[opendtv] Re: Samsung Rep admits HDMI won't support 1080p until
- From: Olivier Houot <olho_avatar_i@xxxxxxxxxx>
- To: opendtv@xxxxxxxxxxxxx
- Date: Fri, 11 Nov 2005 02:31:13 +0100
Craig wrote
>You are hung up on the big numbers, not the
>factors that must be observed to deliver high
>quality images at any spatial/temporal resolution
>to ANY display.
It is not the love of big numbers for their own sake. The numbers stem
naturally from the 30° viewing angle and 30 cpd resolution. Of course,
if you lower the resolution to 22 cpd or below, 1280x720 fits right in,
except when a red blinking light will actually be displayed, in which
case the viewers will become aware that the system is operating with
NEGATIVE safety margins.
You say a 100 inches screen is needed to satisfy both the 30° angle
immersion requirement and the need of the viewer for the most
comfortable viewing distance. Fine by me, this is exactly the kind of
size i would settle for. I don't know of many living rooms where the
walls are not at least 100 inches wide, so there would be a place on at
least one of the walls for such a screen (and a 16/9 screen would be
only 56 inches high).
Current living rooms are arranged for smaller SDTV sets, but if such
home theaters become fashionable, people will find clever ways to
accomodate it and hide the screen when not in use.
So a 100 inches screen is still, or will be, a consumer display.
And with a video projector, you can always zoom down the picture to
avoid eye tracking when doing computer work.
So much for the big numbers hung up. You see, i am not clamoring for
UHDTV in my living room, though the technology already exists (And
yet...A 120° viewing angle would go a long way towards perfect
immersion. THAT would require a redesign from scratch of the living room
(unless you use individual spheres or goggles). I would suggest building
a hemispherical ceiling that would serve as the screen. You would watch
it lying comfortably on a bed, and the whole perimeter of the room would
still be accessible).
>I have a 50 inch HDTV display and there is a HUGE
>difference between SD and HD sources. With HD you
>actually want to move the sofa closer to the
>screen.
I agree that a 50 inch HD picture is stunning (some demos available in
european supermarkets, for some time now).But is still feels like
watching television, as it is not immersive. The threshold is above that
value.
>You need big screens with lot's of pixels when
>the viewer is seeking the HD viewing experience
>you defined above. But you do not need the source
>to have the same resolution. It is a question of
>eliminating the perception of the raster.
I understand that once the final display has taken over the job of
hiding the pixels by providing them in sufficiently large amounts, you
can display an upscaled lower resolution source on it, and it will
probably look fine. But an optimal audiovisual chain should have the
goal of just saturating the human senses, with an adequate safety
margin. If the screen size/pixel count/ viewing distance is designed to
be at the limit of the viewer's eye capability, then your lower
resolution source will look less realistic.
>You are smoking something really good if you
>think that we can deliver the level of fine
>detail that can be exploited by the 1920 x
>1080@60P format via the MPEG-2 distribution chain
>at less than 20 Mbps. MPEG is a low pass filter,
>and the first thing to go will be those fine
>details.
This is an interesting point. I've gone through some parts of the
document you have pointed out.Unfortunately, this statement is in the
"future work and other issues" section :
7.4.1 What is the proper tradeoff between compression quality, data
rate, image resolution, image dynamic range, image color fidelity, and
image temporal rate? How should these tradeoffs be resolved for
different industries and applications (e.g., medicine, scientific
visualization, production, videoconferencing, transmission)?
It would be nice to have a study that reliably says , for a given
channel bandwidth, which resolution / compression ratio compromise
produces the best visual impression, and whatreally remains of the
original source after that transformation.
It reminds me of the Pixel+ concept : given the bandwidth of a 100 Hz
CRT, which combination of scanning rate and picture resolution will
produce the best picture? And it turned out the optimal point was not at
one of the usual extremes.
There would be a need to explore the continuum of resolution/compression
settings, and not only at the usual 720 and 1080 points.
However, i would have the following remarks :
- if 720p can be transmitted today using 15 Mbps in Mpeg2, it is not
unreasonnable to think that a somewhat superior visual result can be
achieved with 1080p using 24 Mbps in H.264.
- compressing a picture by a 10:1 ratio using JPEG will not visibly
degrade it. Reducing the picture resolution by a factor of 10 will
degrade it significantly. So is resolution downgrading really the best
tool in that endeavour?
- the 20 mbps limit is tied to terrestrial channel transmission, which
is but one of the possible medias nowadays. DBS using DVB-S2 can achieve
60 Mbps per transponder, if memory serves. Cable operators, having
control over their bandwidth and the set top boxes used to receive it,
could aggregate several channels together to provide a sufficient
datarate for the purpose. And let's not forget distribution of content
through material media like discs or tapes. Current VAX technology tapes
can provide 96 Mbps for more than 3h30 (160 GB)in a small form factor,
and there are holographic discs coming....
>The justification to avoid production in
>this format is that it is simply overkill -it is
>both economically unjustified, and unnecessary in
>terms of the laws of physics.
If we only ask for 1080@60p tolerance, no element of the chain is
compelled to operate at 60P right now, except the decoder IC. When
production in that format becomes sufficiently cheap to be a profitable
differentiator, then the relevant links can be upgraded, without the
need to worry about the installed base.
However, if really the human eye is limited to 21 or 22 cpd , which
means the 30° FOV can be satisfied with 1280x720, i will have to agree
that 720p is not a bad choice, but i feel there is really not much
safety margin. In a digital progressive world, perhaps the transmission
format should be at some intermediate point like perhaps 850p or 900p,
with capture occuring at 1080p or above.
----------------------------------------------------------------------
You can UNSUBSCRIBE from the OpenDTV list in two ways:
- Using the UNSUBSCRIBE command in your user configuration settings at
FreeLists.org
- By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word
unsubscribe in the subject line.
Other related posts:
