How does the white balance work?

Duckys54 · January 31, 2007

I've tried understand it but I don't get it. How does it work? What does it do? Why do you need it?

Phyxius · January 31, 2007

Light is measured by temperature, a unit called Kelvin. I did a quick google search to get a chart for you. Click Here. Going off memory this looks accurate, I've never memorizes the light temperatures.

If a picture is warm it generally has a orange tint, conversely if it's cool the image has a blue tint. I see in your signature that you shoot film. Film camera users would rely on filters. Digital users can set white balance in their camera. Different types of light have different temperatures (as you can see in the above link), so setting a preset white balance allows you to customize the "tint" of your image. (Nikon calls manual white balance "Preset" as you set it BEFORE you take the picture. Most camera manufacturers refer to presets as a group of predetermined white balances. For example, incandescent or tungsten light bulbs that are commonly found in homes create an orange color cast. So if you set the in camera white balance to tungsten the camera will add a blue tint, balancing the colors. This is much the same as if you had placed a blue filter over your lens.

It's important so that your images don't have color tints...think of seeing the world through rose colored classes....or blue glasses....or orange glasses, etc.

**Also, to add to all that, many of today's cameras also offer RAW files. These are uncompressed, unprocessed images. What that means is that the camera does not process the image, it records it exactly as the sensor sees it. You can then edit the white balance using a RAW editor such as Adobe Camera Raw (ACR). In my opinion the ACR allows better, more accurate WB than setting a manual WB. Lights tend to work in cycles, especially flourscent lighting, and I can view an image much larger on my monitor than I can on a viewfinder. But, that's me.

LiquidAir · January 31, 2007

A white object (say a sheet of paper) reflects red, green and blue equally well. However, your camera doesn't acually see the paper. Rather it sees the color of the light reflecting from it so that sheet of paper takes on the same as the color of the light shining on it. If the light is reddish, the paper will look reddish. If the light is bluish, the paper will look bluish.

Normally in a fully color balanced digitial file, we want white objects to have equal level of red, green and blue. However in your capture the color of a white object is going to be the same as the color of the light. White balancing is the process of adjusting the color channels of your capture so white objects in the scene have equal levels of red, green and blue.

The most common example of this is the difference between incandescent light and daylight. Standard tungsten house lighting has quite a bit more red and less blue than daylight. So if you calibrate your color so daylight looks white, your house lighting will look quite orange. Correspondingly, if you calibrate yoru color to make house lighting look white, daylight will look blue.

Commonly the color of a light is described by a color temperature, e.g. 2800K for 60W house lighting, 3400K for halogen lighting and 5500K for noon time sun. The temperatures used to describe light come from the physical theory of hot lights: the temperature of the light determines the color. As the world moves away from inefficient hot lights to more exotic technologies (flourescent, mercury vapor, sodium, LED, etc) color temperature becomes less useful as a way of describing a light color. What you most often see in software is actually two adjustments: a yellow-blue and green-magenta corresponding to the LAB color shifts.

claudermilk · February 1, 2007

To split hairs, what you mean is a neutral object reflects equal amounts of red, blue, and green light. The common 18% gray card does that & gives you a middle tone to work off of between white and black. Slightly OT, many of the really bright white papers are actually bluish-toned as we interpret that as whiter, so isn't a perfect WB target.

Essentially what we are doing is trying to get the camera to see the light as if it were 5500K or 6500K (daylight) since that's what we perceive as normal. So setting WB in a digital camera is the same as swapping film types to account for the light temperature (e.g., putting in tungsten-balanced film for tungsten light, or adding color-correcting filters for fluorescent light).

wmstumme · February 2, 2007

Duckys54 wrote:

... Why do you need it?

I had exactly the same question, and just skipped through that part of my manual. Then I hit this horribly lit gym. Understanding custom white balance was the only thing that let me get any reasonable shots.

http://www.dgrin.com/showthread.php?t=50897

I only learn by making a lot of mistakes (I guess I get a lot of learning opportunities...)

Regards

Will

colourbox · February 2, 2007

Here is a quick white balance exercise that should put all the theory into reality.

Go outside during the day. Take the same photo twice, the first one with your camera set to Daylight white balance and then set to Tungsten white balance.

At night, go inside and take a photo of a room in your house. Again, take the same photo twice, first set to Daylight white balance and then set to Tungsten white balance.

Then compare all four with their settings.

