Mein kleines Mondlicht,

‘Thou shall use The Histogram! Or else… all your photos will suck.’

It is not uncommon to regularly come across on the Web –and even on printed publications– the specious recommendation encouraging the use of ‘The Histogram‘ as the best method to determine photographic exposure.

Such advice is: B-a-l-d-e-r-d-a-s-h!

In fact, the much hyped histogram is one of the worst tools to use for exposure determination. Allow me to tell You why, and why You should ignore those who strongly suggest its use in order to achieve ‘perfect exposure’. Yea, take no heed of their advice, even if they deliver it in such a confident and smug a manner that they might even resemble John Macdonald holding out the BNA.

 

What is the Histogram anyway?

You probably remember these graphs from your days taking math and statistics at the varsity, Liebchen. Histograms are simple graphs made up of lines. Each of these represents the amount of pixels the image contains for each of the 256 values that encompass the grey tonal scale, from the blackest black to the whitest white. When we’re making colour images, as is the case with digital cameras, the basic histogram is a composite of three histograms –one each for Red, Blue and Green (see Plate 2).

In actuality, there are two histograms:

  1. Camera histogram: digital cameras have a feature that displays a bar chart identifying the distribution of tones across all tonal levels (Plate 2).
  2. Image histogram: the tonal distribution bar chart as displayed on the RAW processor and digital editing programs (Plate 3).

Plate 2. A photograph’s histogram(s), as displayed on the camera’s LCD screen after taking, or reviewing, the image.

 

Plate 3. A photograph’s histogram(s), as displayed by the RAW processing progam on the computer.

In the present epistle, whenever ‘the histogram’ is referred to, the one on the camera is meant, as it is the one we’d use when taking photos. Furthermore, so as not to confuse You, I have also capitalised it –to indicate the (unwarranted) mighty status some give it–, and typed it out in italics –to jocularly reflect the snotty inflection with which the words would be uttered —’The Histogram‘—, as if the thing were something out of the immutable and venerable and unassailable truths of the Universe. (Just think of the dignified stance of Mr Macdonald in his statue and You’ll have the idea 😉 ).

Illustration 1 shews You a typical histogram and tells You how to read the graph.

Illustration 1. A typical brightness histogram as seen on a camera’s LCD screen.

The X-axis is the brightness scale, going from 0 (pure black) to 255 (pure white), whilst the Y-axis plots the amount of pixels falling into each level of brightness. The extent of the histogram from the darkest pixel to the brightest one is the tonal range. The tonal range can fall within the 0-255 limits or exceed them; when the limits are exceeded, clipping –the absolute loss of tonal value information– occurs.

 

Using The Histogram

Overall, the advice from the Histogramers goes as follows:

First, shoot an image and review it on your camera’s LCD screen. The report from the camera looks like Illustration 2. (*Note that the histograms in the illustations are not accurate, but have been simplified for the sake of exemplification.)

Illustration 2. Information reported by the camera on the LCD screen can include the image’s histogram –either the Brightness histogram, or the Brightness and RGB histograms, as in Plate 2. In this case, only the Brightness histogram was selected for display.

 

After shooting the image, chimp and look at its histogram: if it’s centred like the one in Illustration 3, the photo is well-exposed.

Illustration 3. A centred histogram –the signature of a ‘well-exposed image’, according to the Histogramers.

 

However, if The Histogram is not centred, but rather looks as in Illustrations 4-7, the image is badly exposed! (Tut-tut.) Apply compensation until The Histogram is centred, like Illustration 3.

Illustration 4. A histogram leaning to the left –i.e. towards the shadows. (The image was ‘underexposed’, according to the Histogramers.)

 

Illustration 5. A histogram leaning to the right –i.e. towards the highlights. (The photo was ‘overexposed’, the Histogramers would say.)

 

Illustration 6. A histogram leaning to the left and clipped –i.e. detail has been lost, as a number of pixels have turned full black. (The photo was ‘grossly underexposed’, the Histogramers would cry out.)

 

Illustration 7. A histogram leaning to the right and clipped –i.e. detail has been lost, as a number of pixels have turned full white. (The photo was ‘grossly overexposed’, based on the Histogramers’ opinion.)

 

Why is The Histogram unsuitable for exposure determination?

The simplest reason why we shouldn’t use The Histogram to determine exposure is because it is always different for each image.

This is obvious and expected, but the problem begins –or is compounded– when those who swear by it start telling You about the ‘ideal histogram’ (Illustration 3) and how a properly exposed image (!!) should match it, or at the very least approximate it. This is just plain wrong!

“Be very, very careful what you put into that head, because you will never, ever get it out.”
―Thomas Cardinal Wolsey

Plate 3. A centred histogram.

Thus, based on The Histogramers chief advice, a histogram like the one on the right (Plate 3) belongs to a ‘well exposed’ image.

