No matter what you photograph, there is one thing you should realize about light. Not all light is created equal. I’m not talking about the quality of light, but rather the color of light. What you might see as white light from different sources can actually have different colors, or what are referred to as color temperatures. Direct sunlight at noon (which I’ll just refer to as sunlight) is considered to be a “normal” color temperature, so all light sources are compared to this as the standard. For example, light from an incandescent light bulb appears to be more orange than sunlight. On the opposite side of the spectrum, shady areas appear to be more blue than sunlight. In photography, we refer to these differences as being “warmer” (or more orange) and “cooler” (or more blue) than our neutral sunlight reference point. In this article, we will go over the basics of white balance and color temperature, topics that can be a bit intimidating for beginners to understand.
So how does this apply to photography? Have you ever taken a photo that came out looking too orange or blue? When you looked at the scene with your eyes, it probably didn’t look orange or blue. It looked normal. That’s because our brains compensate for different color temperatures so that we just see normal colors. Our cameras don’t automatically compensate for different color temperatures. Instead, unless you use a setting that compensates for different color temperatures (which we’ll discuss soon), cameras capture the light and color temperatures that are actually in a scene, not what your eyes see.
1) What Is Color Temperature?
Let’s talk a bit more about color temperature. Color temperature is measured in units of Kelvin (K) and is a physical property of light. There is a large margin for variance between different light sources, even if they appear to be exactly the same. For example, maybe you’ve been in a room with rows of overhead fluorescent lights and noticed that there were some bulbs that were a slightly different color than the others. Maybe they were older or a different brand of bulbs, but regardless of why, they had a different color temperature than the rest of the bulbs. Similarly, sunlight at noon can have a different color temperature than it does at sunset.
A neutral color temperature (sunlight at noon) measures between 5200-6000 K. You’ll find most external flash units come set from the factory in that range, which means they are basically trying to imitate sunlight. An incandescent light bulb (warm/orange) has a color temperature of around 3000 K, while shade (cool/blue) has a color temperature of around around 8000 K. Here’s a chart that gives you a few different light sources and their typical range of Kelvin measurements:
| Light Type | Color Temperature in Kelvin (K) |
| Candle Flame | 1,000 to 2,000 |
| Household Lighting | 2,500 to 3,500 |
| Sunrise and Sunset | 3,000 to 4,000 |
| Sunlight and Flash | 5,200 to 6,000 |
| Clear Sky | 6,000 to 6,500 |
| Cloudy Sky and Shade | 6,500 to 8,000 |
| Heavily Overcast Sky | 9,000 to 10,000 |
1.1) Color Temperature of Different Light Sources
Photographically speaking, things get tricky when the scene you are photographing has multiple light sources with different color temperatures. This situation is known as mixed lighting. Take a look at the photo below:
This scene had chandeliers hanging over the tables that had incandescent bulbs in them, while indirect sunlight was coming through the windows behind me. After adjusting the white balance for the tungsten overhead lighting (which I’ll explain how to do in a bit), the sunlight that is lighting the side of the tablecloth and the flowers on the right looks blue.
1.2) Color Temperature of Different Light Conditions
It’s not just different light sources that can give you different color temperatures. Different lighting conditions can also have different color temperatures. Take a look at these two photos:
They were taken only moments apart, but between the first and second image the sun went behind a cloud, creating shade and giving a cooler color temperature. The light source (the sun) didn’t change, but the conditions did.
2) What is White Balance?
Now that you know what color temperature is, white balance should be fairly easy to understand. As the name suggests, white balance balances the color temperature in your image. How does it do this? It adds the opposite color to the image in an attempt to bring the color temperature back to neutral. Instead of whites appearing blue or orange, they should appear white after correctly white balancing an image.
2.1) In Camera White Balance
Most cameras come with the option to manually set or adjust white balance. Typical settings include “sun”, “shade”, “tungsten” and “fluorescent”. Some cameras come with the option to manually set a color temperature by choosing a specific Kelvin value.
Let’s take a look at a few examples:
In the image on the left, you can see how orange the light bulbs look when I have my camera set to a neutral white balance, but once I change it to the color temperature of the bulbs (either manually or with a preset white balance), they look normal. Why is that? My camera is “cooling” down the color temperature of the bulbs by adding blue to the photo, giving us the appearance of white light. Notice that while the light bulbs now look white, the bokeh in the background now looks blue.
