(The fix begins at mid-page below if you want to skip some standard overhead basics stuff.)
If not familiar with white balance, then to understand the problem, start by simply taking two pictures using intentionally wrong white balance. Best try to see this is to take a picture outside in sunshine using Incandescent white balance (it will be very blue). Take another picture indoors in incandescent light using Daylight white balance (will be very orange). See? The lights are simply different colors, and the right idea is to use proper white balance to match the light color, to get good color results in various lighting. But the problem is that there are many types and colors of lighting, which affect the color in our photographs. We don't know more than a rough guess about the right match. So usually the color in our photos won't be quite right, it could be better.
We may not realize the degree of this. Our human brain tries to fool us. Or rather, our brain amazingly knows to adjust our own white balance, and it can ignore these differences (in common situations). Humans recognize the color that subject things should be, and we often see what we expect to see. So when we come indoors, from out of cloudy 6000°K sunlight into incandescent 3000°K light, we don't realize that incandescent is orange. This (and more) is named the Land Effect, explained by Edwin H Land (of Polaroid fame). But cameras have no brain, and they do see and capture the color of real life, which we then can see incandescent is orange, which becomes a problem in our photos. There's a big difference, which photographers have to deal with. Auto WB does try to fix this, halfway well, but it has no clue about the right color either (Auto WB is just making some guesses, which may not be correct). We have ways to do it better ourselves.
So this subject of White Balance Correction is about when you set the camera right, and still get poor colors (color at least a little wrong is very common, it's not an exact system). This article is about how you can fix it and salvage a perfect picture. It's not difficult, but it is something we must tend to. Some users think their camera always gets color right. IMO, they have not realized yet that it's not. :) But once we start learning, and start seeing good color, correct color will become extremely important to us. White Balance is just a fact of life in Photography. It is a difficult problem, but we have ways to make it easy.
This animated histogram at right shows the concept of what White Balance actually does (the image is from page bottom of page 2). Red and blue channels are shifted oppositely, here to align (balance) the RGB component peaks of a white card (which we know should be white) to remove a blue color cast. We can't see the photo here, but an image "too blue" is made less blue and more red, to become white. Because we have the known white card in the photo, we can see that in the histogram. White is a bright gray tone where neutral is defined as equal RGB components, no color cast (specifically, white is defined as bright and equal RGB components). The large data spike toward the right end of this histogram is a white card in the photo photographed with flash (the 5300°K result). As the WB is adjusted up towards 5300, the red and green and blue components converge to equal, and white, and perfect white balance (perfect because the card is known to be right, and the equal RGB result is white). The "equal" aspect is root of the "balance" we seek. The sliders are discussed on the next page, but the Temperature slider had a large effect, the Tint slider much less here. This image is repeated here from the last example at bottom of next page.
Some definition seems required before we start. In physics, color temperature (see Wikipedia) is the color of light radiated by an ideal black body heated to high temperature. In the way that steel or lava becomes red hot at high temperature, their colors of red light are described by the temperature of the source radiator. Same idea for tungsten filaments of incandescent bulbs, and for our Sun. Other sources like fluorescent lamps are not black bodies nor heated, but their color is described as the temperature of a black body that would generate the same color of light. Black body temperature can go much higher than red, to white, and then blue. The surface of our Sun is 5800°K, which is between the red and blue, which we call White. These temperatures are used to describe the colors of light, and color is used to describe the temperature.
Historically, the art world thinks of blue as a cool color (water and ice), and red and orange are warm colors (fire or Sun). Which seems very reasonable to us, however, science eventually learned that numeric Color Temperature of heated black bodies in physics is measured to run opposite, which while the "color" blue may appear cool in art, but in physics, blue is instead a higher "temperature" than red, which is a relatively cooler Temperature. So we learn there are the two concepts, of Temperature, and of Color, which seem to run backwards of each other. We learn that lowering White Balance Temperature makes the Color appear more warm.
The camera provides the following White Balance choices to match the color of the light in our photo scene.
These numbers are standard ball-park guesses, a bit vague and arbitrary. The camera cannot know what color the light actually is, it is just a standard guess that might be close. Flash color varies with power level. Open shade (under clear blue sky) can be 10000K or more. Lighting varies so much, many colors exist of each. So except maybe for clear direct sunlight (but our Earth's atmospheric conditions can also affect the sunlight, like humidity, haze, angle from overhead, etc), it's extremely doubtful we will ever know any precise temperature anyway. Very unlikely that any of the camera White Balance choices exactly match whatever light we actually see. So in the camera, we really don't have much chance of getting it closer than ballpark (other than maybe direct sunlight from our one Sun comes out close). We instead need to see the result first, to then see and know what it actually needs. Then it can become an easy problem, which is the subject of this page.