Bob&Glennie · February 3, 2007

If I read correctly Trevor is a film shooter. So, to put it terms that you'll find familiar, White balance is like changing film for different types of light. Just as you would use a "Daylight" film for shooting outdoors on a sunny day, so, we set our white balance to "sunny" so that the camera responds to that color temperature accordingly. If it's a cloudy day chances are you'll use a "daylight" film with an 81a ... we can fine tune our white balance the same way.

If you were shooting under tungsten light you could use a tungsten balanced film or you could use a daylight film with an 80a or b. We can also change our white balance to suit.

In practice, if documentary accuracy is the goal, a digital camera white balance can be be changed or fine tuned to accommodate almost every situation we get ourselves into. Digital shooters no longer need to carry correction filters and if we want to get funky we can use the "wrong" white balance to simulate using colored filters for film. For example, I can use White balance for tungsten under daylight sky or with my flash to simulate using an 80b with daylight film.

hope that helps
Bob

jatz · April 19, 2008

white balance
hi im new to all this photography aswell but loving every minute of it the one thing that is stumping me is white balance. i have a panasonic lumix dmc-fz50 and i find it a good camera for now. but on it it has two different white balance settings it has whit balance and then it has white balance adjustment.i have read this form and understand white balance as far as daylight settings and that but i dont get the white balance adjustment it has four colours in it orangeish, buleish, greenish and redish. all i want to know what is the use of this setting well thanks and i hope someone can give me the right answer to understand this stuff thanks take it easy bye

Photoskipper · April 20, 2008

I alway go to www.about.com to find the answer.
Just make it easy to understand. The visible light is EM Wave. The color of the object depends on what color is the light source. We use K as the unit to describe the light-color.
The lower is warmer, such as tungsten light is yellowish. The higher is cool, such as the blue light.
the camera just a machine and need someone to tell it what is white light source then it automatically adjust the object color accordingly. during the film era, we use the filter to correct the light going into the lens then the film. For digital, we can do it by turning the dial or press the white balance and let the sensor do the job.
For quick reference, good light under the direct mid-day sun is 5250K, Flash gun is 6000k, Tungsten light is 3200K. Shade is cooler so it 6000K. During the cloudy day, only the shorter wavelenght light can penetrate, so the light is even cooler - 7000K.
You can go to the car accessory shop and check on the head lamp bulbs demostration. The original headlamp usually about 3500K, the halogen is up to 4000K and looks more white. But HIB can reach 6000 which is pure white or even a bit blue. Take a color picture and put in front of those lights and you will find the difference.

Martyn · April 22, 2008

Is there a way to measure the light temperature using a dslr?

ziggy53 · April 22, 2008

Martyn wrote:

Is there a way to measure the light temperature using a dslr?

If you have a custom WB capability in your camera, most do, capture a known white target with the camera and then set the WB to the target using your camera's procedure. Some cameras will even read out the apparent color temperature of the custom WB in the camera.

Otherwise ...

Plant a fairly large white target centrally in a scene, set your camera to AWB and snap the scene in RAW mode.

Review the image in a RAW image processor that allows both custom and camera (As Shot) WB. Both WB should agree if you measure the white target and set that as your custom WB.

If you have nothing else at your disposal, RawTherapee has both a "Camera" WB and a "Spot WB" tool which seem to work well.

Measure a couple of known light sources like a fresh tungsten filament light bulb (3200K) and a manually fired electronic flash (5500K generally). (Don't use an attached flash because the camera may sense the flash and set itself to what it thinks should be the correct WB.)

Try not to confuse the issue with multiple color temperatures at first. Also no filtered or shaded light, it should be direct from the source to the white target. Do not spot measure a region which shows RGB levels above 250 in any channel.

As a "for instance", I just shot a white target (sheet of inkjet paper, HP Multipurpose, 96 Brightness) using a GE, 60W, soft white bulb and measured 2942 with the spot sampler and 4151 with the camera setting from AWB.

(The discrepancy is because Canon cameras typically do not do well measuring tungsten WB and I'm not sure why. Also a 60W bulb will not burn as hot as a 100W bulb. I just couldn't find a 100W bulb to test.)

An electronic flash (Sunpak 383 Super) with the same sheet of paper measured 5794 with the spot measure and 6300 from the camera. I am actually surprised that these two figures are that far apart.

The test camera was a Canon 40D with the Canon EF-S 17-55mm, f2.8 IS USM.

I am investigating further.

Photoskipper · April 22, 2008

Martyn wrote:

Is there a way to measure the light temperature using a dslr?

