[opendtv] Re: 1080p@60 for MPEG-2
- From: Craig Birkmaier <craig@xxxxxxxxx>
- To: opendtv@xxxxxxxxxxxxx
- Date: Mon, 12 Nov 2007 07:58:40 -0500
At 7:10 PM -0500 11/11/07, Manfredi, Albert E wrote:
Craig Birkmaier wrote:
The ITU paper said that the advantage of using MPEG-2 in the 1080 at 60p
mode was cost. Quoting:
"A large number of display and capturing devices will support
1080@60p/50p in the near future and it is urgently desired to add
support for these video formats in the existing MPEG standard.
Complexity is one of the biggest issues for such applications,
especially for software solution. Although AVC(14496-10 | H.264)
supports such video formats, complexity is too high for some
applications, especially simple software based applications, because of
high demand of CPU and memory resources. In addition, it is not always
possible to assign the entirety of the available CPU resource to the
video decoder in some applications. We believe that there is a great
demand for MPEG-2 video, especially for software based solutions. MPEG-2
video is the appropriate format for such applications, in terms of both
complexity and picture quality."
The ITU is correct that there are (will be) displays and cameras that
can deal with 1080@50/60P.
It is important to note however, that many, if not most of the 1080P
displays on the market today could not handle 1080@50/60P because the
internal image processing circuitry cannot handle the clocks
necessary for input formats with these pixel clocks. This may not be
a problem, IF, the display has an HDMI input that can handle the
higher clock rates directly. For example, Sony notes that 1080P
sources bypass their Digital Reality Creation (image processing)
circuitry.
What this means is that it may well be possible to deliver
1080@50/60P to some existing displays via an external box that can
connect via HDMI. Sony and Panasonic both note that the only way to
do this today is via their Blu-Ray DVD players, and at least for now,
they are delivering 1080@24P, which is already covered by the MPEG-2
standard.
As for thew complexity issue and software codecs, I am already using
a software codec to play H.264 720P files. I have seen a variety of
tests and demos that involved using the MPEG-2 tools for higher frame
rate sources. These tests were shot on film at frame rates of 36, 60,
and 72 frames per second. The good news is that you can use longer
GOPs with more B-frames, which helps with coding efficiency relative
to 1080@30i, but you have much more detail to encode.
I have major doubts that it would be possible to deliver 1080@50/60P
using MPEG-2 at bit rates that are useful to broadcasters. The peak
bit rate requirements will peak to beyond 30 Mbps consistently. This
suggests that the RIGHT WAY to deal with 1080@50/60P in MPEG-2 land
is to resample to 720P for emission...but we have been through all of
this before.
I suspect that what is REALLY happening here is directly related to
intellectual property issues, and to the desire to extend proven
MPEG-2 designs to higher clock rates. This might be worthwhile, IF
there were practical applications for the new MPEG-2 level that the
ITU is suggesting.
I have participated in ITU activities related to the 1920 x 1080
Common Image Format. This is a very political process, driven by some
politically powerful people with the desire to have one HDTV format
to rule them all.
This is wrong-headed, but certainly not something new. This battle
has be raging for nearly two decades, and will likely still be the
subject of international political wrangling for another decade,
until 1080@50/60P becomes an economically attractive proposition.
If you open the other zipped files, you'll see that they are also adding
AVC modes. For studio or ultra high def purposes.
"Legacy technology" is *always* considered, Craig. It's called "good
engineering."
There comes a time when enough is enough.
Interlace WAS and IS a very crude analog compression technology that
has no place in our new digital world. This did not stop the ATSC and
the CE industry from embracing a legacy technology that afforded them
some competitive protection, while allowing them to load up both
MPEG-2 and MPEG-4 with a bunch of unnecessary intellectual property.
This was not good engineering, but rather, protectionist engineering
making it easier for the CE industry to defend its turf from the
advancing Moore's Law tidal wave.
Frankly, getting rid of an interlaced format that should never have
existed (1080i), by moving to progressive HDTV formats that have
little practical use for emission is a waste of time.
It's time to forget about formats and focus on proper image
processing and encoding practices that could care less about the size
or frame rate of the raster that is being processed. The MPEG
compression codecs are limited only by performance and complexity
issues - they could care less about formats. That is why there are
levels and profiles to place some upper bounds on an implementation
level.
Thanks to Moore's Law, it is now possible to consider higher clock
speeds and increased codec complexity. That is why the ITU and ISO
created the Joint Video Team and developed the H.264 codec rater than
extending MPEG-2 to higher clock speeds. By the time 1080@50/60P
becomes a practical format for acquisition and emission coding, there
will be a replacement for H.264. That's the way things should work in
our digital world.
Extending legacy designs is not a good engineering practice. It is
just another attempt to delay the application of good engineering
practices to protect a legacy.
Regards
Craig
----------------------------------------------------------------------
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:
