4 ms·
> That you can zip a large movie file, for example, owes to the fact that pixel colors have a statistical pattern, the way English words do. Engineers can build
by zbobet2012 4y ago
> That you can zip a large movie file, for example, owes to the fact that pixel colors have a statistical pattern, the way English words do. Engineers can build probabilistic models for patterns of pixel colors from one frame to the next.
This is, in fact, incorrect generally. Movie files are generally compressed via _lossy_ compression, which escapes the Shannon entropy limit. Entropy based encoding, as described in this article, is generally only the final stage in a video "codec" (algorithm to compress, and decompress video). The first stage relies of finding information in the visual field that humans are unlikely to notice the loss of and discarding it. The second stage does interframe (between frames) compression by simply searching for blocks of pixels that match. Entropy coding is generally the final stage.
This, by the way, is why zipping a video file can make it... larger.
- domador 4y agoThat lossy compression "escapes" the Shannon entropy limit doesn't seem like an accurate way of putting it. Lossy compression throws information away, so there is less information to compress. The entropy limit is still in place. I agree with the overall point, though, that movie files aren't "zipped" up, but are compressed using lossy compression (with few exceptions).
- kidme5 4y agoLossy compression is something Shannon specifically addressed with Rate Distortion Theory https://en.wikipedia.org/wiki/Rate%E2%80%93distortion_theory https://en.wikipedia.org/wiki/Rate%E2%80%93distortion_theory
- abetusk 4y agoThis is a pretty unkind reading. The author almost surely was using zip as a synonym for compress. Some people use "hoover" as a proxy for vacuum, even though they're not literally using a Hoover vacuum to clean. Movies aren't "escaping" the Shannon entropy limit, they're falling well within the Shannon entropy limit for the appropriate domain of human vision. Whether it's throwing out high frequency noise at one stage or using moving blocks and keeping delta changes at another, there's an underlying assumption of the "codeword" space is versus the source space. Even by the time it gets to the codec it's already "compressed" by throwing out almost all the spectra not visible to us. These are implicit assumptions about the domain that you've dismissed.