Still having trouble understanding what’s going on? Take a look at these images of the same scene that were taken in daylight:
Now that it’s daylight, you can see that 5500 K is the correct white balance for the color temperature. What happens if I set my white balance to 3050 K in daylight? The image turns blue! This is how much blue was added to the photo of the orange incandescent bulbs to balance the orange and make the color temperature of the incandescent light look normal.
2.2) Post Processing White Balance
If you don’t want to worry about changing the white balance in your camera for different situations, as long as you’re shooting RAW you can always adjust the white balance of your images with post processing software such as Adobe Photoshop or Lightroom. This is sometimes referred to as “color correction”. In your software you’ll probably see a panel that looks something like this:
Just like setting the white balance in your camera, you can manually set the white balance either by adjusting the temperature value or by using the eyedropper tool on the left side and clicking on a neutral or white part of the image. Similar to your camera, you can also choose a preset white balance:
Here is the same image from above, both straight out of camera and with the white balance adjusted in Lightroom. Compare it to the image where white balance adjustment was done in camera:
Remember, this is only possible if you are shooting RAW images. If you are shooting JPEG, you will be able to make slight white balance adjustments to your images but will not be able to make drastic corrections.
Here’s another example of adjusting white balance in post processing. This photo was taken during an engagement session that started in daylight, so I set my camera to a white balance of 5500 K. As the sun started to set, the light got warmer and warmer, giving this image a very orange feel.
Now some people might like a sunset image to be this warm (it personally doesn’t bother me that much), but I thought it needed cooled down a bit to really bring out the pinks and blues in the sky, not to mention make their skin look a bit more normal. In Lightroom, I adjusted the white balance so that the temperature was at 4500 K, giving me this image which I feel looks more natural:
3) The Relation Between Color Temperature and White Balance
Now that you know the difference between color temperature and white balance, you should be able to see the relationship between them. They are opposites! Unfortunately, since photographers mainly work with white balance, we sometimes get confused when referring to color temperature values.
Personally, I almost always shoot at a fixed Kelvin of around 5500 K. When I photograph interiors that are lit by incandescent bulbs, my images all look orange. Since I shoot RAW, it’s not a problem! In Lightroom I just “cool them off” by changing my white balance to around 3000 K. For shade, I “warm up” my shady images by changing the white balance to around 6500 K. In my head, higher Kelvin values start to equate to warm light and lower Kelvin values start to equate to cool light.
But remember, I’m balancing the color temperature! What I’m actually doing is adding the opposite color temperature to my image. I’m so used to thinking of 3000 K as cool, when I see that the color temperature of an incandescent bulb listed as 3000 K, it takes me a minute to remember that the 3000 K I think of as a cool color temperature is actually a cool white balance.
So now for the big conclusion you can draw from all of this. If you set your white balance to the color temperature of the scene you’re photographing, it should look great! If you’re photographing light bulbs that have a color temperature of 3000 K and you set your camera’s white balance to 3000 K, the light should look white! Now here’s the bad part. There’s really no way to measure color temperature, so you’re left approximating or adjusting in post processing.
4) Using Auto White Balance
If you prefer shooting JPEG or just don’t want to worry about color correction after the image is taken, most (if not all) cameras and post processing software come with the option to use auto white balance, or AWB. With AWB, your camera evaluates the scene that you’re photographing and decides on the best white balance to use. It will typically reference a neutral color in your scene such as white or grey to determine the correct white balance. Depending on your camera and the scene you’re photographing, your results will range from perfect to not very close at all.
4.1) Using Auto White Balance In Camera
When using auto white balance in camera, your results will vary depending on the lighting conditions you’re shooting in. For example, if you’re shooting in daylight the white balance of your photos will typically look correct. Unfortunately, mixed lighting can really give AWB troubles, so you might still end up needing to adjust the white balance in post processing.
Even daylight can fool auto white balance. Here’s a set of images of an orange fox on a brown table that really demonstrate how inconsistent auto white balance can be without a neutral color in the image for your camera to reference:
All of these images were photographed in the same light. You can see how much of a difference the background makes when using auto white balance. In the third image, simply adding a white background helped the camera get the correct white balance. You can see the last image is the same as the 2nd image, only with a correct white balance (adjusted in Lightroom).
