Home » Opinions » Digital Photography Tutorial - Dynamic Range And How To Fake It

Dynamic Range and how to fake it

by | Go to comments

Share:

In my camera reviews I often mention something called dynamic range, and how some cameras have more of it than others, however some of you may not know what dynamic range is or why it's important, so I’ll try to explain it in some detail, and also show some ways in which dynamic range can be improved in digital images.

In short, dynamic range is the difference between black and white. All visual recording media, be they film, CCD or CMOS sensors, or even the human eye, have a limited range when registering degrees of brightness. Beyond a certain level of brightness, the incoming light will register as pure white, and below a certain level it will register as pure black. It is the range of available tones between these two points that is the dynamic range (sometimes called the exposure range) of the medium.

The best way to illustrate this is with an example. In this photograph, which was shot using standard matrix metering, the light coming through the window on the left is too bright for the sensor, and registers as pure white, with no visible details. The highlights have been ‘burned out’. Meanwhile the darker shadows in the ceiling vaulting are too dark for the sensor, and register as pure featureless black.





It has traditionally been believed that the dynamic range of digital sensors is not as great as that of photographic film, and indeed in some cases this is still true, but over the past ten years continued developments in sensor and image processor design have improved the dynamic range of digital cameras considerably, to the extent that some top-end professional DSLRs now have superior dynamic range to most popular types of film, and even standard compact digital cameras have about the same dynamic range as their film-using counterparts. Whichever medium you use, the total effective dynamic range, the difference in exposure value between black and white, is unlikely to be more than six or seven stops, due mainly to limitations in the ways that images are stored, processed and displayed. Again, the best way to illustrate this is with an example.

This shot was taken using spot metering to expose the interior walls correctly. As you can see there is detail visible in the shadows, but the highlights are badly burned out. The shot was taken at an aperture of f/8.0 and a shutter speed of half a second.





This next shot was also taken using spot metering, but this time exposing for the window highlights. There is now plenty of highlight detail visible, but the rest of the scene is reduced to featureless black. This shot was also taken at an aperture of f/8.0, but at a shutter speed of 1/200th of a second. The difference in exposure values between these two shots is approximately seven stops.




Digital cameras are very good at capturing detail toward the shadow end of the dynamic range, and it is possible to bring out some shadow detail by adjusting the image using an editing program, but in extreme cases like this, levels adjustment is unsatisfactory because it results in amplification of image noise and a breakdown of colour reproduction, as seen in this example. This is the same shot as the previous example, but with the levels adjusted in Adobe Photoshop.





What is clearly needed is some way to combine the shadow detail of the first example and the highlight detail for the second example into the same shot. However, as I mentioned, the way in which digital images are usually stored and processed limits this possibility. The majority of consumer digital cameras, most image editing programs and the monitors and printers used with home computers, store and display digital images in the 24-bit “Truecolor” format. What this means is that for each pixel of the image, 8 bits (bits are the 1s and 0s that make up all computer data) are used to describe each colour channel of each pixel, giving a total of 256 gradations of red, green and blue in each pixel, using 24 bits (3x8) per pixel for a total of 16.7 million possible colour combinations. This might seem like a lot, and it’s certainly enough for most common applications, but in nature there are no such gradations, and to capture the full dynamic range of a scene like the one above we need a more powerful data coding system.

Go to comments
comments powered by Disqus