Bit Depth
A Background to Understanding

A "bit" is a computer switch that can be in only one of two states "on" or "off" and this is represented by "I" or "O" respectively. Bit depth is the numbers of "bit" used to define the numbers of shades of colours (or shades of grey) required for a particular image. To understand if more fully, it would be helpful to refer to "The Binary Number System" as this explains how bits are used to represent data. Assuming you have an understanding of this, you also need to know something about "light" and the human eye and this is explained below.

Mixing Light - The RGB Colour Space

As you know daylight is composed of the colours in the rainbow namely, Red, Orange, Yellow, Green, Blue, Indigo and Violet. However, the primary light colours "Red, Green and Blue" (also known as RGB) are able to produce most of the other colours.

Colour monitors produce their colours by mixing red, green and blue on a black background. Each colour can be varied to suit. When the red, green and blue guns fully fire at one pixel on the screen, this turns to white. The screen is usually black so no colours result in black. So colours are produced starting from black (no colour) by adding and mixing red, green or blue colours. For this reason "light" colours are known as "Additive Colours" as when "added" they produce different colours. This colour system is known as a "colour space" and is one of a number e.g. CYMK (Cyan, Yellow, Magenta and Black) is another colour space. It must also be remembered this refers to "light" colours and what happens when light colours are mixed.

The above does not work with mixing artist paints or printing as the light is not "added" it is "reflected" and hence different rules apply.

The Human Eye

We also need to know something about the ability of the human eye. The human eye is very well adapted to picking out slight changes in the shades of colour. However, it can only detect variations between the same colour (e.g. shades of grey from black to white or all shades of red or all shades of blue, etc) of approximately 250 or more shades. So we need a computer system to identify around 250 different shades of any one colour. For reasons explained in the section headed "The Binary System", this is most conveniently handled at 256 shades, (or 8 bit depth).

Scanning in Lineart (One Bit Depth)

Lineart are those images having only black lines on a white background e.g. drawings, etchings, plans etc. In common usage we have lots of technical drawings plans, artists drawings, etchings, etc that are normally black lines on a white background. If we consider these simple black and white images i.e., only two states, either black or white, are required to represent the image. Consequently one bit is sufficient to show this as it is either "black" or "white" i.e. "on" or "off" hence I or O. Translating this into scanning, we can see that a bit depth of "one" is all that is required for scanning text or drawings i.e. solid black on a white background. Lineart and text files are quite compact as they contain only one bit per pixel.

Scanning in Greyscale (Normally 8 Bit Depth)

"Greyscale" is a type of image that has lots of shades of grey as will as black and white. Scanning black and white photographs etc, i.e. where the image has several shades of grey in addition to solid black and white, requires "Greyscale". Greyscale is normally 8 bit and result in 256 levels of grey from black to white. So, to scan a black and white photograph, you need 8 bits to provide 256 shades of grey. More or less bits could be used. If less bits, the image will be of a lower quality as there would be less shades of grey. If more, very high quality black and white images with lots of shades can be produced.

Scanning in Colour (Normally 24 Bit Depth)

When it comes to colour its important to remember that your monitor shows colours made up of three colours red, green and blue. Remember also, the human eye can't differentiate between much more than 250 or so shades of any one colour. Consequently, all that is needed is, say 256 shades of each of the red, green and blue colours and, mixing these colours produces most colours we require. Of course, the monitor screen is black and all three colours firing together will produce white. To obtain black, no colours would fire.

If we now consider the mathematical approach, if each of the three colours (red, green and blue) have 256 levels (or shades) of colour, we need (8 bit) three times, i.e. 28x28x28=224 bits. This means we require a 24bit depth to define most colours. The resulting combinations from these 24 bits would be approximately 16.7 million colours. This arrangement is known "True Colour".

Scanning at Greater Bit Depths (36 and 42 Bit Depth)

If you have a scanner, you may have seen scanning offered at 36 bit or even 42 bit. These provide additional colours for fine tuning images and allow some unnecessary colours to be discarded before a final image is produced. Hence one reason for having greater a bit depth is simply to provide more colours to obtain the best effect on the final image.

This not only applies to colour but also greyscale. A greyscale image could be scanned at a bit depth greater than 8 bit. This might be used for high quality black and white photographs scanned for enlargement.

There are also reasons why you may wish to scan in colour in less than 24 bit. This could be an image required for a web page. Old graphic cards still work on 8 bit and this bit depth is sufficient for a colour picture comprising 256 colours. So 8 bit colour images for the web will load more quickly and will also be able to be read by old graphic cards.

© 2001 Hamilton Ltd - issue 30/10/01