A primer on audiovisual compression
Workshop: How do they do that: A primer on audiovisual compression
George Blood
The vast majority of on-line audiovisual content is distributed using compression. Even in the developed world where high speed broadband is readily available video and audio is generally only available in compressed formats. While universally unacceptable for preservation, most users experience a/v content by way of internet access made possible by perceptual coding techniques of compression. This paper explains how file sizes and bandwidth can be reduced significantly by utilizing various evolutionary adaptations of human perception. From masking and persistence of vision to logarithmic compression and chroma subsampling we’ll explain how careful analysis and clever mathematics enable us to put 100GB of content into a 10MB package.
The presentation will include side by side demonstrations of varying levels of compression, examples of how easily human perception can be fooled, and the delta between the source and compressed signals.
The goals of this presentation are
- to demonstrate why all compression is bad for preservation
- to provide content distributors information to better prepare their materials for a superior end-user experience
- to help end users understand what they’re missing and that under some circumstances they should seek access to high quality preservation sources
- to provide less tech-savvy archivists the information and vocabulary to advocate for uncompressed formats for preservation