Menu Close

More than a number: a closer look at dynamic range, Part 1

This image was shot with an exposure that protected the significant highlights then had its shadows lifted, exploiting the fact the camera captured much more dynamic range than it would typically include in its JPEGs. But can a single number capture this capability?

Dynamic range figures are widely quoted and sometimes discussed as if they are a measure of image quality. In reality dynamic range is an aspect of image quality but one that doesn’t come close to telling the whole picture. We’re going to look at what dynamic range is, and why putting a number on it isn’t a particularly good way to understand camera performance.

Image quality as a whole encompasses a range of properties, such as color and resolution (and hence considerations such as white balance and sharpness). But even if you’re just thinking about the tonal response of a camera or sensor to different brightness levels, dynamic range numbers still might tell you less than you expect.

What is dynamic range?

At its simplest, dynamic range is the range of subject brightness that your camera can capture, from the brightest (where the information ‘clips’) to the darkest usable tone. It can be a good indicator of how flexible the Raw files coming out of your camera are. But it doesn’t tell you much more than that.

At its simplest, dynamic range is the range of subject brightness that your camera can capture, from the brightest to the darkest usable tone.

Differences in dynamic range generally exist in the shadows of your image. Digital sensors have a hard cut-off in the highlights: any additional signal simply gets recorded as the maximum Raw value. Nothing beyond this is recoverable, so there’s limited scope for extra DR at the bright end.

If a new sensor design has more tolerance for light, its base ISO gets rated as being lower to encourage the use of more light: it still clips roughly the same highlights. So it’s in the shadows that additional dynamic range is found in and where the differences between sensors become apparent.

Noise makes things messy

Meanwhile, in the highlights

In practice there’s a little wiggle room for dynamic range at the top too, as the red, green and blue channels of your sensor won’t all clip at the same brightness level. But once one channel has reached clipping, there’s very little hope of reconstructing color-accurate highlights.

Your camera will generally do all it can to make use of as much highlight information as possible (leaving headroom unused would mean that the camera could have tolerated more exposure, which means every tone in the rest of the image is noisier than it needs to be). The amount of recoverable highlights in a Raw file generally comes down to a question of luck around whether the channel that’s clipped is critical to the color of the part of the scene you’re trying to recover, and how sophisticated a guess your chosen Raw converter can make, when trying to reconstruct the missing data.

Note that our definition above sets the lower cutoff as the darkest ‘usable’ tone. It’s a subjective judgement. There’s a technical definition of DR, often called ‘engineering DR,’ that uses the point where the noise level is equal to the strength of the signal from the scene (a signal-to-noise ratio of 1). This is much noisier than most people will accept, visually. Alternatively, Bill Claff at Photons-to-Photos has come up with his own cutoff, which he calls ‘photographic DR,’ but it’s still not necessarily the threshold you, personally, would find usable.

In addition to there being an element of taste to how much usable dynamic range your camera has, your workflow will also have an impact on the answer. Because the lower limit of DR is based on noise, it’s also affected by scaling images and by noise reduction. So if your workflow regularly includes downscaling your images for viewing, or applying some sophisticated noise reduction, this too will change how much usable dynamic range you find your camera to have, and this may not relate to any number you’ve seen quoted.

What dynamic range isn’t

But, even with all this ambiguity about where you should assess the lower limit of DR, there’s a bigger concern. DR numbers describe the distance between the brightest captured tone and the darkest usable one, but they don’t tell you anything about the quality of any of the tones between those two points. Even once you’ve decided your lower cutoff, two cameras that measure as delivering 12 stops of DR don’t necessarily have the same performance at stop 11, 10, 9… and so on.

Judging a camera by its DR numbers is a little bit like being blindfolded and grasping an animal’s tail. You can guess at some things based on how far off the ground it is and how it feels, but ultimately you can’t be entirely sure if the tail is attached to an elephant or a donkey.

In Part Two of this article, we’ll have a look at the factors that have an impact on dynamic range, which mainly means focusing on that messy bottom end. We’ll look at what noise is, including why the measured noise values in your image can be in the part of the image you perceive as being cleanest.

Author:
This article comes from DP Review and can be read on the original site.

Related Posts