On Mon, 25 Aug 2008 19:45:33 +0100, Andy Champ 
wrote:
>
> Java Jive wrote:
> > 
> > On Sun, 24 Aug 2008 22:23:32 +0100, Andy Champ 
> > wrote:
> > >
> > > So the CRT scan line runs over the edge of the dot 
> > > mask/phosphor spots and interpolates the intervening lines in the 
> > > analogue domain.
> > 
> > This is rubbish!  You don't seem to understand how CRTs work!  Why
> > isn't the shadow mask preventing it straying off the phosphor dot? How
> > can it 'interpolate' when not hitting a phosphor dot  -  what can
> > cause the screen to emit light if there's no phosphor at a given
> > point?
> 
> Typical 19 inch CRT:  18 inch visible, that's ~450mm, which becomes 
> ~1600 pel across the diagonal, as it's 4:3, that's 960x1280.

Let's look at some actual specs ...
	http://tinyurl.com/6dkn88
... standing in for ...
http://h18000.www1.hp.com/products/quickspecs/north_america_01-1999/50143_na.PDF
... the first 19", your choice, is the V90.

19", or as you say, 18" visible equates to about 366x274mm visible at
4:3. The horizontal dot-pitch is 0.22mm, so there are about 1664
'native' pixels to display 1280 screen resolution ones. 366/1280 is
0.29, so you seem to be saying that 0.29 - 0.22 = 0.07mm is sufficient
for the separation for the holes in the shadow mask, which whatever
its actual value will also be the gap between the physical pixels, AND
the overlap of which you speak?  

Supposing for the sake of peace I accept that, it's not actually what
you said, and my remarks were a reply to what you actually said, which
gave me a completely different impression from that which you seem to
have intended.

> The beams aren't going to track precisely on the phosphor dots, so a 
> single colour pixel will probably be made of a bunch of dots with 
> different illuminations, dimmer the further they are from the ideal 
> centre of the pixel.

You seem to be forgetting about the thickness of the beam.  No matter
what resolution is being displayed, the picture is made up of
*contiguous* pixels  -  there are no dimmer or, with lower and lower
resolutions, ever widening black bits, in between the pixels making up
the picture.  Right now my CRT is displaying a CCTV image of much
lower resolution than broadcast tv, but there are no acres of real
estate between the pixels.

So, given that all the 'physical' pixels (which merely for the sake of
accuracy I will remind that in this TV are actually phosphor stripes
being crossed by the beam) are in use, the mechanism is entirely
comparable to an LCD.

> Nevertheless, even if the resolution selected precisely matches the dot 
> pitch, visible pixels will be made from a blend of phosphor dots around 
> the desired position.  The image will therefore be slightly fuzzy.

I think any CRT designer, and if they're honest most of the pro-CRT
people here, would take issue with that.  The whole art of adjusting a
CRT seems generally to consist of minimising, preferably eliminating,
such effects.  If the resolution matched the dot pitch but the beam is
straying over a neighbouring pixel, then the wrong pixel is being
illuminated, which is a flaw to be corrected.

> In the case of an LCD or Plasma display, the pixel will be just that: 
> Precisely one will be lit.

You are contradicting yourself ...

On Sat, 23 Aug 2008 12:02:27 +0100, Andy Champ 
wrote:
> 
> Most CRTs are running in the native resolution of the broadcast 
> pictures.  Most LCDs aren't - they tend to have some silly native 
> resolution like 1440x900 (That's a Samsung LE19R 86BDX just because I 
> found it first in Google, not for any other reason).

... and my measurements of my own LCD shows similar, so precisely one
pixel will not be lit.

> As you change the scan away from the native resolution of an LCD or 
> plasma, it too has to generate visible pixels from a blend of real 
> pixels.  Imagine a scan rate of 2/3 the native lines, and a horizontal 
> edge black above, white below, ending halfway through a native line. 
> The pixels which are on the desired edge will be at 50% gray;  the edge 
> will be fuzzy.  I suggest you take a drawing package, and draw a 
> slightly sloping line, then look at it in different resolutions.  You'll 
> see some damn odd effects.

All the above also applies to a CRT ...

> In the case of a CRT, that edge is *already* fuzzy, and doesn't look any 
> worse as you turn the resolution down;

The mischievous in me is loving this, you know.  Almost every post you
write, you contradict something said by other pro-CRT people before
you, usually something that's taken as gospel.  Now you're telling
them that CRTs are better because they're out of focus!

> Now back to my original point:  Why on earth would I want to buy an LCD 
> or Plasma that was 900 lines to display 576/720/1080 pictures?