There are two values in White Balance, Color Temperature increasing flom yellow to blue, and Tint from green to magenta (see next Part 2 page). Adobe adds magenta Tint to their values above (+10, except Tungsten is 0 and Fluorescent is +21, correcting green). The cameras don't mention Tint, but they do it too (note on next page). Still it is all a guess about light as yet unseen. But there a a few ways (below) to use a technique to compare to an exact measured value.
We can set our camera to Incandescent or Fluorescent or Flash or Daylight or Shade WB (whatever, as appropriate), but the problem is, everything varies, and we don't know the actual color temperature of the specific light in front of us. Or Auto WB can take a stab at it, which is just another guess, because it also has no clue about the actual color of the light either.
Common types of photo lighting
Fluorescent, including CFL lighting has some unique photography downsides.
Incandescent / Tungsten bulb types vary in color too, similar different types and colors (from 2600°K household to 3200°K photo, and some are towards Daylight). Color varies with type and watts and age. Most are Orange, but have a complete spectrum. So incandescent is the definition of theoretical CRI 100, so they correct well (if we can determine their color). Adobe and Canon call this Tungsten WB, Nikon calls it Incandescent WB, which are the same thing (incandescent bulbs have tungsten filaments).
Daylight (direct sunlight) through the atmosphere on a clear day is about 5200°K, and pretty much the only constant color we have, always the same sun, same as expected by the camera Daylight WB. Color might vary slightly with angle in the atmosphere, due to latitude or season or time of day. But cloudy and shade and sunsets, etc., all become a different story.
Open Shade is "Open" to the sky (like at the side of the house, shaded from the sun, illuminated by the clear blue sky above) might be near 10,000°K. The sky looks blue because the red and yellow has been filtered out, but color temperature is given as the temperature of a black body that would generate the same color of light. Shade in a forest or under a cover is still blue, but less blue.
Flash WB is about the same color as Daylight, but flash color varies with flash power level (from maximum to minimum, see the second page here). Also brands vary with specific flash tube, and with design and age and use of that flash tube. Flash color depends on degree of the Xenon gas ionization, which depends on the flash tube current value (amps through tube), which varies with voltage and flash tube diameter. Nikon recommends service for our speedlights every two years, which means flash tube replacement. That is for heavy professional use which changes faster than most of us see, but the flash tube color drifts up with years of use, towards blue. And some flash brands start off towards bluish.
But for speedlights, low power flash is more bluish, and high power flash is more reddish. With only a few exceptions (that use speedlight design), most studio mono-lights are the opposite, being a different voltage-controlled design where low power is instead more red. Speedlights use an IGBT (thyristor-type) design where low power is always the full power level, but with the flash duration cut-off short (which starts with the high-intensity leading edge, but cuts off the decaying red tail). Sometimes portraits might like a bit of red for a more warm tone, but too much is a problem. Bounce from beige walls can take on sickly color (unless corrected), but when we get both exposure and white balance right, the good image jumps out, bright and vibrant. But realize that the cameras Flash White Balance setting is just a general WB guess, not related to your specific flash type and power level and age.
What about printing lab corrections for digital? There are many print labs and some offer options. One issue is that many of us do our own WB correction, and we will get angry if the lab changes what we carefully provided. I choose a print lab that offers a "no corrections" option (usually meaning they just turn off the automatic correction, but the rare best case means a human operator actually looks at it). But automatic machines do everything today, and correction simply means they let their own Auto WB have a go at it. I could have done that, but Auto WB failure is why many of us correct our own work. And not all photos are printed anymore anyway.
So if good photo color is important, learning how to correct White Balance is a necessary thing to know. Fundamentally important to good picture results, it can make a really big difference. I argue that since we never know the actual color of the light, setting WB in the camera (or using Auto WB) is merely an approximate guess, If Auto WB is "close enough" is not really debatable for critical users. It seems obviously better to see and measure and correct the results precisely later. It is not difficult.
This WB fix is easiest and best done if using raw images, where we have really good tools to do it, much better than the camera has. I would argue that actually seeing what we're doing is the overwhelming advantage. It is the reason to use raw images. Frankly, if you have a WB card in a session test shot (in the same light), then you could not care less about the WB setting in the camera. You have a raw image, and you're going to set it later anyway, and WB is simply not a concern yet.
