Page #169 (Chapter 17. Correcting Brightness and Contrast in a Photograph)

136. About an Image's Histogram

An old Polaroid print whose yellows have faded.

Suppose that such a chart represented the relative redness of pixels throughout an image. All the pixels whose red component is 0 can line up in the leftmost column, all those pixels whose red is 255 can line up in the rightmost column, and all the intermediate values can fill in between the two. So, you now have a plot of dark reds on the left, bright reds on the right. Combine this with a plot of relative greens and relative blues, and the chart now gives you a complete picture of the distribution of all the colors in the available spectrum, throughout your image.

What do you learn from looking at such a chart? When you know there's something wrong with the appearance of a photograph but you just can't put your finger on what or why, the histogram helps to direct your finger to specifically what and why. For instance, old color photographsespecially Polaroid printstend to look dreary and muted. After you've scanned one, the histogram of its color channels will probably reveal a dip in its green levels on the right end of the graph. Red and green together, optically, produce yellow; and on a printed Polaroid photograph, yellow is the ink that tends to fade the soonest (cyan next, whereas magenta always seems to outlast the other two inks). A histogram of the same photo's brightness channel will probably also reveal deep chasms along both the left and right sides, indicating an image with neither sharp brights nor deep darks. Shifting the weight of these plots to restore balance to the green channel and to fill in the side gaps on the brightness channel can restore most, if not all, the original print's luster and brilliance. With the Photoshop Elements Levels command, you can restore an image by directly tinkering with the histogram; you can actually change the graph to alter the image.

NOTE

Although yellows tend to fade first in old Polaroid prints, the cyan inks fade first in old Kodak prints, leaving the reds, pinks, and yellows.

This histogram of a faded Polaroid print shows a classic separation between the channels.

Making the proper changes to an image can be as simple as adjusting the histogram so that the left side of the "mountain" touches the left edge of the graph, the right side touches the right edge, and the peaks are more toward the middle than to one side or the other. This is true if you're changing overall brightnesswhich affects all three color channels at onceor one color channel at a time. When you're working with individual color channels, making the proper changes can also involve moving one channel in the graph so that its peaks and valleys fall in line with the other two channels.

There is, however, one tricky point to histogram adjustment, and it involves neither the left side nor the right side, but the middle.

Specifically, where should the middle be? In optics, the richest or brightest color is not the lightest color, but instead what artists and engineers alike call the midtones. When Photoshop Elements is plotting a histogram for the histogram palette or for the Levels dialog box, it's answering two questions: First, how soon is the image progressing from darks to midtones? Second, how much space remains for the progression between midtones and lights? This is important because the very lightest shades attainable for any color are all represented in the HSB scale by white. The whiter a shade is, the less saturated it is andbecause all the other color channels use white as their lightest point as wellthe more likely your eyes are to mistakenly perceive that shade as a color other than it truly is.

KEY TERMS

Midtones Pixels with a luminance value near the middle of those alloweda value of about 128. The midtones in an image are an object's true color, without shadows or highlights (reflected light).

Gamma The measure of the contrast of an image or imaging device, such as a monitor. Adjusting the gamma of an image affects the contrast within the midtones without affecting the contrast within the shadow and highlight areas. If you plot the power needed by your monitor to produce a range of values from pure white to pure black, you'd see a curve. In mathematics, the gamma is the point at which the curve bends.

So, the most important point on a histogram is not the peak but the point that represents the midtone for all channels together, as well as for each channel separately. This point is called the gamma.

No single adjustment is more effective at restoring midtones to an image than repositioning the gamma point. When an image is first scanned in or imported from a digital camera, the Editor assumes a gamma for that image of 1.0. This value represents the bias that the Editor is applying to the image it's processing now; keep in mind that your camera might already have applied its own bias when it originally shot the photo. The adjustments you make to an image's gamma in the Levels dialog box affect the way the Editor reinterprets the image.

In the middle is the original image. On the left is a copy with gamma decreased; on the right is a copy with gamma increased.

With the Levels command, you can make the adjustments that restore the balance and continuity to a scene that your mind told you it had when you first received the impulse to take a picture of it. You can make the lights lighter and the darks darker, and in so doing, vastly improve the contrast of an image while losing none of the detail. Yet these types of changes are not at all the same as turning up or down the contrast and brightness knobs. Those other one-touch changes (one of which you'll see in 137 Improve Brightness and Contrast) affect the whole picture equally and proportionally in all places. And as the histogram will demonstrate to you, they don't actually restore or create balance. Because everything is adjusted equally, these controls simply move the imbalance of an optical factor from one point to another, less noticeable point.

"Twin peaks" help illustrate the clear and natural contrast between midtones and highlights in this photo.

