This is part one of Ming Thein‘s series on Understanding Metering.
An image from my recent Introduction to Wildlife workshop, and a very tricky metering situation – more importantly, do you know why, and what to do in a situation like this to achieve the desired exposure outcome?
One of the more important – yet almost always overlooked – aspects of camera operation is metering. Simply put, the meter determines what your final exposure is, and how bright or dark your image looks relative to the scene. Unless you are shooting manual – and even then – the camera’s exposure is determined by the meter. Add the fact that the eyes of a viewer tend to go to the brightest and/ or highest contrast portions of an image first (i.e. this should be your subject) – and it’s clear to see why it’s absolutely critical to understand both how metering works as a fundamental concept and any camera-specific peccadilloes that might exist. The last thing you want is to find that your camera drastically underexposed a once-in-a-lifetime shot of some critically important event because you didn’t know (or forgot) that the meter was extremely affected by point light sources*.
*This can actually happen. The meter in the Leica M8/9 is extremely sensitive to direct point light sources, and can often yield nonsensical readings of say 1/1000s ISO 160 for a shooting aperture of f4 at night – that’s because it’s picking up a street lamp. One can only hope the new M is less affected by this – the only solution to the problem I’ve been able to find is just go 100% manual at night.
How meters work
Depending on which exposure mode your camera is in, the meter will try to find a combination of settings that creates an image that averages out to middle gray in luminance, i.e. the histogram average is around level 127 or thereabouts. There are three exposure parameters the camera can use to control the amount of light reaching the image processor – note that the sensor is also now involved in the process – shutter speed, aperture and digital gain, i.e. ISO. If you fix any one of these variables manually – say by shooting aperture priority at a set ISO – then the camera varies the remaining parameters according to a set of rules in order to achieve the ‘correct’ exposure. If the correct exposure is out of adjustment range – e.g. the required shutter speed for a given aperture is too high – then you’re going to land up with an over or underexposed image.
In program mode, the camera controls both aperture and shutter values depending on its preset program; the photographer can usually shift the program to a different combination of values which still yield the same net amount of light hitting the sensor. In shutter priority, the user fixes the shutter value manually, so the camera alters the aperture. In aperture priority, it’s the other way around. In manual mode, the user fixes both values – the only thing the meter can do is display how far off the manually chosen exposure is from the correct exposure, or alter the ISO or flash. If auto-ISO is activated, then the camera will always default to the lowest possible ISO within the specified range in order to keep the shutter speed at or above a certain value – either user selected or 1/ focal length in second. (Note that for some cameras, using manual shutter and aperture values will cause the camera to shift the ISO rather than display the variance from correct exposure.)
Simple enough, right? So why are there so many different metering methods? My OM-D, for instance, has no less than five: matrix, centerweight, spot, spot low and spot high. The differences are down to the area of the frame the meter evaluates when deciding what the correct exposure should be. Note that in all situations, it will still try to expose the considered portion of the frame to middle gray – except this area might change. One uses the different metering methods in different subject situations. We’ll get into that in more detail later; first, there are a few more things that need explaining by way of background.
The meter itself is a photovoltaic cell, or combination of cells, whose output voltage over a certain area varies depending on how much light lands on it. The more light, the higher the voltage, which is translated into a brighter exposure. A particular chemistry’s electrical response is a fixed property of the material, and therefore consistent across different situations and cameras. Note that some meters require power to give a readout – this is because a base voltage must be applied across a semiconductor for it to respond to light, or to amplify the signal to a point where it gives an output that can be displayed – CCD meters are like this, for instance – other types of semiconductor photovoltaics do not require power because they already produce current on their own the minute light hits them. (Solar cells, for instance, fall into the latter category.)
Note that not all cameras have built-in meters; very early film cameras generally did not, and required the use of a separate handheld meter, or a particularly sensitive eyeball. My Nikon F2 Titan, for instance, comes standard with the unmetered/ plain DE-1 prism/ finder. Early Leicas are the same. A whole variety of hotshoe-based clip on meters are available, as well as handheld types. Modern digital cameras either use a separate metering CCD, usually located in the viewfinder (for an SLR) or use the imaging sensor (for any live-view based cameras) – this is obviously the most accurate possible method of metering given that the metering sensor also perfectly represents the response of the imaging sensor. (This was not always the case with film and separate meters; it was therefore highly important to know the characteristics of your particular chosen film.)
