Archived from groups: alt.tv.tech.hdtv (
More info?)
>>Ok, I admit this doesn't address your point directly (i.e.
>>'WM9 vs MPEG-2 at standard-definition bit-rates'), but have you
>>seen the high-definition 'tech-demos' put out by Microsoft?
>
>
> Yes, and those are the ones that prove that they only get about 25% extra
> compression, because most are filtered down to *far* less than 1920x1080,
> or use 1920x1080/24p, which takes *far* less bandwidth than 1920x1080/60i.
>
>
>>Again, the quality is superb to my eyes, and at a bitrate half of
>>OTA/ATSC.
>
>
> Again, not a big deal. Do the math:
>
> For 1080/60i:
> 1920 * 1080 * 60 / 2 = 62,208,000 pixels/sec
>
> For 1080/24p:
> 1920 * 1080 * 24 = 49,766,400 pixels/sec
>
> Just to start with, the raw pixel rate is 80% of that of 1080i. Thus,
> you get a savings of 20% without even doing anything. So, if the actual
> bitrate is 50% of the full 19.3Mbps ATSC MPEG2, they're only saving about
> 35% off what MPEG2 would take to encode 24p data. Plus, they don't have
> a single example of realtime compression.
>
> We all know that the very best DVDs can look astonishingly good at a "mere"
> 5Mbps, yet 5Mbps is barely enough for good realtime encoding of MPEG2
> without spending literally *millions* of dollars on encoders.
Ah, for some reason my basic math was failing me, and I just now see
your point. Indeed, accounting for the inherent bitrate-reduction of
24fps vs 30fps, the compression-gain no longer looks as 'fantastic' as
it once did.
> Then your local ABC affiliate is doing something hideously wrong. 24p
> encodes wonderfully into the 60p transmission that ABC does. You get
> lots of "repeat frame" flags that allow all those extra bits to go towards
> dealing with every problem.
You're probably right. The Los Angeles TV stations were among the early
adopters of ATSC, so they all got stuck with early encoders.
Our local stations never use the repeat-frame flag (at least, not
according to my ATSC-captures), but I've seen other reports that say
other stations and the cable-channels (HBO-HD) do make proper use of the
repeat-flag for 3:2 material.
>>movies could be encoded offline using multi-pass and the highest
>>quality possible, and still deliver excellent quality at just
>>4-5 MBps.
>
>
> This won't happen because there isn't a TV station in the world that
> can deal with pumping out the bits unchanged. That may happen far in
> the future, but it's actually quite an expensive proposition, and who
> is going to pay for it at those small-time stations that are selling their
> bandwidth to USDTV?
Sorry, I probably didn't explain myself clearly. Let me try again.
USDTV (and others) are investigating a broadcast-based 'movie service.'
I guess it's a lot like satellite or cable pay-per-view systems. The
cable-operator transmits a pre-selected batch of movies (rotated on
some schedule.) A customer orders a movie, and his 'box' is
allowed to decode it.
In USDTV's scheme, the transmission medium is the ATSC broadcaster
(instead of satellite or cable.) Each customer has a USDTV
receiver with a hard-drive. The customer orders the movie the night
before, then the movie is 'streamed' over the nighttime.
The following day, the entire movie is on the DVR's hard-drive,
and ready for play.
I guess the idea/model isn't really new. The 'twist' is that
companies want to buy ATSC bandwidth to operate this service over
the TV-airwaves.
My point is, USDTV would arrange for the delivery of their
'service bitstream.' I would think the ATSC TV-station doesn't
care what the bitstream contains; live programming or 'streamed'
movie packages. It's just a separate feed they plug into their
station multiplixer.
But since I don't work in industry, I hvae on clue whether this
is 'easy' or 'difficult.' I think the very few ATSC-transmitters
that are already sending out USDTV-content, could easily accomodate
USDTV's propsed PPV movie-service.