Uncompressed YUY2 AVI video is 100% decoded video frames, 100% of the original image information. Lossless compressors such as
huffyuv and Lagarith do not re-encode video -- rather, they're more like ZIP or RAR lossless compression but much faster and specifically designed for video. They compress, but they don't re-encode. What you get back after you decompress lossless AVI is exactly what you put into it in the first place.
Video
encoding + lossy compression is a different story. Lossy codecs such as MPEG, h264, XVid, QuickTimne, DivX, etc., involve multilple levels of compression. Video is encodd into Groups of Frames (GOPs), each GOP containing at least one or more key frames. A key frame is the only frame in the GOP that is a complete, 100%, full-sized digital copy of an original frame. The other frame types inside a GOP are not complete images: they contain only data about how the frames have change since the last key frame in the GOP.Moreover, lossy codecs use various levels of additional compression that discard data that the encoder decides is unnecessary, or that the encoder can't accurately maintain at insufficient bitrates. Video encoded with very long or large GOPs up to 250 frames are very lossy encodes that are difficult to edit without serious data loss. Smaller GOPs such as the ~18-frame limit in PAL/NTSC are easier to edit with smart-rendering editors, but they still are encoded in a lossy manner that loses anywhere from 15% to 60% of the data that went into it.
Each modification of lossy video involves lossy re-encoding. What gets lost from earlier encodes is never retrieved. It's gone. Kaput. Nada. Re-encode again, and you lose more. Losslessly compressed AVI as discussed here contains 100% of the original, full-sized images that you started with. But a method of "compression" is used that's more like ZIP, where similar occurrences of data are combined into shorter codes. No data is thrown away.