# Problem in white balancing photographs taken in sun

When a photograph is taken in sun, the whole photograph becomes slightly orange in color. Specially the skin tones become too orange. The photographs are taken from simple mobile phone high megapixel camera. So, when I try to white balance during post-processing, correction made on skin tones leads to slight tint in other objects. I would like to explain this with an example.

Say I have a photograph having a person standing in front of some grayish stone, clicked around 3 pm. In the photograph, initially, the person looks too much orange in color, while the stone is neutral gray. I am telling you this because I check RGB values with eye-dropper tool. Now, to decrease orange in the skin tones, I drop the red curve slightly more than I drop green at appropriate luminous levels respectively. Now the orange cast is gone, but when I inspect RGB values of stone, it is more towards blue, which is obvious. So, I try to adjust the color of the stone to make it neutral back again. But, the image and consequently the person becomes orange in color more or less as he looked initially.

So, is it not possible to white balance a photograph in this scenario, with the help of a single curve adjustment layer, without using masking? This raises other questions: does our eyes (more precisely brain) white balance different objects in a single scene separately? In that case, it should not be possible without masking, I guess. Or is my camera faulty?

• No, our eyes don't "... white balance different objects in a single scene separately?" Our brains do, though! – Michael C Sep 17 '14 at 0:49
• Please post an example photo. – Philip Kendall Sep 17 '14 at 8:11
• " ... eyes and brain tuned ...". It can be argued that the actual colour changes when the irradiating source changes, as the response of the various target components to various wavelengths change. When we try to "colour balance" the image we are attempting to make it look as it would with illuminated with a standard source ("tungsten incandescent") and in fact if we look at the scene at the time the photo was taken it does not look to our eye as it would if we illuminated it with tungsten. Why should we be able to or want to modify the image in ways which do not represent reality. ... – Russell McMahon Sep 17 '14 at 8:19
• ... (And, yes, I know we often do want to). Light at various colour temperature may fall on the black body/ Plankian locus and therefore be "white" but there is white and there is white. As I know we all know. – Russell McMahon Sep 17 '14 at 8:20
• @PhilipKendall, I have uploaded an example photo. – user2118622 Sep 17 '14 at 9:18

First, the color balance looks plausible in the picture you post, but of course I wasn't there and I don't know what the guy's skin color really is.

Second, stone is a bad gray reference unless you have specifically measured it. I think your basic mistake is assuming the stone is supposed to be gray. Since you give not justification for that, I'll assume you just made it up. When a scene doesn't have a known gray or white object in it, you have to take a separate picture with such object.

Third, the real problem with your picture is elevated black level.

Here is a slightly processed version:

The black level is now black. For gray reference I used the white part of the pigeon at the left edge of the picture. That's not a good gray reference either, but should be better than assuming some part of the stone is neutral gray. The man's skin tones look plausible, especially for someone of Indian decent, but of course I don't know what they are really supposed to be.

There is a haze over the picture, particularly in the upper right corner. I suspect there was bright sky there just outside the picture, and the lens wasn't all that clean. Fingerprints or similar smudges on the lens can cause this kind of effect.

• Yeah, it is almost there. Thanks for your help. The processed photograph looks good. But can you explain the process a bit. Did you do this without masking? However, the skin tone is still orange little bit and the stones appear blue-greenish in the darker areas. I don't know... you may have stylized the photograph a bit. – user2118622 Dec 4 '14 at 13:25
• Or you may be right when you say the stones are not neutral gray. – user2118622 Dec 4 '14 at 13:37
• @user: All I did was shift the black level to make the darkest point true black (0, 0, 0), then adjust the white balance to make the pigeon a shade of gray. I also applied a little bit ofnon-linear brightening of the dark areas, but this does not change the color balance. Quite likely the stones are not pure gray. – Olin Lathrop Dec 4 '14 at 14:48

Here are the curve adjustments that I used to get the white balance that I think you want.

``````Input/output
``````

Red 62/56 84/84 129/130 244/255

Green 82/84 129/130 251/255

Blue 60/56 83/84

• Thanks for your effort, man. However, this is not what I wanted exactly. May be it is not even possible without masking. May be you were right when you commented previously that neutral background is throwing my color perception off. Meanwhile, after posting the question, I searched for how to get standard skin tones for people from different countries. The main trick is to use cmyk channel to view your target color area, keep y channel more than c and m channels for skin tones. The problem in this photograph is that the input levels for both skin and grey stones are same place. – user2118622 Nov 3 '14 at 16:47
• Maybe Dan Margulis's action to reduce skin staturation would work. Look at moderncolorworkflow.com (I think) that will have it a part of a free panel for Photoshop. – JenSCDC Nov 5 '14 at 23:47