There was a bit of an argument on Faceplop/ Melter (colour me surprised), about the right way to meter light for photography. So I thought I’d weigh in with my own version.
The reason for even thinking about exposure is because what it means is getting the right amount of light on your sensor/ film. To do that you need to measure the amount of light there is. To do that you need some form of meter (which is better than guessing).
Actually, it’s a two stage process. The first step is to measure the amount of light correctly. The second stage is to decide how you want to use that information. Let’s start with step one.
The amount of light emitted by the sun is effectively constant. The amount that reaches the ground (or the subject) varies with time of year, time of day and the weather conditions. This is why a meter is better than guessing.
Sounds easy though: point the camera at the subject and either press the shutter or change the camera settings to what the meter says. Most of the time this works, and the better or more modern the camera the more likely it is to work pretty well. If you have one of these cameras and you get good results, that’s the end of this article.
But… some cameras don’t have meters, or don’t meter light well, or the subject lighting is not ‘average’. This is where we need a better way to meter the light. The obvious tool for the job is a separate hand-held light meter. But there are two basic types and they work in different ways, which was the cause of the online argument.
The first and most common type of meter is a copy of the one you find in a camera. You point it at the subject and it measures how much light is being reflected back towards the camera. Providing the bright and dark areas in the subject average out, the reading is good to use. A reflected light meter can struggle if the view contains lots of bright sky, or is backlit, or is a bright object on a dark background. There is a specialised version of the reflected light meter that has a very narrow angle of view, so lets you meter on a single small part of the subject. These are useful if you can’t get close, or for measuring the brightest and darkest spots to calculate the full range of brightness (the reason for this is in step two, below). These spot meters are expensive though, and don’t give you the general average reading you also need.
The second type of meter measures how much light is falling on the subject. This type is an incident light meter. The idea is that light falling on the subject is the correct middle point that you need to expose for. The brighter bits of the subject will reflect more light and be brighter, the dark bits darker. Providing the total range of brightness fits within the sensitivity of your film or sensor, then this works very well and is immune to scenes that are not an average mix of light and dark. It does need you to measure the light falling on the subject though, which can be difficult if the subject is distant to you or under different lighting.

Which type of meter is best? The one you have with you, obviously. Both types work, providing you understand what they are measuring and if they might need some interpretation. The reason for the interpretation is step two.

Your sensor or film can record a certain range of brightness. Too little light and it won’t record. Too much and it will record as pure white with no detail. Ideally the brightness range of the scene will match the sensitivity range of the sensor, and it usually does (because sensors and film were developed to match the average range of brightness we encounter). So the average reading that a light meter gives you is intended to provide the mid-point of the camera’s range. How that average reading fits onto the range of a digital sensor or a film is shown below.

So despite all the noise about 18% grey and metering for the shadows or highlights, what you are trying to do is to find the average brightness and set it at the midpoint of the camera’s range. If possible you also set the camera so that the range of brightness in the scene or subject matches the range that the sensor or film can record. If the range of brightness in the subject is less than the range of the camera you can choose to move it up or down the camera range by giving it more or less exposure. Sensible people give as much exposure as possible, without the highlights going off the top of the scale. To be more accurate, the highlights in which you still need to see detail should be on or just below the top of the scale. If the sun is in shot, just accept that it will be overexposed. But if your subject has a white shirt or dress you may want any highlights to show a bit of detail and tonality and not be featureless white. This pegs the maximum exposure you can give. Alternatively, if the scene is low contrast (has a small range of brightness) you may want to give it more than the average exposure. This shifts the whole scene up the scale and will reveal more detail in the shadows. This is exposing for the shadows.
But what if the range of brightness in the scene is too great to get both the highlights and shadows within the camera’s range? You have options. One is to accept that part of the scene will not record. So you could let the highlights or the shadows fall off the scale. Most people keep the highlights and let the shadows go totally black, but it’s up to you.

