I’ve written about metering before but I thought I’d recap an important point about the light meter in your camera: it’s a stupid, stupid device. Which isn’t to say that it’s useless. Far from it. Your camera’s light meter is an extremely useful tool. But, you have to know when you can trust it and when you can’t.
The light meter in your camera doesn’t know what your subject is or how contrasty your scene is. It doesn’t know if you’re in an evenly lit room or if your taking a photo of a bright moon in a black sky. All it knows is that a certain amount of light is entering the lens and it needs to adjust the aperture and shutter speed to capture a certain amount of it. Typically it will do pretty well. Front-lit scenes, even illumination—these are situations that fit the light meter’s view of the world. But take a very contrasty scene and the light meter is quickly fooled. It doesn’t see the bright and dark areas separately. Take the following photo, for example:
Notice the white shirt in the background? That white shirt was reflecting so much light into the camera that the light meter suggested a relatively fast shutter speed (since I was in Aperture Priority mode, I set the aperture and the camera chooses the shutter speed). Because of the shirt, it saw a lot of very bright light entering the lens. And to compensate, it turned down the exposure. The trouble is that the whole scene wasn’t that bright. The child’s face was much darker than the shirt and because of the fast shutter speed he would have been underexposed.
So here’s the counter intuitive result: when metering a scene that is very bright and contrasty, you need to add exposure to get the scene exposed correctly. That’s because the light meter will tend to underexpose scenes like this.
So I added nearly two full stops of exposure compensation. I could have switched to manual and changed the shutter speed to 1/100s instead (leaving the aperture at f/6.3). But my camera has a handy exposure compensation control that basically lets me do the same thing without having to change modes. I could also have tried switching to center-weighted or spot metering which wouldn’t have included the shirt in the light meter’s calculations.
If I hadn’t “overexposed” (according to the light meter), the subject would be a dark, underexposed figure. By adding nearly two full stops of exposure compensation, I’ve exposed the subject properly. Using the light meter’s suggested exposure would have saved the highlights in the shirt but also would have left the shirt light grey in appearance (underexposed). This is a perfect example of when blowing a highlight is acceptable. The shirt adds nothing to the photo and “saving” the shirt would have meant underexposing the real subject of the photo.