Because IME an LCD tv of about 867 horizontal pixels will display
broadcast tv of 720 hp much better than a CRT tv of 516 hp, and an LCD
monitor of 1280 hp displays both 1280 & 1024 hp better than a CRT
monitor of 1664 hp will display 1280 hp, so an LCD looks a fair bet
over a CRT.

Java Jive wrote:

> Because IME an LCD tv of about 867 horizontal pixels will display
> broadcast tv of 720 hp much better than a CRT tv of 516 hp, and an LCD
> monitor of 1280 hp displays both 1280 & 1024 hp better than a CRT
> monitor of 1664 hp will display 1280 hp, so an LCD looks a fair bet
> over a CRT.

LOL, So essentially your final conclusion appears to be that apples taste 
different from oranges

...but I'd be interested to know how you get an SD tv to display a 720hp 
broadcast source as implied above...

The CRT TV argument of course goes that a SD CRT displaying SD pictures 
looks better than a 720 or 1080 HD TV using the same SD source.

I believe that for most people any TV displaying in its native resolution 
will look better than when using upscaling of the source.  Because of this 
many LCDs do indeed look worse than a standard CRT because the source is 
normally SD.

cheers

David

On Tue, 26 Aug 2008 10:12:17 +0100, DM  wrote:

> Java Jive wrote:
> 
> ...but I'd be interested to know how you get an SD tv to display a 720hp 
> broadcast source as implied above...

If it's done properly, it should be done by mathematical interpolation
as described in a post further up.

But I'm not trying to claim that having the number of pixels raised by
Andy is necessarily a good thing, I was merely answering his question.

> The CRT TV argument of course goes that a SD CRT displaying SD pictures 
> looks better than a 720 or 1080 HD TV using the same SD source.

Well, the arguments that Andy and Rod were putting forward, if
extrapolated to include HD, would be exactly opposite to that.  

Andy has been claiming that the mechanism of a CRT interpolates the
picture, and therefore, although he didn't use this phrase I think
it's a fair summary of what he claims, they scale better to different
resolutions.

Rod seems to have been claiming that since CRTs have no native
resolution, they scale better.

I think both are wrong, but find myself agreeing more with Andy,
because he does at least accept that CRTs have a native or physical
resolution.

Having had a chance to sleep on the implications of my earlier
measurements, I think one has to distinguish between three situations
where:
1)	Physical resolution exactly matches picture resolution
2)	It is less
3)	It is greater

Each of these cases is actually quite different.

2) 	Let's deal with this first, because it's non-controversial.  My
CRT has less physical hp than SD, so now we know why it is not as good
as my LCD, and the reason for all that detail loss I painstakingly
logged in the CRT vs LCD investigation.  Basically, regardless of
technology, any display built like this is cr*p.

1)	This could easily be achieved on an LCD, but would be very
difficult to achieve in practice on a CRT, but assuming it could
actually be done, then this is a one-to-one situation, and the picture
should be displayed without loss.  However, this thread has shown that
with TVs, this situation doesn't seem to occur with either display
type, even though it is commonplace on LCD monitors.

3a)	If mathematical interpolation is used to achieve an optimum
compromise rendition, mathematically speaking there is a functional
relationship and the situation is more like a less good 1)
b)	Otherwise, the physical picture elements are now sampling the
picture elements, so by the sampling rule both H'y & V'y the physical
pixels should be at least double the picture pixels for faithful
rendition.

Consequently, on a multi-resolution display, it could be argued that
the physical pixel resolution both H'y and V'y needs to be double the
maximum resolution it is meant to display.  This is not the situation
with any display of any technology that I know of, so we are totally
reliant on how well interpolation is done.

> I believe that for most people any TV displaying in its native resolution 
> will look better than when using upscaling of the source.  Because of this 
> many LCDs do indeed look worse than a standard CRT because the source is 
> normally SD.

I don't think you can have been following this thread very carefully -
one of the interesting things to come out of it is that there is no
real difference between SD on SD and upscaled SD on HD in this
respect, none appear to be displayed by an identical physical
resolution.

The effects you describe with upscaling I covered earlier, I suggest
you reread those posts.

Java Jive wrote:
> On Tue, 26 Aug 2008 10:12:17 +0100, DM  wrote:
> 
>> Java Jive wrote:
>>
>> ...but I'd be interested to know how you get an SD tv to display a 720hp 
>> broadcast source as implied above...
> 
> If it's done properly, it should be done by mathematical interpolation
> as described in a post further up.
> 

Sorry I did not make the question completely unambiguous.  I'm familiar with 
interpolation or in this case decimation algorithms.

