Dynamic Range

Dynamic range is a term that is used in photography to indicate the tonal range, the range of brightness that exists within a scene and that a recording medium can capture in a single image. In some respects exposure range might be a more appropriate and understandable term to use for most ordinary camera users although it could then be confused with exposure latitude, which arises out of it but is not the same.

Dynamic range is important because whilst the range that an image recording device can capture, whether it is an electronic sensor or film, is fixed and cannot alter, that found in any scene to be captured can, and vary widely. In general, when you take a picture of a scene, you are trying to record all the tones present in it. This means for an ordinary typical outdoor daylight scene, a shot that has a range of tones from nearly white to almost black. But the actual lighting conditions can vary enormously. Take a foggy day for example. All that might exist is a few shades of grey. No deep shadows and no bright highlights. By contrast, on a bright sunny day in summer there may be a huge difference, with the most brightly lit places being many times brighter than those in areas of shadow. It is not uncommon to find a shadow to brightness ratio of over 200x and sometimes an exposure range of over 13Ev stops.

When the scene exposure range is less than that of the recording medium then a range of different exposures can be used all of which will capture the full range of tones present. When the ranges match, then only one exposure will correctly cover and capture all. And when the scene range exceeds that of the recording medium then choices have to be made as to whether to forfeit detail in the highlight or shadows areas because no exposure will record all the range of tonal brightness. So the wider the dynamic range of the equipment you are using the more flexibility you have - exposure latitude - and the better the chance of capturing all you want in terms of tonal quality in an image as the scene dynamic range increases.

Sensor response to light

A problem that arises with digital camera use is that electronic sensors respond to light in a different way to that of film. It is called tonal response. And it means that if you use your digital camera in the same way as you have used film cameras in the past, then the results you get will not always be the same.

Basically the chemical grains on film that capture the light - absorb it - do not do so at a constant rate. When they are first exposed the rate is slow. Then it gets quicker until the grain is nearly full when the rate slows down again. And when the grain is full it stops absorbing any more, whether it continues to be exposed to light or not. And each grain is separate and self contained. What happens to one does not affect those adjacent to it. Obviously the amount of light absorbed depends on how much light fell on the grain and was collected. No light collected means the pure black tone results. The maximum amount collected means the pure white tone results. And various amounts collected result in the different tones between. 

Electronic sensors are different. They gather light at a much more even rate. It still starts slow and then increases but the biggest difference is that although it slows down again as the pixel reaches the maximum level that it can record it doesn't stop, it can't 'switch off', when it has reached it's maximum, and this is where the problems with digital capture, and digital camera dynamic range lie. In practical terms the light hitting the pixel when it is full 'spills over' and affects the pixels surrounding it. And this has a 'domino' effect. Each pixel affecting others around it. It's called 'blooming', and it's an effect that is often see in bright areas of digital images where the dynamic range was exceeded. In other words where the scene tonal range, the levels of brightness, exceeds that which the sensor can cope with. And It's more commonly referred to as blown highlights simply because it's an effect that only affects the lighter end of the tonal range.

If you were to make a graph to represent the rates of capture, from empty to full, it would look something like the one below. It's not accurate but just an illustration to show the differences clearly.

Some sensor makers have attempted to make adjustments to sensor design to reduce or overcome the difficulties that occur with the highlight range by using gates that 'drain' away excess light capture passed the maximum, a bit light an overflow pipe on a water tank or toilet cistern, whilst others like Fuji have incorporated different sizes of pixels. One for the main light capture and one for highlight capture. The smaller the pixel is the slower it is to collect light, so the small pixels can produce highlight information lost by the bigger pixels. Camera makers have also attempted to alter the tonal value of the output files, and thus the tonal curve, by processing algorithms.

What needs to be understood in relation to this is image contrast. Lowering contrast increases the tonal range, increasing it decreases the range. Low contrast equates to a wider tonal range, finer detail, more tonal steps but less distinction between them. High contrast produces a narrower tonal range, a loss of finer detail graduation, less tonal steps but with greater distinction between them (Posterization). Slide film is high contrast and high density 'reversal' film - the image produced is positive as with a digital camera image - because it is designed to have very bright light shone/transmitted through it continuously (and resist the subsequent heat produced) so the image can be displayed at a large size on a screen or wall etc. As a result the colours look bright and the image sharp because of the greater delineation between tones.

By their very nature print negative films are of lower contrast than slide film because they are intended solely for printing, using a short burst of less bright light, not continuous, and their overall density is also therefore lower. B&W print film has a wider range than colour print film because it comprises of just a single emulsion layer instead of three. Producing prints from slides is harder than from negatives, not only because they are positive images, but because of their combination of high contrast and high density. 

The problem with digital images is that they are required to be both viewable and printable. So a reasonable balance between the two is needed. The wide DR and tonal range that would result from low contrast images would not suit most digital camera users.

Another point is the human visual perception of tonal values. When we look at a range of tones our eyesight attempts to display it all. But although we have a wide dynamic range ability, we can see in moonlight and in bright sunshine, our eyes need time to 'adjust' to the light levels, the tonal values present, and we can't cope with a wide range variation at the same time. There is also a limit to the maximum brightness that we can cope with. It's why we wear sunglasses. The bottom line is that when we view a scene that has a mixture of brightness in it, and the levels are wide, the lighter tones, the brightness, tends to swamp the darker one's. 

