Few things improved my photography more than learning when and how to set the exposure manually. That knowledge allows us to get good exposures in situations that automatic exposure can't handle. Setting the exposure manually also encourages us to make conscious, creative decisions about exposure.
I've heard some photographers say that they don't see any reason to use manual exposure. If that's your view, here's why I think you should reconsider.
To understand why manual exposure can be helpful, we need at least a rudimentary understanding of how a camera's reflected-light meter works. I've read several technical explanations of this subject. I've become convinced that it's impossible give an explanation that is brief and comprehensible while also being technically accurate in all regards. This explanation will sacrifice technical accuracy in favor of brevity.
A light meter samples the tones in the viewfinder. Based on that sample, it calculates the camera settings that are required to make the average* of the sampled tones medium gray*. (*Well, not quite in either case, but the details won't matter to most of us.) Although we habitually speak of medium "gray," color is irrelevant to the meter. Only the luminosity matters. Medium gray is a constant tone, also known as 18 percent gray. It looks about like this:
It's important to understand that medium gray is not the middle tone of whatever scene you're viewing. Medium gray looks like the sample above regardless of what is in your viewfinder. In a very dark scene, medium gray may be one of the brightest tones. In a very bright scene, it may be one of the darkest. Whether the scene is bright, dark or average, the meter will calculate settings that place the midtone of the resulting image near this tone.
When you use evaluative metering, the meter takes a number of samples from all over the scene. It proposes an exposure that will place a weighted average of those samples at medium gray. In many situations, that approach works very well. That's why I mostly shoot in aperture priority with evaluative metering. In some situations, though, evaluative metering doesn't work at all. If you've followed this discussion so far, you can probably figure out what evaluative metering will do with a scene of this sort:
When evaluative metering samples tones all over this scene, it's mostly going to find very dark tones. To get an exposure that places the average of the sampled tones at medium gray, the meter will therefore propose a longer exposure than is really appropriate. Evaluative metering would have given me a result about like this:
The result has blown-out highlights and a lot of unwanted shadow detail. I went to a lot of trouble to take this shot during the very brief window of time when the tree and ridge would be lit but the shadow detail would be suppressed. Evaluative metering would have nullified my effort.
If we were instead looking at a snow scene, in which a few darker tones were surrounded by mostly very bright tones, we'd have the reverse problem. The meter would try to turn all those bright tones to medium gray. It would therefore propose too short an exposure. You'd get gray snow.
The solution lies in partial metering. Partial metering does what evaluative metering does, except that it samples only the tones in a small region at the center of the viewfinder. I used partial metering on the scene above. I placed the pine tree's foliage in the center of the viewfinder. The meter dutifully proposed an exposure that would render that foliage medium gray. Since I wanted the foliage to be a little brighter than that, I increased the exposure a bit over what the meter recommended. I then composed, took the shot and checked the resulting histogram in the LCD. I was satisfied. If I hadn't been, I'd have adjusted the exposure up or down as needed.
Once we start seeing the world as a meter does, we realize that the world is full of scenes that will mislead evaluative metering. They include snow scenes, bright subjects surrounded by deep shadow, and mixtures of bright light and deep shadow. With partial metering, you can usually find a good exposure in such situations with either one or two test shots.
How does manual mode fit into this? You could, of course, use partial metering in aperture priority. In practice, however, that would mean:
(1) metering on the selected tone
(2) locking the exposure before you recompose
(4) repeating steps one through three for every shot you take. It's far simpler to set the exposure once in manual mode and leave it there. As long as the light doesn't change, your exposure will be correct.
You may be wondering how you know what tone to meter on when you're using partial metering. The answer depends entirely on how you want the image to look from a creative standpoint.
Suppose you want a conventional exposure, with the highlights fully exposed but not clipped. At this point, my own approach has become too intuitive to articulate. You might start by deciding what the brightest area in the scene is, and then metering on a tone that you estimate to be two to three stops darker than the brightest area. Check the histogram of the test shot in the LCD, and adjust the exposure if need be. With a little practice, you'll get surprisingly accurate at making these judgments without any conscious calculation. That's how I set the exposure when I photographed this canal at dawn a few weeks ago—another scene that evaluative metering would have mangled.
The dynamic range was too wide to get all of this scene in one exposure. Accordingly, I exposed first for the sky and water, metering a tone that I estimated to be a few stops darker than the highlights. Where I metered was an area of the sky roughly where the white dot appears here:
When I reviewed the result, the histogram told me that the exposure was about where I wanted it...a tad "underexposed," but underexposure enhances color, and the colors were what this image was all about. I then exposed again for the ground and combined the exposures to get the result.
It bears repeating that there is no such thing as an objectively "correct" exposure. The correct exposure is the one that produces an image that reflects your creative intentions. Most of the time, I like to have the brightest tone approach pure white, and the darkest tone approach pure black. If we use that approach automatically and invariably, though, we'll produce results that look automatic and unvarying. We'll also get some results that wreck the mood of the scene we saw.
Once you get used to setting the exposure manually, you realize that you can alter the "key" of an image through your choice of the tone you meter. If you meter a brighter tone, you'll get a lower-key image. If you meter a darker tone, you'll get a higher-key image. Few of us will want to push images very far in either direction on a regular basis. Many images, though, look better if they are somewhat "underexposed" in a mechanical sense.
We also encounter some scenes that call for an exposure that varies significantly from what a mechanical approach would call for. The brightest tone in this next image, for example, falls well short of white—and ought to do so, in my judgment.
The biggest advantage of using partial metering in manual mode is that doing so almost forces us to make conscious, creative decisions about exposure. In many situations, that's the only way to capture the image we're after.
Don Peters' photos can be seen here.
GeorgeL, thanks for the comment. All I have is the in-camera meters.
Don, thanks for the very informative post. Do you ever use incident meter or do you stick with the in-camera meters?
Leroy, thanks very much for the comment. Your method would also work, and just as well. I just learned the other way and it became habitual.