Metering and Exposure Compensation

In any shooting mode except for manual, the camera calculates at least one of the shooting parameters for you. To do so, it measures the light coming in Through The Lens (TTL). Cameras have several ways of measuring and calculating the amount of light in a scene, which are called Metering Modes. Which modes are available depends on the camera model, but most olympus cameras have the following modes: Digital ESP, Center-weighted, Spot, Highlight spot, Shadow spot. These will be discussed later on but first we'll look at how metering works in general.

The Matrix

No, I don't mean the SF movie. A matrix is a mathematical table of values which can be used for calculation. This is relevant here, because cameras measure the light in many points to create such a matrix. They then assign a weighting value to each cell and calculate a (weighted) average. The difference between the metering modes is the weights that are applied and how the resulting average is used.

Recent olympus cameras use a 49-zone light meter. This means the image is divided up into 7 rows and 7 columns (7*7=49), presumably laid out like the leftmost image below. Other camera makers use a very similar system. Each zone here measures the amount of light it receives separately. Now, when you're taking an image, each zone will cover part of the scene (image 2).The light meter then averages out the light over each zone. Image 3 below shows the same scene as viewed by the light meter and image 4 shows the matrix this results in (I have arbitrarely used a scale from 0 to 100 but in reality the meter will get very different values, partly because the human eye sees light intensity logarithmically. For simplicity's sake, I'm ignoring all that as it doesn't affect the principle).

The unweighted average of all the values in this example matrix is 33. This means that if there were a metering mode that treated every spot as equal (there isn't), the measured brightness of this scene would be 33. The camera would then set the shooting parameters (aperture and/or shutter speed and/or ISO) in such a way that a brightness value of 33 would be "middle gray" or 50% gray. The fact that this image averages out to 33 and not 50 is proof that the metering mode used was not an unweighted average.

Olympus metering modes

If not an unweighted average, then what? Well, to be honest, nobody really knows. Camera makers have a tendency of being very vague about the specifics of their metering algorythms. However, while we don't have the exact numbers, we can describe them in big lines.

Digital ESP

ESP stands for "Electro Sensitive Pattern" which I'm sure makes it a lot clearer. Olympus describes it as a full-matrix (using all zones) pattern that "measures the brightness of the central area and the surrounding areas separately." This, again, doesnt say much, but we can presume that it means that the center zones have a different (probably higher) weight than the edge ones. Olympus performs a few more tricks on this (it's not a straightforward weighted average) but in essence that's what happening. The following image is a complete guess and shouldn't be taken as gospel, but this is roughly how I'm guessing the weighing is done, with brighter reds signifying a higher weight:

Important note: "Centre" doesn't necessarily mean centre here. If you have your camera set to [ESP+AF] it will "centre" around the chosen autofocus point. To set this up, go to the camera 1 menu, chose metering, go to the top value (ESP) and then press the right arrow. You can now chose between normal ESP and ESP+AF. For all other modes, centre is really the image centre (around the central AF point).

Centre-weighted

 The second matrix mode uses quite different weights than the ESP mode above. As the name suggests, much more significance (higher weights) is given to the zones near the center of the image. According to olympus, most of the data comes from an area of about 6% of the image. This means the metering pattern probably looks a bit like this:

A centre-weighted pattern is useful in situations where you have a relatively large central subject which contrasts significantly with the background or where you want to ignore the background for other reasons. Examples are a well-lit subject in a dark room (say a studio portrait with Rembrandt lighting) or a dark coloured subject in a bright area (say an elephant at noon).

Note that while the weights for the background are low, they're not zero so they influence the metering a little bit.

Spot

Again the metering area shrinks. In this case we can't even speak of a multi-zone matrix anymore. According to olympus spot metering uses around 2% of the image area. Since there's 49 zones, each zone would be around 2% of the total area if they were all equally large (which this seems to indicate, but isn't a certainty). The metering pattern is simple:

With spot metering you chose an area of the scene that is neutral, you put the central AF point indicator over that area, half-press the shutter button to lock the illumination, recompose and then take the picture. This may seem complex, but it does let you decide exactly what is neutral in your image (provided you have a neutral area). Of course, half-pressing the shutter also locks focus, which is annoying if your neutral area of choice is not supposed to be the center of focus. To get around this you can use the AEL function, but that lies outside the scope of this article.

Highlight/Shadow spot

These two modes use the exact same 2% pattern of spot metering. The difference is how the result is interpreted. As stated before with the othering metering modes the measured light is presumed to be 50% gray: a neutral value. If you use highlight spot then the measured value is considered to be about 90% gray: it is presumed you are measuring the highlights of the scene. Obviously he exact opposite is done for shadow spot.

You can use highlight and shadow spot in much the same way as normal spot, but instead of metering on a neutral area you should measure on a bright or dark area respectively. Highlight spot is a good way of avoiding blown highlights, especially for those cameras that are a bit susceptible to them (like the E-510).

Easily fooled

It's important to realise that no matter what metering mode you use, the light meter is essentially stupid. It will use the same calculation for every scene. More importantly it will use the same presumption for every scene, namely that on average the light is 50% gray (or about 90% gray for highlight spot and 10% gray for shadow spot). This is rarely the case, however. Look at the following images:

In the first of these, the scene consists of 43% black, 43% white and 14% middle gray. Using our unrealistic unweighted metering mode from above, this results in a nice 50. The image will look just like the scene. Great. The next image, however isn't as balanced. Here there's 14% black, 72% white and 14% middle gray. The resulting average brightness is 79. The camera will then try to get that 79 in the scene to be 50 in the resulting image, for example by selecting a faster shutter speed, and thus darkening the final image. The end result will be that the white in the scene will be light gray in the image, and the middle gray will become dark gray.

Luckily we have a tool to partially override the light meter: exposure compensation.

Exposure compensation

 So far we've seen how the scene brightness is measured by the camera, and how the metering mode then interprets these measurements. Exposure compensation lets you override what the meter is telling you and should be used when you know the camera will not do a good job on its own (after all, it's easily fooled) or when you want to over/underexpose for an artistic reason. The exposure compensation (also called exposure bias) can be dialed in by pressing the +/- button on the top or back of the camera and then turning the wheel, or by turning the dedicated wheel in cameras with dual wheels (like the E-3 and E-30). When you do so, the display will show the exposure compensation value. This value is expressed in EV which we saw at the start of this guide. Remember, +1 EV is doubling the light, -1 EV is halving the light. Now, say you dial in an exposure compensation of +1 EV. This tells the camera to do its measurements as per normal but then increase the exposure by this value. For instance if you are in aperture priority mode it might switch from f/4.0 (the measured value) to f/2.8 (one EV brighter). Conversely dialing in -1EV would go from f/4.0 to f/5.6 (one EV darker).

In practice, knowing the right exposure compensation needed for a particular picture is an essential skill which relies heavily on experience. Luckily, such experience can be acquired pretty easily by looking at an image you've taken and following these rules:

• If the image is too dark, increase the exposure compensation
• If the image is too bright, decrease the exposure compensation

 Or:

• Positive exposure compensation brightens the image
• Negative exposure compensation darkens the image

Here's some more guidelines for specific situations:

Use a positive exposure compensation when ...

• You are shooting a (smallish) bright (e.g. white) subject against a dark background
• You want to create a high-key image

Use a negative exposure compensation when ...

• You are shooting a (smallish) dark subject against a bright background (e.g. a black sheep in a field at noon)
• You want to create a low-key image
• You are shooting fireworks
• You are shooting into a light source (e.g. a sunset/sunrise)
• You are using an E-510. Just dial -0.7 EV in there as default.
• You are experiencing clipped highlights