I'm not aware of any hardware that actually does this though in this 
context.  i.e. I'm not aware of any SD crt that attempts to take in a HD 
signal and decimates to its own lower resolution, which makes your example a 
little superfluous.


> But I'm not trying to claim that having the number of pixels raised by
> Andy is necessarily a good thing, I was merely answering his question.
> 
>> The CRT TV argument of course goes that a SD CRT displaying SD pictures 
>> looks better than a 720 or 1080 HD TV using the same SD source.
> 
> Well, the arguments that Andy and Rod were putting forward, if
> extrapolated to include HD, would be exactly opposite to that.  
> 
> Andy has been claiming that the mechanism of a CRT interpolates the
> picture, and therefore, although he didn't use this phrase I think
> it's a fair summary of what he claims, they scale better to different
> resolutions.
> 
> Rod seems to have been claiming that since CRTs have no native
> resolution, they scale better.
> 
> I think both are wrong, but find myself agreeing more with Andy,
> because he does at least accept that CRTs have a native or physical
> resolution.
> 
> Having had a chance to sleep on the implications of my earlier
> measurements, I think one has to distinguish between three situations
> where:
> 1)	Physical resolution exactly matches picture resolution
> 2)	It is less
> 3)	It is greater
> 

yes that would be reasonable.

> Each of these cases is actually quite different.
> 
> 2) 	Let's deal with this first, because it's non-controversial.  My
> CRT has less physical hp than SD, so now we know why it is not as good
> as my LCD, and the reason for all that detail loss I painstakingly
> logged in the CRT vs LCD investigation.  Basically, regardless of
> technology, any display built like this is cr*p.
> 
> 1)	This could easily be achieved on an LCD, but would be very
> difficult to achieve in practice on a CRT, but assuming it could


hmmm why do you think that this is so hard with a normal run of the mill old 
fashion crt. what makes you think an old crt tv does not display the 
transmitted resolution


> actually be done, then this is a one-to-one situation, and the picture
> should be displayed without loss.  However, this thread has shown that
> with TVs, this situation doesn't seem to occur with either display
> type, even though it is commonplace on LCD monitors.
> 
> 3a)	If mathematical interpolation is used to achieve an optimum
> compromise rendition, mathematically speaking there is a functional
> relationship and the situation is more like a less good 1)
> b)	Otherwise, the physical picture elements are now sampling the
> picture elements, so by the sampling rule both H'y & V'y the physical
> pixels should be at least double the picture pixels for faithful
> rendition.
> 
> Consequently, on a multi-resolution display, it could be argued that
> the physical pixel resolution both H'y and V'y needs to be double the
> maximum resolution it is meant to display.  This is not the situation
> with any display of any technology that I know of, so we are totally
> reliant on how well interpolation is done.


> 
>> I believe that for most people any TV displaying in its native resolution 
>> will look better than when using upscaling of the source.  Because of this 
>> many LCDs do indeed look worse than a standard CRT because the source is 
>> normally SD.
> 
> I don't think you can have been following this thread very carefully -
> one of the interesting things to come out of it is that there is no
> real difference between SD on SD and upscaled SD on HD in this
> respect, none appear to be displayed by an identical physical
> resolution.

I'd be very interested in the references (real ones not newsgroup waffle) 
that supports your statement here. Upscaling algorithms for simple static 
pictures can be complex and produce considerably different results never 
mind for tv.  Do you really understand what goes on when you take a low res 
image and increase the number of pixels.

Personally I've seen acceptable results and really pretty poor pictures with 
SD on HD, which leads me to believe that there isn't a standard 
implementation.  OF course some algorithms will be better suited to certain 
image types so this could also be a factor


> 
> The effects you describe with upscaling I covered earlier, I suggest
> you reread those posts.

So many of these posts are just such complete waffle with little information 
  that I hardly think it would be beneficial to read any of them again. 
Entertaining the first time around but I wouldn't go back to it.

cheers

David

On Tue, 26 Aug 2008 17:31:42 +0100, DM  wrote:
>
> So many of these posts are just such complete waffle with little information 
>   that I hardly think it would be beneficial to read any of them again. 
> Entertaining the first time around but I wouldn't go back to it.

If you are too lazy to take in the salient points of what previously
has been said, then you have no right to expect others to make any
effort to reply to you.

In article , Java Jive 
wrote:
> Rod seems to have been claiming that since CRTs have no native
> resolution, they scale better.
> 
> I think both are wrong, but find myself agreeing more with Andy,
> because he does at least accept that CRTs have a native or physical
> resolution.