OK, there are ifs and buts... I am no fan of Auto WB, but it is halfway good to use with raw, not because it has any effect on the raw image data (it doesn't), but only to see the camera rear LCD preview image, and for the camera histogram, which these are affected by camera WB (these features necessarily show a JPG embedded in the raw file, and JPG is affected by camera WB, and Auto WB is good enough for that, better than ignoring setting a proper choice). Otherwise, WB does not matter yet, there is no WB in the raw data. This camera WB setting does not affect the raw data. Instead of setting a hopeful WB in camera before the shoot, just simply set WB afterward, when you first view your shots, and can see what you're doing. There are several methods, and any try seems better than no try.
The problem with correcting white balance in JPG images is that white balance can be a substantial data shift which needs more than 8 bits. JPG processing does the white balance in the cameras 12 bits, and then creates the 8-bit JPG. That's fine if white balance was set accurately in the camera, but if not (and nothing except maybe direct sunlight ever can be precisely known), that 8-bit conversion is irreversible. However raw files are processed by always using the original 12 bit raw data, when white balance and other extreme data shifts (like Vivid) have less problem, so changes over much greater range is easier.
Flash, regarding Auto WB: There are two very different methods used by Auto WB, the one just above, and this one for flash. Flash color simply varies with flash power level (the spectrum of the ionized Xenon gas varies with amperes of current through the diameter of the flash tube). Meaning, no WB you can set in advance can be precisely correct. Nikon has a CLS feature called "Flash Color Information Communication". Canon has this too, in some models. If these flash and cameras can communicate through the hot shoe, then the flash reports a corrected "probable" white balance based on flash power level. And if camera mode is Auto WB (so that WB can change), then the image is set to that value of WB, instead of whatever Auto WB might compute from guessing at the image. This does Not affect Raw image data, but the Exif still does contain this new WB value. The flash does know it used high power or low power, so the designers programmed WB information that might match, which it reports. If you take a few flash pictures around the room (hot shoe flash using Auto WB), you will see high power at farther range reported warmer, but becoming cooler when closer, when power drops. But flash tubes do age and vary, the image color result still varies, even with constant Flash WB.
All methods using a white balance card are independent of the subject colors, instead substituting a neutral reference. The white balance is adjusted to make any color tint on the neutral card actually become neutral (equal RGB components).
Neutral: A neutral color is one with equal RGB components (no color cast). "White" is RGB(255,255,255), bright and neutral. "Black" is RGB(0,0,0), neutral. Black might work for WB, but it is really way too dark to show much color tint. "Gray" is the intermediate numbers, for example RGB(180,180,180) or RGB(127,127,127). But a neutral gray has equal RGB components, i.e., no blue or pink color cast as is commonly seen in gray. Then the idea of a White Balance tool is to make this known neutral color actually come out to be neutral in our photo image. That will be perfect color. Even if you might intend to intentionally add some color tone, this is a very good starting point to know what you're doing.
For this Custom WB, the WB card should fill the entire camera frame (exact focus is not actually necessary, it's about color). Then it lets Auto WB correct that view of that card (of only the card), in the same light that is on the subject. This works because the card is a known-neutral color (and in the same light as the on the subject). Then all subsequent shots will use this custom WB (until it is changed). The problem with Auto WB in general is that it has no clue about the actual color, but it works here using the neutral card, when any color cast is incorrect and should be removed.
Instead of the neutral card, a more expensive alternative is a diffusion filter placed on the camera lens (like Expodisc) for the Custom adjustment. The camera is aimed at the light source, and Auto WB neutralizes that light to determine White Balance (neutralizes the actual light, independent of subject colors).
This in-camera Custom method apples to shooting JPG, and it would be good for JPG since JPG has less range to readjust it later. The method seems fiddly to do to me, and to me, Custom WB makes no sense for raw, since raw data is not affected by WB. Raw editors probably could recover WB from the Exif and copy it into the final RGB, but the point of raw is to be able correct white balance after we can see what we're doing.
The card need not be very large in the scene, just enough to click on reliably (in the post processing image). It must be in the same light as the subject. The portrait subject can hold the card, or an assistant might go hold the card at the scenic subject. Then remove the card, and shoot away (in the same light), and we can fix it later (preferably using raw images). I use camera Auto WB here with raw, not because Auto WB is good or even affects the raw image, but for the only purpose to be a reasonable enough approximation just for viewing the cameras rear LCD view and histogram. The color will be fixed later. IF it will still be in the same light as the subject, we can sometimes just hold the card at arms length in front of the camera (clearest focus not required, it is only about color). Then later, the raw editor will have a good White Balance Tool, we simply click the card in the first test shot, which can simultaneously correct all frames in that same session (same light on the subject). Ought to be perfect color, no color cast. Accurate, easy as pie, and really good. See the WB Cards below.
