FW: 8K TV

[William Fenn]

Kinda sounds like "GENERATION DEGREDATION" from the old 2" tape days.

-----Original Message-----
From: tacos-bounces+bfenn=cox.net at amrad.org
[mailto:tacos-bounces+bfenn=cox.net at amrad.org] On Behalf Of Rob Seastrom
Sent: Sunday, April 14, 2013 7:33 AM
To: wb4jfi at knology.net
Cc: tacos at amrad.org
Subject: Re: 8K TV


Here's my $0.02, coming at this from a totally different direction
from Terry (I'm an IP network guy who happens to work for a big MSO
supporting, among other things, the infrastructure that runs their
"watch TV on your tablet or IP stream to your 'smart tv'" apps.

I think it's fair to say that *everyone* is doing stream splicing for
local ad insertion.  It may be expensive (not even factoring in the
cost of confusion caused by deliberately obtusely named companies like
This Technology and thePlatform) but it's less expensive than creating
a separate stream at headquarters for every possible market and then
shipping them there.  Which is in turn less expensive than deciding to
turn one's back on local ad revenue.

One thing to keep in mind is that *everything* south of the camera
output is a lossy compression format.  Think there are no generational
losses in digital like there are in analog?  That's only true if you
are doing straight up bit moving and not re-encoding.  You lose
information when you re-encode, which results in rather different
artifacts than having a frame that's damaged due to bit loss.

Mezzanine formats, despite being really high quality (and bit rate)
and pretty are still lossy.  My somewhat un-nuanced understanding of
mezz formats is that their distinguishing factors are that they tend
to (a) be all-I-frame (i.e. flipbook, no predictive or
backward-looking delta frames) and (b) have marginal interoperability
between implementations/vendors.  One fairly popular format is Motion
JPEG 2000, but there are others based on all-I-frame H.264, VC1, etc
in containers including Quicktime, Windows Media, MXF, MPEG2, and more.

So, from uncompressed video out of the back of the camera through
editing you'll get at least one mezz format but possibly more than
one, into post production and then you have a finished work product
that's in a mezzanine format ready to be turned into mpeg2 and put on
the satellite (or spit out terrestrially in a mezz format to a cable
or direct satellite company).  It's too expensive to put mezz formats
on a satellite; while satellite is very efficient from a fan-out
perspective and the cost to add a new recipient is limited to the cost
of the dish and receiver, the sunk costs per megabit are extremely high.

So now it's mpeg2 on the bird and gets downlinked somewhere, but since
the I frames are probably not precisely where they are wanted for
splicing in the ads the stream gets decoded and re-encoded (which
Terry mentions but does not emphasize below).  Now, it may get
re-encoded with a mezzanine source ad inline or it may just get
flash-cut (which is easy once the frames are aligned just right).

That's for national channels.  Locals on cable may have another step
or two thanks to over the air broadcast, acquire, re-encode, etc.

The generational losses are annoying independent of how much crunch
one tries to put in relative to the amount of motion in the frame.
Talking heads are easy.  NASCAR is worst case.  Basketball is about as
bad due to the camera panning.  Everything else is in-between.

Cable companies are actually acutely aware of video quality issues,
particularly the oversubscription problem Terry mentions below, so it
should come as no surprise that with the proliferation of settop boxes
that are capable of at least SCTE-55 and maybe DSG interaction with
the upstream plant, the model is evolving to more one of spatial reuse
via switched video and less always-on
compressed-within-an-inch-of-its-life video.

Then there's terrestrial acquisition of mezzanine feeds, which cuts
out a couple of interim generations particulary when the final product
is going to be mpeg4 not mpeg2 - not just for streaming; see the new
whizzy STBs outlined above.

-r

<wb4jfi at knology.net> writes:

