During today’s informative lesson folks I’ll show you the possible benefits of eyeballing all three RGB histograms whilst editing your photos. Once you start to understand the meaning of the relationship between the three, you can use them to help gauge what does or doesn’t need to be done to correct colour.
In Photoshop you can choose to display all three RGB histograms individually, which is a useful thing to have going on the side-burner. In Lightroom no such thing is possible, but lately you can cycle through the RGB channels in curves, which provides the same insight.
A technique was once introduced to me that involved adjusting each of the RGB histograms individually in levels: the red, green, and blue shadow or highlight sliders were routinely moved to meet with the edge of the pixel data.
This is an intrinsically harmful thing to do if the colours in your photo are completely non-neutral, i.e. there are no natural white or black points, because you’re effectively altering the relationship between RGB values and creating a colour cast. In those situations the best means of assuring a corrected white balance—should you want that—is to introduce a neutral tone into the scene such as a grey card and use that to create a custom white balance in-camera or to aid adjustment in post processing.
If your picture does contain natural candidates for neutral colour, however, such as a cloud for white, aligning the three histograms will remove any colour cast in your photo. Again, this may not be something you actually want, because it tends to destroy quality of light, but in other instances a colour cast will totally stifle the natural hues in your photo and be undesirable. So a judgement call has to be made really based on where the strengths of the picture lie. It’s worth remembering, too, that colour casts are more noticeable in highlights, so the right-hand side of the histogram might be considered more critical in terms of colour correction. Sometimes colour casts can shift in nature, or they can exist in one area of the picture but not another (shadows, mid-tones, highlights).
The aim of this post is not so much to provide colour-correction method, however, which I’ve done before to some extent, but rather to give a practical demonstration of changing RGB histograms in relation to the appearance of a single picture. See below:
This may not appear to be a 'corrected' photo, but it is, because the colour of the paper is a fair representation of how it appears in real life. I used a grey card to swing the colour away from the camera's auto white balance and to achieve an accurate neutral colour. You can see that the histograms do not align at all in the highlights, but this is nothing more than a true representation of the photo. If I adjust the overall RGB composite sliders I can affect tonal range, but I don't necessarily want to correct the white of the paper, so I leave the individual red, green, and blue sliders alone. Note that neutral colour correction and non-neutral colour correction are related, but separate—achieving the latter would require profiling the camera. An analogy might be made with the quite separate acts of white point calibration and profiling in a monitor. The saving grace is that there's usually no absolute 'right or wrong' in photography!
This is what we get if we accept the camera's auto white-balance setting, which ordinarily might be a more natural colour for a newspaper, but it's less natural in one that's 115 years old! Note though how the highlights have been 'neutralised', which is evident by the far closer alignment of the right-hand sides of each RGB histogram. Whatever colour correction you do to an image, if you watch the three histograms you can see it taking effect in real-time. One of the aspects of 'correcting' the white of the paper I don't like is that areas of strong discoloration and staining tend to be left with quite harsh transitions in tone. But still, unless you're involved in some colour-critical endeavour it's all a matter of taste.
To assist further in our understanding of RGB histograms, observe this black and white conversion. Everything in this image is now neutral, from blacks to whites, and all in-between. There are no non-neutral colours. The three histograms are perfectly aligned as a result.