White things are common in images, if you will stop and look and realize it. Not all whites are equal, off-color cars or white wall paint are not the best choice. But just click some actually white object (intended to look white) for generally good results, not fail safe, but it works often pretty good. Far better try than nothing, better than before. The white thing must be in the same light as the main subject. If you have no other plan or opportunity, why not try it? The worst that can happen is that you might have to click Undo or Cancel, but you may like it. If you like it, keep it.
Clicking the WB tool on the white thing or on the neutral card tells the computer "This spot is neutral, make it be neutral" (meaning equal RGB components, no color cast). White works, and gray cards work too, and 18% cards work, but are just designed to reflect 18%, and are pretty dark to show slight color cast well (and even black sort of work in some cases, if no other choice. Black and white are just extreme shades of gray, which implies no color tint). But neutral cards designed for actual white balance are either white or light gray. If we know the card is a neutral color, and if we click it for the computer to make it actually be neutral, then that white balance leaves no color cast in the image. And it works great (unless this card color was not supposed to actually be neutral). Exposure variations can still make it brighter or darker, and camera profiles (like Vivid) can change colored things too, but we can also see that result in raw, and easily fix that too. Just make it look like you want it to look, but precisely correct color relies on using a known neutral card. But a plain white paper or dish, or white dots on pajamas work well too, and a pleasing color result is good too. Far better than no try to fix it.
#1 and #2 are the cause of our problem, simply wishful guesses, not precise. For #1, we often don't know which precise WB to set, and so unfortunately, many of us frequently simply don't do anything (leaving white balance setting at whatever it might already be, which is no solution). Auto WB is an approximation, maybe better, sometimes, but I rule out Auto WB for any serious effort that matters. It's too easy to do it good with just a little attention.
Other than #1 and #2, the rest are good solutions. #3 is for JPG, #7 is for Raw, but #4, #5, #6 work for either. I use only 4 for setup studio-type portrait situations (it seems fail-safe). And I often lazily rely on 5 for general outdoor snapshots (and then 6 if that fails to show correctly). I shoot Raw, and batches of similar images in the same lighting can copy one measured white balance to the entire batch.
#3 can be precise, WB specifically determined from an accurate known reference in the same light. It might be more work for one picture, but it is a suitable method for a batch of JPG pictures in the same fixed lighting situation.
#4 is precise, and is very easy and convenient, especially in a studio situation. My preference is that #4 is the definitive white balance solution, and I use it with raw images, especially for indoor flash where the subject location is so easily accessible. To me, #4 seems mandatory for portrait sessions.
#5 can often be "quite good enough", a classic way to recover and fix things when there was no prior planning for it. When it is all you've got, it can be a magic solution. But it is no go if there is no suitable white color in the image.
#6 can be pleasing, a guess based on seeing it and preference, but it is not a measured precision like #4 or #3.
#7 is also an approximation, with outdoor lighting situations often easier than indoors (incandescent and fluorescent can be more difficult to match, and #3 or #4 might be better tries then). Raw allows very easy correction, specifically when we can see and judge the results to know what we're doing.
Frankly, walking around taking random sightseeing snapshots on vacation makes the best #4 or #3 a bit awkward, extra work for every shot, so depending on importance (OK, meaning without any preplanning), I do largely depend on #5, #6, or #7 then (fixing it later in raw). Shooting raw makes corrections easy later. But for something like a fixed portrait session, involving many important photos all in the same lighting and situation, we'd be stupid not to invest this slightest extra effort (#4, white balance card in first test image). You only have to try it once to realize how easy, and how fast, and how good it is. Really easy. Really really good.
My own preferences are clicking white things (#5, casual work with no preparation, but I do try to remember to look for any white object nearby that could be included in the frame), especially if clicking on a specific planned White Balance Card (#4, critical important work). Or even adjusting later by just looking by eye if necessary (#7 or #6). But for careful indoor studio work, it is overwhelming important to me to place the #4 known neutral WB card into the scene (at the subject, for first test picture). For outdoor snapshots, I keep an eye out for suitable white things (maybe even sometimes an extra shot to include a good one if in the same light), and there is also a WB card in my camera bag for important difficult cases encountered.
Intentionally including a planned neutral white balance card in a (test) photo (#4 above) is best for known full accuracy (more on next page). But many images do already naturally contain something suitably white (#5)... not off-white, but something intended to look actual real white. Maybe a T-shirt or a porcelain dish or white paper or a sign. Better photo editors have a White Balance Tool to click on something neutral in the image, and then the software makes that spot actually be neutral (which removes the color cast from entire image). The free Irfanview or Faststone editors don't have this white balance tool, but some editors (like the free Irfanview) will accept the free Photoshop White Balance Adobe 8BF plugin (which I have never tried, but these seem well liked).