Incident vs reflective
All cameras’ built-in meters are of the reflective type. This is to say that they measure the amount of light reflected from the subject and hitting the camera; the advantage is that you don’t have to stand in the same light as your subject in order to obtain a reading – potentially problematic if your subject is say, a landscape that’s several kilometers away – but at the same time, they suffer from the disadvantage of not being able to obtain an accurate reading for very reflective subjects. Incident meters are always handheld (but handheld meters can be either incident or reflective) and are placed inn the same light as the subject in order to obtain an accurate exposure reading. The photosensitive portion of the meter is covered by a matte white dome in order to ‘average out’ the light measured by the meter.
The use of exposure compensation is simply translated into an offset of the zero point of the meter. For instance, if you dial in +1 EV exposure compensation, then the meter will add this to the calculated exposure value before displaying the final settings.
Flash meters a slightly more complicated. There are two ways to determine how much flash power is required to achieve the correct exposure. The first is by using an incident light meter next to the subject, firing the flash, and setting the camera with the valued displayed on the meter. This is the most precise method, but again, is often impractical if you do not have time to repeat a shot. The second, more common method, uses a very short duration and low-power preflash of known output in conjunction with the reflective meter to determine how much additional power is required to make up the gap between the trial exposure and a correct exposure, with the given camera settings. The adjustment to flash power is made almost instantaneously and a second, correct power flash is fired along with the exposure. This entire process is so fast that there is almost zero added lag. The disadvantage again is that partially transparent or reflective objects may not be correctly exposed as the metering type is reflective-only.
Histograms and expose to the right
The exposure histogram represents the evolution of the light meter into the digital age. It not only shows you what the average exposure should be over the entire frame, but how that exposure is distributed. For instance, it is important to know whether you have one uniformly gray area across the entire frame, or say two halves of the frame divided into 100% black and 100% white areas. A simple exposure meter that evaluates the entire frame would give identical readings for both situations. However, in the second situation, you would probably expose for the highlight areas to prevent loss of tonal detail. This would actually result in a final exposure that is slightly darker then what the whole-frame evaluative meter would suggest. Learning to read a histogram, is therefore a very useful tool for digital photography. Histograms and digital actually come with two others very useful tools. The first is the ability to display areas of the image that are overexposed – usually in the form of a flashing highlights warning; the second, is the ability to redraw the histogram based on the specific area of the image displayed. Note that availability of both of these functions depends very much on the camera you’re using. Some cameras are able to display histograms and overexposure warnings for individual color channels, as well as overall luminance.
Metering is actually much more critical in the digital age, simply because of the tonal response characteristics of the imaging medium. With film, there was a degree of nonlinearity and reciprocity era which translated into a little bit of latitude in photosensitivity; for negative film, this may vary by as much as 1 to 2 stops: the same exposure with different batches of film, even if the same emulsion type, may not necessarily result in the same final luminance. Add variation in the developing chemistry to that mix, and you can see why having high precision wasn’t all that critical. (Slide film is a different story; it’s very sensitive to over or underexposure.) However, digital photography is nothing if not repeatably consistent. Two identical cameras with identical exposure settings will yield an identical image under any fixed given situation. Changing the exposure by as little as a sixth of a stop will be consistently visible.
There’s also one additional characteristic of the digital medium that we need to take into consideration. This is to do with signal amplification and noise, and also the origins of the ‘expose to the right’ motto. Exposed to the right refers to ensuring that the histogram graph touches the right-hand (highlight) side of the scale, but does not exceed it. The reason for this is to capture as much total information as possible, with as little noise as possible. Underexposure in a digital image may be corrected for by increasing brightness. This is achieved by amplifying the signal; doing so also amplifies any uncertainty in the signal, which translates into increased amounts of digital noise – obviously not a desirable characteristic in an image. The advantage of exposed to the right is that we maximize the amount of single and minimize the amount of noise. The brightest tonal values in a digital image also contain the most information simply because of the way digital sensors respond to light. This translates into maximizing latitude for post processing, higher color accuracy, and less noise – in short, making the most of your image quality potential.