Different cameras have different auto white balance capabilities. As with any technology, more recent cameras seem to be more accurate than older cameras. You’ll also typically have more advanced capabilities in more expensive cameras. That’s not to say that the AWB systems in entry level cameras are not good. For example, my iPhone 6 does a good job with AWB, but chances are my Nikon D810 does a better job getting it right more consistently.
4.2) Using Auto White Balance In Post Processing
Most, if not all post processing software such as Adobe Photoshop or Lightroom comes with an auto white balance option. In my experience, this is never as accurate as shooting with AWB in camera, but it can serve as a good starting point if you’re trying to adjust the white balance in your image and just can’t seem to get it right.
Here’s an example of using auto white balance in Lightroom on a simple, sunlit photo:
I don’t know about you, but I personally don’t think Lightroom got the white balance right in this image.
5) Tint
In addition to color temperature, light can also have a tint. While color temperature ranges within the orange/blue spectrum, tint ranges within the green/magenta spectrum. Major tint adjustments are typically not necessary when color correcting images taken in daylight. If you tend to photograph subjects that are lit by artificial light sources such as tungsten, fluorescent, LED or mercury vapor lights, you’ll find yourself adjusting tint much more than with natural light.
Here’s an example of a scene that was lit by fluorescent lights:
You can see that the first image has a very strong green tint. By changing the tint (adding magenta) but leaving the color temperature untouched, the white balance has been corrected.
Tint is not meant to compensate for color that is reflected onto your subject from nearby objects, such as in this image:
I photographed Melissa while she was facing a green door specifically to demonstrate this scenario. Although the white balance of in the image is correct, there is a green tint to the shadow side of her face that has nothing to do with color temperature. It is a reflection of light off of the green door and can not be corrected by adjusting the tint of the image.
Just like reflected light that has picked up a color cast, colored or gelled lights on your subject are also not easily corrected by adjusting the tint of your image. For example, this image from a New Year’s Eve wedding has everything you’d want in an image taken at the stroke of midnight: streamers, celebrating, happy green faces on the bride and groom… wait, what?
Unfortunately, this is not something that can be fixed by adjusting the white balance. Fortunately, it still looks pretty good in black and white.
6) Consistency (or lack thereof) Between Camera Makes and Models
As if getting the white balance correct between all of your images wasn’t difficult enough, using different cameras while photographing the same scene introduces a whole new dimension of complexity. This is something that wedding photographers know all too well since they typically work with another photographer who rarely has the same make and model of camera, but it’s something any photographer who works with more than one camera will encounter.
Take a look at these sets of images of the amazing sensor gel stick taken in the exact same light but with different cameras.
Below each image you can see what my white balance and tint were set to in Lightroom. Notice that, although they are identical, the images look completely different! This tells you that different camera models, even if they’re from the same manufacturer, affect the final look of the image. The reason for this lies in the software used for RAW processing. What’s that mean for you if you use different cameras? It means that if you want to copy the white balance settings from one image and paste them onto another that was taken with a different camera make or model, your final images might not look the same.
7) Conclusion
As long as you shoot in RAW, knowing the color temperature of different light sources is not that important. Simply understanding the basic concept of color temperature should be enough for most photographers. What is important is knowing how and when to adjust white balance, either in camera before you create an image or in post processing software after you create an image.
Once you get comfortable adjusting the white balance of your images, you can start to use white balance creatively in your images, either warming them up or cooling them down to change the feel of the entire scene. Although that’s not something we’re going to go into for now, keep an eye out for a future article on creative use of white balance.
If you have enjoyed this article, please check out our in-depth Level 1 Photography Basics Course, where we explore all the basics of photography in much more detail. It is an intensive, 5+ hour course with enough material to not only get you started today, but also to serve as a reference material in the future.
If you’re interested in the science behind it, you can read the Wikipedia article for more information on color temperature.
The different look in your gel stick example is because colour temperature and tint are calculated in Lightroom from Adobe-defined reverse colour matrices, ColorMatrix1 and ColorMatrix2.
White balance exists in camera metadata/makernotes in the form of white balance multipliers, not in the form of colour temperature and tint. Colour temperature and tint are calculated in a converter, and there is no precise way of doing that, especially if ColorMatrix1&2 are imprecise. In other words, setting the same colour temperature and tint does not mean the white balance is set to the equivalent values.
Different sensors have different colour filters in front of them, and the same light causes different values of white balance multipliers.