Raw editors, for example Adobe Camera Raw or Lightroom excel in white balance, and they can also do JPG images too, also in a lossless fashion. For raw images, raw editors have the same tools the cameras have, plus more and better too, plus we can actually see what we're doing then, and can choose among different tries. And Raw software like Adobe Camera Raw (ACR, in Lightroom or Photoshop or Elements) has its own better White Balance tool (more modern tool, and with wider range - more on Part 2 page). However, since JPG images already have some white balance in them, we cannot now just specify Daylight or Cloudy (the camera already did that).
Alternately, the actual method that I advocate (especially in critical studio work) is to use a known accurate white balance card for corrections (in the first test picture, much of what follows here). If we fail to include that card, we can always expect white balance problems. Using an included KNOWN white card works best, but the point here is that utilizing whatever white things we find in the scene can be a big help then. Something actually intended to look actual white. But do realize that many things are intentionally off-white, while others may be intended to look really white. Again, maybe a white envelope or white piece of paper, or a table cloth, or a porcelain china plate or dish, or white plastic bottle cap, or a sign, or a church steeple, or a picket fence, or a white T-shirt, or a shirt collar, or the white polka dots on the kids pajamas, etc, etc.. If you see white, consider it for White Balance.
I am NOT saying clicking everything white or gray works well, since there are many tints of white or gray which don't work. Walls are not a good bet. But many pictures include white things that are intended to look neutral white, and do work, or at least well enough to get you closer. Maybe white clouds if not overexposed from clipping. Small dots on the kids pajamas can work. Even whites of eye balls are not a bad try, so do stop and look for something white. The white thing must be in the same light as the main subject. Try a few different things, and click on a few different places on it. You will see the one you like, and will get much better at it real soon. A big improvement may not be a total fix, but it can often be much better than no improvement. Getting it close first helps when tweaking by eye too. When adjusting manually by eye, go too far one way, then too far the other way, to see the span of the middle, to be sure you see the good color pop out at you. Try both sliders. There will be a good spot.. which might be pleasant instead of accurate, but that's good too.
When serious about good color, there are inexpensive White Balance Cards which can be added to the scene in the first test shot (in same light), which provide known accurate color comparisons. Using the known source, then you know it will work, very accurately. Very convenient for fixed studio sessions with many photos under same lighting, but it works anywhere (insert card into first test shot, then remove it for final shot, and apply same correction to both). An 18% gray card works reasonably well, but it is designed for exposure (and makes no claim about neutral white balance). Its gray color is less well controlled, and it is also pretty dark to show white balance. There are "white" white balance cards, and there are "digital gray cards" of a much lighter shade of gray which are supposedly controlled to actually be neutral, and these are designed for white balance purposes. There are many such devices sold, and some seem overpriced.
Note that White Balance has two components, Temperature, a yellow to blue slider, and Tint, green to magenta (see more on next page). These are simply the two axis of the CIELAB Lab color system (Wikipedia).
These next examples are using the White Balance Tool in Adobe Camera Raw. The purpose is to compare accuracy of white things. Note that clicking on these things is very easy, not rocket science at all, anyone can do it. :) We can see it then, and all we have to do is to choose the best resulting color, judged by the neutral WB card.
The WB numbers below are the reported WB result (Adobe ACR shows them for raw images). Then using that WB, the White Card RGB color is shown. It being a special neutral card, its correct RGB components should be equal (neutral, no color cast). The RGB Spread is the difference of the largest and smallest RGB components, i.e., how much Not equal that result (to help sort the numbers).
The first three lines of the table below are the methods called #1, #2, and #4 above. All the following lines are method #5. Perhaps you don't see much difference, and don't really care. But if you do want correct color in your images, it's not a hard effort.
Examples (clicking white things in your images. A known WB white card is the most reliable.)
|Clicking White Things||White Balance|| White Card|
|Camera Incandescent WB||2850K Tint +0||(202,194,127)||75|
|Camera Auto WB||2850K Tint +22||(199,189,164)||35|
|White Balance Card||2650K Tint +24||(189,189,189)||0|
|The plastic cabinet||2750K Tint +22||(194,190,173)||21|
|Window shutter behind clock||2550K Tint +24||(184,188,197)||13|
|Cap of plastic bottle||2650K Tint +24||(190,189,187)||3|
|Painting above it||2600K Tint +25||(187,189,193)||6|
|White paper in basket||2600K Tint +20||(188,190,184)||6|
|White plate under Card||2600K Tint +26||(187,189,195)||8|
|Cup under Card||2650K Tint +26||(188,189,195)||7|
|Label on blue lid||2700K Tint +26||(192,189,185)||7|
|Rectangular plate, left||2650K Tint +26||(190,189,190)||1|
|Plastic oval lid, right||2650K Tint +28||(189,188,194)||6|
|Small dish above it||2650K Tint +26||(190,189,190)||1|
This example above was lighted with only the overhead room light, which used two 13 watt compact fluorescent bulbs trying to simulate incandescent temperature. Incandescent WB was not correct, but Warm White Fluorescent WB seemed worse. We often really don't know what white balance to set.
