[opendtv] AW: Re: De-interlacing with HQV high quality video processing
- From: "Hoffmann, Hans" <hoffmann@xxxxxx>
- To: <opendtv@xxxxxxxxxxxxx>
- Date: Thu, 6 Oct 2005 20:28:40 +0200
Mark,
I agree to your analysis. We used also some LDK6000 for our test shootings
of sport in 1080i/25 and 720p/50. Quite interesting results.
Could you forward the 702p reference? Should be interesting.
Regards,
Hans
-----Ursprüngliche Nachricht-----
Von: opendtv-bounce@xxxxxxxxxxxxx im Auftrag von Mark Schubin
Gesendet: Do 06.10.2005 16:36
An: opendtv@xxxxxxxxxxxxx
Cc:
Betreff: [opendtv] Re: De-interlacing with HQV high quality video
processing
Hoffmann, Hans wrote:
>But the real issue is what happens in the broadcast window.
>The big contributor to the visibility of image artefacts is the
compression
>format such as MPEG-2, H.264-AVC, proposed SMPTE VC-1 (you my name
all of
>them). All work better when they have to compress progressive
pictures.
>Putting MPEG-2 aside as history, we also see that the modern
compression
>systems in the area between 6 to 18 Mbit/s perform much better with
>progressive input signals.=20
>
>Let's discuss this......I am curious on some views.
>
>
I'll throw something into the ring.
On March 30, I worked on a live PBS show from a theater in the Time
Warner Center in New York. The producer knew we would be shooting
in an
HD production unit, so he contacted a high-end electronics showroom
in
the facility to see whether they'd want to take an HD feed (even
though
PBS would take the show only in SD) and host a party. The showroom
was
delighted.
Then, as the show got closer, the producer realized that, with a
number
of SD roll-in tapes and SD graphics, it made sense to downconvert
the
cameras and run the switcher is SD. But what about the showroom?
It
was too late for them to cancel their party.
So we took the SD output of the switcher, upconverted it back to HD,
and
sent it to the showroom as HD-SDI. Reports from the party were,
"This
is the best HD we've ever seen!" I attribute that to lack of
compression artifacts.
So, what about the lost resolution?
It's true that downconversion throws away more than half the
resolution
both horizontally and vertically. But the psychophysical sensation
of
"sharpness" is proportional to the square of the area under an MTF
curve
(Schade). As the MTF curve typically drops near zero long before
the
limit of resolution, the bulk of the resolution loss in
downconversion
loses practically no area under the curve.
As we have seen here with Tom Barry's pictures, there appears to be
a
point at which the downconversion DOES make a significant
difference.
But, if we downconvert to the point where it doesn't make a
significant
difference prior to compression, we reduce the amount of information
the
compressor must process. Perhaps that's a good way to reduce
compression artifacts.
Another point from the same show: We were using LDK-6000s, which
can be
set to 1080i or 720p native (thanks to sub-pixel sensors on the
imagers). We were in the middle of a rehearsal when we decided to
check
the cameras. They all looked the same, but one camera kept
indicating
on the monitor that it was 720p. Sure enough, it WAS 720p.
So, why did it look the same as the others? The material we were
shooting -- people with relatively slow movement -- did not lend
itself
to making interlace artifacts visible. NHK research showed that
702p
(not a typo -- 702, not 720) active is approximately equivalent
visually
to 1080i active. That is attributed to an interlace coefficient.
But
1920 is greatly more than 1280, and there is no horizontal
equivalent to
the interlace coefficient to explain the difference. So 1280 x 720
cameras look softer than 1920 x 1080 cameras. Even 4:3 525-line SD
cameras have typically been 1350 x 486 to increase sharpness; it's
silly
for 720p cameras to have no more than 1280 sensors horizontally.
But,
in 720p mode, the LDK-6000 is 1920 x 720p. The horizontal sharpness
is
identical in either 720p mode or 1080i mode. And that explains to
me
why, even very close to a professional quality-control monitor, we
didn't see a difference between that camera and the others.
TTFN,
Mark
----------------------------------------------------------------------
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.
**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager.
This footnote also confirms that this email message has been swept by
MIMEsweeper for the presence of computer viruses.
www.mimesweeper.com
**********************************************************************
-- Binary/unsupported file stripped by Ecartis --
-- Type: application/ms-tnef
-- File: winmail.dat
----------------------------------------------------------------------
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:
- » [opendtv] AW: Re: De-interlacing with HQV high quality video processing