Coast - awful filmic effect

wrote:

John Russell wrote:

That explains my disappointment with my new LCD. With my previous
Panasonic CRT movement was fluid, now everything jumps. It's as if
the CRT triggers an expectation in the brain that it's seeing fluid
movement and the brain perceives it as such.

If it's jumping, then the frame rate is wrong. Many flat panels
apparently use 60fps (only) internally - useless for our 50fps
sources.

It could also be a case of a bad deinterlacing algorithm.

Normal, fluid, 50 Hz fields-based interlaced video becomes distractingly
stuttery if it is deinterlaced to a 25 fps frame-based progressive
video - and that's exactly what software-based DVD players running on a
PC screen often misguidedly do if they encounter interlaced material. I
wouldn't be surprised if there are some LCD panels out there that had
taken this approach.

Of course, the flat-panel tv set _could_ generate 50 fps progressive
frames out of the original fields - with the help of spatial
interpolation and motion estimation / compensation / phase correlation
(there's a BBC paper about this [1]). That would at least retain the
original temporal resolution even if synthesizing the "missing" lines
can be difficult.

But as I see it, the real problem is that no-one really knows what kind
of nasty image processing (or mishandling) is happening inside a
Chinese-made LCD panel. The manufacturers surely don't tell you any
technical details that would matter. You would have to feed carefully
prepared test signals to the panel yourself and shoot the screen with a
high-speed special camera to see what's actually going on, and even then
it would be largely just guesswork.

_____

[1] http://www.bbc.co.uk/rd/pubs/reports/1996-09.pdf

--
znark

Jukka Aho wrote:
wrote:

John Russell wrote:

That explains my disappointment with my new LCD. With my previous
Panasonic CRT movement was fluid, now everything jumps. It's as if
the CRT triggers an expectation in the brain that it's seeing fluid
movement and the brain perceives it as such.

If it's jumping, then the frame rate is wrong. Many flat panels
apparently use 60fps (only) internally - useless for our 50fps
sources.

It could also be a case of a bad deinterlacing algorithm.

Normal, fluid, 50 Hz fields-based interlaced video becomes distractingly
stuttery if it is deinterlaced to a 25 fps frame-based progressive
video - and that's exactly what software-based DVD players running on a
PC screen often misguidedly do if they encounter interlaced material.

There's a deinterlacing setting on the MPEG-2 decoder - auto, blend,
bob or weave. Auto (which is the default) uses blend far too often on
some decoders. The least bad for 50Hz field-based interlaced video is
bob. For such content, weave gives 25fps with mice teeth/combing while
blend gives 25fps with the mice teeth/combing replaced by blur.

Some software doesn't give access to this setting, but it's there!

I've never seen an LCD which runs at 25fps, but I don't discount the
possibility.

Of course, the flat-panel tv set _could_ generate 50 fps progressive
frames out of the original fields - with the help of spatial
interpolation and motion estimation / compensation / phase correlation
(there's a BBC paper about this [1]). That would at least retain the
original temporal resolution even if synthesizing the "missing" lines
can be difficult.

Most just bob the content to 50fps - that's trivial.

But as I see it, the real problem is that no-one really knows what kind
of nasty image processing (or mishandling) is happening inside a
Chinese-made LCD panel. The manufacturers surely don't tell you any
technical details that would matter. You would have to feed carefully
prepared test signals to the panel yourself and shoot the screen with a
high-speed special camera to see what's actually going on, and even then
it would be largely just guesswork.

I agree that it's a problem, but I don't really care what they do if it
looks horrible! However, few are doing anything very clever and it's
usually quite easy to spot a bob, weave, blend, or smart bob using a
standard test card with moving zone plate at various rates. Frame rate
conversion is obvious with fast smooth pans or just scrolling text.

[1] http://www.bbc.co.uk/rd/pubs/reports/1996-09.pdf

That's a good forward thinking paper. I believe there are lots of other
approaches now.

Cheers,
David.

Alan Pemberton wrote:

Why couldn't an lcd say display a 50i source as one field followed
by the next?

The concepts of scanning, refreshing and 'fields' fall apart with flat
screens. In order to display a range of brightnesses bits of the
screen are turned on and off, and the contents updated, ad hoc.
That's one of the reasons movement looks a mess.

I've been toying with the idea that interlaced scanning (and the
non-ideal focusing of the electron beam in a CRT-based set, with the
"hot spot" spanning over multiple picture elements on the screen) could
be simulated on a SED display matrix.

That, of course, would require that the panel's firmware "paints" the
picture on the screen with a 15 kHz line rate, just like a CRT would,
and lets the "painted" phosphor sites decay after each refresh, instead
of just keeping the image "on" all the time.

Even though that method would reintroduce the 25 Hz line twitter and the
50 Hz flicker (which some people find objectionable), it would be great
if this kind of "emulation" of interlaced scanning could be optionally
turned on.

--
znark

"Jukka Aho" wrote in message
...
Alan Pemberton wrote:

Why couldn't an lcd say display a 50i source as one field followed
by the next?

The concepts of scanning, refreshing and 'fields' fall apart with flat
screens. In order to display a range of brightnesses bits of the
screen are turned on and off, and the contents updated, ad hoc.
That's one of the reasons movement looks a mess.

I've been toying with the idea that interlaced scanning (and the non-ideal
focusing of the electron beam in a CRT-based set, with the "hot spot"
spanning over multiple picture elements on the screen) could be simulated
on a SED display matrix.

That, of course, would require that the panel's firmware "paints" the
picture on the screen with a 15 kHz line rate, just like a CRT would, and
lets the "painted" phosphor sites decay after each refresh, instead of
just keeping the image "on" all the time.

