pull ing apart the pixel s
Many Nikons can shoot RAW files of different bit depths – here’s why it matters
Computing works with ‘bits’ – basically zeros or ones – and most of the images we see on screen use eight bits to describe each pixel, for each of the three colours, red, green and blue. Eight bits, also known as 28, gives 256 possible values for each of these colours. When these three colour ranges are multiplied together (ie 256 x 256 x 256) the number of possible colours in an image, from pure black to pure white, is 16.7 million. As these numbers can get out of hand when we talk about higher bit-depths (well into the billions), it’s normal to use bit-depth as a kind of shorthand. There’s the potential for confusion depending on whether we’re talking about bits per channel or the overall number of bits, but per-channel is usually used (eight bits per channel means 24 bits per pixel – eight for red, eight for green and eight for blue).
As you can see from the table below, even an eight-bit JPEG has more colour options than we need for viewing an image – it records 16.7 million colours, and we can only see about 10 million. That would be more than enough were it not for the work that needs to be done on digital images. If you make big changes to brightness or colour, for example, you are essentially increasing the differences between adjacent pixels. If the bit-depth is low when you do this (i.e. if there aren’t enough colours and tones to work with), there’s likely to be a noticeable jump, and this can appear as a hard line separating two areas. This is called posterisation, because the results can look like blocky poster art. it occurs across the image, but is most obvious in areas of subtle variation in tone, where it manifests itself as obvious banding.
There are two occasions when this happens. It always happens when converting the freshlycaptured image from linear to viewable (in other words, when applying the 2.2 gamma curve, described on page 73), and it may happen if you process an image extremely, for example when trying to save an under-exposed shot.
This is where having a higher bit-depth than eight-bit makes a difference. The more colours you have to play with, the less damage is done when they’re stretched out by the gamma curve. That’s why some Nikons, like the D810 and D4s, offer you the choice of 12-bit or even 14-bit capture. Both take up more space, and can slow down the shooting a little, but they ensure better image quality, especially in the dark areas. RAW, and in particular 14-bit RAW, definitely has the edge with very high-contrast scenes.
Because a RAW file contains more data, it still looks good after boosting
When we boost the tones in the JPEG version, we get horrible posterisation
Here’s an image straight out of the camera – it looks a little bit flat