Well at least I now understand what we disagree about, though it still 
makes no sense to me. A flat panel display has a definite number of 
very separate light emitting elements as part of its physical 
structure, but a CRT can have any number of scanning lines (within 
reason) depending on the signal it is fed with, and this is not part of 
its physical structure. The shadowmask and phosphor dot pattern of a 
CRT has much less effect than you seem to think.

Rod.
-- 
Virtual Access V6.3 free usenet/email software from 
http://sourceforge.net/projects/virtual-access/

OK, pardon the slow response I didn't get to this newsgroup until too 
late at night for reasoned comments.  And don't take me as a blind CRT 
lover (IYSWIM) - see below.

Java Jive wrote:
> 
> Let's look at some actual specs ...
> 	http://tinyurl.com/6dkn88
> ... standing in for ...
> http://h18000.www1.hp.com/products/quickspecs/north_america_01-1999/50143_na.PDF
> ... the first 19", your choice, is the V90.
> 
> 19", or as you say, 18" visible equates to about 366x274mm visible at
> 4:3. The horizontal dot-pitch is 0.22mm, so there are about 1664
> 'native' pixels to display 1280 screen resolution ones. 366/1280 is
> 0.29, so you seem to be saying that 0.29 - 0.22 = 0.07mm is sufficient
> for the separation for the holes in the shadow mask, which whatever
> its actual value will also be the gap between the physical pixels, AND
> the overlap of which you speak?  
> 
> Supposing for the sake of peace I accept that, it's not actually what
> you said, and my remarks were a reply to what you actually said, which
> gave me a completely different impression from that which you seem to
> have intended.
> 
>> The beams aren't going to track precisely on the phosphor dots, so a 
>> single colour pixel will probably be made of a bunch of dots with 
>> different illuminations, dimmer the further they are from the ideal 
>> centre of the pixel.
> 
> You seem to be forgetting about the thickness of the beam.  No matter
> what resolution is being displayed, the picture is made up of
> *contiguous* pixels  -  there are no dimmer or, with lower and lower
> resolutions, ever widening black bits, in between the pixels making up
> the picture.  Right now my CRT is displaying a CCTV image of much
> lower resolution than broadcast tv, but there are no acres of real
> estate between the pixels.
> 
> So, given that all the 'physical' pixels (which merely for the sake of
> accuracy I will remind that in this TV are actually phosphor stripes
> being crossed by the beam) are in use, the mechanism is entirely
> comparable to an LCD.
> 
>> Nevertheless, even if the resolution selected precisely matches the dot 
>> pitch, visible pixels will be made from a blend of phosphor dots around 
>> the desired position.  The image will therefore be slightly fuzzy.
> 
> I think any CRT designer, and if they're honest most of the pro-CRT
> people here, would take issue with that.  The whole art of adjusting a
> CRT seems generally to consist of minimising, preferably eliminating,
> such effects.  If the resolution matched the dot pitch but the beam is
> straying over a neighbouring pixel, then the wrong pixel is being
> illuminated, which is a flaw to be corrected.
> 
The most important point I'm trying to make is that the phosphor dots on 
a CRT are not a precise match to the pixels in the digital image you're 
trying to show.  I was surprised at how close the resolution of the 
phosphor dots to the actual pixels was;  but I don't believe that the 
beam scanning is precise enough to put a set pixel of source image onto 
a single phosphor dot.  If it was, we wouldn't need convergence controls!

>> In the case of an LCD or Plasma display, the pixel will be just that: 
>> Precisely one will be lit.
> 
> You are contradicting yourself ...

ah ... not on purpose...
> 
> On Sat, 23 Aug 2008 12:02:27 +0100, Andy Champ 
> wrote:
>> Most CRTs are running in the native resolution of the broadcast 
>> pictures.  Most LCDs aren't - they tend to have some silly native 
>> resolution like 1440x900 (That's a Samsung LE19R 86BDX just because I 
>> found it first in Google, not for any other reason).
> 
> ... and my measurements of my own LCD shows similar, so precisely one
> pixel will not be lit.

... If I run a CRT in a non-maximum resolution, the beam scanning will 
be at the resolution of the incoming picture.  The phosphor dots of 
course won't move;  but the spread of the beam (deliberate spread, if 
the CRT is smart enough to notice the lower input resolution) will 
result in an input pixel being shown as some blend of dots around the 
target.

This is cheap and easy;  it probably just falls out of the design.

