owen James

[Jeremy Cavanagh]

David,

What is your conclusion about the quality of LEDs overall versus incandescents for lighting film and TV (e.g. spiky spectrum)? I'm a broadcast engineer and in the past spent many, many hours 'shading' ('racking' here in the UK, CCU in Australia, etc) broadcast cameras for all sorts of TV shows and I hated CFLs in studios because of the effect the spiky spectrum had on reflecting from different coloured surfaces. No matter what you tried the flesh tones of a presenter, etc sitting under CFL lighting to me always looked like they were lit by flourescents, but, to be fair, that may have been the types bought for TV studio grids rather than for location or film work. I haven't had an opportunity to see LEDs much in operation and am expecting them to be far better than CFLs for television work but wouldn't know what they would be like with emulsions so would value a DOP's opinion.

[David Mullen ASC]

Right now I think most DP's still feel that flesh tone still look best under incandescent light rather than an LED or fluorescent built for 3200K, but some LEDs and fluorescents are "good enough" depending on the shooting scenario. In daylight balance, I think the LEDs and fluorescents are better in color rendition and real daylight situations have such a mix of color sources that there is rarely a 5500K reference anyway to compare against. It's really in the 3200K world where it's hard to beat the color of incandescent light on skin but a good LED or fluorescent isn't bad at all.

[Adrian Sierkowski]

Personally, and I am becoming more and more a minority in this, I always find non tungsten sources, on Caucasian skin especially, to feel very washed out, almost jaundiced, or sickly-- they almost always render slightly cool or slightly green even when I just stick my hand under them and look @ my own skin. Of course, if you're gong to be doing something "modern" then I think this is fine, but woe to me if i were ever given a period film and had to use an LED, I think my head would explode seeing them. Of course, no one else would probably notice.

[David Mullen ASC]

A lot of stills photographers shooting portraits for decades have been using daylight balanced electronic flashes and seemed happy with the color of flesh tones.

[Adrian Sierkowski]

Quite true; but I wonder if that is not more a product of most stills film being daylight balanced, and the same true with digital systems now. Sadly I've never had the chance to shoot T based stills film despite always saying "oh next roll"

[David Mullen ASC]

I wouldn't be surprised though to hear that when flashbulbs were replaced by electronic flashes if some photographers thought that some of the "warmth" (literally and figuratively) was lost.

[Phil Rhodes]

I don't know how deeply into the technical side of it we've got before, but it remains a fact that most LED lighting at the moment has a spectral output that is - if you'll excuse the term - beyond the pale.

 

Most of the stuff that's available is rated only in CRI, which is an almost absurdly inadequate way of characterising something that has a spectral output which looks like a cross-sectional map of a significant mountain range. I hate to wear my most accustomed hat here, but it's no great leap to the conclusion that better approaches to test and measurement are being overlooked because they'll make more or less everyone's products look pretty poor.

 

I'm starting to get the feeling that almost anything that uses a single type of LED will suffer this problem, and I'm equally suspicious of combinations of colours plus white which seem rather specifically tailored to passing the tests. Nobody's really pretending that RGB-only devices are really intended for anything but effects work.

 

In short, it appears to be a bit shonky because it all too often is.

 

I'm not sure if there's a colour temperature difference between burning magnesium wool and a xenon flash tube, but it would seem a bit convenient if there wasn't. Magnesium flash bulbs were often blue-tinted in my recollection, but not universally. Perhaps their native CT was nearer tungsten.

[Mathew Collins]

Do you realize that the correction for 3200K to 5500K loses you 2-stops, whether it's full blue gel on tungsten lights or a full blue filter on the camera? At some point, it may be worth the money to rent some HMI's and daylight Kinos and LED's.

 

Yes, to match daylight, your tungsten lamps will need Full CTB gel. However, you can only partially correct the tungstens if you want a warm light afternoon sunlight effect, or leave them uncorrected for an sunset orange look.

 

You don't need a warming filter if shooting in the shade, you can fix that in post.

 

Why are you shooting in color reversal film? It doesn't give you much latitude for correction. Is this for a fantasy or flashback effect?

 

David,

 

>Yes, to match daylight, your tungsten lamps will need Full CTB gel.

 

My understanding is tungsten lamps emits broken spectrum. Does CTB gel converts 3200K to 5600K?

 

>However, you can only partially correct the tungstens if you want a warm light afternoon sunlight effect, or leave them uncorrected for an sunset orange look.

 

Isn't Tungsten is a warm light? Is there any correction needed to match with afternoon sunlight effect if we use Tungsten lights?

[Mathew Collins]

Right now I think most DP's still feel that flesh tone still look best under incandescent light rather than an LED or fluorescent built for 3200K, but some LEDs and fluorescents are "good enough" depending on the shooting scenario. In daylight balance, I think the LEDs and fluorescents are better in color rendition and real daylight situations have such a mix of color sources that there is rarely a 5500K reference anyway to compare against. It's really in the 3200K world where it's hard to beat the color of incandescent light on skin but a good LED or fluorescent isn't bad at all.

 

David,

 

>Right now I think most DP's still feel that flesh tone still look best under incandescent light rather than an LED or fluorescent built for 3200K, but some LEDs and fluorescents are "good enough" depending on the shooting scenario.

 

Is incandescent light=> tungsten light?

[David Mullen ASC]

Yes.

[Mathew Collins]

Do you realize that the correction for 3200K to 5500K loses you 2-stops, whether it's full blue gel on tungsten lights or a full blue filter on the camera? At some point, it may be worth the money to rent some HMI's and daylight Kinos and LED's.

 

Yes, to match daylight, your tungsten lamps will need Full CTB gel. However, you can only partially correct the tungstens if you want a warm light afternoon sunlight effect, or leave them uncorrected for an sunset orange look.

 

You don't need a warming filter if shooting in the shade, you can fix that in post.

 

Why are you shooting in color reversal film? It doesn't give you much latitude for correction. Is this for a fantasy or flashback effect?

 

 

>Yes, to match daylight, your tungsten lamps will need Full CTB gel.

 

My understanding is tungsten lamps emits broken spectrum. Does CTB gel converts 3200K to 5600K?

[Adrian Sierkowski]

Tungsten light is most certainly not a broken spectrum-- it's why it's the most pleasing light to light with-- all other lamps, however, discharge as they are or LED most certainly DO EMIT a broken spectrum. A Tungsten filament is a VERY good analog for a Plank Black Body Radiator, and yes, full CTB is designed to bring 3200K to 5600K.

[Mathew Collins]

Tungsten light is most certainly not a broken spectrum-- it's why it's the most pleasing light to light with-- all other lamps, however, discharge as they are or LED most certainly DO EMIT a broken spectrum. A Tungsten filament is a VERY good analog for a Plank Black Body Radiator, and yes, full CTB is designed to bring 3200K to 5600K.

 

Thank you Adrian.