Redundancy and How Much Compression You Get

Compression works by spotting repetition and storing it more cheaply. A photo of clear blue sky is mostly the same pixel value repeated thousands of times, so an algorithm can replace all those repeats with a short instruction like 'this color, 50000 times.' Television static has no pattern at all; every pixel is a surprise, so there is nothing to shorten. The amount of redundancy in the original data sets a hard ceiling on compression: no algorithm can shrink pure randomness much, no matter how clever it is.

Compression ratio measures the result: original size divided by compressed size, with both sizes in the same unit. A ratio of 8 means the compressed file is one eighth the size of the original. This number depends entirely on how redundant the source data was, not just on which algorithm you ran.

Lossless compression keeps every bit needed to rebuild the file exactly, while lossy compression permanently throws away detail the eye or ear barely notices, trading accuracy for smaller size. Smaller is not automatically better: the right choice depends on whether the file must be reconstructed perfectly. This sets up the next question in the unit, which is how image and sound formats decide what counts as safe to discard.