With a plasma or LCD, as the resolution drops you have to do something 
drastic, probably involving lots of software running at high speed, to 
calculate the intermediate screen pixels.  At the very least this will 
involve a store of several picture lines that can be accessed 
simultaneously by the logic handling the input picture and the logic 
drawing on the screen.  And the two sets of logic will be running at 
different speeds.

> 
>> As you change the scan away from the native resolution of an LCD or 
>> plasma, it too has to generate visible pixels from a blend of real 
>> pixels.  Imagine a scan rate of 2/3 the native lines, and a horizontal 
>> edge black above, white below, ending halfway through a native line. 
>> The pixels which are on the desired edge will be at 50% gray;  the edge 
>> will be fuzzy.  I suggest you take a drawing package, and draw a 
>> slightly sloping line, then look at it in different resolutions.  You'll 
>> see some damn odd effects.
> 
> All the above also applies to a CRT ...
> 
>> In the case of a CRT, that edge is *already* fuzzy, and doesn't look any 
>> worse as you turn the resolution down;
> 
> The mischievous in me is loving this, you know.  Almost every post you
> write, you contradict something said by other pro-CRT people before
> you, usually something that's taken as gospel.  Now you're telling
> them that CRTs are better because they're out of focus!

I think a CRT is better in one specific way;  when it is running at a 
non-optimal resolution it will get much closer to its best picture than 
LCD/Plasma screens.

OTOH, if you are running at that native resolution, or an easy fraction 
of it, the LCD/Plasma will give a sharper picture.
> 
>> Now back to my original point:  Why on earth would I want to buy an LCD 
>> or Plasma that was 900 lines to display 576/720/1080 pictures?
> 
> Because IME an LCD tv of about 867 horizontal pixels will display
> broadcast tv of 720 hp much better than a CRT tv of 516 hp, and an LCD
> monitor of 1280 hp displays both 1280 & 1024 hp better than a CRT
> monitor of 1664 hp will display 1280 hp, so an LCD looks a fair bet
> over a CRT.

No argument that an 867 line LCD (where did you find that?) will do a 
better job than a 516 line CRT with a 720 line picture.

A 1280 line LCD vs a 1664-line CRT, with a 1024 picture?  I suspect the 
CRT, if set up correctly, will do a better job. On a 1280 picture, the 
LCD will be sharper.

I think a 720p native LCD or plasma will do a better job than a 768 or 
900 native unit.  I also suspect the 720 unit will do a pretty good job 
on a 1080 picture;  scaling 2:3 is a darn sight easier than scaling 
1080:900 (= 6:5) or 1080:768 (= 45:32) which are LCD resolutions I've 
seen recently.

I'm using a 1920x1200 LCD as I type this.  Far better than my previous 
CRTs.  Just for interest, I've dropped it back to 1680x1050.  Yuck.  It 
didn't stay like that for long.  I was recently given a laptop for work, 
which looked all fuzzy.  Guess what - not set to maximum resolution. 
This is an effect I have NEVER noticed on ANY CRT.

Andy

Java Jive wrote:
> On Tue, 26 Aug 2008 17:31:42 +0100, DM  wrote:
>> So many of these posts are just such complete waffle with little information 
>>   that I hardly think it would be beneficial to read any of them again. 
>> Entertaining the first time around but I wouldn't go back to it.
> 
> If you are too lazy to take in the salient points of what previously
> has been said, then you have no right to expect others to make any
> effort to reply to you.

You miss the point completely.

There are very few salient points in the vast majority of newsgroup posts 
and this thread is no different

There is a great deal of opinion and little fact.

While you suggest I reread something about upscaling and do not 
differentiate between interpolation and decimation was a little amusing.

cheers

David

On Tue, 26 Aug 2008 22:53:18 +0100, Roderick Stewart
 wrote:
> 
> A flat panel display has a definite number of 
> very separate light emitting elements as part of its physical 
> structure, but a CRT can have any number of scanning lines (within 
> reason) depending on the signal it is fed with, and this is not part of 
> its physical structure.

a)	But that's only over the vertical dimension of an aperture grille
CRT with striped phosphors.  Over its horizontal dimension, or on a
shadow mask CRT where the phosphor is arranged in triads, I can see no
essential difference between a CRT TV and an LCD TV in this respect,
except, perhaps, that LCDs seem to have a greater number of picture
elements, physical or 'native' pixels, call them what you will ...

