NTSC Takes on All Comers

You might not have noticed that NTSC wasn't the world's first television broadcasting standard. And, yes, I'm talking about the first NTSC -- not the one that gave us color in 1953, but the one that gave us 525 interlaced scanning lines 30 times a second in 1941.

That's 62 years ago, making NTSC old enough to collect Social Security. My, my. Howdy, old-timer! We fogies need to stick together!

Where was I? Oh, yeah, the teevee standard that came before NTSC was System A (NTSC is System M; don't ask why). I'd give System A a big "Howdy," too, except that it's dead; it just couldn't hack it in the new world.

System A was 405 scanning lines per frame, 25 frames per second, AM sound (oh, yes, indeed!) and positive modulation. It was born around 1936 or 1937, and spread across Britain, crossed the water to Ireland, and even found its way to Hong Kong. That was in the good old days.

The 1941 NTSC prevented System A from being quite as widespread as the British Empire (when Canada got around to joining the U.S. in 525 lines), but that wasn't what killed System A. What killed System A was the 625-line, FM sound, negative modulation, color-friendly standard that got introduced in England in 1964. A mere 21 years later, not quite 50 years old (so young!), System A was completely dead in Britain.

The FCC (aka Our Beloved Commish) was sort of hoping that NTSC could be murdered even faster. The U.S. DTV rules came out in 1997, indicating NTSC transmitters are supposed to be shut down by the end of 2006.

Just in case you've temporarily lost your marbles, allow me to remind you that the end of 2006 is about three-and-a-half years from now, or so. There is less than one percent of more than 100 million U.S. homes able to receive U.S. DTV signals. There are maybe 30 million TV sets sold each year. Still think NTSC will be dead by the end of 2006? What? Your marbles? Don't worry. Those nice men in white coats have already taken them to your new home.

Anyhow, my point, if I happen to have one this lunar cycle, ain't that NTSC is, thus far, the longest-lasting video standard. Heck, VHS is on its way to being the longest lasting videotape standard-and the ones it's out to beat are U-matic and quad.

No, my point, if I happen to have one, is that it might be worth having a good look-see at NTSC-or any other long-lasting standards: Railroad tracks, power-line frequencies, the International Postal Union, etc. Maybe we could learn a thing or three.

Why is a standard like a bath in liquid nitrogen? That's on account of they're both good at freezing stuff.

Yes, that's what a standard does; it freezes stuff. Before the soundtrack got added, movie film was shot and projected at all sorts of rates. Then it got frozen at 24 fps. What's the result? 2/3 pulldown judder. Okay, so maybe that wasn't the best example I could give.


Let me go back to NTSC, (just try and stop me; it's my rant!). Back when the committee was meeting in 1940, some members proposed 1,000-line frames. There were proposals for widescreen. CBS was already pushing color. I am not making this up. But the committee froze things at 525 lines, 4:3, monochrome and monophonic.

That was the bad news. The good news was that, with everyone working on the same frozen standard, it could be improved.Color got added about 1953, without having to drop to lower line and frame rates as CBS wanted.

Stereo sound got added, too. Widescreen? Hey, have you checked out DVDs lately? Ghost-canceling? Methinks that was the very first ATSC standard.

Closed captions? Check. Other data transmissions? Hey, why not? At last year's NAB, Telisar even showed two movies being multicast on a single NTSC channel. Yeah, you heard it right, two movies being multicast on an analog NTSC channel, not a DTV channel. And then there are those Microsoft Actimates, the "Barney" dolls that watch TV with you and make comments based on data transmissions in the NTSC channel.

Dang! I knew I should have quit while I was ahead. The court reporter will please strike everything in the preceding paragraph after "DTV channel," and the jury is instructed not to remember what I just said.

"What are we not supposed to remember?"

The "Barney" dolls that watch TV and talk.


Never mind.


So, how does Telisar stick two movies into one broadcast NTSC channel? Well, there are a bunch of data-carried-in-NTSC folks (besides Telisar there are Dotcast, Wavephore, etc.), and the most optimistic of them figure they can squeeze maybe 4.5 Mbps or so into an analog NTSC channel that's still carrying analog NTSC teevee. That seems like a more-than-reasonable quantity of bits to carry a standard-definition movie via MPEG-2.

Ah, yes, MPEG-2. There surely is a lot of MPEG-2 out there. Every DTV box in the world, methinks, is MPEG-2, whether it's an ATSC box in Canada or a DVB-T box in Australia; whether it's a digital cable box in Los Angeles or a digital satellite box in London. DVD is MPEG-2. Sony's Betacam SX and IMX are both MPEG-2. But Sony's Digital Betacam, DVCAM, HDCAM and HDCAM SR ain't.

That last one, HDCAM SR, is kind of interesting. It uses MPEG-4. MPEG-4 can compress more efficiently than can MPEG-2. So can AVC, the Advanced Video Codec (aka H.264). So can On2's VP5. So can JPEG2000. And I could go on listing even more.

Hey, does anyone remember MPEG-1? It's still around, but it definitely ain't the compression algorithm du jour. So, how long is MPEG-2 going to be king of the hill-as long as NTSC or as long as MPEG-1?

You might want to ask the ATSC. They've been asking for proposals for advanced compression.

Does that seem odd to you? The very same committee that froze DTV with its MPEG-2-based standard is now looking into advanced compression.

Well, now, they sort of have to on account of another project they've been working on. It's called robust VSB.

VSB, you will recall, is the kind of broadcast DTV we've got here in the good old U.S. of A., and, when it works, it's pretty danged great! We use the 8-VSB flavor, which gives us 19.4 Mbps-when it works. But there are a couple or three places where it doesn't work.

Enter robust-VSB. It gives some bits vitamins to make them big and strong enough to be received in more places. But those big, strong bits take up more room-around twice as much room.

So, if you wanted to send maybe 7.5 Mbps robustly, they'd take up as much room as 15 normal, emaciated Mbps. Subtract that from 19.4 Mbps, and you're left with about 4.4, pretty close to that reasonable figure to use for standard-def MPEG-2 for legacy-box viewers.

Meanwhile, what do you do with the 7.5 big, strong Mbps? Well, now, they only get received by robust receivers, which don't exist yet. Ergo, they can use something other than MPEG-2, such as, for instance, AVC, or JPEG2000, or Windows Media 9, or something else like those there. And, lo and behold (though I still don't know how to lo), 7.5 Mbps might just be enough for HDTV, using one of those more-efficient compressors. It sure as heck ain't enough for any HDTV other than a stationary test pattern using MPEG-2.

Now, then, it ain't that MPEG-2 has been fixed in quality any more than NTSC has been. Improve the encoders, do some pre-processing, and so on, and you might be able to get decent MPEG-2 SDTV down to two-point-something Mbps, which means HDTV might get down to maybe as low as 10 Mbps. But that ain't good enough. Two times 10 is 20. 19.4 minus 20 is negative 0.6 for MPEG-2 SDTV to legacy-box viewers. So it's mighty unlikely that MPEG-2 will be able to challenge NTSC for long.

Ain't math fun?