[opendtv] Re: Overscanning on LCD TVs
- From: Kilroy Hughes <Kilroy.Hughes@xxxxxxxxxxxxx>
- To: "opendtv@xxxxxxxxxxxxx" <opendtv@xxxxxxxxxxxxx>
- Date: Mon, 16 Nov 2009 17:42:57 +0000
You could use a Blu-ray player and Joe Kane's "Video Essentials" (and similar
from Stacy Spears, etc.) as and HDMI/YPbPr test generator in the $150 range.
Hundreds of specific test patterns have evolved to check for many of the things
that go wrong with video, including errors commonly found in decoders when they
convert 4:2:0 to 4:4:4, 10-bit to 8-bit rounding bias, requantization, aspect
ratio distortion, setup, gamma slope out of black ... as well as all the usual
calibration stuff like color primaries, color temp, gamma, gray scale over the
entire gamma curve, etc. You can always see errors, but it helps to know which
are in encoding, decoder/signal processing, interface, display processing, and
display.
Odds are that something like color banding is happening in all four places, so
you have to do mental subtraction to judge encoding quality (for instance) on
LCD displays that don't have the contrast and quantization granularity to
accurately represent 8-bit gradients (and will introduce dithering noise
instead to mask that). People disappearing into the shadows (setup and gamma
errors relative to viewing conditions and encoded render intent), and parts of
the picture getting hacked off (overscan) are a little easier to measure, but
still subjective in the case of black levels and display gamma for particular
viewing conditions.
Kilroy
From: opendtv-bounce@xxxxxxxxxxxxx [mailto:opendtv-bounce@xxxxxxxxxxxxx] On
Behalf Of dan.grimes@xxxxxxxx
Sent: Saturday, November 14, 2009 11:28 AM
To: opendtv@xxxxxxxxxxxxx
Subject: [opendtv] Re: Overscanning on LCD TVs
"The thread started on "overscan", but evolved to other stuff that is often
wrong: setup/black level, sample aspect ratio pixel aspect ratio, cropping,
primaries/color temp, matrix, gamut, gamma, progressive encoded as interlace
(with 3:2 pulldown), etc."--Kilroy
When I originated the thread, I started with one of the many locations that I
am having the translation errors. The thread developed into a second problem,
and here am going to introduce a third that I am experiencing.
In order to save money in our new production facility, we used consumer level
monitors where production quality monitors were not critical. We used AJA Hi5
HD-SDI converters to convert HD-SDI to HDMI to go into the Sony Bravia monitor.
The monitor is overscanning the picture and there could be some color space
issues, too. This was the first issue in the original post.
The second problem involves trying to get a VGA or DVI (of all types) into an
HD-SDI signal that can be used in the production chain. We have Gefen
equipment that translates VGA to DVI-D and then DVI-D to HD-SDI. There are
problems at each of the translators. I'd like to state for the record that I
did not specify or purchase the equipment.
I have been able to make adjustments to both converter boxes that make for an
HD-SDI signal that is somewhat useable. By adjusting the converter so that
black is at 16 and white at 235, the HD-SDI signal loses the green banding
"noise" and becomes somewhat stable on a display. However, when looking on an
HD-SDI analyzer, the overshoot and noise in the signal is still translating to
noise in the numbers. Here is where my knowledge runs out: I didn't think it
was possible to have numbers below 16 and over 240 within an HD-SDI but the
HD-SDI waveform monitor shows these distorted and overshooting edges
(originally from the analog signal) clearly above and below those numbers and
the picture shows green distortions/noise. When to excess, the HD-SDI signal
is not usable to equipment and cuts in and out. So clearly the Gefen box is
not putting out a compliant SMPTE 292M/296M/274M signal.
And now to introduce a similar problem as with the first issue, all is not
perfect in the production world, either. We have Sony production monitors that
have a 1680x1050 display and we are producing in 1280x720. The monitors have
the following settings:
Normal: the full raster is seen but there are black areas above and below.
The picture appears slightly squished vertically to me(the proper test signal
could confirm but I haven't found one in our test signal generator).
Over: The picture is slightly overscanned at the horizontal yet black bars
remain above and below.
Native: 1280x720 is mapped to 1280x720, displayed 1:1, where the picture is
only taking a fraction of the screen with black all around; the image appears
to be natural (not distorted vertically or horizontally).
Full: The picture fills the screen but part of the edges to each side are
cropped off and the full image is top and bottom, no black.
Clearly, the 16:9 picture on a 16:10 production monitor is demanding the need
for the various settings.
While I don't have aspect ratio test signals in our HD-SDI sources, I do have
test signals in my computer and I would love to use them to find out what is
going on with each of these conversions but I don't have much of chance without
being able to measure and normalize the outputs (VGA and HDMI) to see if they
are proper before I use them as a test source. There is something to be said
for having a good test signal generators and analyzers of each type. Wish I
could afford all those items.
I think modern TV production test signal generators need to include test
signals that would help flush out some of these issues. Greyscales and some
standardized test signals are usually available to figure out many of the color
space issues but with the plethora of rasters, aspect ratios and other issues
that Kilroy mentions above, some signals that handle these issues need to be
in the engineer's toolbox, too.
Dan
Other related posts: