Lossy Codecs
Most video codecs are necessarily lossy, because it is usually impractical to store and
transmit uncompressed video signals. Even though most codecs lose some information
in the video signal, the goal is to make this information loss visually imperceptible.
When codec algorithms are developed, they are fine-tuned based on analyses of
human vision and perception. For example, if the human eye cannot differentiate
between lots of subtle variation in the red channel, a codec may throw away some of
that information and viewers may never notice.
Many formats, including JPEG and all varieties of DV, use a fairly complicated algorithm
called DCT encoding. Another method, called wavelet compression, is starting to be used for
popular codecs, such as the Apple Pixlet video codec. DVDs, modern digital television, and
formats such as HDV use MPEG-2 compression, which not only encodes single frames
(intraframe, or spatial compression) but encodes multiple frames at once (interframe, or
temporal compression) by throwing away data that is visually redundant over time.