The term ‘dynamic range’ is used in all manner of situations, but an understanding of what it means within the context of photography is essential for getting the best results out of your digital camera
The technical definition of dynamic range, as the term applies to digital photography, is the ratio between the darkest and lightest tones that a camera’s sensor is capable of reproducing in a single exposure.
To understand dynamic range, it might help to forget about digital photography for a moment and think about how we experience sound. A great number of frequencies of sound exist in the ambient ‘noise’ of the world at all times, with infinite variances between the frequencies. Our brains, however, can only register frequencies within a relatively small range – roughly 20-20,000 hertz. Some people can distinguish a wider range of sounds than others – the dynamic range of their hearing, in other words, is wider.
To understand dynamic range as it relates to digital imaging, it is important to remind ourselves of what digital photographs actually are, and how they are created.
Back to basics: digital photography
A digital camera’s sensor has millions of photosites, which convert light coming through the camera’s lens into electrical signals. Above the photosite ‘layer’ is a colour filter array consisting of a pattern of red, green and blue filters. There are twice as many green filters as there are red and blue. Each photosite, therefore, records the luminance level of one of the three primary colours. These ‘raw’ RGB values are then combined via interpolation, to make a true colour, which is assigned to a pixel in the resulting photograph.
Each pixel in a digital photograph has a combined RGB value that describes its colour. A pixel that has RGB values of 184, 136 and 116, for example, is a typical Caucasian skin tone.
When the light arriving on the photodiode drops beyond a certain threshold, the signal generated is so low as to create a black pixel (with RGB values close to, or at, 0, 0, 0). Likewise, when the luminance rises above a certain threshold, the pixel will be white (with RGB values close to, or at, 255, 255, 255). If all three values are 255, the pixel is completely white and, again, contains no recoverable colour data.
In the same way that our brains can only process a limited range of audible frequencies, a digital camera’s photosites can only make use of a limited luminance range when creating a digital photograph. If subtle highlight tones are reproduced simply as white, and subtle shadow detail is ‘blocked out’ to black, we say that the camera has a restricted dynamic range.
Even cameras that have a relatively limited dynamic range will cope well with low-contrast scenes, like a foggy landscape with a narrow tonal range. The problem comes when a scene’s tonal range is wider than the camera’s sensor is capable of recording.
A daylight landscape with the sun in the frame is a good example of where the sky is significantly brighter than the foreground. The difference in luminance between the brightest and darkest areas of the scene may be too great for the camera’s sensor to cope with in a single exposure. This is why landscape photographers traditionally rely on neutral density graduated filters to ‘hold back’ bright skies in images that have been exposed for the foreground. Alternatively, multiple images of different exposures can be combined to create a single image with a higher dynamic range (see the High Dynamic Range Image section below).
High Dynamic Range Images
One way to get around the often limited dynamic range of a digital camera is to create a High Dynamic Range (HDR) image. The aim of HDR photography is to take a series of images that capture more highlight and shadow detail than could possibly be recorded on a digital camera sensor in a single shot. These images are then blended together to create a single image that more closely matches the scene as you saw it.
What your eye can see
The human eye has a far greater dynamic range than the sensor of a digital camera. The exact range depends on an individual’s eyesight, but it can be as much as 24EV, which is almost double the 11EV or 12EV dynamic range common among digital imaging sensors. Part of the reason for this is that the pupil of an eye acts like a constantly adjusting aperture, opening wider to let in more light when looking at a dark scene, and closing so as not to damage the eye when viewing a particularly bright scene. If you are in a forest on a bright day, you will be able to see the forest floor and, if you look up, the detail in the sky. However, the restricted dynamic range of a camera generally means that any image of this scene will result in either a completely burnt-out sky but with well-exposed trees, or a detailed sky and very dark trees. By creating an HDR image, both highlight and shadow details can be captured.
Creating a good HDR image begins with having a good set of pictures to work with, which means having a series of shots that provide the final HDR image with a full range of tones. To achieve this, it is vital to have one exposure where there are no large burnt-out highlights, and another where there are no large completely black shadow areas. This allows you to create a final image that shows detail in both these areas.
Most digital cameras have the ability to shoot bracketed exposures of at least ±2EV for three shots, with many capable of bracketing for an even greater range than this. However, in all but the most high-contrast lighting situations, three exposures set to -2EV, 0 and +2EV should be enough to capture sufficient detail in both the highlights and shadows.
Obviously, a sturdy tripod should always be used when shooting an HDR sequence. This will ensure that the camera does not shift its position slightly between exposures.
Size matters: photosites and dynamic range
Due to the way in which photosites work, the smaller they are, the lower the dynamic range of their sensor is likely to be. Imagine a photosite as a bucket that collects light, rather than water. The larger the bucket, the more light it can collect in a given time, and the less liable it is to overflow.
Remember that the amount of light a photosite collects determines the amount of charge it is able to generate. Once full, a single photosite cannot deliver any more charge and a white pixel will be the result. Light might still be coming through the lens, but it ‘overflows’ and goes to waste.
A larger photosite fills more slowly, so more light is required before a white pixel is recorded. This has an effect in bright areas, where cameras with larger photosites tend to do a better job of reproducing subtle tonal gradations. In low light, larger photosites pay off in increased sensitivity and a wider range of high ISO speed settings.
Older DSLRs tend to have larger photosites than their modern counterparts because their sensors are less densely populated. However, this does not automatically mean that older DSLRs deliver images with a wider dynamic range. Improvements in sensor design mean that today’s highly populated sensors have a better dynamic range than many of the older cameras, despite the photosites being smaller.
Dynamic range and file type
In order for JPEGs to be usable straight from the camera, they need to be relatively contrasty. However, if contrast is increased, dynamic range suffers because subtle tonal gradations are lost. Raw files typically need considerable work post-capture to bring them to life, but retain a wider tonal range because they are ‘raw’ and unprocessed.
As JPEG is a ‘lossy’ file format, images are made smaller by reducing the number of tones contained within them. In other words, data is lost for the sake of file size. Compression is especially noticeable in highlight and shadow areas, where subtle tonal gradations can be lost.
A camera’s dynamic range in raw mode is difficult to measure, because different raw converters apply different amounts of contrast expansion to raw images when they are opened.
A histogram is, very simply, a graphic representation of the tonal spread of a digital photograph. In a scene containing mostly bright elements, much of the data ranged along this line will be to the right of the scale, towards 255 (white), and in a scene containing mostly shadows it will be further to the left, towards 0 (black). The height of the graph represents the proportion of tones of a particular value. A sharp ‘spike’ somewhere on the scale represents a lot of data in a narrow tonal range.
Camera metering systems are designed to deliver accurate exposures from an average (mid-grey) subject. In a ‘perfect’ histogram, therefore, most of the tonal information in a scene is in the midtone areas.
When the range of tones in a scene exceeds a camera’s dynamic range, picture data ‘falls off’ either end of the histogram. If the histogram appears to continue off the graph, data has been lost.
Dynamic range optimisation
The dynamic range of a sensor is fixed, but many DSLRs now feature some sort of dynamic range ‘optimisation’ function. Sony’s Dynamic Range Optimiser, Canon’s Highlight Tone Priority and Nikon’s Active D-Lighting are all designed to help photographers get the most out of their camera’s dynamic range.
All these functions work in a similar way and involve the camera applying either a blanket or selective tone curve(s) to images before they are written to the memory card and saved as JPEG images.
The radial graduation target shown below was printed onto Ilford photographic paper. It shows a complete tonal spread from White 0-255 Black, and is intended to test both the ability of a camera to deliver smooth tonal gradations and to demonstrate the effect of dynamic range optimisation functions.
By setting the metering and white balance from a grey card beforehand, it is also possible to see how the calibration of a camera’s metering affects tonal reproduction in the images. Because it is a reflective rather than transmissive target, though, the black areas in the corners of this radial graduate are likely to be reproduced as a dark grey, rather than 0 Black, by all cameras.