To process a compressed bitstream and recover the original data (if lossless compression), or an approximation of the original (if lossy compression).
Decompressed video is only used over cables for production equipment, though there exists equipment that will capture and record raw HD-SDI video streams (and similar) from professional equipment.
All video codecs use data compression, of which there are two kinds: lossless and lossy. Lossless data compression means that an approach is taken that allows each video frame to be restored exactly as it was captured. Lossy data compression means that the frames are analyzed and compacted in such a way that it is impossible to recover the video frames exactly as they were captured, but you instead get something close.
Audio codecs are either uncompressed (lossless) or lossy. In either case, the audio codecs are attempting to represent a continuous analog wave with a digital approximation. Besides the lossless/lossy factor, the number of bits used per sample, the sampling frequency, and the number of channels (mono, stereo, surround) dramatically affect the quality of the audio.
CinemaDNG is probably the most widely supported lossless video format, which more or less stores each frame as a separate image file. Sometimes the frames are stored in an MXF file (MXF itself is a container and it can have many different codecs, but for “RAW” video, that would be a bunch of DNG frames).
There’s a lossless version of the MPEG-4 h.264 codec (the codec itself is widely supported, but the lossless mode is rarely used).
Another format is raw HDMI video frames stored in a QuickTime container (some Nikon cameras can do that).
Furthermore information about the videos you can read our articles;