COMPRESSION: HDTV's Dirty Little Secret

[Jerry Jones]

http://www.pcworld.com/reviews/article/0,aid,122629,pg,3,00.asp

"The ATSC standard includes support for MPEG2 video encoding, but it says nothing about compression levels."

"Broadcasting an uncompressed MPEG2 video would require 885 mbps (for 720p content) or 995 mbps (for 1080i content)."

"A station that broadcasts a single HD program can devote only 18 mbps to it, HDTV consultant Peter Putman says; and to get that, broadcasters have to use a compression ratio of 49:1 for 720p and 55:1 for 1080i."

Jerry Jones

http://www.jonesgroup.net

[Dr Mordrid]

Saying that the ATSC standard says nothing about compression levels shows they didn't do their reading.

The trick to squeezing that much data into a 6mhz band is in how you modulate the signal, and 8-VSB is the way ATSC decided to do it. Read this;

http://www.broadcast.net/~sbe1/8vsb/8vsb.htm

In the alphabet soup world of digital communications, there are two big names to remember when thinking about the complete DTV system: 8-VSB and MPEG-II. 8-VSB is the RF modulation format and MPEG-II is the video compression/packetization format. To convert high definition studio video into a form suitable for over-the-air broadcast, according to DTV standards, two stages of processing are needed: MPEG-II encoding and 8-VSB modulation. Accordingly, two major pieces of equipment are required: an MPEG-II encoder and an 8-VSB exciter.
>
>
Although MPEG-II compression techniques can achieve stunning bit-rate reduction results, still more tricks must be employed to squeeze the 19.39 Mbit/sec DTV Transport Layer signal into a slender 6 MHz RF channel and transmit it (hopefully without errors) to the eager consumer waiting at home in front of the TV set. This is the job of the 8-VSB exciter.

Dr. Mordrid

[Jerry Jones]

Regardless of whether ATSC says anything about compression or not, the key point is that there is, indeed, COMPRESSION - and that means DEGRADATION of the picture received by the end user, the consumer.

These compression methods, unless I misunderstand, are *not* LOSSLESS.

They are LOSSY.

So there can be no debate about the end result:

HDTV over the air broadcasts CANNOT and DO NOT deliver the true image quality of the high definition format.

In this regard, the high definition DVD formats will be one of the few ways to actually deliver the full potential image quality of HDTV.

Or - as an alternative - one can play video/audio directly out of a high definition camcorder, which could be attached to the HDTV television set using a high quality connection.

This reminds me of the early days of Mini DV.... when consumer DVD formats were not yet available.

In those days, one had to attach the Mini DV camcorder directly to the analog TV set using an S-Video cable in order to actually see the quality of the DV format.

We're in a similar situation now, it seems to me.

With no high definition DVD formats, we can only see the full quality of HDV, for example, by attaching the HDV camcorder to the high definition TV set using a high quality connection.

The only other way a consumer could display true HDTV quality on an HDTV television set would be through one's computer connection.

One could then play either a native HDV MPEG-2 file *or* one could play such a file after conversion to an HD .WMV or H.264 file.

Jerry Jones

http://www.jonesgroup.net

[wkulecz]

Am I missing something? MPEG-2 is lossy compression from the git go, as is WMV and H.264.

--wally.

[Brian Ellis]

In this regard, the high definition DVD formats will be one of the few ways to actually deliver the full potential image quality of HDTV.

Not at all. In fact, this is nonsense. As you say, uncompressed HDTV approaches 1 Gbit/s. A 30 Gby DVD will therefore give you 4 minutes of HDTV. Your HDTV will always be compressed with MPEG-2/4, with current technology.

IMHO, HDTV has been rushed through before the technique was mature and fully developed. True HDTV was designed as 8192 x 4096 and the BBC did a lot of research with this format, including compression methods. Unfortunately, this was abandoned for cost reasons (each camera cost millions!).

[Jerry Jones]

By "full potential," I meant that after the *initial* compression to MPEG-2, there would be no further compression required by the HD DVD formats.

With over-the-air or DBS, one must...

a) compress to MPEG-2 initially;

...and then...

b) compress the MPEG-2 *again* for broadcasting to end-users.

That's my understanding, at any rate.

Yes, if you're talking about UNCOMPRESSED HDTV, then there's really no practical way for a person to view that... quite true.

Jerry Jones

http://www.jonesgroup.net

[Jerry Jones]

Am I missing something? MPEG-2 is lossy compression from the git go, as is WMV and H.264.

--wally.

The *INITIAL* compression to MPEG is lossy.

But when this material is broadcast, it undergoes a *SECOND* round of compression, which is *also* lossy, as I understand it.

Broadcasters who send an HDTV signal, 19.28 Mbps of data in a single 6 Mhz channel, have to reduce the data rate if they also wish to transmit additional standard definition signals, simultaneously.

Jerry Jones

http://www.jonesgroup.net

[Dr Mordrid]

Please note: 8-VSB includes eight amplitude levels that support up to 19.28 Mbps of data in a single 6 Mhz channel. There is also a 16-VSB mode that has 16 amplitude levels and supports up to 38.57 Mbps of data on a 6 Mhz channel.

Point of reference: HDV = 25 Mbps, and that's for editing...not distribution. Encode for the channel and no re-compression on broadcast is necessary. In fact in 16 bit mode they can multicast (2 programs at once) or do data-casting (program + data)

Dr. Mordrid

[Jerry Jones]

Please note: 8-VSB includes eight amplitude levels that support up to 19.28 Mbps of data in a single 6 Mhz channel. There is also a 16-VSB mode that has 16 amplitude levels and supports up to 38.57 Mbps of data on a 6 Mhz channel.

Point of reference: HDV = 25 Mbps, and that's for editing...not distribution. Encode for the channel and no re-compression on broadcast is necessary. In fact in 16 bit mode they can multicast (2 programs at once) or do data-casting (program + data)

Dr. Mordrid

"A station that broadcasts a single HD program can devote only 18 mbps to it, HDTV consultant Peter Putman says; and to get that, broadcasters have to use a compression ratio of 49:1 for 720p and 55:1 for 1080i."

http://tinyurl.com/cdgft

So, Doc, are you saying Mr. Putman is incorrect?

Obviously, he's claiming broadcast MPEG-2 at 720p requires 49:1 compression.

That's a lot more compression than what you've stated.

In addition, Dr. Alvy Ray Smith's figures are on this table...

http://tinyurl.com/9aud9

...agrees that 720p requires 49:1 compression.

Is he incorrect, as well?

Jerry Jones

http://www.jonesgroup.net

[Jerry Jones]

Also from PC WORLD's article:

"If a station uses its bandwidth to broadcast both an HD show and a standard-def show, the HD program has to fit into 13 or 14 mbps."

"And a station sending out two standard-definition channels along with an HD channel must compress the HD signal to roughly 13.5 mbps, which entails compression ratios in the vicinity of 66:1."

"Such high compression produces artifacts that might not be noticeable on a small CRT, but can be quite obvious on a big fixed-pixel display."

"These include mosquito noise, an effect in which small dots seem to surround a person's head; and macroblock errors, similar to what a fast-moving video game looks like on a PC with too little graphics power."

http://tinyurl.com/cdgft

Jerry Jones

http://www.jonesgroup.net

[Jerry Jones]

As far as I can tell, 8-VSB is a *transmission* standard.

16-VSB doesn't seem to be in use; 8-VSB does appear to be adopted here in the U.S.

So to fit an HDTV program into that narrow pipe - along with standard definition programs - it would seem to me some additional compression is either required *before* transmission or *during* transmission.

So, I'm not certain on this point.

When they talk about those high compression ratios, are they talking *only* about the *initial* compression of video to MPEG-2?

Or are they talking about *additional* compression following the creation of MPEG-2 video?

It seems to me additional compression may be required, especially if an individual broadcaster is attempting to multicast more than one program at a time.

Jerry Jones

http://www.jonesgroup.net

[Jerry Jones]

Ultra High Definition!

http://tinyurl.com/yuyml

Jerry Jones

http://www.jonesgroup.net

[Jerry Jones]

Peter Putman, CTS (Certified Technology Specialist), points to the ***refresh rate*** of consumer HDTVs as another weak link in the broadcast chain.

He makes a compelling point, in my view.

Why?

Because 1080/24p content is, in reality, ***down-converted*** to 1080/30i when it is broadcast.

He acknowledges that it is theoretically possible to broadcast 1080/24p content, ***however,*** current HDTV technology makes that unrealistic.

Why?

Because "none of the consumer HDTV sets out there would support the non-standard horizontal scan rate required."

"And you sure wouldn’t want to watch 24Hz video for any length of time; the flicker would drive you crazy after a few seconds."

http://tinyurl.com/8or4o

What about 1080/60p?

Putman makes another compelling point...

"To cut manufacturing costs, most HDTV sets run their horizontal scan at a constant 33.8 kHz, which is what’s needed for 1080i (or 540p)."

"1080p scans pictures twice as fast at 67.6 kHz."

"But most of today’s HDTVs don’t even support external 720p signal sources, which requires a 44.9 kHz higher scan rate."

Here's another ripper:

"The leading manufacturer of LCD TVs does not support the playback of 1080p content on its own 1920x1080 products, whether the signal is in the YPbPr component or RGB format."

"Only the INDUSTRIAL monitor version of this same LCD HDTV can accept a 1920x1080p RGB signal."

More bad news:

"To show a 1080i signal, many consumer HDTVs do the conversion from interlaced to progressive scan using an economical, “quickie” approach that throws away half the vertical resolution in the 1080i image."

"The resulting 540p image is fine for CRT HDTV sets, which can’t show all that much detail to begin with."

"And 540p is not too difficult to scale up to 720p."

"But a 540p signal played back on a 1080p display doesn’t cut the mustard."

"You will quickly see the loss in resolution, not to mention motion and interline picture artifacts."

(And I've seen these artifacts, myself, in HDTV demonstrations.)

"Add to that other garbage such as mosquito noise and macroblocking, and you’ve got a pretty sorry-looking signal on your new big screen 1080p TV."

What about DLP (digital light processing) sets?

According to Putman:

Your "1080p TV may not have full horizontal pixel resolution if it uses 1080p DLP technology."

"The digital micromirror devices used in these TVs have 960x1080 native resolution, using a technique known as “wobbulation” to refresh two sets of 960 horizontal pixels at high speed, providing the 1920x1080 image."

"To summarize: There are no fast refresh (30Hz or 60Hz) 1080p production or transmission formats in use, nor are there any looming in the near future – even on the new HD-DVD and Blu-ray formats."

"The bandwidth is barely there for 1080i channels, and it’s probably just as well, because most TVs wouldn’t support 1080p/60 anyway – they’d just convert those signals to 1080i or 540p before you saw them."

What about 720p.

More bad news:

"The 1280x720 progressive-scan HDTV format, which can be captured at full resolution using existing broadcast cameras and survives MPEG-2 compression better than 1080i, doesn’t make it to most HDTV screens without first being altered to 1080i or 540p in a set-top box or in the HDTV set itself!"

"So what chance would a 1080p signal have?"

It does make one think.

Jerry Jones

http://www.jonesgroup.net