Getting to Home Base With A Cast of Multi

You might not have noticed that the term HDTV doesn't necessarily guarantee quality. For instance, take an HDTV camera. Aim it at a wall. Start transmitting its pictures. See how long your ratings stay above an asterisk.

Indeed, content matters. So do such other trivial pursuits as lighting, makeup, lenses, and stuff like that there.

Methinks Canon's Larry Thorpe still points out the importance of contrast ratio on image sharpness; someday you must go through the frightening exercise of calculating contrast ratios for ideal reconstruction filters for various HDTV-imager resolutions. Then you can look at diffraction, too. There are good reasons for those beyond-HDTV cameras with giant imagers, even if the ultimate outlet for the pictures is broadcasting.

Anyhow, those are exercises for you to perform when you're done reading this month's rant; they ain't why I shoveled coal into my computer this lunar cycle. No, my cause of the moment is multicasting.

Perhaps you have heard of the term. The idea of forcing cable operators to carry broadcasters' digital multicasts was put on the agenda for last month's meeting of Our Beloved Commish (aka the FCC) and then-over a weekend-pulled. Religious broadcasters raised a ruckus, and don't think it was on account of the item's deletion on a Sunday. The NAB and NCTA are among others weighing in on the issue.

That issue ain't got much to do with quality. Broadcasters want cable ops to carry whatever they transmit; cable ops don't want to have to carry more than a broadcaster's primary service.

The cable ops say they ain't got room, but that doesn't make any sense. If a broadcaster they've got to carry transmits HDTV in 18 Mbps, they've got to carry that 18 Mbps; if the same broadcaster divides the 18 Mbps into six, 3 Mbps standard-definition multicasts, just how does that require more capacity from the cable system?

On the other hand, if a broadcaster's trying to compete with cable by transmitting multicast channels (for instance, see USDTV's off-air channel line-ups), it's hard to understand why a cable op needs to encourage the competition. If a broadcaster wanted to hold its annual picnic on a cable op's premises, that might not cost the cable op anything extra, but why should they have to provide the space?

Anyhow, those are issues best left to bureaucrats and politicians so they'll keep busy and not do things that can really screw things up. I've also been wondering a lot about the multicast economic model for commercial broadcasters. They get paid for ads based on the viewers they provide. If they divide their channels, they provide fewer viewers per subchannel. Meanwhile, they pay for programming based on market size. Pay more; earn less. Methinks I'm missing something.

On account of my not being able to figure out the economics of commercial multicasting, I figure I'll use noncommercial stations for my 20 cents, or, as fancy-word users like to say, pair o' dimes. For instance, try KQED-DT in San Francisco.

According to their schedule, they carry HD on channel 9.1. They carry something called Encore on channel 9.2. Something called World is on channel 9.3. A subchannel called Kids is on channel 9.5 (something called Life is on channel 9.4, but not when HD is on).

They have a couple more channels-as well as video on demand-available on cable only (see, broadcasters and cable can be friends). The analog channel could also be important; a bunch of public stations carry stuff on their analog service that never shows up on the digital. But I want to concentrate on the digital multicast.

SMPTE's HD-SDI carries about 1.5 Gbps. Strip 1080i HDTV down to its 8-bit, "4:2:2" minimum, and you've got 1920 x 1080 x 29.97 x 8 bits x 2 (for 4:2:2). That's still up over 994 Mbps. But Our Beloved Commish allows broadcasters only around 19 Mbps, including audio and housekeeping signals-figure around 18 Mbps maximum for HD video. So, that's a compression ratio of more than 55:1.

That's one whole heck of a squeeze, but that's if a broadcaster is using the entire capacity of a digital channel for HD. KQED ain't. They've got three standard-definition multicasts running full time with the HD. If they happen to devote 3 Mbps to each, that drops the HD to just 9 Mbps, a compression ratio over 110:1! Zounds!

Methinks maybe 9 Mbps ain't quite enough to deliver quality HD for some noncommercial-type programming like maybe a ballet. They could also drop the standard-definition subchannels to maybe 2 Mbps each, leaving a whopping 12 Mbps for the HD and screwing up the quality of the standard-definition multicasts to boot.

The aforementioned USDTV is sending some standard-definition stuff at about 1.5 Mbps, and it looks reasonably okay, thanks to their encoding it as MPEG-4 Part 10 (AVC) instead of MPEG-2. They can do that on account of their supplying the boxes used to receive their signals. For older boxes, they made an MPEG-4 USB dongle; the hard part was getting it to work with USB 1.1.

Unfortunately for non-USDTV broadcasters and viewers, none of the other set-top boxes or digital TVs that have ever been sold includes MPEG-4 decoding. Our Beloved Commish requires broadcasters to use MPEG-2 for their primary video, but they could go to MPEG-4 for the multicasts unless they're picky and want to have viewers.

Before the 100 percent 25-inch-and-up "tuner mandate" kicked in on March 1 of this year, there were only a few million of those non-MPEG-4 boxes and digital TVs to worry about. These days, manufacturers sometimes ship close to 700,000 to U.S. dealers in a single week.

Here's a recap: You can screw up HD with multicasting. You can screw it up less by screwing up the multicasts, too. You could screw it up least with MPEG-4 multicasts, but not if you want viewers for them. Got it?