I have just spent some time trying to find out the full specification
of a normal, commercially available CRT TV, including dot-pitch, but
despite searching over 10s of pages on each of a number of likely
terms, I've failed  -  if they give any worthwhile specifications at
all, which is quite rare with TVs but relatively commonplace with
monitors, when manufacturers are selling TVs they refer to Grade A or
B tubes, but when they are selling monitors then they refer to such
things as dot-pitch; I've never convincingly found both terms in the
same sentence.

Accordingly, I'll have to give two answers ...

If we are discussing monitors, it's definitely not my experience that
CRTs scale better to different resolutions.  Rather it is my
experience that as long as the resolution is 4:3, either type of
display will display it fine, but the further away from 4:3 the
resolution is, the worse it looks on any type of display, largely
because fonts look terrible.

If someone tells me different, I wonder if they've either got a bad
monitor, or are using the wrong refresh rate (both CRTs and LCDs when
supplied a wrong refresh rate often display a picture, but badly  -
this can be a common problem in Linux, where the default refresh rates
for, say, 1024x768 often look terrible, and this has happened to me
even on a 1024x768 LCD monitor, whereas correcting the refresh rate
has always cured the problem).

I, and doubtless others will, note the willingness of myself to
support the above view with photographic evidence and the reluctance
of others disagreeing with it, including yourself, to support their
claims, with, well with anything really.

As far as TVs are concerned.

I wanted to marry up the dimensions of a commercially available TV,
preferably 2 or 3, to dot-pitch because I wanted to know how generally
applicable the results I obtained for my CRT TV were/are.  Clearly,
any TV with fewer picture elements along one dimension than the
resolution it is built to display is not going to display the
resolution as faithfully as one that has the same number or more, as
it seems that LCDs have.  However, the dearth of hard information
means that we are left with the hard results of my own measurements on
my TVs, and some more hearsay ...

http://www.dansdata.com/gz029.htm

Accepting that it's just hearsay, and therefore not worth a point by
point rebuttal, I would nevertheless be interested in hearing your
comments on the following quotes:

1	"Non-high-definition ("Standard Definition", or SD) TVs have a
hard time resolving even 640 by 480 pixels. Some expensive SD CRT TVs
can manage a vaguely plausible 800 by 600."

2	"TV manufacturers seldom even quote dot pitches for their sets,
but you can bet on a mainstream $AU700/$US500-ish 68cm/27" set not
being significantly finer than 0.7mm.

If it's got a nice bright aperture grille tube, its effective dot
pitch is likely to be even worse. Sony ramble on about how their
current Trinitron aperture grille CRTs and rear projection TVs have
finer stripe pitch than the old ones had, but they still won't tell
you what it is.

The highest quality SDTV CRTs manage a dot pitch of around 0.4mm.

A 0.7mm-pitch 68cm TV has a 27 inch viewable diagonal, 1.5 times
bigger than the 18 inches of a "19 inch" monitor (TVs, unlike CRT
monitors, don't generally rip you off by one inch under their quoted
diagonal size), but 640 by 480 will still be the approximate limit of
its resolving power."

3	"Big nasty secret: There does not exist a colour CRT tube, whether
shadow mask or aperture grille, TV or computer monitor, that can
clearly display all 1920 horizontal pixels of full-res HDTV.

There are plenty of computer monitors that can accept input
resolutions of 1920 by whatever, or higher, but none of them have
enough phosphor units to actually clearly display them. TV tubes
aren't even in the race."

> The shadowmask and phosphor dot pattern of a 
> CRT has much less effect than you seem to think.

Mmmm ... See above!

There is a way you or others reading this may be able to help clarify
things a bit though, if anyone from the broadcasting trade can find
out what the terms 'A Grade' and 'B Grade' when referring to CRTs mean
in terms of actual specifications such as dot-pitch, that would be
useful to know.

In the meantime, my own measurements of my own TVs are still the only
hard facts we've got to go on, and they support the idea that LCDs are
better.

On Wed, 27 Aug 2008 22:56:54 +0100, Andy Champ 
wrote:
> 
> The most important point I'm trying to make is that the phosphor dots on 
> a CRT are not a precise match to the pixels in the digital image you're 
> trying to show.

Which by my measurements and the Samsung LE19R 86BDX specifications
supplied by none other than yourself is also true of the physical
pixels in LCD TVs.

> ... If I run a CRT in a non-maximum resolution, the beam scanning will 
> be at the resolution of the incoming picture.  The phosphor dots of 
> course won't move;  but the spread of the beam (deliberate spread, if 
> the CRT is smart enough to notice the lower input resolution) will 
> result in an input pixel being shown as some blend of dots around the 
> target.

Which by interpolation is potentially also true of LCDs.