Correcting on the added known White Balance card (center) was the best result, essentially automatic, eliminating those concerns. But clicking on ANY of these other white areas was nearly as good — often 2600K or 2650K, and Tint often +24 or +26. All were better results than Auto or Tungsten. The chart shows their result, and the resulting White Card RGB color. White or gray is Neutral when the three RGB colors are equal (no color cast). Compared to the first two lines, the other chart results here are very nearly neutral. This assumes the selected white actually OUGHT to be neutral color. Clicking an off-white wall may not be a great choice.
The point is, objects that are actually white are often present in the scene, and if no white card was planned, you really would like clicking any of these items much better than simply selecting Incandescent or Auto.
Daylight WB is simply wrong here (blue). Cloudy WB helped this one. Do you set cloudy when needed? Actually, this one seems cloudy open shade, but actually is Raw, WB adjusted later when result could be seen.
Clicking WB on the white water splash (at red X) is about right (make sure any clicked white is not clipping however). The big difference here is that it reduced the green cast significantly. Not bad at all, it has pluses, but I chose Cloudy WB.
The back of the white shirt (red X) ought to work, but it's in the shade, so it matched that shade result. So specifying Shade WB directly was about same. The clicked white needs to be in the same light as the rest of the image (but this time was still better than Daylight).
This is a RAW image, and I'm playing in ACR (Adobe Camera Raw software, Lightroom, Photoshop, Elements), but they can do same on JPG files too, maybe with less extreme range than Raw allows.
These are multiple choices, to look at and consider and chose after you can see the result. So when White Balance is giving trouble, look for something white in the image. However random white objects are certainly not guaranteed — if that object's true color was not actually neutral white (or gray), this can cause another color cast — maybe mild and still better, but maybe not. Even if not perfect, you can see the resulting color, and can try others around that value a bit. You would like "accurate", but you are looking for "pleasing". Often easy as pie, it just works. Sometimes not always so easy — look for other spots then. And try a few clicks around the same spot, to see if one is best. Yes, this may be hit or miss, and yes, certainly we can fault this, explaining how it does not always work, but yes, it can often be just the thing for pleasing results (consider the alternatives). If your problem picture has some color cast, and you realize there is also something white there, then why not click it? In tough situations, you can be very pleasantly surprised — but maybe not always. Then you can always Cancel out, or Undo. We don't always realize the color is bad, until we see it better, which then may be a Wow experience. There's nothing like correct white balance. You ought to go back to your old pictures, and try clicking the WB tool on some white area, just to see how this works. You definitely should try this for yourself. This is a very mainstream procedure.
It is a good plan to start with a monitor calibration system, to be sure you see things right. I still use an old DataColor Spyder 2 Express (most basic version), which seems more than enough. It is a light sensor that reads known colors and intensities on the monitor screen, and adjusts video response so that they display as expected. I bought an inexpensive Dell IPS monitor, and it's easily the best I ever had.
The Raw White Balance eyedropper tool is better than the Levels tool (wider range), but the Levels eyedropper works fairly well for minor shifts. I tend to say a white object, white seems more common, but it can also be gray, so long as it is actually neutral color — with no color cast in the original itself. Neutral colors are black, gray, and white, those pure tones with equal RGB components (no color cast by definition). Black is RGB (0,0,0) and White is RGB (255,255,255), which are just extremes of gray. Gray is any intermediate tone of equal RGB values, for example (100,100,100). But not every gray object is fully neutral color, blue and pink tints are common, which would not have equal RGB components. Even near black can possibly work here, but there is not much difference possible in channels at value zero. There is more difference detectable at higher numerical values, the same percentage of color cast is easier detected.
The card needs to be known to be an actual neutral color. Neutral gray is a special colorless color, and white and black are special cases of gray brightness, but neutral specifically means a color with equal red, green, blue components. Certainly there are off-whites, and also blue or pink grays, but equal RGB is neutral (no color cast). We click this card in the image, and the White Balance Tool makes that spot actually be neutral (equal RGB), which removes any lighting color cast (allowed since we know that spot on card ought to be neutral). White is much brighter and shows any cast much better than black where it may be hardly detectable (near zero). An 18% gray card can work (if actually neutral), but it is a dark color, not bright.
