[opendtv] Re: Microsoft Exec: 1080p HDTV Is Meaningless
- From: Craig Birkmaier <craig@xxxxxxxxx>
- To: opendtv@xxxxxxxxxxxxx
- Date: Wed, 23 Aug 2006 10:39:03 -0400
At 11:46 AM -0400 8/22/06, Tom Barry wrote:
Craig -
I'm still not sure why you single out non-nyquist filtered images as
being a special problem for non-square pixels.
Because virtually all computer applications and web pages use the
convention of non-Nyquist filtered square pixels to represent
graphics on a display (computer or any entertainment display that
also can display this type of images.
To me, the main problem with non-nyquist images is the values at the
designated sample points are correct but there is an ambiguity about
what the values are between those points. Especially with moving
images any interpolation or scaling to calculate the missing values
will likely disagree from frame to frame.
It is rare that "moving images" are created with non-Nyquist filtered
samples. The main exception is simple graphics like animated GIFs
that typically are not filtered. The very nature of moving images
requires proper filtering to create natural motion.
When information is created using non-Nyquist filtered samples the
intention is almost always to maximize contrast, as is the case for
the text that you are looking at as you read this (and yes, there are
circumstances where text is now being filtered as the on-screen
rendering resolutions increase (e.g. 72dpi versus 96dpi or 110dpi).
Theere is no ambiguity about the values between the points in
non-Nyquist filtered samples - the intention is to display them
WITHOUT filtering to maximize contrast. When they ARE filtered all
kinds of bad things happen:
- Aliasing
- reduced contrast
- blurring of the information
But this means a problem with all scaling or shifting, not just on
non-square pixels. Do you just mean that non-square pixels implies a
higher likelihood of scaling?
When we are dealing with samples that are properly filtered, the
pixel geometry can be arbitrary - in essence the proper filtering
characteristics/rise times were determined at the time the image
(still or moving) was sampled. It is important to note that to create
fluid motion, especially at lower frame refresh rates like 24P, we
may also need additional blurring of the edges of moving object to
create the illusion of fluid motion. Another way of looking at this
is what happens when you acquire a still image at a slow shutter
speed - the moving objects will be blurred.
Any pxel geometry can be used with non-Nyquist filtered samples. You
may recall that this was the case with early computer display formats
like CGA and EGA. The problem is that when these samples are
displayed at other pixel aspect ratios (without filtering) the image
will be distorted. This is the reason that the computer industry
quickly adoped square pixels as the lingua franca for non-Nyquist
filtered samples.
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: