Most Photoshop images are going to be in 8 bit pixel mode or 16 bit pixel mode. In a digital image with stills photography, a pixel is a small square of digital information.
The pixels are laid out in a mathematical grid and do not deteriorate when copied - the mathematical pattern is simply reconstructed.
So what does this mean in terms of image size and quality?
A bit is a single unit of colour information. A 1 bit pixel can be one of two things in the computer binary system; it can be 1 or 0, on or off. In practice it can be black or white. In simplistic graphic terms, imagine a pixel as a rod with the colour on the end, so this 1 bit pixel can only represent black or white.
An 8 bit pixel has greater pixel depth or bit depth and can contain and display much more colour information. As each bit has two choices, 8 bits have 2x2x2x2x2x2x2x2 or 256 choices, which means 2 to the power of 8 or 256 colours. A complete image which is only 8-bit is likely to be a greyscale or an indexed colour image.
A colour image in this category will have 8 bits on each of the red, green and blue channels.
The total pixel content of the image combine to make the file size. An 8 bit pixel uses one 'byte' of computer memory and a thousand bytes amount to one 'megabyte' of memory.
So what is the difference between 8bit and 16bit and does it matter?
The greater the bit depth or pixel depth in this 'rod' of digital information, the greater will be the number of colours and information stored. Therefore, the digital image will give a more accurate representation of the original.
Photoshop CS can handle most functions in 16 bits per channel, or a 48 bit image. Earlier versions of Photoshop can handle only a limited range of tasks in 16bit, while Elements will work in an 8bit per channel image.
Many scanners can scan film in 16bit per channel mode and this will capture considerably more information and hence more colours, tones and details. In actual fact, PS works in 15 bit + 1 unit, which amounts to 215. This produces 32,769 values, which is still considerably more than in standard 8 bit mode.
Converting from 8-bit to 16-bit mode in this way is mainly for the purpose of illustration. By converting back to 16-bit, there are not more tones than in the original 8-bit image, but there are more values available for the editing. This can help prevent degrading the image further in some cases; it depends what editing is being done.
Some edits consider the surrounding pixels (bicubic), while some edits do not look at surrounding pixels (bilinear).
The very fact that so much information is removed when converting from 16 to 8-bit, means converting back to 16 bit is of very limited value. It's a bit of a fudge and if major surgery is to be done on an image, it's best to go back to the original Raw file or the original scan and start again.
All the text and photographs on this page are the copyright of the author, Duncan Wherrett. Any unauthorised use of this material on other sites will result in the page being delisted by Google.
More about 8 bit and 16 bit
Every digital capture and film has its dynamic range, which is a scale showing the difference between the lightest point and darkest point that that particular digital file or film can record. Stating a definitive value for the dynamic range of film is risky.
Transparency can have a range of 8-9 stops with an intensity range of 1 : 500 although it will vary for different makes of film and different film speeds.
Colour negative might have a range of 12 stops, with an intensity range of 1 : 4000.
The same applies for scanners and digital cameras. They will also produce their own dynamic ranges, which will be the difference between its darkest and lightest points. Digital cameras produce a range comparable with colour negative film or more with the better cameras, with the advantage of capturing more detail in the shadows.
For post production treatment, the darkest point will be assigned a value of zero and the lightest point will be assigned a value of 255.
The number of tones between these two points will depend on the bit depth of the digital file.
A camera file or scan of an 8-bit depth will have a tonal range of 256, 0-255, which is 2 to the power of 8.
A 16 bit Photoshop file will be in 15 bit + 1 unit, which is 215 + 1, or 32,769 colours.
Bit depth will determine the number of steps between the maximum and minimum values, so a 16 bit image will produce much smoother tonal gradations. A very large dynamic range will need more bits to reproduce all the tones within that range.
A raw file is what the digital device produces when it is sqeezes as much information as it possibly can out of the subject or the film it is scanning. For best quality, the raw file should always be in 16 bit. Every tonal adjustment made in post-production throws away information. A 16 bit file can cope with this loss much better than an 8-bit file.
The original histograms of both the 16bit and 8bit files look the same. After editing, however, we can see gaps in the 8bit file histogram. These gaps represent a loss of digital information.
An 8-bit file has only 256 levels as mapped on the histogram.
Moving the highlight triangle in to the 200 level and the shadow triangle in to the 10 level where the information begins (or doing a curves adjustment to change contrast), will mean 190 levels will be spread out over the 256 spaces.
This produces the comb effect in the new histogram, because of the loss of data.
In a 16 bit file, starting with 32,769 levels, doing the same adjustment will still leave some 24,250 levels. Converting this to 8-bit will give the full 256 levels.
If another adjustment is made later, more information will be permanently lost. If the changes are large enough, image quality will suffer and the image can start to posterize, creating blocks of colour rather than smooth transitions.
The 16 bit file, on the other hand, shows no noticeble loss of information. There is so much data in the 16bit file that it can cope with the loss without degrading the image.
With all the extra detail in the file it is possible to make a better choice as to which parts of the image require emphasis - shadows, midtones or highlights, bearing in mind that the print will have a much smaller range.
A 16 bit file does not actually have a larger gamut or contain more colours than an 8 bit file. It is that there is greater detail within the tones of the file and the image will not degrade as much with adjustments.
When Photoshop converts from 16 to 8 bit, a small amount of dither is applied. This smooths the results and avoids posterisation and banding which would show blocks of tones rather than smooth transitions.
The moral is: