Image formats

Posted on 3 Jul 2002 at 17:20

• 
• 
• 
• 
• 
• 
• 

 1 of 7

Gallery

Whether it's for publishing images on the web or simply sending funny pictures to friends, you want to be sure the format you're using is viewable and not too large. Kay Ewbank puts you in the picture

When you save a word processor document or a spreadsheet, you're usually offered a default file format choice n .xls for a Microsoft Excel spreadsheet, for example. When you save an image, you're offered a bewildering array of options n TIF, PCX, GIF, PNG and so on. There's obviously some difference between the choices, otherwise they wouldn't be on offer. So what are the advantages of each and how can you be sure you're choosing the right one for your particular task?
Colour
When you create a picture using an image-editing package, you have to choose how many colours you're going to use in the image. There are four main options: black and white, 16 colours, 256 colours or True Colour, also referred to as 24-bit colour. It's called True Colour because, by using 24 bits to represent a single dot on the screen, you can represent 16.7 million unique colours. As your average human can only distinguish a few million colours, software with 16.7 million colours at its disposal can produce images that look 'real' in terms of the colours n hence True Colour.
It would seem, then, that we should always use True Colour images. The reason we don't is, not surprisingly, because of memory and file size. Let's start with the memory. To control the colour of a single dot (otherwise known as a pixel) on your PC's screen, the operating system has to use some memory in order to remember what colour the dot is supposed to be. The memory used is usually called the Video Ram or VRam.
If you're working with a screen showing only black and white, you need only a single bit of memory to work out the colour that should be shown on the screen n 0 for black, 1 for white. A bit, in this sense, is the smallest element of memory you can work with. Suppose you're bored with your black and white image and want more colours. If you use four bits, you can get 16 colours. This is because 16 is two to the power of four (24) so you can combine your four bits of memory in 16 different ways to represent 16 different colours n 0 is black, 1 might be blue, and so on. In the same way, if you use eight bits of memory per pixel, you can achieve 256 colours. Finally, if you use 24 bits of memory per pixel, you get up to 16.7 million colours, which should be enough for anyone.
It's not just displaying the image that takes memory, though. When you save an image and try to send it to someone else, you'll quickly realise that an image file that is fairly small when saved in 256-colour mode becomes much larger when saved in 16.7 million-colour mode. As a single point in an eight-bit (256-colour) file uses eight bits of memory where a True Colour image uses 24 bits, files are three times larger. This isn't always the case in reality because of image compression techniques, but that's the point from which you start.
You might wonder how a 256-colour image can ever look realistic if we can distinguish between several million colours. One reason why this works is because of the use of colour palettes. You will have seen these if you've ever chosen a 256-colour option when saving an image. The idea is that many images can be represented fairly accurately with just 256 colours, particularly if the palette is optimised so only the most widely occurring colours within that particular image are used. You might also see the palette referred to either as a colour index or a colour lookup table (CLUT).