Even though that method would reintroduce the 25 Hz line twitter and the
50 Hz flicker (which some people find objectionable), it would be great if
this kind of "emulation" of interlaced scanning could be optionally turned
on.

--
znark


It does seem like an awful lot of effort to reproduce the "apparent" motion
blur created by a CRT which leads to smooth motion. Wouldn't it be easier to
actually capture motion blur in the first place? DVD's look great on my new
LCD TV and if you pause them during motion what you have is a blurred image
i.e captured motion blur. If I pause my SKY+ I never get anywhere near the
same amount of blurring, it's as if they are treating TV like a cartoon, a
series of unblurred still images.

The message o.uk.invalid
from lid (Alan Pemberton) contains
these words:

====big snip====

Not at all. I found a lot of useful information in the article linked to
above, and it's a very common misconception that tv crts have
long-persistance phosphors that store the image between scans. Some
computer monitors do/did, and any movement (scrolling text for example)
looked awful.

That's it exactly. the phosphors used in TV CRTs have such a short
persistence, they might as well have zero persisitence. It's persistence
of vision (in the eye itself) that does the 'frame storage'.

--
Regards, John.

Please remove the "ohggcyht" before replying.
The address has been munged to reject Spam-bots.

John Russell wrote:

[emulating interlaced scanning and a sweeping electron beam
on a SED display matrix]

It does seem like an awful lot of effort to reproduce the "apparent"
motion blur created by a CRT which leads to smooth motion.

I'm not sure 1) why you're bringing motion blur into this discussion
(since I didn't mention anything about motion blur in my original
message), 2) why do you think that CRT and motion blur would be somehow
inherently related, and 3) why do you think that only blurry motion
would look smooth.

I'm only proposing interlaced scanning on SED displays because they're
(supposedly) the flat-panel display technology that is, in many ways,
most similar to CRT-based tvs. Adding an emulation of interlaced
scanning on top of that technology would, in theory, make them like flat
CRTs, and help in displaying interlaced material as it was originally
intended to be displayed when it was shot.

Even if the previously-described emulation of interlaced scanning _was_
implemented in a SED panel, it would not have to be always on. It could
be a user-selectable menu option, or something that is only switched on
when needed (when genuine interlaced material is detected on the inputs,
for example.)

--
znark

"Johnny B Good" wrote in message
...
The message
o.uk.invalid
from lid (Alan Pemberton) contains
these words:

====big snip====

Not at all. I found a lot of useful information in the article linked to
above, and it's a very common misconception that tv crts have
long-persistance phosphors that store the image between scans. Some
computer monitors do/did, and any movement (scrolling text for example)
looked awful.

That's it exactly. the phosphors used in TV CRTs have such a short
persistence, they might as well have zero persisitence. It's persistence
of vision (in the eye itself) that does the 'frame storage'.

Simple proof can be obtained by photographing the TV screen using a fast
shutter.

Persistence of vision varies from person to person, and I have known people
find their new TV set very flickery. If you go into a shop where there's a
large display of TVs you can demonstrate to yourself that persistence of
vision is less in the peripheral areas of the visual field.

Bill

John Russell wrote:

It does seem like an awful lot of effort to reproduce the "apparent" motion
blur created by a CRT which leads to smooth motion.

I think you need to read the thread, and the Poynton links, John -
that's not right at all!

Wouldn't it be easier to
actually capture motion blur in the first place? DVD's look great on my new
LCD TV and if you pause them during motion what you have is a blurred image
i.e captured motion blur. If I pause my SKY+ I never get anywhere near the
same amount of blurring, it's as if they are treating TV like a cartoon, a
series of unblurred still images.

That's nothing to do with DVD vs DVB. It's all about shutter speed. You
were probably comparing a DVD of a 25fps movie (with ~ 1/30th shutter
speed) with a Sky+ of a 50fps video (with anything from 1/50th shutter
speed to 1/5000th).

Films, on the whole, do tend to include more motion blur, because of
the lower frame rate. They don't have to, but increasing the shutter
speed make motion visibly "strobe" at 25fps before it does at 50fps,
for obvious reasons.

Cheers,
David.

On Wed, 15 Nov 2006 22:05:48 +0200, "Jukka Aho"
wrote:

I've been toying with the idea that interlaced scanning (and the
non-ideal focusing of the electron beam in a CRT-based set, with the
"hot spot" spanning over multiple picture elements on the screen) could
be simulated on a SED display matrix.

There's an experimental system around which uses an LED matrix as the
backlight for an LCD panel and 'scans' the matrix in this sort of way.
The LEDs are about ten times bigger than the pixels, so they tend to
light up a diffuse area. Apparently this gives a similar effect to a
traditional CRT in the way that highlights tend to blow-out etc.

I haven't seen this myself, so I can't comment on the results

Steve

The Doctor Who Restoration Team Website
http://www.restoration-team.co.uk

Steve Roberts wrote:

I've been toying with the idea that interlaced scanning (and the
non-ideal focusing of the electron beam in a CRT-based set, with the
"hot spot" spanning over multiple picture elements on the screen)
could be simulated on a SED display matrix.

There's an experimental system around which uses an LED matrix as the
backlight for an LCD panel and 'scans' the matrix in this sort of way.
The LEDs are about ten times bigger than the pixels, so they tend to
light up a diffuse area. Apparently this gives a similar effect to a
traditional CRT in the way that highlights tend to blow-out etc.

Interesting. Can you tell any more about this experimental system? For
example, is it being considered for commercial production?

The reason why I only mentioned the SED technology (and not LCD panels)
in the above was because SED panels could - at least in theory - use
exactly the same kind of phosphors as CRT-based tv sets. In other words,
the colour rendition, rise and fall-off times, and afterglow
characteristics could be made practically identical to CRT-based sets -
the only real difference being that the screen is flat and would never
need any geometry adjustments.

--
znark