D810 Daylight White Balance, Red = 1.95, Green = 1, Blue = 1.34
D800 Daylight White Balance, Red = 1.97, Green = 1, Blue = 1.38
With proper white balancing the resulting images must be extremely close in terms of colour, final look should not be affected. But with imprecise colour matrices and colour transforms in a raw converter they may look very different.
So I had this problem at an event shooting with D750 and D810 similar settings. Had a hard time matching in LR. What should I do before importing to LR to get similar results during Copy/Paste?
The workaround is to create custom colour transforms for your cameras, using, say, http://www.ludd.ltu.se/~torger/dcamprof.html
Great information, Iliah. Thanks for sharing!
Always informative, concise and factual Iliah, a couple of other ‘regulars’ shout take note when getting involved in their own little ‘mine is bigger than yours’ arguments and driving people away in the process.
Iliah, just one question to the expert. Assume I shot a raw image with a certain WB setting of the camera. If I used a different WB setting would the actual image data differ or would only the metadata differ? The second seems to be implied by the beginning of your second paragraph above.
The raw data in the current cameras does not depend on the white balance.
The only notorious example that I know of where white balance actually changes the raw data is Nikon D1 original.
These days white balance is recorded in metadata, and applied to embedded and out-of-camera JPEGs, and Nikon sRAW files — those Nikon sRAW files are also JPEGs, but of higher bit-ness, effectively 11 (but advertised as 12) bits instead of regular 8. “Normal” raw data is not affected by the white balance.
However, automatic exposure and photographer’s decisions based on in-camera histogram may be affected by the white balance settings. Canon shooters can use MagicLantern firmware to overcome this — cameras with this firmware installed can display raw histograms, free of the effect of white balance. Another way to cancel white balance is to use UniWB, which sets the white balance coefficients to all 1’s. This is possible with all camera brands, however with various amount of effort. Jim Kasson lists shortcuts to UniWB here: http://blog.kasson.com/?page_id=2466 and offers a nice and easy procedure for the cases where shortcuts do not work here: http://blog.kasson.com/?page_id=2481
Iliah, thank you for your confirmation! I wondered about this for a long time, and not even Thom Hogan in his really helpful camera guides states this simple but essential fact on the almost 30 pages he writes about WB. So, although the WB setting has an effect on the in-camera JPEG-based histogram need for assessing exposure, as long as I get an ideal exposure I do not lose anything by being sloppy w.r.t. WB setting. Good to know.
Also, thanks a lot for the hints to Jim Kasson’s postings about UniBW.
Do you have any idea why the manufacturers do not provide in-camera raw histograms? I cannot imagine this is due to missing processing power given today’s ASICs like EXPEED 5. So, is it just ignorance?
Hi John, I took some photos in Disney Springs at night but the Red neons, like The Boathouse sign, became Orange Neon. Do you know which white balance configuration should I use to take photos of Red neons?
Hi Ricardo,
Since neon light isn’t something you encounter very often, there is no white balance setting in your camera for it. You could always set a custom white balance in your camera before you take the image, which I don’t describe how to do in this article, but you can find in your camera’s manual.
Hi John,
Good article, touching briefly on a massive subject.
Mr Adobe’s settings and his “one size fits all” is not necessarily suitable for colour critical work.
Most of us use Adobe products, and they have a lot to offer but there are many times the software just doesn’t get it right. In critical cases you would be better making a proper profile for the camera you’re using. The accurate colour white balance of your camera is good start, but gets reheated once you bring that image into your processor/computer. So begs the question why bother getting it accurate in camera? Well mainly not to clip any channels.
No best colour is from your own profile for your own camera etc.
YCMK was a preferable workplace to RGB, and days past we had to submit work corrected in the subtractive YCMK for a good reason.
VHS won yippee, as did RGB, but sadly that doesn’t make them the best. Your ink-jet printer most likely agrees.
Thank you Andrew.
This is a massive subject and is simply meant to be an introduction to the basics, not necessarily anyone interested in color critical print work.
So true, the best format isn’t always the winner. We just have to make the best of what we have to work with. :)
Get it right in the camera but if you don’t it is easy to correct in post processions. Then, why bother getting it right in the camera in the first place?
If you get it right in the camera, post processing software will see it differently and you will still have to correct wb and temp.
If you get it right and your monitor shows perfect picture, other people’s monitor will be different and they will see it in wrong colour.
