The two issues most often discussed with regard to digital video compression (more properly called bit-rate reduction) are image quality and compression ratio (or bit rate). Unfortunately, the most important issue is probably money.
A paper presented at an NAB engineering conference once stated digital TV broadcasting would be impossible. The figures checked out.
Nyquist’s rule for digitization says signals must be sampled at a rate higher than twice their top frequency. Broadcast video in the U.S. goes to 4.2MHz. Twice that is 8.4 million samples per second. Code it at even just six bits per sample (far below the eight commonly used today) and there’s an apparent need to transmit more than 50 million bits per second (50Mbps) in a standard 6-MHz broadcast channel.
Not even the latest digital TV technology used in a cable TV system free from outside interference can cram that much into a single channel. Yet digital TV broadcasting works. The secret is compression.
The form of HDTV known in the U.S. as 1080i has 1920 “active” (picture carrying) pixels per scanning line, 1080 active scanning lines per frame, and 29.97 frames per second. Counting both gray scale (luma) and color information, it has 16 bits per pixel. That comes to just under a billion bits per second. Yet it is commonly transmitted in a digital TV broadcast channel with a maximum capacity of just over 19Mbps.
That demands a compression ratio of more than 50 to one. And MPEG-2 delivers it—or even more. Some HDTV is transmitted in as little as 12Mbps, a compression ratio of roughly 83:1. Unfortunately, that might not be enough.
Last month’s column noted that KJZZ in Salt Lake City carries its own programming as well as four channels of USDTV programming (and the USDTV guide) on its digital TV station. Given that standard definition TV can easily be compressed to about 3Mbps, there’s no problem—unless KJZZ wants to air its programming in HDTV.
In that scenario, compression ratios seemingly beyond the capabilities of MPEG-2 would be required. Need to squeeze HDTV onto a DVD? Again, advanced compression is needed.
Fortunately, there are quite a few organizations offering advanced video compression technologies. Two of the most discussed are the video coder/decoder (codec) used in Microsoft’s Windows Media Player 9 and the advanced video codec (AVC) developed by the joint video team of the International Telecommunications Union (ITU), the International Standardization Organization (ISO), and the International Electrotechnical Commission (IEC). It’s referred to by either its ITU designation, H.264, or its IEC/ISO designation, MPEG-4 Part 10.
According to thoe who have been performing comparative testing on the two, there’s no strong reason to choose one over the other based on the quality or technology. But then there’s the licensing.
MPEG LA, a patent-pool licensing organization, has a license-fee structure on its website calling for 25 cents per consumer decoder or encoder with a $1 million cap per entity per year and the first 50,000 per year free. But then there are use fees for broadcasters of roughly a penny per 30 minutes for relatively recent programming (half that for older material). For a multichannel programmer, that can add up.
Microsoft offers a more favorable fee for encoders and decoders, with no use fees. But MPEG LA wants to license the technology in Microsoft’s codec, suggesting that Microsoft’s fees might not be the last word.
Fortunately, MPEG LA has now offered to license at least some H.264 broadcast encoders for one-time payments. In some circles, those are considered terms of endearment.