Most camera do not provide the color temperature reading in K, (I believe there is none), They usually let you set it according to the light source such as tungsten, daylight, shade, cloudy and ????.
But I usually don't care too much about the color temperature, just set it to auto and shoot in RAW. Come back and tune it if necessary.

I am not sure whether it is true to use the white card for the color balance test. There are very expensive "calibrated grey card" on sale for white balance.

Sometimes, the wrong Color temperature can get some funny result - I call it creative

claudermilk · April 22, 2008

Don't get too hung up on the numbers. I know of no DSLR that will tell you what the numbers it's reading are, the only way I am aware of getting that is a color light meter--which is expensive & more for cinematography. The only time I ever pay attention to actual Kelvin numbers is when manually setting for theater lighting since I know what those lights output. Otherwise, I use a WhiBal card (any digital WB/grey card will work) and set custom WB.

NeilL · April 22, 2008

As well as the great info that others have written above, my understanding, possibly flawed, is that WB is the electronic photographic way to do what our eyes and brain do automatically.

Imagine that you get a sheet of white A4 paper and carry it around to a variety of locations, in the sun, in the shade, in an artificially lit room, and wherever. Probably you will see and describe the sheet as being white in all those different light conditions, a kind of persistence of perception of the whiteness of the sheet. Your perceptual system adjusts in the different lighting conditions to give you unchanged information about the sheet, that is that it is white. This is referred to as color constancy.

However if you measured the light reflected from that same white sheet in all those different light conditions your readings would tell you that the sheet is not white, but the colors of the light reflected off it in each of the different locations.

In the same way, your camera will see the same white sheet as this color or that, depending on the color of the light it is reflecting in a particular light condition. In the photograph the "white" sheet will be that color, which you immediately recognise as being not white and therefore "wrong".

WB restores the quality of whiteness to the sheet, so mimicking the color constancy which your eyes and brain produce in actual live vision, and making the white sheet look "right" under the different lighting conditions. When you select a WB the software adjusts anything in the picture whose native color is white to look white in the photograph. All other colors in the photograph are simultaneously adjusted, removing the tone cast by the light source in the photograph. To our eyes, the photograph better corresponds to what we saw because the camera has now made an adjustment to tone which parallels the adjustment our eyes and brain made in that situation.

18% gray can be used as the reference as well as white, and you can manually stipulate the color value (temperature) of the reference, as others have already described.

I am ready to be corrected

Neil

claudermilk · April 23, 2008

At it's root, WB is saying a neutral-value tone is R=G=B. Any gray or neutral will work, so long as it's truly neutral (18% was for setting exposure in the film days before AE). This is why I don't use all the old cheapskate tricks (palm of the hand, Pringles lid, copy paper, coffee filter)--I don't trust that they are truly neutral. A proper, calibrated neutral target such as the WhiBal, Expodisk, etc. ensures that I am really set properly. Just an expansion on Neil's last comment, the rest AFAIK is accurate.

ziggy53 · April 23, 2008

I ran some more tests, this time with the Canon 1D MKII and the Canon EF 50mm, f1.8. Table, hopefully, below:

Flash and Flash WB (Sigma 500 DG Super attached to camera)
Default WB.....Custom WB
ACR_6100......ACR_7250
RSE_7650......RSE_8250
RT__6088.......RT__7116

75 Watt light bulb - AWB
Default WB.....Custom WB
ACR_4650......ACR_2900
RSE_5100......RSE_2950
RT__4737.......RT__3036

75 Watt light bulb - Tungston WB
Default WB......Custom WB
ACR_3300......ACR_2900
RSE_3400......RSE_2950
RT__3270.......RT__3005

Abbreviations:
ACR=Adobe Camera RAW
RSE=RAW Shooter Essentials 2006
RT=RAW Therapee
AWB=Auto White Balance (camera)
Default WB=RAW converter default values
Custom WB=RAW converter re-sampled values

All numbers in Kelvin

ziggy53 · April 23, 2008

Comments about the tests,

It does appear that the paper I used is potentially reacting to the flash light. This is probably a UV brightener. It does look like I will be purchasing one of the Kodak cards to test.

The results from ACR and RT seem to jive pretty well.

While DPP doesn't produce WB numbers, so I could not show anything in data, I was surprised at how "poorly" it handled tungsten sourced images. From my tests I do not recommend DPP for either mixed WB or tungsten lit images.

These results are quite different from tests that I ran years ago between a Kodak Gray Card, using the white side, and ink jet paper. Those tests showed that the two different white targets imaged very similarly in the studio (Kodak DCS460 dSLR and Bowens studio monolights.) As a result, I never bothered purchasing a genuine Kodak gray card for my own purposes.

