Practical applications of two commonly-used terms that are often misunderstood by photographers.
When used in imaging, bit depth quantifies the number of discrete colours (hues and tones) that can be reproduced in a digital image. The term ‘bit’ refers to the way computers handle data in ones and zeros, while ‘depth’ implies a range. In computer language, a single bit can only represent 1 or 0; in a pixel it would be either pure black or pure white.
Continuing in computer language, each time another bit is added, the number of possible combinations doubles; so for two bits you get 00, 01, 10, and 11, with three bits it’s 000, 001, 010, 011, 100, 101, 110, and 111 – or eight possible combinations. In general, the number of possible choices is 2 raised to the power of the number of bits. In that context, 8-bit depth represents 256 separate intensity values.
When you’re working in black and white (‘greyscale’) you’re only dealing with one gradient – from pure black to pure white. The illustration on this page shows how increasing the bit depth gradually changes our perception of the image from a series of separate blocks at low bit depths to a visually continuous range of tones by 8-bit depth.
Note how the visible steps in tones at bit depths between 1-bit and 4-bits gradually fade to become continuous tones by 8-bit depth. This explains the choice of 8-bits for JPEG images; tonal transitions are effectively invisible to human eyes.
Because colour images require separate channels for the red, green and blue bands of the visible spectrum we generally discuss colour depth on the basis of the bit depth per channel. Since JPEGs are 8-bit files, this means 256 tonal values will be recorded for each colour. Combining all three colour channels results in 8-bit JPEGs that contain more than 16 million (256 x 256 x 256) colours.
However, even 16,777,216 distinctive hues and tones may not be able to record the full tonal range you can discern on a bright, sunny day, especially when on a beach or a snow-clad hillside. That range is much higher than most camera sensors can capture so if you meter on the average mid-tones, the highlights will likely be ‘clipped’ to pure white and the shadows ‘blocked-up’ to pure black.
Cameras that can record raw files will save them at higher bit depths, usually 12-bit or 14-bit. which will provide scope for extending this range. Some cameras provide a choice of two bit depths. Image editors usually open a 12 or 14-bit raw file at 16-bit depth because there are significant advantages in being able to work at higher bit depths, even though you may not see a difference between 8-bit JPEGs and 16-bit TIFFs on the monitor’s screen, especially if it is 8-bit only.
Most image editors let you examine colour channels separately so you can check each one’s tonal balance. In each display, the dominant hues will appear lighter in tone, while the areas with the lowest level of the selected colour appear dark. (Colour saturation plays no role in tonal distributions.)
A typical colour image split into the three colour channels. Since its purpose is to show tonal distribution, each channel appears in greyscale.
How many colours can you see?
Normal human vision can discern approximately 10 million different colours – which is between 16-bits and 24-bits for each colour channel. So if you want to work with colour and see the true reproduction of subtle hues and tones, 8-bit per channel JPEG files may not be good enough. Interestingly, the new HEIF (High Efficiency Image File) format can support colour depths of up to 16 bits at similar file sizes to JPEGs.
While bit depths beyond 10-bits per channel are pointless if you’re only concerned with how well subtle tones are retained for viewing or printing, when it comes to editing images, higher bit depths will provide much greater scope for adjustments.
Why high bit depth is needed
If we only viewed our images on screens we wouldn’t require files with high bit depths. Most consumer TV sets have 8-bit displays, which is fine since live TV broadcasts are also 8-bit. OLED screens can support 10-bit colour and some streaming services offer 10-bit videos in HDR10 and Dolby Vision, usually on their premium plans.
Photo editing is where hidden differences will show up, so the best monitors are 16-bit displays and the best image editors should handle 16 bits per colour channel.
Image histograms can illustrate the benefits of starting with 16-bit files. Any adjustment that ‘stretches’ the histogram can produce posterisation – which causes continuous tones in an image to separate into distinct bands, as shown in the illustration on this page. Such posterisation occurs when the colour depth is unable to sample continuous tonal gradations.
The adjusted 8-bit JPEG image, which has a limited tonal range, is shown at the top with its associated histogram. Below it is a similarly adjusted 16-bit TIFF file from a raw original. In both images, the image tones have been ‘stretched’ by pulling in the ends of the original histogram. The resulting posterisation is clear in the JPEG image, where the sharp peak in the histogram (circled in red) shows where tones have ‘dropped out’, leaving detail-free bands in the sky. No banding is seen in the raw file below, which retains its full tonal gamut, as shown in its histogram.
Posterisation is relatively rare with present-day technologies – and normally absent when you’re working with raw files since with 16-bit TIFF files you can access up to 281 trillion RGB values. In addition, shooting in one of the wide-gamut colour spaces, such as Adobe RGB and ProPhoto RGB – both of which produce the best results with 16-bit images – virtually guarantees you won’t encounter it.
This pair of images shows how to identify posterisation. The image on the left is the original file; on the right is the same image with tones ‘stretched’ until they separate into bands, clearly seen in the drooping flower petal but also visible as dropped-out tones in the upper petals.
Bit-depth and colour space
Bit depth and colour space are different: bit depth specifies the number of possible tonal and hue increments, while colour space defines the maximum values (or ‘gamut’) of the hues. Having more shades and tints for each hue is equivalent to having a higher bit depth, whereas having a higher saturation is like having a wider colour gamut from pale pastels to the richest, deepest hues. It’s possible to have a high bit depth with low saturation – or vice versa.
These two photographs show how bit depth and saturation operate independently. Both images originated as 14-bit raw files that were converted into 16-bit TIFF format for editing. The bit depth is the same but the top image is highly saturated, while the saturation in the lower image is relatively low.
In practice, increasing the bit-depth reduces the risk of banding by creating more increments, while expanding the colour space gives you access to more extreme colours. But beware of pushing saturation up too high as over-saturated colours usually look unnatural.
Caution is also required with wide-gamut colour spaces like ProPhoto RGB because even though it encompasses more than 90% of the colours we can perceive, it doesn’t include some of them – especially in the green band of the visible spectrum where human eyes are most sensitive. It also contains hues we just can’t see, notably in the blue-green region.
The chromaticity diagram for the various colour spaces used in photography. The colours normal human eyes can see are contained within the coloured area; anything falling outside of that area is effectively invisible to our eyes. The Colormatch RGB profile shown in this diagram is close enough to sRGB to have become largely irrelevant, while the SWOP CMYK profile is used exclusively for printing. (Source: BenRG and cmglee – CIE1931xy blank.svg, CC BY-SA 3.0)
These ‘imaginary’ colours are included as a way to extend the actual gamut beyond the limits of sRGB and Adobe RGB. This is only really useful when editing images on a wide-gamut screen because, as you can see from the chromaticity diagram, the colour gamut of most printers is actually smaller than that of Adobe RGB.
If you use a restricted bit-depth and expand the colour gamut it will probably result in banding, especially if it stretches the bits so they are thinly distributed.
When working in 8-bit/channel sRGB you can normally expect smooth gradients with no perceptible banding. But those 16.8 million RGB values are spread out to fill a larger ‘container’, making them more vulnerable to banding than you’ll encounter when working in a much larger 16-bit colour space.
Output space
The output space is the colour space used to output images for viewing and sharing. Each of the main colour spaces will suit different applications, with sRGB being ideal for images that will be shared online and viewed on regular screens, while Adobe RGB and ProPhoto RGB are preferable for editing images.
For printing it pays to know the limits of the printer that will be used and its ink set. Some printing services and many basic multi-function printers with CMYK (Cyan, Magenta, Yellow and black) inks will only accept sRGB images, because of their limited gamut. For the same reason, sRGB files will also look reasonably good on screens that aren’t colour managed.
No printer can reproduce all of ProPhoto RGB’s real colours; but almost every photo printer can print some colours outside the sRGB gamut. The ink set will define how many colours can be printed. Dedicated photo printers with six or more inks should be able to reproduce images in the Adobe RGB colour space because they have a wider gamut. Most wide-gamut monitors will mimic Adobe RGB more closely than sRGB so they will look much better than sRGB screens.
ProPhoto RGB should be seen as primarily an editing colour space and images for printing should be converted directly from ProPhoto to the printer’s ICC profile – which will depend on the ink set and media on which the image is printed. All printer and paper manufacturers provide ICC profiles for specific printer and media combinations that are either supplied in the printer driver (in the case of media with the same brand as the printer) or downloadable free of charge from the paper manufacturer’s website.
Article by Margaret Brown (see Margaret’s photography pocket guides)