> With a plasma or LCD, as the resolution drops you have to do something 
> drastic, probably involving lots of software running at high speed, to 
> calculate the intermediate screen pixels.  At the very least this will 
> involve a store of several picture lines that can be accessed 
> simultaneously by the logic handling the input picture and the logic 
> drawing on the screen.  And the two sets of logic will be running at 
> different speeds.

Yes, but this is just reiterating what I've already posted, and still
doesn't tell us anything from real knowledge about how it's actually
done in practice.  Also, you've contradicted yourself again ...

> You don't need floating point, you only need fixed point.  And IAC the 
> price of a graphics chip capable of this kind of work is *plummeting*. 
> It's one of the bits of the computer world that is exceeding Moore's 
> Law; price:performance is halving more often than every 18 months.
> 
> > I have no idea how broadcasters actually do it at the hardware level,
> > perhaps someone here can enlighten us, but I don't suppose their
> > equipment has FP coprocessors.
> 
> I do.  My son's new graphics card will do about 2 teraflops.  2 million 
> million floating point operations per second.  Surely they can manage that?

So in one post it's "drastic", and the other it's "Surely they can
manage that?"!!!!!

> I think a CRT is better in one specific way;  when it is running at a 
> non-optimal resolution it will get much closer to its best picture than 
> LCD/Plasma screens.

But the only hard *evidence* available, which comes from my own
photographic investigations, points to an opposite conclusion.

If you want to convince people, you have to find some actual evidence,
not just endlessly go round in circles repeating these same
unsubstantiated assertions, which is an especially dangerous habit for
you to indulge in, as you seem to have a knack of contradicting
yourself when doing it, which just results in you looking faintly
ridiculous.

Give us some facts, or give us a rest ...

In article , Java Jive wrote:
> > A flat panel display has a definite number of 
> > very separate light emitting elements as part of its physical 
> > structure, but a CRT can have any number of scanning lines (within 
> > reason) depending on the signal it is fed with, and this is not part of 
> > its physical structure.
> 
> a) But that's only over the vertical dimension of an aperture grille
> CRT with striped phosphors.  Over its horizontal dimension, or on a
> shadow mask CRT where the phosphor is arranged in triads, I can see no
> essential difference between a CRT TV and an LCD TV in this respect,
> except, perhaps, that LCDs seem to have a greater number of picture
> elements, physical or 'native' pixels, call them what you will ...

In a flat panel display, the light emitting elements are completely separate, in 
that they are separately addressable, so it is possible for one to be emitting 
light and the one immediately next to it to be dark, or at a different 
brightness level. When they are used for computer displays, showing a lot of 
electronically generated material consisting of a lot of straight lines, this 
probably happens quite a lot, as there will be many boundaries which require the 
brightness level on one side to be completely unaffected by the brightness on 
the other side. This is easily achieved with absolute perfection if the number 
of separate digital numerical values in the signal fed to the display is an 
exact match for the number of physical light emitting elements in it, so that 
each light emitter is controlled by one and only one numerical value and is not 
affected by the ones next to it.

(In a colour display, for "light emitting element" you might read "group of 
three light emitting elements", but the principle is the same).

In a CRT, the phosphor dots are not related to scanning lines at all. Ideally 
the phosphor dot structure in a CRT would be a lot finer than the electron beam, 
though in practice it is comparable with it because of maufacturing 
difficulties. Effectively the finite size of the electron beam performs a sort 
of interpolation as it passes over several phosphor dots at once, and brightens 
them  by varying amounts depending on the area of coverage. (The same sort of 
interpolation that would have to be performed electronically if an LCD is 
required to show a number of pixels which is different from its number of 
display elements). The scanning lines need not even align perfectly with the 
rows of dots, which is just as well because this would be very difficult to 
achieve. A shadowmask CRT would work just as well - perhaps better in some 
respects - if the dot pattern were not a pattern at all but completely random, 
because there would then be no Moire patterning, which is a kind of beat 
frequency pattern between the regular pattern of the scanning lines and the 
regular pattern of the phosphor dots.

The phosphor dots could be regarded as equivalent to grain size in film, which 
determines the maximum resolution the film can carry, but without each grain of 
emulsion being separately controllable, because typically the grain structure 
will be finer than the resolving power of the lens. And film grain is 
essentially random. We don't usually talk about film having a "native" grain 
structure at which it gives best results, just a maximum that can be gradually 
approached.

Rod.
-- 
Virtual Access V6.3 free usenet/email software from 
http://sourceforge.net/projects/virtual-access/

All this is either assertion stated for the umpteenth time as though
it were fact, or facts that I'm not disputing.

