I have read numerous articles on the internet explaining the theory and use of histograms.  They have all treated the subject in great detail, but I feel they are much too difficult for the average photographer.  This scares people off, and drives them away from using their histograms.

Histograms are HUGELY VALUABLE and should always be used as they offer the photographer information that would not otherwise be known.  Here is my attempt to make your histogram easy to understand and to use.

The best way to understand a histogram is to construct one using a simple example.  Let’s construct a histogram using the ages of the people in my fictitious photo club.  Let’s assume there are 54 people in my club and their ages are shown in the figure below which surprisingly looks like a full frame 36 mm x 24 mm digital sensor having an aspect ratio of 3:2.

To construct the histogram, we first identify the bins on the horizontal axis, e.g., ages 36 to 40 fall into the bin labeled “36-40”, ages 41 to 45 fall into the bin labeled “41-45”, etc.  Once the bins are defined, we then count the number of ages which fall into each bin.  This is shown plotted in the graph below.  The vertical axis shows the integer number of ages in each bin.

This simple example is completely analogous to the histogram on your camera, but instead of counting the number of ages in each age bin, it counts the number of pixels having brightness values ranging from the darkest of dark tones to the lightest of light tones.

The level of effort for pixel-by-pixel brightness level counting is huge, as there are 21,026,304 actual pixels in a Canon 5D Mark II full frame sensor.  Given the physical size of this sensor, 36 mm x 24 mm, this means there are 24,336 pixels per mm2.  Wow!  So, in order to construct a histogram on the rear of your camera, over 21 million calculations must be made.   

The place to begin our understanding of the histogram is your camera’s sensor.  It is generally agreed that a DSLR’s sensor range of light sensitivity is approximately five to seven stops.  For the sake of argument, let’s say it is five stops.  A histogram then, as explained above, is a graphical count of the number of pixels falling in bins ranging from the darkest of dark to the lightest of light brightness levels.  This is described by the following fictitious brightness histogram.

The histogram above reflects the brightness distribution of a properly exposed image spanning approximately 3 stops, and biased ever so slightly to the right.  I use the word “properly” because the brightness of all pixels were found to be within the 5-stop range.

What if the brightness of all pixels were found to exceed 5 stops, or the histogram ran off the left or right end of the scale?  These three possibilities are shown below.

The parts of the histogram which fall outside the range of brightness bins are cut off, i.e., your sensor cannot detect them even though your eye can see them.  Once pixels have been cut off, they are gone forever.  No matter how hard you try during post processing, your printed image will never look quite right.

As an experiment, purposely print a photo where the highlights have been cut off and you will see portions of the print that have no ink on the paper.  This is because the affected pixels contain no data.

With the above as background, let’s take another step forward in understanding histograms.  Let’s examine the histogram on the back of your camera in replay mode.  How did it get there and what is it a histogram of?

First, I know I don’t need to ask what format you used to capture your image.  It’s RAW - never any format other than RAW.  

Capture your image in RAW format.  Not JPEG.  Use full RAW format.

Let’s start with the photo you captured that is now shown as a histogram on the back of your camera.  

Replay is JPEG, not RAW

You will ALWAYS postprocess RAW

More data to the right

Expose to the right

Demonstrate two methods – regular and live view where you move around your point…

Now, with all this knowledge, how do we put this to practice in the field?  First, set up your camera and tripod and compose a nice landscape.  Stay away from bright sunlight and dark shadows for this learning example.

Metering style – average, point, etc.?

Make sure you are shooting on Aperture Priority, and go ahead and capture Image A.  Look at your histogram.  Chances are you’ll see something like this on the left.

This is a ‘nice’ histogram because you haven’t clipped-off highlights or shadows, but it could be much better.

Retake the photo leaving everything the same, except correct your exposure one full stop to the right.  Your new histogram for Image B will look something like on the right.

This is a near perfect histogram since you’ve exposed right up to the right hand side without clipping any highlights.  You have captured the maximum amount of data available to your sensor.

Now, compare JPG Image B to JPG Image A.  Image B looks washed out, or overexposed.  This is true, but remember, you will post process with your RAW file, not the JPG.  So, the lesson to be learned here is, use replay to check for composition only, and use your histogram to check for exposure.

After a day of shooting, and someone proudly shows you a landscape image (JPG) taken directly from their camera that is beautifully exposed, you now know why their ability to achieve the best photo possible during RAW post processing will be limited.

Now, if you are not shooting RAW and will therefore use the JPGs straight out of your camera, then abandon the above process and get the best exposure you can from the JPG replay on the back of your camera.

Let’s investigate the file sizes of Images A and B.  The RAW file size for Image B is approximately ??% larger than for Image A.  This is because you have captured a lot of data at the highlight end of the histogram.  This means Image B will have this much more data to post process with.

Kit and caboodle

In this case, if the real exposure range is greater than five stops, then the tones either lighter and/or darker than the five stops will not be captured (often referred to as “clipped” or “cut off”) by your sensor.  The figure below shows a histogram where highlights have been cut off.