Every image is made up of a range of tones from darkest on the one end of the scale to lightest on the other end. If we had to create a graph of those dark, mid-tone and light values, we would have what we call a histogram. Histograms are very useful tools for understanding what is going on in an image and particularly how to correct an image that may have been underexposed or look rather flat.
The beauty of getting to understand histograms is that they give you a scientifically accurate method of correcting images, which is important if you are to maintain a professional standard. Many photographers simply correct by eye adjusting an image until it looks right. It may not be right, however, and the mess that is made may only really show up down the line when the images is sent out to print. If you understand histograms, you can avoid this eventuality and editors and clients can come to rely on your professionalism, because you know what you are doing.
Introduction to histograms
A histrogram is just a graph showing the range of tones that occur in an image from the darkest tones on the left to the lightest tones on the right.
Figure 1 The image above has a range of tones from the dark areas in the top left hand corner to the lightest areas in the pillar bottom right. The histogram below is the map of the number of pixels at each tonal value.
Figure 2 The graphs above relate to the image above. Each graph is a histogram. The black histogram is the combined Red, Green and Blue tones and then the red, green and blue graphs show the histogram for each colour channel, the Red channel, the Green channel and the Blue channel.
The horizontal scale of the histogram is the range of tonal Levels from 0 to 255, making 256 in all. With 16-Bit images, which have many more tones, the histogram is still shown with the 256 levels. The vertical scale is the number of pixels at each level from 0 to the largest number (which can vary greatly). This means that the vertical scale is not always the same and separate RGB histograms with the same height may not represent the same number of pixels.
Let's look at a grayscale image for a moment just to help simplify a description of the histogram. Think of the histogram as a wall constructed not of bricks, but of pixels. All of the available pixels are sorted into how light or dark they are and placed into columns from black to white. This image has more in the highlights than the shadows, as can be seen by the size of the two 'hills' in the range.
Figure 3 Above is a grayscale image with lots of light areas. Light areas are commonly known as 'highlights' and dark areas as 'shadows'
Figure 4 Above is the histogram for the grayscale image above it. There are two peaks in the highlight areas meaning the image is quite a light image. The illustration of the histogram below it gives you the idea that these are piles of pixels of a particular tone stacked on top of each other. The higher the stack, the more pixels of that tone there is in the image
The image below is made up mostly of pixels that are dark in tone, which we call 'shadow detail'. The 'hill' in the histogram on the left hand end shows this clearly. The remainder of the range may look like it contains few pixels, but the whole histogram scale has been reduced vertically so that the peak fits. Since the grayscale covers the full range from black to white it means that the image has a full tonal range.
Figure 5 The grayscale image above has most of its pixels in the darker tones.
Figure 6 Above is the histogram for the grayscale image above it. The peak in the darker end of the tonal range means that most of the pixels in the image are on the darker end of the scale.
Histograms give you an immediate insight into the health of an image. This is why picture library editors and others receiving your images may make an immediate inspection of the histograms of the images you are delivering. If they don't like what they see, they are likely to reject the entire batch.
Figure 7 The image above is flat. This can be seen because the histogram is bunched in the middle. The shadows of the image don't stretch all the way down to the far left of the histogram and the highlights of the image don't stretch all the way up to the right of the histogram. For a richly toned image, you want the tones to stretch right from the shadows to the highlights. You can fix this image, by stretching it out. Sometimes if you are needing to stretch an image out wide, you can get jumps appearing in the tonal range which we call posterization. A 16 bit image has so much more information in it that this tends not to happen, this is why it is best to have scans done at 16 bits. The image can then be corrected and then converted to 8 bits for use
Figure 8 This version of the image above is damaged and can't be repaired. See how the pixels are stacked up against the highlight end like a wave hitting a wall? That is because someone has forced lots of the highlight pixels to pure white which means that lots of the detail in the highlights has been lost - they are just pure white
There are many image faults that can be read from a histogram. Some of these may be obvious to the eye, such as colour cast. The histogram tells you where the problem is and gives clues as to how to fix it. The histogram can also tell you if an image fault is fixable or if it is beyond repair.
In the Image Errors course and the Colour Correction courses we will go into recognizing and fixing problems in greater detail.