Balancing White Light

Reading the Image

FIGURE 1

Looking at the image above you should see that it is a little flat; there are no deep blacks or bright whites.  But the biggest problem is the bluish/green, cyan, tint that is covering the image.  This known as a "colour cast" and the usual action is to get rid of this discolouration.  But the question is, how? 

You might say that you don't know what the true colour of the arches in the image are meant to be, so correcting the colour cast can only be guess work.  But you do know one thing: the scene is lit by daylight.  It is not night time, it is not at sunset, there are no coloured lights illuminating the arches.  It is daylight, and even if the day is overcast it means that to the human eye the light is "white".  But now look at the Histogram to get the information that you need as it will tell you what to do.

NOTE:  Your eye will notice colour cast a lot easier in the highlights than in the shadows.  This is only natural as the human eye is more sensitive to brighter colours.  Don't therefore make the mistake of thinking that the colour cast only exists within the highlights.  Do NOT go selecting those areas that most obviously have a cast and start correcting them one by one.  There is only one light sourse here, namely daylight, and so the light that is illuminating the highlights is also the light that is illuminating the shadows, even though the amount varies.

As a general rule, if you see a colour cast then it will usually be a universal problem across the whole image.  The soloution will therefore also be a universal one.  There are exceptions to this, such as with mixed lighting where you may have a photograph of an indoor room lit by tungsten lighting and with daylight coming in through the windows.  These situations will be dealt with in a more advanced course.  Meanwhile, we will be starting with images that have a single light sourse, and take things just one stage at a time.

FIGURE 2

The Histogram shows that the pixel values cover most of the available tonal range, but they do fall short in both the shadows and highlights.  As with the Greyscale image in the previous section we will use this Histogram to tell us how to correct the image.

NOTE:  You will see that at each end of the Histogram there is a thin line running down to the "0" shadow level, and up to the "255" highlight level.  This has no formal name, but will be refered to in this digital traing course as a Histrogram "string".  However, we need to keep things simple to begin with, so these "strings" will be ignored at this stage.  Histogram "strings" will be very important with some images and this will be dealt with in a more advanced course.

Correcting the Image Contrast

FIGURE 3

Open the Levels panel (Image > Adjustments > Levels) and you will see the same Histogram as shown in the top ofd the Histogram palette.

FIGURE 4

Ignoring the strings, pull in the shadow and highlight sliders until they meet the point where the Histogram starts to lift above the string level.  If it helps to visualise things: then think of the Histogram as being a tropical island and we need to bring the two sliders up the beach to the high tide mark. 

Leave the middle slider alone at this stage.  It will automatically move to sit at the mid point between the shadow and highlight sliders as they are moved.  This slider will be looked at later.

FIGURE 5

The image has had a small increase the the contrast, which is an improvement, but the big problem here is still the cyan colour cast which has not been dealt with.

FIGURE 6

The Histrogram explains why and tells you how to fix the colour cast.

The Black Histogram at the top is simply a combination of the three separate RGB Histograms below it.  The verticle fracture lines show that the pixel values have been stretched so that the tones now covers the full possible range.  But the Black Histogram only shows that at least one of the separate RGB Histograms has reached the '0' value in the shadows, and that at least one of the separate RGB Histograms has reached the '255' value in the highlights.  It does NOT mean that ALL the three RGB channels have nessacerraly reached the ends.

NOTE:  When the shadow and highlight strings were ignored, meaning in fact that they were clipped, it does not mean that those few pixels were deleted; just that those pixel values were forced to '0' or '255'.  Those pixels that were in the strings before are now seen in the small verticle spikes at each end of the Histogram.

Looking at the shadow parts of the Histogram in the three separate RGB channels you can see that all three have in fact meet the '0' point.  So the shadows in the image are now balanced.

The same can not be said of the highlight end of the three separate RGB channels.  The Green has hit the maximum '255', the Blue has just reached it, but the Red channel has been left short.  Remember, we are ignoring the strings at this stage.  The result is that those highlight areas of the image, that have values higher than those in the Red Histogram, are only made up of Green and Blue.  The Red is missing the upper highlights, and so the highlights are not balanced.

One of the key points from the section on Colour and Light was that White light is made up of Red, Green and Blue.  Remove any one of these prime colours and you can no longer have White light.  The light coming through the arches in the image is only made up of Green and Blue, which is Cyan, and so there is a Cyan colour cast.  The Histogram was telling us this all along.

What we were doing by correcting the combined RGB in the Levels panel was to only improve the image contrast.  But the image levels need to be balanced to make sure that the light in the image is indeed White.  To do that we need to start again.

