Image Compression Formats

A major element of the computer graphics industry is image compression. The largest and most complete format for an image is the bitmap form, for which each pixel has a unique color value stored. Ultimately, every image becomes a bitmap when it is displayed, but often it is convenient to store and transfer images in a compressed format, in which the file size is smaller than that which would arise when stored in a bitmap format.

It is important to know a little about image compression formats, because their characteristics can determine the quality of a final displayed image, and some are more appropriate for different types of images than others.

GIF (Compuserve) - Reduce the Color Depth

GIF files compress images by two methods applied one after the other. Firstly, the palette is picked, and has an upper limit of 256 colors (8-bit), but can be stored at a color depth of anything from 8-bit to 1-bit. The entire graphical information is then further compressed using LZW (Lempel-Ziv and Welch) compression, which is a lossless compression technique. Only the first compression method is important to the creator of the image, since it may require the color depth to be reduced.

The GIF format is good for diagram-type images, and images which are already in 256 colors. It can also incorporate transparency (where the area ‘behind’ the image can be seen through part of the image), animation, and interlacing. Interlacing is the process by which the downloading of a GIF can be streamed, so that it is does not have to be downloaded completely before being partially displayed. A low-resolution version of the image can be presented before the full higher-resolution version is downloaded. This is a feature designed specifically for use in Web graphics.
	An animated GIF

A full color image	The image in 256 colors
The image in 16 colors	The image in 2 colors

JPEG - Encoding the Changes

JPEG files are a 24-bit alternative to GIF files. JPEGs use a ‘lossy’ compression technique which compares the next pixel value to the current one, and attempts to preserve the color gradient from one to the next. Whilst the compression technique does not produce an image which is identical to the original, it does retain the color depth. The full-color image above is an example of a JPEG file.

This format does not, as yet, support transparency or animation. A form of JPEG called Progressive JPEG does, however, allow for interlacing, and also has better compression than non-Progressive JPEGs. JPEG is a good format for photographs, and other such images with smooth color gradients. The format is not very good for line-art or diagram type images.

The Fractal File Format

This is a very recent file compression format, which uses the Fractal Transform. The Fractal Transform was created by mathematician Michael Barnsley, and is implemented by his company Iterated Systems. It works by splitting the image into small regions and identifying parts of the image which are similar to other parts, and can be obtained by a set of transforms.

The fractal file format has excellent compression properties, retaining almost all of the detail of the original image, and the color depth. Unfortunately, it is currently quite expensive to buy, and still unsupported by most graphics products. Look out for its introduction over the next few years.

A portion of the Koch snowflake - a fractal demonstrating self-similarity

Contents | Previous - Color Palettes | Next - Techniques in Pixel Graphics