In a fixed studio session, or even walk-about bounce in the same room (any situation with several pictures with same lighting with the subject), a truly great solution is to simply include a known true white card in the first test picture (in the same light). I normally like to use this Porta Brace White Balance Card, 5x7 inches, near $6 at B&H or Adorma (both have reviewer comments). Plastic, durable, washable, inexpensive, accurate, quite good, and it is all we need to make White Balance be easy. Buy a couple and cut one to shirt pocket size. Or a Whibal brand white balance card is also good ($20 or $30, the small one seems adequate to me, at least at indoor distances). The WhiBal brand card is technically perhaps more accurate, since each card is said to be individually tested for neutral color. However, it's a little cooler color, and I in fact prefer to use the Porta Brace card. Or when it's the only choice, I'm also glad to use plain copy paper or white envelope. Pictures in bright direct sunlight don't have much issue with white balance, but otherwise, any kind of planned session surely can include the right kind of card in the first test shot.
Here are a few white balance "cards", and when using Adobe ACR (to correct WB of this Alien Bees studio flash light), I measure corrective WB to be:
|foam board||5050°K +2 Tint|
|5700°K -8 Tint|
|Fun foamy||5750°K -1 Tint|
|#10 envelope||5700°K -5 Tint|
|Porta Brace 1, $6||5650°K +1 Tint|
|Porta Brace 2, $6||5650°K +1 Tint|
|WhiBal 1, $20||5550°K +1 Tint|
|WhiBal 2, $30||5550°K +1 Tint|
Hard to be very specific, because clicking around at different places on any of them will often vary the reading at least ± 50°K and/or ± 1 Tint, but which is only one step of the possible adjustment. Those are very slight differences, and any click is good enough, with very nice improvement. I've listed some of the most common numbers, and the Porta Brace 5650°K +1 Tint choice is shown here. Percentage-wise, 100°K is a big change at 2800°K, but it seems near nothing at 5500°K.
Cheap cards likely have no manufacturing color control, and make no claims about color. I'd guess white must be easier to control in plastic than colored gray ink. And alternately, even plain white printer paper (cheap copy paper) works fairly well too, perhaps better than expensive papers with added brighteners which can be cool (blue). The cheap copy paper or envelope or business card are pretty good, and certainly far better than no try at all. If no card, look for something intended to be true white already in the picture, a shirt collar or an envelope or a sign or a porcelain dish, etc. These are good, certainly vastly better than nothing (but avoid off-white walls and trim paint, anything not intended to be true white).
These different WB cards have different manufacturing philosophies. The Porta Brace card is just white plastic. It has no color pigments to be controlled and checked, and is very inexpensive. The color probably is titanium dioxide (the "whitest white") used in plastics like PVC pipe, and surely cranked out the same way. The WhiBal card adds light gray pigments to the plastic, to be a "controlled color", which color then has to be controlled, but they claim to measure and verify color of each card, to ensure it has no color cast. This costs more. But this is about minor variations though, and the $6 Porta Brace card works great too. I have a couple of each (Porta Brace and WhiBal), all are fine, they compare not precisely identical, but well enough. THe WhiBal card claims to be carefully controlled and tested, and it is a bit cooler than other choices. Which conceivably could be very correct, but I normally use the Porta Brace card, which is nothing special, it's just a good white color, very adequate and I really like the result.
The 18% dark gray cards also do work pretty well for white balance, but 18% is about exposure, not color, and different brands are variable with respect to color. They use gray ink, primarily designed to reflect 18% of the light, and possibly do, but without heroics about color control (no color specs are ever mentioned). Better ones do work for WB, usually well enough, but will vary a bit more, and 18% is really too dark for WB (dark does not show color cast so well). Actual white balance cards are white or light gray. For white balance purposes, I would suggest either the Porta Brace or WhiBal cards.
FWIW, I didn't like any of the cheap three card sets I have tried (Vello or DGK, which are three black, gray and white cards, probably usable, but all three colors have different WB, and are different from the next set too). The classic problem with having more than one measuring instrument is then wondering which one is correct? :)
The philosophy of use is that if we photograph the neutral card, it will take on the color of the light shining on it, and then when the WB tool clicks it, it makes R, G, B values of the spot clicked be equal again (neutral) in the RGB system (which is also a and b zero in the Lab system). If we click on a known neutral item, then presto, no color cast in our image. It doesn't get any more straight forward than that. The card does have to be in the same light as your subject, which means placed with the subject. The card can be placed at the edge, cropped off and out of view. Or raw makes it easy to include in only the first test shot, and then use that value for all subsequent images.
