[opendtv] Re: Philips ClearLCD technique for motion sharpness
- From: jeroen.stessen@xxxxxxxxxxx
- To: opendtv@xxxxxxxxxxxxx
- Date: Thu, 20 Jan 2005 08:51:33 +0100
Hello all,
Craig wrote:
> Sounds like Philips just can't let go of scanning screens.
??
This is not by choice. The fact is that all LCD screens are
refreshed top to bottom, using the largest possible fraction
of the frame time. It simply can not be done any faster.
It takes time to charge each pixel capacitance to its opposite
polarity. Trying to do it too fast leads to uniformity errors,
a "dirty screen".
So the bottom of the screen is refreshed some 15 ms later than
the top, and the pulsing of the backlight must match that delay,
in order to sample the transmitivity of the LCD at the moment
when it is most stable. It is the only proper way...
> It will be interesting to see if this produces a similar
> psycho-visual response to that of a scanning CRT.
It always has been with LCDs, the only thing that we change
is that the always-on behaviour, a.k.a. the temporal sample-
and-hold, is replaced by a pulse of (e.g.) <= 25% duty cycle.
This greatly reduces the motion smear on the retina of the
tracking eye. But it re-introduces frame flicker, indeed.
And then we solve that. The demo at CES was still without a
solution for the flicker, just so that you know that.
> This may be useful for legacy content, but it seems from here that
> the correct approach is to deal with the shortcoming of the source,
> rather than the limitations of the display to deal with sub-optimally
> sampled sources.
This is totally a non-issue. At worst, a mismatch may lead
to a geometric distortion of moving objects. Big deal.
> Or you could just move to a higher frame rate at
> the source...
Motion-compensated frame-rate up-conversion is one obvious
solution, for ALL flickering displays. (And a proper display
IS a flickering display !) This includes CRT, OLED, FED, SED,
and any light valve display with a blinking or scanning light.
However... too many LCDs are not specified to work above 60
frames per second. There may be immediate non-uniformity, or
the non-uniformity may occur too early in its life time (due
to degradation of the TFT switches). Increasing the spec would
reduce the yield, and that is something that the manufacturers
with their multi-billion dollar factories are not waiting for.
Also, applying frame rate conversion to 2 megapixel signals
costs a hell of a lot of processing bandwidth.
> Actually, Jeroen has already told us about another Philips
> initiative to deal with this issue.
> Sorry Jeroen but I cannot remember the trademark you are using
> for the techniques Philips is using for frame rate interpolation
> at the display.
"Natural Motion".
> At one time the TriMedia processor was doing this at SD
> resolution.
That was a simplified version: "Layered Natural Motion".
This effort has been continued as "Software Natural Motion",
now also with significant hardware assistance. Sony is doing
something similar, with some help from Philips Research.
Any form of Natural Motion is a significant improvement of
the quality of moving images, but you still have to stay
within the limitations of the display, of course.
> As Bert may recall, there have been several recent announcements
> about new chips that do motion compensated prediction to improve the
> temporal presentation of lower frame rate sources such as 24P (film
> or video).
Yes, it is very important for film sources, at least if you
believe that it was not a deliberate artefact, intended by
the director... "Suspension of disbelief" and all.
But it is still a major hurdle to do this for HD resolutions
and beyond, as many displays are now offered with 1920x1080p.
That is a pixel bandwidth of at least 120 MHz at 60 Hz frame
rate. And then there's that issue that the manufacturers just
won't specify these panels at more than 60 Hz. Yet.
> Just another example of old truism that there's more than
> one way to solve any problem.
Just like there is more than one way to overlook any problem... ;-)
> As for re-writing the ATSC or DVB standards, it is mostly an
> editing job, and a rather easy one at that.
> You simply delete all the stuff after the modulator and
> definition of the transport stream - the market place is
> driving the rest.
And in the process you lose compatibility with only a couple
of million receivers and displays out there... :-(
Best regards,
-- Jeroen
+-------------------------------+----------------------------------------+
| From: Jeroen H. Stessen | E-mail: Jeroen.Stessen@xxxxxxxxxxx |
| Building: SFJ-5.22 Eindhoven | PHILIPS Applied Technologies |
| Phone: ++31.40.2732739 | Visiting & mail address: Glaslaan 2 |
| Mobile: ++31.6.44680021 | NL 5616 LW Eindhoven, the Netherlands |
| Pager: ++31.6.65133818 | Website: http://www.pdsl.philips.com/ |
+-------------------------------+----------------------------------------+
----------------------------------------------------------------------
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:
