The Histogram — Explained

The image histogram is one of the most powerful tools available to a digital photographer and, to the beginner, one of the toughest to grasp. The good news is that the histogram is actually very simple to read. Once a few simple concepts are grasped, it will enable you to take much better pictures.

What the Histogram Tells Us

The histogram is a tool for judging exposure. It shows the distribution of tones throughout the image - black on the extreme left, white on the extreme right and shades of gray in between. The amount of each tone is shown by the height of the histogram at that point.

	Histogram with gradient beneath. This image has a tiny number of black tones, and no pure white.

Above, you see an example image and its histogram. I have added a gradient to the bottom of the histogram to show you what's happening. Pick a point along the gradient and the amount of that tone in the image can be read by the height of the part of the histogram directly above it.

This image is perfectly exposed. The histogram extends all the way to the left, but it does not 'climb the wall' on that side. This means that we have no large black areas in the scene. Likewise, it extends almost all the way to the right, but stops short of the very edge. That means that there are no large areas of pure white in the image. No pure white at all, in fact.

Why is this a good thing? If more than a tiny fraction of the image is black or white, then you have lost detail in those areas. Typically, this is not desirable - a photograph is usually about capturing detail!

Below are two more shots of the same scene and their histograms. The image on the left is underexposed, and the one on the right is overexposed. You can see the underexposed image's histogram is piled up towads the left hand (dark) side of the box, and the overexposed one is piled up towards the right (bright) side.

In and of itself, this doesn't mean bad things for the image. If both histograms trended down to the bottom just before the left and right edges of the box, then we would just have a very dark and a very bright image - but they would both have detail. As it is, both histograms are right up against the wall.

The underexposed image has a large peak on the very left hand side, indicating a lot of pure black in the image. This would correspond to the dark area in the foreground, leaving it a featureless blob.

The overexposed image has a peak extending halfway up the right hand wall, indicating that there is a large area of pure white in that image. This would correspond to the blown out (white) area of sky above the mountain. We have lost the delicate detail of the sky and clouds in that area, again leaving it a featureless blob.

It's worth noting that there is no such thing as the right shape for a histogram - if you have a tall, narrow histogram it just means that the scene you're photographing is low contrast. If you have a smooth hump of a histogram extending from left to right, that means that the scene you're photographing is high contrast, and has a high proportion of midtones. The histogram will be whatever shape it will be due to the nature of the scene.

The only problem is when you have anything more than the tiniest line going up the extreme left or right sides - right at the edge of the histogram. That indicates excessive black or white. If there is a peak near either side, it just means there's lots of dark or light tones - which is not a problem.

In Practice

The histogram is a useful tool in post-processing, letting us know if we have over brightened or darkened an image, and giving us clues as to what sort of processing the image needs. However, the most important place to consult the histogram is on your camera when you're taking the shot.

Looking at the image on the camera's LCD alone is often not enough to judge if the exposure is correct. To ensure that you haven't lost detail in important areas of the image, you should examine the histogram also. In fact, when I go out photographing, I rarely look at the image itself on the LCD - all my attention is on the histogram.

Any camera worth its salt will display a histogram of the image just taken - in fact, many point-and-shoot compact cameras can display a live histogram on the LCD as you are composing your shot. Check your camera's manual for instructions on how to bring up this display.

Below is the first of the three images taken above, shown on the LCD of a Canon 5D.

This image is properly exposed. If it wasn't, the histogram would show data 'climbing up' either wall and we would use the camera's exposure compensation feature to darken or brighten the scene, bringing the histogram closer to the middle.

Luminosity vs. RGB Histograms

The histograms we've looked at so far are termed 'brightness' or 'luminosity' histograms. They represent an average of the tones in the scene, but they don't always tell the full story.

To create a colour image, the pixels in your sensor are filtered so some are sensitive to red, some to green and some to blue light. Each set generates a monochrome image made up of only that colour light, and all three are composited together to produce the final colour image. This is done because digital sensors don't know about colour themselves, all they know is how much light they saw - how dark or bright that part of the image was.

Below is a colour image with its luminosity histogram. Underneath it are three images, each showing just one of the three colour channels which make up the final colour image, along with their histograms.

	Normal image, colour channels composited together, luminosity histogram.
	Red channel, with its histogram.
	Green channel, with its histogram.
	Blue channel, with its histogram.

You see that the histograms for the red and green channels look pretty similar. The blue channel, however, is much further to the right; indicating that it is brighter overall than the other two. This can be clearly seen in its corresponding image.

In this particular shot, that is not a problem. But what if the exposure had been brighter? The blue channel might have lost detail in the highlights and 'climbed up the wall' of the right side of the histogram. Looking at the luminosity histogram on our camera, we might not see this happen. The luminosity histogram is an average of the three channels - so if one is slightly over the edge and much brighter than the other two, the average might still look ok.

That is why some cameras have the option to show an RGB histogram instead of (or as well as) the luminosity histogram. If your camera has this option, you should turn it on. The principle is the same - if any one of the three channels 'climbs up' the left or right wall of the histogram, then you need to adjust your exposure to compensate.

Here is the original image from the beginning of the article shown on the Canon 5D's LCD, this time showing the three RGB histograms instead of the single luminosity histogram.

I hope that you have found this tutorial useful. If you'd like to improve your camera or post-processing techniques, I run regular workshops at my studio in West Cork. I am also available to speak at camera clubs and other gatherings. Contact me for details.