If you have clients who want to see images before you have a chance to adjust them, getting the right in camera is a big deal. Likewise, if you decide to shoot JPEG, you’ll want to get everything right in camera.
I completely agree about other people viewing your images on their own devices. As a wedding photographer, I would be willing to bet that 99.9% of people who ever see my images don’t have color calibrated monitors and my images look slightly different to them than they do to me. Unfortunately, there’s not a thing we can do about it.
Still, even if their monitor is not calibrated, they will most likely print at least 1 picture. To me, it`s worth every extra hours in my computer room after shooting a weeding ;)
Even if their viewing device is not calibrated, the picture will look much better after post-processing. In fact, JPEG from the camera is processed…by the camera. I rather choose the setting myself especially for the tricky situation that the camera does not understand.
Great article John. Even if they are getting better with metering, I find this is the thing that camera struggle the most with.
So true, Simon. I still spend many hours color correcting my images, even knowing that different monitors and printers will not be calibrated exactly the same. As professionals, we know that processing and color correction are not optional. :)
Just a question. I have an Led TV slash Pc monitor that I wanna use for photo editing rather than buying Eizo monitor with the calibration tool. Do you think, because it is Led, my pictures will appear very, very different to pictures edited with the help of normal pc monitor?
Phil, as far as I know the type backlight of the display won’t affect the color temperature since it should be color balanced to a normal temperature. If you are simply using a TV as a monitor, though, you’ll probably run into other issues since most televisions are very contrasty and saturated.
If you don’t have another monitor, I’d suggest viewing images after you edit them on other devices to make sure they look the way you want them to look. You might also order some test prints to see how they look when printed.
Hi Phil,
You may also notice your picture are not as sharp as when you look at them on a computer screen because of the Dot Pitch. TVs are great for gaming as they offer a much larger viewing area for the price but they have many downsides for photography. That doesn`t mean you cannot edit picture with it, but it’s a good thing to be aware of the flaws.
Great article, and great examples! I work with the presets, since I have no K scale on mine, and I find that they almost always do a better job than my own manual settings. Very hard to get WB right manually, even with the grey/white card trick, although that sometimes helps. Also, those who shoot RAW – how do you set up WB when you have a lot of pictures to work with? I mean, do you remember what the light looked light in every situation…? I’ve recently started experimenting with RAW and found that getting this right in post is one of the most challenging part – I can’t always recall the exact lighting conditions for each and every picture, so most of the time I just have to go with “this looks right”, but that may not reflect reality correctly. I often have to look at the JPEG counterpart for reference.
I liked the B&W trick in 5.) – if all else fails, B&W can still be a last resort :D
Thanks Judit! Using the camera presets can be the easiest way to get a good white balance straight out of camera. As for shooting RAW, there is no way I would try to remember the light in every single scene. Sometimes you just know, “that’s tungsten”, “that’s daylight”, “that’s overcast”. You could even get in the habit of taking a reference photo of the light source in every scene if that helps. Usually it’s just adjust the WB until “that looks right”.
And yes, when all else fails getting a good white balance, B&W works every time. :)
It would be nice if cameras also supported setting the tint besides the Colour Temperature…
Mark,
You can actually do this with a Nikon body, only not by using the WB button an dials but via the menu system. Your camera manual will tell you how to do that.
Why do one need colour temperature or tint? Because that is what Adobe are using? Or because looking at tint one can estimate the quality of light? Or maybe for some other reason, like those CCT and tint numbers are easier to understand than the white balance numbers that characterize per channel underexposure?
Illiah, my answer to your question would be: yes, because color temperature is an easy to understand model of what is going on physically, known for color film for almost a century now. Even if Pete A’s Mired calculations below seem to be complex, they are actually quite simple, if you understand the color temperature model. Ok, it would be nice if ACR’s calculations of color temperature and tint would be more precise, but this doesn’t mean these parameters are not helpful.
The WB multipliers, though, seem to depend on the spectral properties of the sensor which differ from camera model to camera model, and even from camera to camera. So, even if they are easy to understand for a raw expert like you, this doesn’t seem to be a universally applicable (i.e. camera independent), easy to understand model.
Exactly. They are wrong, they are obfuscating the matter (how, based on colour temperature and tint, one is going to determine the amount of underexposure?), but they are easy to understand, so let them be.