Plate 4. A histogram pushed to the left and clipped.

By extension, if we were to believe them, histograms such as those in Plate 4 and 5 would belong to badly exposed images –the one being underexposed and the other overexposed. They would have You reshoot the pictures in question after dialing in the proper compensation, till the histogram is centred (or leaning to the right [more on this later]).

Plate 5. A histogram pushed to the right and somewhat clipped.

However… the shocking truth is that both images were actually carefully and very well exposed! Plates 7 and 8 shew these ‘badly esposed’ photos. Plate 6, on the other hand, would be deemed ‘the good one’ by the Histogramers, because it has a centred, ‘good’ histogram. However, this scene just happened to have all its tonal values comprehended within the camera’s dynamic range, and they all naturally fell into a centred histogram. Exposure for Plate 6 was determined using an incident hand-held lightmeter, and not The Histogram.

 

Plate 6. This image (whose histogram is shewn in Plate 3) simply happened to have a well-balanced tonal range and a dynamic range that did not exceed the camera’s own. I.e., The Histogram was not used to determine the proper exposure. (24mm, 2 min @ f/22, ISO 50)

 

Plate 7. This photo of smoke was shot on a black background, which means a large percentage of its pixels are pure black (i.e. have been clipped), hence its ‘bad’ histogram (Plate 4). Nevertheless, the exposure is spot on, regardless of what The Histogram might ‘say’. (100mm, 1/200 sec @ f/16, ISO 200)

 

Plate 8. The white background and the predominance of white and light hues in this high-key image naturally pushes its histogram towards the right, even to the point of clipping –i.e. turning some pixels pure white, with no detail in them. Its ‘bad’ histogram is shewn in Plate 5. Again, the exposure is correct, in spite of The Histogram‘s crying out loud that the image was ‘overexposed’. (100mm, 1.3 sec @ f/11, ISO 100)

 

Is the Histogram useful at all?

With all of the above, You might wonder now why do we have histograms in our cameras if they are so useless. Well, thing is, camera histograms are not useless, only inadequately used, with the grossest misuse being exposure determination.

The histogram is useless for exposure determination.

 

However, the histogram is useful for tonal distribution analysis and to check for clipping.

Given that a histogram is nothing but a tonal distribution map, and that every image has, by nature, unique tones which are uniquely distributed, it follows that You cannot rely on The Histogram to properly expose your photos –unless your brain can compute with lightning speed the tonal values in the scene and know exactly where in the histogram they should fall, that is. I’ve yet to meet someone with such ability –though a number of geologists might possess it (see Plate 9).

 

Plate 9. A chart used by geologists to determine the approximate volume percentage of a given mineral in a rock. A similar approach could be used by a photographer to determine the percentage of a given hue in an image or scene: such skill might make the histogram more useful for exposure determination, as the photographer would know where on the graph the tones are supposed to fall, and would easily adjust if/when they did not.

 

However, what You can certainly do is to use the camera histogram to see if any clipping occurred. Based on your Photographic Intention, such clipping might or might not be acceptable. If the latter, then you can make adjustments to prevent it (fill-in lights, shading screens, &c.).

Looking at the histogram can also tell You whether a particular scene is high-key (Plates 5 and 8, Illustrations 5 and 7) or low-key (Plates 4 and 7, Illustrations 4 and 6), or if it falls or exceeds the camera’s dynamic range.

At length, there’s no right or wrong histogram, though there are right and wrong exposures.

 

Since the Histogram is hardly reliable, how then can one obtain good and consistent exposures?

The most reliable methods to attain accurate exposure all the time remain:

1. Using a hand-held light meter. This is the best method to determine accurate exposure.

And here I must mention how funny and hardly coincidental it is that The Histogram is much touted about by those who have an irrational aversion to using hand-held lightmeters. Quite a number of Histogramers regularly bash the venerable and extremely useful incident lightmeter as a ‘thing of the past’, and/or a total ‘waste of money’ in the days of über-advanced camera technology and super-smart electronics. When incident lightmeters are brought up in an exposure determination discussion, that’s when the words ‘The Histogram‘ are uttered in the smuggest and snottiest fashion –not infrequently by those who are too lazy to learn the use of the device, too miserly to invest in one, or too ignorant to know anything about photographic exposure. You should neither fear nor heed these fellows.

2. Using key tones and the camera’s reflective spot metering function.

This method would take an epistle of its own to describe fully, so I won’t even attempt to explain it here. Essentially, though, all You need to do is to find a patch in your scene whose tonal value is middle grey or equivalent to middle grey, point your spot at it, adjust the camera controls till the exposure needle in the viewfinder is centred, and –voilà!– there You have your accurate exposure.

So, welcoming You to the Histogram-free Zone, Liebchen, I take my leave and remain,

Ever Yours,

HVG