We therefore want to expose the image as brightly as possible, and then adjust the tonal map later in post processing – or do we? The reality is that in most situations this holds true. However, due to the nature of the total response of some sensors, there may be situations under which we do actually want to underexposed overexpose slightly in order to create a particular look due to the nonlinearity of tonal response. Of course, if you are a JPEG shooter and do not post process at all, you should expose at your intended final output level.
Note that this is much more of an issue for digital cameras than film ones, as the tonal response of film is non-linear – however, underexposure in a digital camera will usually result in undesirable noise when the luminance value is brought up to the desired level because it can only be done by amplifying a small signal. This in turn amplifies the uncertainty in the signal, i.e. noise.
One additional complication brought upon the digital photographer has to do with white balance and color temperature. Different colors have different luminosity values even under identical illumination; this is to do with the wavelengths that are reflected or transmitted to the imaging device, and their associated energy (luminance) levels. From a perceptual point of view, we see this as different brightness**. White balance is an important setting that comes into play here: it acts as the zero-offset point for color, effectively adding or subtracting different amounts of exposure compensation from the various channels to compensate for the ambient light. (This is how whites can still be rendered as white under extremely warm incandescent light if the correct white balance is used.)
**Nikon’s color matrix metering system has long compensated for this by using a metering CCD that had a color filter array over the top, both to aid scene recognition as well as increase exposure accuracy when presented with strongly colored subjects – for instance, yellow objects always render brighter than reds or purples of a given reflectance even if they’re illuminated under identical light – the color matrix meter compensates for this by increasing or decreasing the exposure if a scene is predominantly of one color or another.
We have several considerations here. The first and most obvious is of color accuracy – even so, this can be compensated for with the eyedropper tool in Photoshop providing we can find something white in the frame to set as a baseline. The less obvious problem is to do with individual channels. If the white balance is incorrect and a channel is overexposed, there is no way to recover this information afterwards. It is therefore important to set a white balance that is in the right ballpark – it doesn’t have to be perfect – to avoid this. Similarly, extreme underexposure of a channel will result in a lot of noise when compensated for afterwards. Generally, the auto white balance function in most cameras will get you in the right ballpark, but you will need to make adjustments afterwards.
The auto white balance function actually works in a similar way to an exposure meter – except instead of trying to average the scene out to middle gray in luminance, it tries to average out the scene to a perfectly neutral color.
The confluence of exposure metering and autofocus
As if the whole metering thing wasn’t complicated enough, DSLR manufacturers have started to use the metering CCDs to aid autofocus – after all, it’s an additional source of information that can be used to help track subjects especially when the mirror is down, and the main imaging sensor is not available. The flow of information is two way and affects both autofocus and metering. The autofocus system uses the spatial and color information from the metering sensor to track subjects by color and location across the frame, especially if they move out of coverage of the autofocus sensor array – the metering sensor always covers the entire frame. The exposure meter uses the autofocus information to determine which area in the frame is being focused on, and is presumably the subject, which the photographer presumably wants to have correctly exposed – in matrix metering mode, exposure is thus biased towards whatever subject is underneath the active autofocus point, or points.
I’m sure you can see there are a lot of presumptions involved. This of course means that the camera doesn’t always get either exposure or focus right when left to its own devices; the metering sensor may lack the resolution to distinguish between the desired subject and another similar-looking one, resulting in focusing errors; or the meter may be too heavily biased towards the area under the active focus area and thus yield erroneous exposures. A situation in which this might happen is say if your subject is much larger than the active focus area, and of a different luminance value. Anything small and reflective almost always causes problems, too.
The bottom line is that it pays to take control of both your meter and focusing system: without this, you can never be fully certain of what your camera is doing; I seldom use auto-anything especially with DSLRs since they do not meter off the imaging sensor (unless in live view, of course).
To be continued in part two! MT
About The Author
Ming Thein is a photographer based in Kuala Lumpur, Malaysia. He specializes in watches, food and architecture / interiors. He is also the author of the Photography Compendium app. You can follow his excellent blog here, his photography on Flickr and his musing on Facebook. This article was originally posted here.