Balancing the Colour

FIGURE 7

With the Levels panel open again you will see that at the top is a drop down list of the combined RGB, and the separate Red, Green and Blue channels.  Now instead of doing the work in the combined RGB, as before, that same work will be done in the three RGB channels separatly.  This will give us individual control of separate Red, Green and Blue shadow and highlight levels.  In this case it means that the Red highlight can be brought up to meet the Green and Blue so that the highlights are balanced.

FIGURE 8

And the image now has White light at last.

FIGURE 9

The Histogram shows that the image is now colour balanced in both the shadows and highlights, and thus returned the light sourse back to White light as it should be.

Shifting the Midtones

FIGURE 10

There is one further small step that can be added here.  In the Levels panel again, the middle slider can be used to lighten or darken the overall effect of the image.  This should be done in the combined RGB so as not to upset the colour balance work done previously.

FIGURE 11

The middle slider was shifted a little to the right here so that the centre weight of the pixel values will move a little to the left, thus darkening the image slightly.  This part is subjective with no help from the Histogram to guide you.  Just move the mid slider a little to the left and right to see if it improves the image.

FIGURE 12

The finnished product.  This work should take you less than a minute after a little practice, but if you have a lot of images to correct then there is JavaScript available that can do this with a click of a button.

FIGURE 13

The mid-tone correction is seen in the Histogram as a few spikes added towards the shadow end.  The tropical island now has a few trees on it.

NOTE:  In the end the use of colour correction is a subjective matter.  If you really do prefer flat images with colour casts, then you can certainly keep them that way.  There are times that a colour cast can add "mood" or "feeling" to an image, but if it is not done well then it will just look like a mistake.  A lot will depend on what sort of photography you are doing, but usually clients and picture libraries will prefer balanced, neutral colours.

What Can Go Wrong

FIGURE 14

Some images are naturally flat, such as views with mist, fog, or heavy air pollution (as in the image above).

FIGURE 15

You can still do the colour balancing with the Levels, but then it is up to you as to which version you prefer.  If the image ends up not looking quite right, then step back.  If the subject of this photograph was intended to illustrate air pollution, rather than the building, then you may well prefer the uncorrected version.

FIGURE 16

Some images are both flat and have a colour cast, such as views with heat haze in the late afternoon.

FIGURE 17

Colour balancing images such as these will strengthen some of the detail, but at the cost of the heat haze effect.  It is up to you as to which you prefer, or whether you want to use a compromise between the two.

FIGURE 18

Some images have a strong colour cast, such as views at sunrise and sunset.

FIGURE 19

The rules about White light break down here as at first the Blue light collapses leaving just the Red and Green, hence the Yellow colour cast.  Eventually even the Green light collapses at sunset, leaving only Red light.  Correcting these images using White light rules are likely to give you results with very dull colours, so it is up to you which version you prefer or if you would like a correction somewhere inbetween.

Images with mixed lighting  will also cause problems as they are more difficult to work with.

We are at the basic levels of colour correction here, so if you find at this stage that the rules about White light, used above, are giving you some odd results, then stop.  You may have encounted some problem that we havn't dealt with yet.

Quick Summary

1.  When colour balancing single channel Greyscale images then you only have the Grey channel to work with in the Levels panel, so you will only be correcting the image contrast.

2.  When colour balancing three channel 'RGB Grey' images then use the combined RGB channel at the top of the Levels panel as all three channels will be the same, so you will only be correcting the image contrast.

3.  When colour balancing colour RGB images then correct the three RGB channels separatly in the Levels panel, and not the combined RGB channel at the top.  This will correct the contrast and balance the colour to White light at the same time.

4.  At this stage ignore the Histogram strings in the highlights and shadows by bringing the sliders up to the point where the Histogram becomes thicker than the string.  This will cause problems with some images, in which case retreat and only bring the sliders up to the start of the strings instead.  Dealing with Histogram strings will be covered in an advanced digital imaging course.

5.  The middle slider on the combined RGB channel (the Grey channel on a Greyscale image) can be used to lighten or darken the image.  This setting is subjective, so be cautious.  A more advanced method of doing this in the Curves panel will bo covered in an advanced digital imaging course.

6. If White light dosn't exist in the image, such as at sunset, then the rules about using White light for colour correction don't apply.

7.  Getting rid of colour cast by colour balancing may be "technically" correct, but if you prefer the image as it is because it adds "mood" or "feeling", then there may be no problem in keeping the colour cast.  Do be careful though as a lot will depend on the type of photography you are doing.  Adding warmth to portrait images can make people look more attractive, especially if the clients are the ones being photographed.  But if you are photographing news and events then adding "mood" and "feeling" to images of a riot or scenes of poverty may be the last thing a client will be wanting.