Another option is to decrease the range of the subject. You can add light to the shadows with flash or a reflector. You can reduce the highlights by changing the lighting or adding some haze or filtering. Moving out of direct sunlight into open shade works well (but beware of the blue cast you will get from the sky).
A third option is to expand the range of the sensor or film. This is what HDR does for digital. With film you can play with different types of film, developer and processing. The aim with both is to be able to squeeze a wide range of subject brightness onto the narrower range of the sensor.
So, to get back to measuring exposure, a reflected light meter is saying ‘this is the average brightness of everything I can see’ while an incident light meter is saying ‘this is how much light there is. If everything in this light averages out, this is the correct exposure’. Which is better? If the full range of brightness in the subject fits into the range of your camera, the incident light meter is better as it can’t be fooled by non-average subjects. But if the brightness range is too great for the camera or you have something special in mind, you will need to set the camera differently to the average, change the lighting or take special measures to widen the camera’s range.
How do you know if the subject fits the camera range? Digital cameras win here if they can display a histogram or the under/overexposed flashies. If you can adjust the camera settings, you want the histogram shifted as far to the right as you can (as bright as possible) without losing any important highlights. With film I’m afraid it comes down to experience, and knowing that negative films, particularly colour negative, can take a bit of overexposure and still produce good results (due in part to that S shaped response, as above). If you have a separate light meter and you are close enough to the subject to be able to measure the highlights and shadows separately, try measuring the range. It works best with a reflected light meter. Take your overall average reading. Then measure the brightest highlight that should still show a bit of texture and tonality. This should be no more than 3 stops brighter than the average. The darkest shadow that you want a hint of tonality in should be no more than 4 stops darker. It is possible to capture a wider range, but this is about what works without taking special measures.
The special measures? There are ways of developing film that can capture a wider range of brightness. With slide film you are really stuck with what it offers. With digital you can try HDR. This combines a set of over-exposed shots (that capture the shadows) with under-exposed ones that capture the highlights. With some techno-magic the best bits of each are combined to compress a wide range of brightness in the subject to fit onto the range that the sensor is capable of recording. It can look strange if it’s done badly, and it often is.
The alternative is to base your exposure on what is important in the subject and let the rest fall where it may. If there are people in the scene, you would normally set the exposure so that you can see their faces. Just be aware that there is range of skin tones around the ‘average’ – don’t be like Kodak.
So if this is all getting too confusing, this is what you do in practice. Most subjects are average. Point your reflected light meter at the scene and angle it down a bit if there is a lot of sky in the shot. Or point your incident meter back towards the camera, with the meter in the same light as the subject. Job done. If the range of brightness in the scene is likely to be too wide, you will need to decide which end of the scale to keep, and it’s usually the highlights. On a digital camera take a test shot and look at the histogram or flashies. Reduce the exposure until the highlights are inside the histogram or stop flashing. With a reflected meter measure a highlight and give it three stops more exposure. It’s harder to gauge the difference with an incident meter but you could try taking one reading with the meter pointing at the camera and one pointing at the main source of light (often the sun). Try setting the camera at the midpoint of the two readings. And bracket – take extra shots with one stop more and one less of exposure. Bracketing is good for learning, as you can tell just by looking at the results that a scene like the one you shot really needs more or less exposure than what the light meter says.
So there you are. The purpose of metering the light is to work out how best to fit the scene onto the sensor. No one type of light meter is best – you need to use your brain with both of them. Incident light meters are less likely to be fooled, so may give more reliable results. Reflected light meters work from further away. The histogram or flashies on a digital camera do the same job. Light meters work best with average scenes, but luckily most scenes really are average (by definition). But look hard at your subject and the light and you will learn what different to average looks like and what to do about it. And then you can join the perpetual squabble on t’interweb about how exposure works. (As an aside, the only comparable geek argument is how countersteering works for motorcycles. So if you really want to start a flame war, ask people how best to expose for a countersteering bike.)
And by the way, you may have heard people either praising or damning the Zone System. All it does is help you try and fit the range of brightness in your subject onto your sensor, just as described above. None of this is magic, or even particularly difficult. It’s all about squeezing what there is into what you’ve got.