Here’s another example of video lingo from different sectors bumping into one another: The way video is streamed on the Web, versus the way it works on cable. (Or broadcast, or satellite, or telco, for that matter.)
It’s hard to keep it all straight. Hence this week’s translation.
Original Internet video streaming dates back to the dial-up days. Streaming meant video bits flowing over the Internet, which could be viewed on the computer screen as they arrived — if, of course, there’d been no dropped packets along the way.
Next came the “progressive download.” The biggie. Adobe Flash, Microsoft Silverlight, and YouTube use it, among many others. The “progressive” part means you watch the video that progressively spills into your PC’s hard drive, as a file, after a comfortable chunk of it had loaded into your browser’s cache.
The new big thing in Web video is “adaptive streaming,” sometimes also called “fragmented MPEG-4.” It mixes traditional streaming with file-based delivery. The “adaptive” part means multiple video files of the same content are each encoded into two- to four-second chunks. That way, the client can switch between files on the fly, at varying bit rates, depending on available bandwidth.
Example: You’re watching an HD video stream on your computer. At the moment, and probably unbeknownst to you, it’s coming in at a throughput rate of around 7 Megabits per second (Mbps). Suddenly, the Internet sneezes — congestion. In the background, your player knows to hop to a 5 Mbps stream, or a 3 Mbps stream.
On the linear cable side, digital video is encoded in one of two ways: Constant bit rate (CBR), or variable bit rate (VBR). Both deal with how the bits are stuffed into a pipe of fixed size (6 MHz, or 38.8 Mbps). Variable bit rate is harder to do, but more efficient; it finds ways to fill the pipe completely, smooshing bits into all blank space. “Zero nil bits,” as they say. Almost all linear video uses VBR encoding these days, thanks to statistical multiplexing.
VOD, on the other hand, still uses CBR, as does switched digital video – meaning that a fixed amount of the available bandwidth is ascribed to each stream (3.75 Mbps per stream for standard definition video; 15 Mbps for HD).
In that sense, Internet adaptive streaming of fragments is like VBR for on-demand. Near-VBR, for lack of a better term, because fixed file sizes (2 Mbps, 4 Mbps, 6 Mbps) are still on the scene. You click to play a stream; it’s sent in a way that optimizes quality, depending on available bandwidth.
That’s the basics of Internet vs. cable encoding, for video streaming. Next week: Your own personal Faraday cage? Really?