With histograms, you see the distribution of light and dark pixels throughout an image (or a selection of a portion of an image, or a particular layer). Boiled down to its technical underpinnings, contrast can be defined as the differentiation between light and dark elements. Thus, arguably, images with the sharpest possible contrasts would use only black and white pixels. The Editor would plot the histogram for such an image with high cliffs on the left and right, and a chasm in-between. More realistic images with very extreme light and dark contrasts would have relatively high peaks on the left and right, with a dip in the middle, suggesting a relative absence of midtones. To make such a pronounced image more pleasing to the eye, one type of histogram adjustment you might attempt is to blunt the peaks and fill up some of the valley. With histogram adjustment, you experiment with altering the image by tinkering with the geometry of the graph.

With the Levels command, you can see whether brights and darks within a selection are pinched together or spread apart.

Using the Levels command and the Histogram palette in conjunction with one another, you can learn why your composition works the way it does. For example, would you believe the smaller rose in this picture is green? From the computer's vantage point, what appears to be a red rose is actually comprised of strong, dark blue-green tones according to the Histogram palette. Why? Because blues and greens combine to form an optical opposite of red, and opposites are used in combination with pure colors to darken them. So, the dark parts of the rose are less saturated as a result. But notice that it's the bright reds (as you see isolated in the Layers dialog box) that stand out on the right side of the histogram. It's this standing out that gives the entire rose its red appearance. If the red and green peaks were closer together on the right side of the graph, the little rose might look more like the big yellow rose at the top of the image. In 138 Improve a Dull, Flat Photo, you'll see how to use the Levels dialog box.

NOTE

Astonishingly, a photograph that's mainly a portrait of a person's face reveals its own signature through its histogram: from left to right, a blue peak, followed by a green one, and then a red one. This is true for faces with any skin tone, dark or light; with lighter tones, the trio of peaks simply appears further to the right. African skin tones register on the histogram with their green peaks slightly to the right of the middle; Norwegian and other light skin tones register slightly to the left of the middle. In any case, the green peak falls between the other two. Knowing this, one way you can correct an apparent dis coloration in someone's portrait is to examine the histograms of other well-balanced portraits of the same person and adjust the graphs of the red, green, and blue channels of the discolored photo so that the peaks fall mainly within the same areas.

The Colors channel reveals the telltale signature of human flesh tones: a blue peak, followed by green, followed by red.

With the Histogram palette, you can view the current statistics for any open imagespecifically, from any chosen layer in that image or from the current selection for a layer in that image. To display the palette, select Window, Histogram from the Editor's main menu. If it's convenient, you can keep this palette open at all times, either docked in the Palette bin or free-floating. You don't have to do anything more; from that point on, the palette shows the graph of your choice for whatever is currently selected.

You can use the Histogram palette to display information for the three color channels (red, green, and blue), either separately or as a whole.

Open the Channels drop-down list at the top of the Histogram palette and choose Red, Green, or Blue. The histogram curve changes to reflect the distribution of pixels in just that color channel. To get a clearer idea of how colors interact with one another, choose Colors from the Channels list. In this display, each color component is shown with its own natively colored curve. The histogram shows cyan where the blue and green curves intersect, yellow where the green and red curves intersect, magenta where the red and blue curves intersect, and gray where all three curves intersect with each other.

TIP

The Threshold filter helps you find the brightest or darkest spots on an image. This knowledge can be valuable later in designating white points and black points for the Levels dialog box so that you can balance an image's brightness histogram (see 138 Improve a Dull, Flat Photo). Select Filter, Adjustments, Threshold. The current image changes to stark black and white, and a new histogram appears, called the threshold, with a slider at the bottom. The slider is initially set at the midpoint of the graph: 128. Any portion of the original image detail that is darker than the position marked by the slider is rendered in black; anything lighter is in white. To find the image's darkest point, move the threshold slider to the left until only the fewest black spots are visible. These are the darkest parts of the picture. To identify the brightest spots, move the slider to the right until only a minimum number of white spots are visible.

The histogram for the Luminosity channel displays an evaluation of the relative brightness of pixels in the image. There's a subtle difference between the Luminosity channel and the RGB channel of the Histogram palette: Commonly considered the primary histogram, RGB displays relative values of the Brightness (B) component of all the pixels in the image; whereas Luminosity is a result of a formula that evaluates the other two components of a pixelHue (H) and Saturation (S). Think of the Luminosity channel as representing the richness of color, as opposed to its brightness or whiteness. Don't confuse Luminosity with luminance, which is a term often used elsewhere to mean brightness, and in some color models, is used in place of Brightness (B).

As you make a change to all or part of an image using a tool such as the Brightness/Contrast dialog box (covered in 137 Improve Brightness and Contrast), when the Preview option is enabled, the Histogram palette shows you the extent of the change you're making as you're making it. For the currently displayed channel (except for Colors), the palette shows a gray curve with a bright peak, representing the current state of the image, along with a black curve representing the state of the image if and when you apply the changes. You can watch the black curve move away from the gray curve as you use the slider.