Feature: Use the Histogram to Avoid Exposure Issues
One of the most often overlooked goodies in your digital camera is its ability to display histograms. The function is frequently overlooked simply because it sounds intimidating. Histogram? That sounds like a cross between something you did in 10th-grade math class and some sort of uncomfortable medical procedure. How can that be useful?
A histogram display, which is included on most digital cameras, provides a quick indication of your picture's exposure quality. Is it overexposed? Underexposed? Just right? The histogram knows, and it's eager to tell you.
Finding the Histogram on Your Camera
As I say, most digital cameras these days include a histogram display. Check your user guide for details. Most cameras overlay the histogram on top of the picture after it has been taken--so you'll see it on playback. On many cameras, you can turn on the histogram overlay with the same control that lets you change the picture you're previewing on the screen. There are even a few models that will overlay the histogram as you preview the scene before you click the shutter release. That's the best option, since you can see even before you take the picture if it will be well exposed.
Understanding the Histogram
A histogram is a graph that displays how light is distributed in your picture. The left side of the graph represents the shadows, while the highlights are on the right. Here's what that means: If the histogram has a high peak on the left, you can deduce that a lot of pixels in the picture are dark, or in shadow. A peak on the right of the graph means that a lot of pixels are bright, or in highlights. Peaks in the middle of the graph represent pixels in the midtones of your exposure.
And here's the real key to unlocking the power of a histogram: There should not be any peaks that get "cut off" at either end of the graph, as if they want to continue past the edge of the graph. When the histogram starts or ends with a peak that's already in the air, then you know that color information has been lost because the camera's exposure settings weren't correct for that picture.
So let's apply this knowledge to some real histograms. Take a look at this one; it's the sort of thing you're looking for. At the left side of the graph--the shadows--the graph starts at the zero point on the horizontal line and then curves upward. On the right side of the graph (the highlights) the little peak trails off to zero before reaching the far right edge of the graph. These shapes tell you that no part of the scene is over- or underexposed. The light is well distributed through the image. Overall, this shot should look good.
If you don't quite "get it" yet, that's okay--this histogram should help you understand the concept. In this underexposed picture's histogram, I see a spike in the shadows that starts "off the graph" to the left instead of rising from the horizontal axis. That tells me that this picture has lost data--perhaps quite a bit of it--in the shadows. Notice that there's also just a few pixels trailing off the right side of the graph, so a tiny bit of data might have been lost there as well. Ideally, the graph would blend into the horizontal axis before reaching the edge of the graph, but this looks pretty small and so isn't much cause for concern.
Here's a histogram of an overexposed photo. Here you can see a pronounced spike in the highlights. Don't worry about the fact that the height of the peaks is somewhat low in this picture; that's not an indication of under- or over-exposure. All you should really worry about is whether they breach the left or right edges of the histogram.
Finally, photos don't get much worse than what this histogram shows. There's very little information in the picture--it was taken in the shadows, without the benefit of a flash. So what do we have? Very little light information through the midtones at all. There's a deadly spike in the shadows at the left edge of the histogram, which guarantees that the picture is underexposed. Using auto-correction tools on this picture will promote a lot of digital noise, so the image isn't going to be very salvageable. But look at the highlight--there's a little spike there as well. This picture has a sunbeam in the middle of the scene, and that sunbeam has "blown out," or overexposed, a small portion of the image. I'd call it a total loss.
Some General Advice
So what should you look for when sizing up your shots on the camera? Obviously, you'd like to keep the histogram from spiking at either extreme end of the graph, where you'll lose data and have under- or overexposed parts of your picture. Also, note that the particular shape of the curve isn't all that important--it can be shallow, curvy, flat, or some combination. The overall shape really just represents the specific light distribution in your photo, and that's as unique as a fingerprint. Finally, you'll get your best results when the graph sits as far to the right as possible--but without clipping the highlights. Overexposure is always worse (and harder to correct) than underexposure.