Possible complicators are:

Different inkjet paper
Different dSLR
Different light (especially different flashes)

claudermilk · April 24, 2008

Yes, those papers have brighteners & lean towards blue. That's why I don't like using them--they aren't neutrtal & a bad target IMHO.

ziggy53 · April 24, 2008

I will soon see just how bad different papers are. I ordered the Kodak R-27 packet which includes (hopefully) a more photographically neutral white/gray surface.

I also ordered the Canon "Digital Color Management Guidebook" which has a mini gray card as well as a color chip chart. For $5 it should be well within the reach of anyone, if it works.

My concern now is that the small flashes that I uae are putting out too much UV. This is something that Baldy suggested some time ago. If the inkjet papers do have a strong UV brightener, it should be possible to do a comparative test of different flashes once I know how much extra bright the paper is over a calibrated test target. (If the Kodak target really can be assumed to be calibrated as neutral.)

I will also use the tests to try to determine if other common papers "are" suitable for photographic white balance. A goal is to have a valid test target that I can tote around in every bag and case that I own and not have to worry about losing it or getting it dirty or whatever.

ziggy53 · April 25, 2008

OK, I got the Kodak cards this morning and just tested:

Equipment:
Canon 1D MKII, EF 50mm, f1.8 lens and Sigma 500 DG Super flash, everything manual mode.

With Kodak card in the middle and white side showing:
AWB default ......6750
Kodak card .......7150
Inkjet paper ......7500
Foam core board 6000

With Kodak card in the middle and gray side showing:
AWB default ......6450
Kodak card ........6050
Inkjet sheet ......7500
Foam core board 5900

Notice how consistant the foam core board is and how it compares to the gray side of the Kodak card. (I don't know the manufacturer of the foam core board.)

Also notice that even the Kodak card white side is (apparently) responding to the UV in the flash, but not as badly as the inkjet paper, which is also remarkably consistant (but apparently "cold").

Moral: Yes, the Kodak gray card is apparently a very good measuring tool for WB, especially using the gray side. It also looks like I need to find a way to filter my flashes' UV output.

I will cease using the inkjet paper for WB setup and start carrying the Kodak cards.

I plan on testing more of my equipment to see how the different cameras respond to different combinations of light etc,m but I probably won't post the results unless they are truly universal and interesting.

NeilL · April 25, 2008

Many thanks ziggy for a very enlightening test series!

Neil

Martyn · May 4, 2008

claudermilk wrote:

Don't get too hung up on the numbers. I know of no DSLR that will tell you what the numbers it's reading are, the only way I am aware of getting that is a color light meter--which is expensive & more for cinematography. The only time I ever pay attention to actual Kelvin numbers is when manually setting for theater lighting since I know what those lights output. Otherwise, I use a WhiBal card (any digital WB/grey card will work) and set custom WB.

I was interested purely out of curiosity. I don't have a practical application for knowing the numbers.

pathfinder · May 4, 2008

Ziggy,

I don't know if you saw my thread in the Technique forum about white balancing. I have started a thread about white balancing tools - including the Expodisk and the SpectrSnap filters, and a gray reflector from BalanceSmarter.com.

I like the incident tools like the Expodisc and the SpectraSnap, but the BalanceSmarter reflector compares very favorably also. It is a matte gray surface without any specular reflections ( I think this is a good thing ), and has a white line that allows one to not have to turn off AF ( you must turn off AF for the Expodisk or SpectaSnap filters )

If your camera has an RGB histogram display ( my 40D is set up to display an RGB histogram), a proper exposure of the BalanceSmarter reflector, should have three spikes in the center of the histogram one on top of the other. If the spikes are not centered, the exposure was off, and if they are not on top of each other, the surface was not really gray. One then uses the jpg captured to calibrate the camera for a custom white balance. It works very well - calibrated targets shot with this custom white balance yield a true white, gray or black image.

Andrew Rodney says the Kodak 18% gray card is a poor choice for white balancing - the card was really printed for exposure, not color balance. It may not be a truly neutral gray. I will include one in my further discussion of white balance to see if my KODAK 18% gray card is a true neutral gray when I return from travelling next week.

My interest was sparked by a thread by Icebear about cheaper alternatives to the Exposdisk - coffee filters, Pringles lids, etc. The conclusion of this post was that the cheaper alternatives were worth even less than they cost. There is a reason to use a true color neutral filter. The folks that make the SpectrSnap filter seem to agree, as they have just updated their filter with a newer, better balanced version.