Hmm, determining the amount of underexposure based on color temperatue and tint has not been my first thought, up to know. Sounds interesting, though. But I wouldn’t know how to do that in the WB multiplier model as well. See also my comment 8.1.1.1.1 below.
This was extremely helpful. One of the best explanations I’ve seen for beginners. Thank you!
Great! Glad you enjoyed it Dee!
Hi John, Your article is well explained and illustrated. I apologise in advance for my long comment.
I learnt colour balance for use with transparency films so I can’t understand the “White Balance in Kelvin (K)” column in the 3-column table. Allow me to explain… When using a daylight-balanced film to shoot a tungsten lit scene, or a tungsten-balanced film to shoot a daylight scene, we need to apply the correct white balancing filter. This is calculated using the mired (micro reciprocal degree) scale: M = 10^6/T where T is colour temperature in kelvin; the SI unit for M is the reciprocal megakelvin (mired). White balancing filters have a constant mired shift value, regardless of the actual colour temperatures involved. The filter required for a daylight film balanced at 5600 K to shoot a 3500 K tungsten-lit scene = 10^6/5600 – 10^6/3500 = 179 – 286 = -107 mired (the filter appears blue).
Similarly, the filter required for a 3500 K film to shoot a 5600 K daylight scene would be +107 mired (the filter appears amber). Now, if we use this same filter on a 5600 K daylight balanced film (or digital camera) it would be suitable for capturing a scene having a mired value of 179 – 107 = 72 mired, which is 13,888 K.
When using a daylight reference point of 5600 K, the colour balance adjustment required for a 3500 K scene is the same magnitude, but opposite sign, of the adjustment required for a 13,888 K scene. The “White Balance in Kelvin (K)” column of the table states 6500-8000 K for a colour temperature range of 2500-3500 K. What happens when we halve 5600 K: 2800 K, which is a shift of 179 mired from 179 mired to 258 mired. When we shift in the opposite direction, 179 mired – 179 mired = 0 mired, which a is white balance adjustment of infinity kelvin, not 5600 + 2800 kelvin.
Best wishes,
Pete
It is the same with digital, you can use balancing filters (I do), only the cameras are balanced not like film. If one wants neutral out of digital camera, he is first to apply a magenta filter, like CC40m, and then mired, to balance light to 4200-5500K, depending on the make and model. The benefit is that now all 4 channels are exposed properly, green is not leading anymore, red / blue are not lagging behind. With digital, not like with slide film. one does not need to be exact with filter values.
In conversion, all the above is applied in digital domain (unless filters are used at the time of shooting, or the filters were not sufficient), but the goal is not what is in the third column. It is much closer to 5500K (usually between 5000K and 6500K) than to 10000K
Thanks for your reply, Iliah. It seems that I was, perhaps, being far too polite while composing my comment to John. Optical physics is one of my passions therefore I more than adequately understand colour balance for both film and digital photography. Unlike you, I do not use colour balancing filters on my digital cameras because my main speciality is night-time and twilight industrial photography. The reciprocity failure of film when combined with the required filters made this type of photography mostly impossible to achieve.
No offence intended towards either you or John, and many thanks as always for your in-depth explanations of why Adobe RAW converters do the idiosyncratic things that they do. There are several reasons why I never use Adobe image editing products for any of my work :-)
White balance is grossly obfuscated by some cameramakers and software engineers; same as it is with film. This went downstream, to evangelists, teachers, forum gurus, etc. We all can fall victims of such obfuscations from time to time.
White balance measurements have very little to do with colour temperature; colour temperature and tint being only indirect, derived numbers.
All white balance is is a couple of multipliers (some times, 3 multipliers are used, if the Bayer filter in front of the sensor has 2 green channels with different responses; and more, if Bayer filter is not a traditional RGGB) to balance the sensitivities of channels to particular “colour of light” in the scene so that neutral areas will have (close to) equal levels of red, green, and blue. One can use simple multiplication because raw data is in gamma=1, linear (not so easy if the tone curve is unknown, as it is in a rendered image like a JPEG).
To get to colour temperature and tint from measurements, look-up tables and different approximations are used with varying degrees of success and accuracy. That is the case even with colormeters, not only our dear hand-held Minoltas/Gossens/Sekonics, but display calibration devices too.
Iliah, sorry for being persistent w.r.t to this, but I have to admit that I do not understand the WB multiplier model.