There are other fancy complicated expensive systems, typically requiring additional setup at each use, but nothing works better, or is easier, than the $6 white card. You click on the card in the first test image to tell the White Balance tool "This spot is neutral. Make it be neutral". And Presto! Whatever color cast there is removed, also from the entire image area (which assumes that the subject and background have the same lighting). The color of the clicked spot is adjusted to force equal RGB components there, which is neutral. Then remove the card, and the same color adjustment will also work when transferred to all similar pictures in the same lighting situation (raw editors make that be trivial too, which you can use with JPG). If you can include the card (or use something true white already found in the scene), it makes white balance be a totally trivial issue.
The Levels tool gray eyedropper (above) only works on that one selected image, but at minimum, it shows you what good color looks like, a visual goal which you can then work to imitate in the other pictures. Raw editors let the one click apply to several/many selected images (selected to be in the same light). Or instead, if only the other standard photo editor color tools are available, start with the blue-yellow slider — if too blue, then try less blue, etc. Raw software is better, offering more features, specific to camera images, including actual White Balance tools, and for example, is good about copying White Balance to multiple images with one click. Also the Raw White Balance tool adjusts both color temperature (blue to yellow) and tint (green to magenta). I usually use the White Balance tool in ACR (Adobe Camera Raw, which is in Lightroom, Elements and Photoshop). I can select all the images in the same session (using the same lighting), and click this card spot in the one test image, and the same White Balance correction is applied to all selected images at once, one click. Again, Presto! Very trivial job, but perfect results. We don't even have to think about White Balance then. It's like magic, simple and easy.
Two Rules for Using the White Balance Card:
Gray cards: 18% cards can work for white balance, but they are pretty dark because they are intended to have a reflection of 18% (for exposure purposes). They really don't make neutral gray color claims, but most are fairly close. See more about gray cards. This 18% value was believed to be what the human eye sees as middle gray, and ink press printers used the card to calibrate their ink in the old days. You should realize that other than this notion, there is nothing 50% about the 18% card, and digital sees it as 18% (when linear, before gamma). So in the 1930s, 18% was popularized as being visual middle gray by Ansel Adams in his Zone System, which was about visual results seen by eye in prints. There was no digital data back then, and 18% is NOT 50% of anything. Light meters are instead calibrated to 12.5% (Sekonic, Nikon, and Canon are, but Minolta and Kenko are 14%. None are 18%.) Kodak always said to open 1/2 stop if metering on their 18% gray card (which converts to 12%). Which again, is about brightness of reflection, and the ink is not specified to be neutral color. The 18% cards are pretty dark for white balance, and today the newer "digital" gray white balance cards are a much lighter shade, using color controlled inks for this White Balance purpose. White is very good for White Balance too.
Incidentally, we can still buy "Kodak" gray cards, but Kodak has not manufactured gray cards in over 25 years. Kodak sold their printing business (includes books and 18% cards) to Silver Pixel Press around 1995, who sold it to Tiffen in 2000. Both have sold "Kodak" branded cards, and Tiffen also makes Kodak Wratten filters now. But any new "Kodak" 18% gray card was not made by Kodak, and 18% cards have not specs to necessarily be exactly neutral (they can vary, but most are usually pretty close). If your goal is white balance instead of 18%, you can get actual white balance cards, with the specific goal to be neutral color. Actual white balance cards are either white, or very light gray.
For JPG images, your camera may have the Custom White Balance feature, where you photograph something known to be neutral white or gray, a close up of a white or gray card held in the same light with subject. Or instead of a neutral card, you can use a strong diffusion filter on the lens, like an ExpoDisc, and aim that at the light source. The camera Custom WB procedure uses Auto WB to adjust the color of that light to be neutral (no color cast). The diffusion sort of makes a gray card from it. Pringles chips used to use a frosted can lid that was said to be really good for this (but lid has since been changed). :) Either way, card or diffusion filter, this Custom plan seems good, because it does not depend on knowing the actual color of the light. It makes no actual photo. Again, Custom White Balance is for JPG, since raw images contain no White Balance correction.
But IMO, there seems lots to be said for a WB card simply placed in the scene for a test shot, it is an easy way. A camera-based Custom system requires you stop and do this right then, right there, every time you make a lighting change. Just sticking a white card in the scene for the first shot takes little effort, and then you can deal with it all later, leisurely. Frankly, I don't even do that in the general case outdoors. Direct sunlight is constant, so not that hard. Cloudy or Shade WB can work well, when we can see it and experiment with it. Also there is often already something pretty much white in many scenes. Indoor sessions however are significantly harder, many types and colors of lighting exist, even speedlight flashes vary from each other, and the white card makes it easy, quite important to me.
Continued - Please visit Part 2 - Lots More about using the White Balance card