Basically our eyesight is tuned to finding the brightest tone in any scene presented to it, however small in volume, setting that as the maximum tonal value to be captured, and then working down the scale from there. If darker tonal values past a certain level cannot be captured so be it. And this translates into how we view the images we take. Bright areas with no tonal detail offend us more than shadow areas because we expect to see detail there. And film capture, print film capture in particular, works along similar lines to our eyesight. Highlight values are retained better for longer in preference to shadow detail.

Blown highlights with film capture - loss of detail in bright areas - although they exist, don't happen to nearly the same degree as with digital for two reasons. One the effect is contained as maximum light level is restricted to individual grains and can't be 'spread around'. And because the rate of absorption slows down considerably towards the grains maximum capacity it's not often that the situation occurs. The exposure usually ends before it happens.

Generally speaking bright highlights also tend to look worse with digital than film because you don't normally look at film with bright light, but as a print. White paper isn't really white, more like almost white, and never at a strength that hurts your eyes, because it doesn't reflect all the light falling on it, the surface isn't smooth enough, only a mirror comes close to doing that, just as black can't absorb all the light hitting it, only a black hole absorbs all energy. So the density of prints, the dynamic range, is not high, between 1.7 and 2 on the density scale, about 6.5 Ev. Looking at digital images on screen makes the highlights far more prominent than when the same image is printed out, and this should be taken into account when dealing with blown highlights, lost shadow detail, and how problematic they are.

Metering

In order to assess the dynamic range that exists in a scene and find the right exposure value - Ev - that will produce an image capturing as many of the tones present as possible an exposure meter is used. Although in the distant past this was done manually by hand meter, today it is normally an automated procedure done by the camera itself, although some still use individual hand held meters. Please refer to  Metering  for full details of the different ways to meter a scene.

Using and relying on the automatic default metering that most camera are provided with these days, some form of multi-point/zone pattern, is not a problem when the dynamic range of the scene is much less than that of the recording device used, in this instance a digital camera, but becomes one as the range differences narrow and especially when, as so often happens, the scene dynamic range exceeds that of the camera's dynamic range abilities. Or should we say that of it's sensor.

When this happens the situation arises, as we have said above, that choices have to be made. Where to sacrifice image detail. In the highlights or shadows. If you rely on automatic metering then it will make that choice for you and you may find later when you come to look at your shots that it wasn't the choice that you would have made for yourself. The reason this happens is that exposure metering is designed to do one thing. Measure the amount of light it receives, not measure the tonal range. It's meant to be the same thing, but it isn't. Which is why many digital cameras tend to produce blown highlights. The metering used has been developed over many years to suit film exposure with it's greater latitude towards highlights.

Each tone reflects a certain amount of light. The darker the tone the less light, the lighter the more light. Obvious really. Anyway metering works by measuring the amount of reflected light it receives. The total volume, and averaging that to find the 'average' light tone for the scene being metered, or the area the metering covers - depending on the metering method employed -  if this is not the whole scene. So it finds an average tone, the midpoint in whatever range of tones exist. And uses that on which to set the Ev. And it uses the amount of light, the brightness value in volume terms, on which to make that decision. 

Do you see the problem? This doesn't, and can't, measure the number of individual tones in a scene. It works okay whilst there is an equal area - volume - of each tone present, but falls down when there isn't. When there is an unequal area of tones, then the smaller volumes get swamped by the larger ones. Think of it as like mixing paint. If you take a lot of black, and add just speck of white what happens? Nothing. The black swamps the white. And a lot of white has to be added just to get a grey colour, an equal volume to that of the black to get a grey that is tonally halfway between black and white. It's the same in reverse. You've got to add a lot of black to turn white grey.

Now this isn't so much of a problem whilst the dynamic range of the sensor is greater than the dynamic range of the scene. Under or over exposure will produce an image that is too light or too dark and it will be dull and flat, because the contrast is low. But this can be altered without loosing any detail at either end of the tonal spectrum. In other words both the highlight and shadow detail can be preserved. This is called exposure latitude. A bit like elbow room to work. But this changes once the exposure difference is greater than that of the dynamic range of the sensor. The dynamic range of the scene doesn't have to exceed that of the sensor for this to happen, just the use of the wrong Ev. When this occurs tones at either one end of the range or the other, and often both, get lost. And of course when the scene dynamic range exceeds that of the sensor, then the problem just gets worse.

Basically what happens is that any tone past that which can be captured becomes either pure black, or pure white, and in either case an absence of any detail. There is no tonal value present to represent the detail in the scene. This often happens when you take a shot of a scene that includes a lot of bright sky. The volume of the brightness skews the camera's metering into under-exposing the scene. And all the darker tones do not record. So you end up with a sky full of detail, and nothing else. Alter the exposure to suit the other tones so they record properly, using the long time method of 'metering off the ground', (in other words using a mid-tone subject - grass or tarmac are often cited), and the sky becomes a colourless and featureless blank white space in the shot. 