It is clear how to white balance the raw data: just multiply every channel with a multiplier (wherever you get these multipliers from). But what is the equivalent of, for example, a color temperature of 5500 K in the WB multiplier model? And how do I change the WB, for example, from 3000 K to 5500 K in the WB multiplier model?
I suspect you haven’t explained this up to now. But this would be needed, if I wanted to do something comparable to Pete A’s Mired calulations above.
> But what is the equivalent of, for example, a color temperature of 5500 K in the WB multiplier model?
We do not have a device that measures colour temperature directly. We do not have 2 devices that agree on colour temperature measurements. We have a multitude of different light sources that convolve to the same colour temperature, but the spectrums and colour of light are different between them. Most important, why does one need to know colour temperature, what exactly does it tell? – human eyes do not allow us to make an accurate assessment of colour temperature too, instead we think in “preset terms” – incandescent, shade, sunset… Given that colour temperature and tint are not even transferrable between cameras or raw converters, each one will display different numbers for the same light or different colour for the same numbers, what is the point?
There are two reliable methods of setting white balance in postprocessing, by numbers (that includes click-on-grey methods and memory colours methods) and to taste. Colour temperature and tint do not even enter here.
The whole idea of changing white balance based on a imaginary rotated scale (as there are no cyan or yellow filters in the vast majority of cameras, so colour temperature slider is an artifact attempting to control all channels simultaneously and in a non-linear fashion ; purple filters are also non-existant in Bayer, so tint slider is yet another artifact) is wrong, and not uniform. When an image presents an orange tint, how exactly one is going to neutralize it, using colour temperature and tint? In multiplier model it is easy, one just needs to know what orange is, too much red and too little blue. It is not that difficult using standard colour correction filters as well (those we use when printing in a darkroom), but such a mechanism is lacking in most of the raw converters.
log2(colour_multiplier) is colour channel underexposure in EV.
> how do I change the WB, for example, from 3000 K to 5500 K in the WB multiplier model
If you need that for some reason, through presets based on white balance multipliers (that is, factory sensor characterization; each camera supplies metadata containing that, or the data is contained in SDK, like it is in the case of Nikon cameras) and not on some approximations based on generic matrices. The important thing, however, is to realize what is primary, and what is secondary: multipliers are primary data, and any explanation must start with them, and not with colour temperature. Explanation must state the goal of white balance, and how it is achieved in conversion. And that is all about the multipliers.
Iliah, thank you very much for being patient and giving all these excellent explanations! Escpecially, I wouldn’t have thought about assessing underexposure this way.
I just realized that FRV is able to display the WB multipliers. I didn’t understand this at all up to now. I am going to investigate it tonight.
Carsten, Keep in mind the fundamental purpose of white balance: it is to map (transform) the white point of the actual scene to the white point of the display medium. Digital television and sRGB specify a monitor screen white point of D65 (6500 K). Printing paper is specified for a lower temperature illuminant, often around D50 (5000 K), but it varies.
Also note that many colour editing profiles (including sRGB and Adobe RGB) have a white point of D65 and a gamma of [close to] 2.2, but some are D50 (including ProPhoto RGB and Melissa RGB) and should be used with a computer and monitor calibrated to D50, and a gamma of 1.8 for ProPhoto.
White balance is to transform from non-colour-neutral space (neutrals are not represented by equal numbers of R, G, and B) to colour-neutral space, where equal R, G, and B numbers represent neutrals.
Working colour space white point temperature and gamma are (dis)connected from monitor colour temperature and gamma with the help of colour management. There is no need to keep them the same, as most of the cases in image processing colour management is in place and properly implemented.
Pete, this is some new information for me indeed. I thought the purpose of a WB is to map the white point of the actual scene to the white point of the color space selected in the camera. And by using a monitor or printer profile this is mapped to the acutal display medium then.
I use Lightroom (ProPhoto RGB) and my monitor is calibrated and profiled to D50, but up to now I would have assumed that my images look the same if displayed on another monitor calibrated to, for example, D65 (well, at least w.r.t. color temperature). But maybe I was wrong.
In raw, selecting colour space in camera has no effect (btw, there is no such thing as colour space for raw data, in the traditional understanding of what colour space is — because raw data, scanner or sensor in a camera, has no gamut).
