cpn connect

Light and Luminance: Considering the Histogram: The Evolving Tools for Digital Exposure

1/18/2013 11:34 AM Eastern

In last month’s column about tools that aid in exposure judgments with digital cameras, I wrote about the waveform monitor, one of the oldest and most prevalent photography tools today. This month I’m going to discuss the histogram.

This image represents several different luminance ranges. At the top, a pure white box—the histogram shows a straight white line all the way to the right. Second is a pure black box—the histogram shows a straight line all the way to the left. The third is middle gray—the histogram shows a straight line right in the middle. Finally, a grayscale from black to white with a large chunk of middle gray. In the representation of the pixel luminances denoted in the histogram, the line at F is significantly taller because there are many more gray pixels in the large gray box than there are pixels in the smaller boxes.

The histogram has a long history dating back to the 1700s, although the term “histogram” wasn’t coined until 1895. Webster’s Dictionary defines histogram as “a bar graph of a frequency distribution in which the widths of the bars are proportional to the classes into which the variable has been divided and the heights of the bars are proportional to the class frequencies.”

Basically, early “histograms” were merely bar graphs—and actually that’s all a histogram is: a bar graph representing a spectrum of data. In our case, the data represented is the luminance distribution in an image from black to white.

Unlike waveforms, the histogram doesn’t represent the image in a pictorial fashion; it merely displays the number of pixels in a given luminance range. On the image histogram, the far left represents black in the image and the far right represents white. If there were a scale on the histogram, the X-axis would be bit values (in an 8-bit system) from 0 (black) to 255 (white). The Y-axis would be a percentage or pixel count.

If the image were nothing but a white screen, the histogram would be one straight line at the 255 point (far right) and the rest would be flat (no reading). If the image were a 50 percent gray box, then the 128 mark (the middle of the histogram) would be a straight line and all else would be flat.

Most images are not one pure tone, however. Rather, they’re a combination of many tones and variations between black and white. And that is what the histogram shows: the number of pixels in an image at different luminance values between black and white.

Histograms can also show individual red, green and blue channel luminance values, either separately or superimposed over each other. This is very handy as it is possible to overexpose one color channel but not the others—especially with flesh tones, which are high in red but not so high in green and blue.

Actress Lisa Jay in Jamie Neese’s “Tranquility, Inc.,” a short film I shot with the Canon EOS 7D. Grayscale luminance values of this image, along with separated red, green and blue Superimposed RGB values of this image, also with separated channel information

The image of actress Lisa Jay from “Tranquility, Inc.” is indicative of my personal aesthetics. The exposure is weighted toward the shadow end, with flesh tones slightly under the normal, and with strong highlights. We can see these choices represented in both histogram options: the grayscale luminance values along with separated red, green and blue, and the superimposed RGB values, also with separated channel information.

The superimposed RGB can be hard to read, which is why many prefer the traditional luminance histogram with separated channel information.

In a “properly” exposed image, there should be a fairly even distribution (weighted toward the center of the histogram) of the luminance range. “Properly” is subjective, however, and greatly depends on the effect you are trying to achieve—and the contents of the image itself.

In this image of actress/model Becka Adams, I achieved a moody, contrasty feeling with rich blacks and underexposed flesh tones.  

In the dark image of Becka Adams stretching, I wanted to highlight the strength of her arms and define the silhouette of her body but keep the image feeling dark and moody overall.

The histogram shows that nearly all of the image falls into the blacks and shadow range, with very little in the mids and highlights. If you looked only at the histogram and did not consider the overall creative choices, the image would appear to be grossly underexposed.

A high-key image of model Alexandra Preda, photographed by Claudiu Gîlmeanu  

In the photo of model Alexandra Preda, the photographer, Claudiu Gîlmeanu, chose to slightly overexpose the skin tone to lend a more dreamlike quality. The photo’s histogram shows a bias toward the highlights—especially in the red channel—in what might otherwise be considered bordering on overexposure.

He chose, however, to push the overall luminance range toward the whites to achieve an effect. Contrast is naturally obtained through the model’s dark hair and eyes, which frame the brighter skin tone very well, making this a balanced image—even though it is weighted toward overexposure.

In this image of a pair of dandelions on my front lawn, I silhouetted the flowers and seeds against a cloudy sky and maintained as much detail in the sky as I could.  

In the image of dandelions on my front lawn, I silhouetted the flowers and seeds against a cloudy sky and maintained as much detail in the sky as I could. Although it appears to be a very shadow-heavy image, because the majority of the sky actually falls into the middle gray zone, the histogram represents a more “proper” exposure overall, with the majority of the pixel information in the middle gray area.

Many camera manufacturers allow you to see a histogram of your image on the onboard monitor or LCD screen. Many HDSLR cameras can do this, but you need to shoot a still photo first because most cannot display a histogram on a “live” or video image.

Histograms that appear on the onboard displays of RED cameras feature two additional indicators: goalposts and traffic lights.

The goalposts are vertical “thermometers” that sit to the right and left of the histogram image and indicate how many pixels in the image are below black or above white. Goalposts rising to a significant level on one side or the other suggests gross over- or underexposure of the image. Balancing the goalposts is a good way to balance your overall exposure.

The traffic lights “light up” to indicate when a portion (generally around 5 percent) of the pixels of a particular color channel are overexposed. This display will show you at a glance if you’re clipping one channel over the others.

Another tool to aid in your exposure choices, the histogram can be very helpful to give you a quick assessment of the overall luminance values in your image.