Here is the problem illustrated in diagrammatic fashion using three groups. The red box is the sensor dynamic range. The grey step wedge represents the scene tonal range. The red line is where the mid-point tone between black and white is, the tone on which the 18% grey card is based. And the yellow spot is where the middle of the sensor dynamic range sits.

The first set represent the situation where the sensor DR exceeds that of the scene. The second, where they match. And the third where the scene DR exceeds that of the sensor. 

As a rule lost tonal detail at either end of the tonal scale is called 'clipped'. So when you read references to 'clipped highlights' or 'clipped shadows' this is what is being referred to. It's been 'chopped off' by the limits of the dynamic range, in other words - thrown away, or perhaps we should we say not recorded, it amounts to the same thing.

 Getting the correct exposure

Overcoming the problem of blown highlight detail is not always as simple as it might at first seem. The first rule would seem to be one of ensuring that the exposure that your digital camera takes is tailored to prevent the highlight detail disappearing. In other words expose for the highlights. Which means in reality less exposure. But doing so in practice isn't quite as simple as it sounds in theory, and relies mostly on how well you know your camera and the light levels it has trouble with. And how good you are at spotting areas in a scene that are likely to become highlights. Spot metering of the brightest parts of a scene don't always work because then the exposure may be too short and prevent darker tonal detail recording at all. Under exposure can be just as bad in some cases as over exposure. So it's getting the balance right that's important.

However over reliance on the multi-pattern type metering alone, without intervention, is certain to lead to problems, the trick is to be able to judge a scene and know when intervention is needed. It really comes down in the end to experience. And this is only gained by using your camera, and experimenting. In the past with film this could end up being costly and you had to keep careful notes to keep track of the different exposures you had tried.

With digital cameras it's easy and costs nothing, just a bit of battery power and your time and effort. All the information about each shot you take is preserved in the EXIF data stored with each image file. Date, time, aperture, shutter speed, ISO etc. So it's easy to alter exposure settings from that recommended, try different metering patterns, under and over expose using exposure compensation and so on. And then view the images later and compare them. Slowly an idea of when your camera starts having problems with exposure will emerge. And what you need to do to offset it. Because no two camera models, let alone makes, react in exactly the same way, and especially when DSLR's are involved because as well as the different processing algorithms, the lens used can also have a marked effect.

For example we often take readings in both multi-zone and spot and compare them to see how wide the difference is. And when we don't have the time to do this we have learnt to set a measure of exposure compensation, either a -½ stop Ev or -1stop Ev, depending on the contrast levels. Earlier this year when we were in China, during the day, when temperatures hovered anywhere between 25º-40º and the light was extremely bright and the contrast high, we just left -1stop Ev dialed in all the time. Whatever metering pattern we used. And most of the rest of the time, day, night, summer, winter, we now leave a -½ stop Ev permanently dialed in.

A small part of the problem arises from how metering is set up to work. The use of the mid-point or average tone. Because of the way sensors respond differently to that of film, that shadow detail is collected more quickly, our basic opinion is that exposures, which have arisen over many years to suit film and it attributes, need to be altered to better suit digital cameras.

In other words metering should not be calibrated to suit the tonal mid-point and film response but to suit sensor response, perhaps by changing to using a slightly lighter tone as the metering benchmark measure. Which is what we are basically doing leaving exposure compensation permanently dialed in. From digital camera reviews that carry out exposure/dynamic range tests it would seem that with the latest DSLR's makers are starting to alter the metering to better suit sensors. The trick is to do this and prevent lost highlight detail whilst still keeping a decent contrast level. 

At the moment with a digital camera under exposure is far preferable to over exposure when the correct exposure loses highlight values. You can always increase contrast in an image if it is too dull and flat looking, which will happen with under-exposure, but although you can also reduce contrast in an image the same way due to over exposure there is one important difference. Highlight detail lost by over exposure cannot be recovered, it just isn't there, it hasn't been captured in the first place because of the way the sensors work. It's true with under-exposure less shadow detail is captured but as we have said digital camera sensors are pre-disposed towards the capture of shadow detail at the expense of highlights, so it's just creating a more even balance, and usually one the average human visual system is more satisfied with seeing. 

There are those that will tell you that this attitude is wrong. That under exposed images corrected and altered by increasing their contrast results in an effect called Posterization, the reduction in the number of tones present in an image. This is absolutely correct, it does. But this is what contrast is all about. The relationship of one tone to another, the difference between them. That's what the word contrast means. To compare/a comparison/ reveal a difference. To make the difference between tones more visible you increase the difference, the contrast. The more you do this the greater the difference becomes, a greater contrast, and this occurs because the tonal steps are reducing. The greatest contrast of all you can have in an image is composed of just two tones, black and white.

So it's really a one way street as far as image exposure and digital cameras sensors are concerned. Certainly at the present time. Expose for the highlights, as and when you can, no matter what the supposed Dynamic range of the camera you are using. And adjust the images later using levels to get the best final result.

To understand more about digital camera dynamic range and how to correct under exposed shots please see, 

Dynamic Range Assessment   and  Using Levels