The problem with “white point adaptation” approach is that XYZ colour space is perceptually non-linear, and hence the results are not as convincing as they are if balancing is performed in camera RGB — see Viggiano, “Comparison of the accuracy of different white balancing options as quantified by their color constancy.” http://www.acolyte-color.com/papers/EI_2004.pdf
> up to now I would have assumed that my images look the same if displayed on another monitor calibrated to, for example, D65
That is correct.
Carsten , In-camera auto white balance does indeed attempt to map the scene white point to that of the colour space selected in the camera, but this is only for producing in-camera JPEGs or TIFFs. Most cameras have a choice between sRGB and Adobe RGB, both of which have a D65 white point. However, when using RAW you adjust the editor’s colour temperature and tint sliders to the white point of the scene, the converter/editor performs the necessary mapping to the white point of the editing colour workspace that you are using (e.g. D50 for ProPhoto).
If you save a ProPhoto image as a JPEG or TIFF, using convert to profile sRGB, then it will display properly on sRGB D65 monitors and it will be compatible with the World Wide Web. But, if you save the image as a JPEG or TIFF, using the ProPhoto profile, many Web browsers are not colour managed so they cannot perform the required conversion to sRGB D65.
Suppose you save an image as a 16-bit TIFF using the ProPhoto profile and that you have two computer systems — one calibrated to D50, the other calibrated to D65 — and each has a colour managed TIFF viewer. Will your image look the same on each monitor? The answer is both yes and no! Yes, both systems will display your image correctly. No, they will not look the same. If you put the monitors side by side and stare at the D50 monitor, the D65 monitor will appear bluish; if you stare at the D65 monitor the D50 monitor will appear yellowish. So, which one is correct? Both. When using just one of the monitors our vision can adapt to its white point via our inbuilt visual ‘auto white balance’.
Caveats: If the D50 system is using a display gamma of 1.8 then the mid-tone brightness and contrast will always look different from a system that is using the more common gamma of 2.2, irrespective of the white point. I have no idea whether Adobe image editors and viewers display a ProPhoto workspace (which has a gamma of 1.8) using a gamma of 1.8 or 2.2. I hope that something I’ve written is useful to you.
Iliah, apparently we have reached the maximum reply depth. :-) I hope this is placed after your 8.1.1.1.1.1.1.1.2.1 post.
Of cause you are right, I realized this in the meantime as well. I was actually struggling to figure out the correct color space, and I am still struggling. My understandig is that the color transform of a certain camera maps the colors of the raw image, i.e. the sensor data, to the working color space provided, for example, by Lightroom. There must be some kind of “camera color space” because the sensor data must be “interpreted” in one way or another.
There is a multitude of colour spaces that can be assigned to raw data, that is why different colour transforms (or profiles) are resulting from profiling the same camera, depending on the light that hits profiling target and profiling algorithm; and of course the goal of profiling. Flare also affects the outcome, when shooting a scene with low flare and with high flare results are different, and call for different profiles. You may also want to read http://www.rit.edu/cos/colorscience/rc_faq_faq3.php#255
Pete, that third column is actually something I was worried might cause a misunderstanding. I actually meant to say that if you photograph neutral light with white balance set to “X”, the image will have this tint. Unfortunately, that is not at all what I said (I meant to clarify before posting and never did).
I think it’s too confusing and not necessary, so I’m going to remove the 3-column table.
John, I think the article is better without that table. Your images illustrate the effects very clearly.
John, when you have spots of green light like that, I believe you can get rid of some or all of it with the Adjustment Brush in Camera Raw or Lightroom, which allows you to adjust Temperature and Tint for selected areas.
Aaron, believe me, I tried. I was able to get it close, but not perfect. In the end, for such an important moment, I decided to deliver the B&W version to my clients instead of a “pretty good, mostly fixed” color version. If it wasn’t right on their faces it probably would have been fine to fix with the brush.
I personally find the B&W version as a lot more character :)
Superb article John! Looking forward to more and also to the post-processing course Level 1.
Glad you enjoyed it, Dean! Nasim and I were just talking about starting the post-processing Level 1 course. Hopefully we can get it out to you sooner rather than later!
I find it much easier to get good item pictures using shaded sunlight the settings are so much easier to adjust and correct plus low ISO and clearer aperture settings. I have a backdrop in the car and just point in the right direction :D. Good studio lighting can get expensive.
Thank you John, you have removed the confusion which I have about WB and Colour Temperature.
You’re welcome Edwin. Glad to hear this helped!