> Yeah, I can see the pixlelization and other artifacts in DTV broadcasts,
but that is
> usually due to bit-starvation thanks to the stupid cable and satellite
delivery
> companies.
>
>  
>
> Where are you seeing analog broadcasts?  Other than some low-power
stations and
> translators, all television âoebroadcastersâY\.. are transmitting
digitally with ATSC
> 8VSB DTV in the U.S.  Also, cable channels are all using digital
transmission systems to
> cable headends.
>
>  
>
> There are LOTS of opportunities for the HD video to get totally hosed up
between the
> original source and your set.  But, if you sat down and looked at true HD
on even a
> decent quality TV (not high-end), you would be hard-pressed to find any
artifacts. 
> Itâ(TM)s easy to justify throwing away a few unimportant bits here, and a
few more
> unimportant bits there, until you are left with a marginal quality
signal.  Then, that
> signal goes into a âoestatistical multiplexerâY\.., which throws up when
too many
> marginal HD signals want too many of the few remaining bits, all at once.
>
>  
>
> WARNING: DANGER WILL ROBINSON!!!  TECHNICAL INFORMATION AND JARGON
FOLLOWS!!  READ AT
> YOUR OWN RISK.
>
> The Fox broadcast network is the worst.  I donâ(TM)t know if they still
use it, but at
> one time they used a technique called stream-splicing at the affiliate
stations to add
> local commercials, IDs, etc.  Most big-three network (ABC/CBS/NBC)
stations decode the
> network A/V stream back down to baseband 1.5Gb/s or so (1080i or 720p),
and then switch
> between HD sources at that baseband level.  Thatâ(TM)s expensive.  Fox
used a technique
> that kept the data stream at a lower bit rate (either mezzanine or actual
19.39Mb/s
> MPEG/ATSC), and tried to flash-cut by building additional MPEG II
âoeIâY\.. frames at
> the switch point.  This technique was demonstrated early on in HDTV by
Harris I think,
> but was deemed too unwieldy, prone to bad artifacts, and caused large
variations of bits
> required.  Someone fixed most of the issues for Fox, and they built their
whole network
> distribution system based on stream-splicing technology.  Interested
parties can look up
> MPEG II and I, P, and B frames.  I can find out if Fox still uses this
technology.  PBS
> is even more complicated.
>
>  
>
> Another problem is caused when a broadcaster wants to use âoeunusedâY\..
transmission
> bits for a second, third, or more channels.  As long as the primary
channelâ(TM)s
> programming is either SD material, or HD content with little motion, that
is OK.  But,
> if the primary contains fast motion (such as hockey or basketball games)
in HD, many
> more bits are required.  If the broadcaster has âoereservedâY\.. a certain
amount of
> data payload for these other channels in their encoderâ(TM)s stat-mux, the
primary
> content WILL become bit-starved, and artifacts show up quickly.
>
>  
>
> Donâ(TM)t get me started on cable companies.  They will take the primary
channel
> bitstream, set a ceiling in a stat-mux, and then multiplex several HD
signals on a
> single cable channel.  Too bad for you if the overall aggragation of bits
is more than
> what they allow on a single cable channel.
>
>  
>
> As far as analog audio recording, I would rather listen to the very minor
CD drum damage
> than listen to constant tape hiss, scratches, and distortion due to poor
dynamic range. 
> I HATE constantly hearing hsssssssssssssss.  But, thatâ(TM)s my trade-off
in life, and
> Iâ(TM)m glad that others think differently!
>
>  
>
> My oxygen-depleted speaker cables just came in, so Iâ(TM)m off to rewire
my Mac.  No,
> not a computer Mac.
>
> Terry, WB4JFI
>
>  
>
>  
>
>  
>
> From: [[Joseph Bento]]
>
> Sent: Saturday, April 13, 2013 8:57 PM
>
> To: [[tacos at amrad.org]]
>
> Subject: Re: 8K TV
>
>  
>
> We have two TV sets in my house.  Both are CRT based, and both are over 10
years old. 
> One is HD capable.  However, I find analog broadcasts more realistic and
natural.  There
> is something in the detail of most HD programming that somehow makes it
look artificial,
> especially when there is fast movement in the characters.  That
'something' I really
> can't explain.  Similar to the percussion track on some CD's - the drums
don't sound
> natural.  There's something a bit jarring in the sound.  That same
irritant is not
> present on an all analog recording.
> Joe, N6DGY
> On 4/13/2013 6:02 PM, Andre Kesteloot wrote:
>
>
>                On 4/13/2013 19:53 PM, Phil wrote:
>      
>
>
>                     "â¦a whopping 16 times the picture resolution as
>           todayâ(TM)s HD..."
>                               
>
>                     Really? My tin eyes can rarely distinguish between a
true
>           HD broadcast and the same rendered in 'regular' resolution.
>           
>
>
>           
>      Phil,
>      
>      although 16 times better resolution appears a bit silly indeed, there
is a
>      marked difference in today's transmissions.
>      In the early 1960s' we lived in Cambridge (UK) and watched British
TV. It
>      looked good enough for me (although was I had seen in 819 lines in
France was
>      definitely superior).
>      
>      Now, our channel 26 WETA transmits some old British programs (from
tapes I
>      presume)  and I must say that some of the Benny Hill re-runs
definitely lack
>      sharpness. (which, because of the content, hum hum, I sincerely
deplore ... 
>      :-)
>      
>      73
>      André
>      
>      
>      
>
>      _______________________________________________
> Tacos mailing list
> [[Tacos at amrad.org]]
> [[https://amrad.org/mailman/listinfo/tacos]]
>
>
>
>
>
----------------------------------------------------------------------------
------------
>
> _______________________________________________
> Tacos mailing list
> Tacos at amrad.org
> https://amrad.org/mailman/listinfo/tacos
>
> _______________________________________________
> Tacos mailing list
> Tacos at amrad.org
> https://amrad.org/mailman/listinfo/tacos
_______________________________________________
Tacos mailing list
Tacos at amrad.org
https://amrad.org/mailman/listinfo/tacos