Instant Exposure Check
How And Why To Use Your Histogram
One of the main benefits of using a digital camera is being able to evaluate an image right after exposure in the LCD on the back. However, being so small, most LCDs are not as critical as you might like for checking sharpness or even expressions. Some allow you to zoom in, making them more useful in that regard. For me, the main advantage of the LCD is checking the exposure. Just as with film photography, correct exposure is key to quality images. In some ways, overexposure is more problematic in digital, as the sensor tends to clip data above a certain threshold. Overexpose can result in lost highlight detail and tonal texture, and while there are ways to remedy this most corrections are at best difficult to obtain. But there is hope for exposure problems if you catch them using the histogram, a graphic representation of the gray scale values from 0 (black) to 255 (white).
"Expose To The Right" Rule
Is there an ideal histogram? It depends on the image contents and how the values are distributed, but in general the optimum histogram display would:
1) Not indicate any overexposure or lost highlight values.
2) Not indicate too much underexposure, as you may lose shadow details but more importantly get
In theory, the "perfect" digital exposure, as far as a histogram read-out is concerned, is as far to the right as possible where there is no chance of overexposure. In practice, obtaining this is not that easy. The overall shape of the histogram does not matter, as it merely reflects the tonal distribution of the scene you photographed. Nevertheless, slight underexpose is recommended to eliminate the risk any highlight clipping, something those who have mastered slide film photography will recognize and appreciate.
Color Channel Clipping
While most histograms in cameras show the histogram as gray scale levels we are dealing with color images, so we really have histograms of all three RGB channels. Some clipping can occur in the one or two color channels, a condition that might not be shown in the typical luminance histogram.
Here are four histogram examples (created in Photoshop) that show different image characteristics and the histogram basics:
Histogram 1: Histogram indicating strong overexposure.
The highlights (right side of the histogram) have just been lost. This would be a candidate for deletion in all but a few very rare cases. Some might recommend to "burn" the photo in Photoshop. This is "faking" details in the highlights that are not there.
Histogram 2: Histogram indicating potential overexposure.
Here there's only a small spike in the highlights and it very much depends on the photo whether this might be a problem or not. If the spike represents a real pure white or an unimportant spectacular highlight then this photo may be OK. Otherwise, you are in trouble.
In these cases, if the camera has the ability to indicate overexposed areas, I would recommend viewing the camera's LCD preview with the highlight/overexposure indicator enabled.
Histogram 3: Histogram indicates good exposure.
Here the highlights are OK. You lose a bit of the dynamic range in the highlight area but Photoshop can correct for this. You should endeavor to get histograms like this. In principle, it is best to get as close as possible to the right.
Histogram 4: Histogram indicates strong underexposure.
Here the data in the shadows is lost. This is not the only problem because digital cameras show much more noise in the shadows. Even when the image is corrected extended noise might show up in the mid range.
Look at the example from the Canon EOS 10D (#1 above) that demonstrates a typical problem with saturated colors (#1). The histogram in Camera Raw 2.2 shows the challenge in the blue channel (#2 below).