I note that you ducked the assertions by others that I wanted your
comment on.

I will assume henceforth that you have nothing new to add, and save
both our time by omitting reply.

On Fri, 29 Aug 2008 10:09:54 +0100, Roderick Stewart
 wrote:
> 
> In a flat panel display ... Rod.

Java Jive wrote:
> On Wed, 27 Aug 2008 22:56:54 +0100, Andy Champ 
> wrote:
>> The most important point I'm trying to make is that the phosphor dots on 
>> a CRT are not a precise match to the pixels in the digital image you're 
>> trying to show.
> 
> Which by my measurements and the Samsung LE19R 86BDX specifications
> supplied by none other than yourself is also true of the physical
> pixels in LCD TVs.

With an LCD or plasma running in its native resolution their is a 1:1 
match between the pixels coming out of the MPEG decoder and the little 
lights on the screen.  With a CRT it is only an approximate match.  I've 
enquired several times why I would want to buy a 900 line TV, and no-one 
has answered!

> 
>> ... If I run a CRT in a non-maximum resolution, the beam scanning will 
>> be at the resolution of the incoming picture.  The phosphor dots of 
>> course won't move;  but the spread of the beam (deliberate spread, if 
>> the CRT is smart enough to notice the lower input resolution) will 
>> result in an input pixel being shown as some blend of dots around the 
>> target.
> 
> Which by interpolation is potentially also true of LCDs.

See below.

>> With a plasma or LCD, as the resolution drops you have to do something 
>> drastic, probably involving lots of software running at high speed, to 
>> calculate the intermediate screen pixels.  At the very least this will 
>> involve a store of several picture lines that can be accessed 
>> simultaneously by the logic handling the input picture and the logic 
>> drawing on the screen.  And the two sets of logic will be running at 
>> different speeds.
> 
> Yes, but this is just reiterating what I've already posted, and still
> doesn't tell us anything from real knowledge about how it's actually
> done in practice.  Also, you've contradicted yourself again ...
> 

Keep looking down...

>> You don't need floating point, you only need fixed point.  And IAC the 
>> price of a graphics chip capable of this kind of work is *plummeting*. 
>> It's one of the bits of the computer world that is exceeding Moore's 
>> Law; price:performance is halving more often than every 18 months.
>>
>>> I have no idea how broadcasters actually do it at the hardware level,
>>> perhaps someone here can enlighten us, but I don't suppose their
>>> equipment has FP coprocessors.
>> I do.  My son's new graphics card will do about 2 teraflops.  2 million 
>> million floating point operations per second.  Surely they can manage that?
> 
> So in one post it's "drastic", and the other it's "Surely they can
> manage that?"!!!!!

Yes.  Surely a broadcaster can manage something the equivalent of the 
hardware that is currently on the PC market for 300 quid.  I'm pretty 
sure a TV manufacturer *cannot do it inside the box*.  300 quid is a 
small fraction of the costs of the BBC, but a large fraction of the cost 
of a TV.

When I play a composite signal through the LCD I have in front of me it 
looks pretty bad.  The screen struggles to upscale the PAL picture to 
the screen resolution.

When I play a PAL image through my computer (DVD) it looks a lot better. 
  This just might be something to do with the HW in my PC having a lot 
more horsepower than the screen.

I can't feed it an HD image, my PC isn't fast enough.

> 
>> I think a CRT is better in one specific way;  when it is running at a 
>> non-optimal resolution it will get much closer to its best picture than 
>> LCD/Plasma screens.
> 
> But the only hard *evidence* available, which comes from my own
> photographic investigations, points to an opposite conclusion.
> 
> If you want to convince people, you have to find some actual evidence,
> not just endlessly go round in circles repeating these same
> unsubstantiated assertions, which is an especially dangerous habit for
> you to indulge in, as you seem to have a knack of contradicting
> yourself when doing it, which just results in you looking faintly
> ridiculous.
> 
> Give us some facts, or give us a rest ...

Your photos don't point to a generic conclusion.  The experiment we need 
involves two computer-quality screens of each technology, comparable 
size, and two different resolutions.  I have no decent quality CRT 
available any more - only standard def. TVs.  None of the LCDs I have 
available are anywhere near this low resolution.

I put it to you that a CRT will not lose as much quality as an LCD when 
fed a sub-optimal scan rate.  I know this from many years of using 
displays of both technologies.  However, as this has been almost 
entirely with computers I am happy for you to provide evidence proving 
the opposite.

My choice of technology, for picture quality, would be:

(1) LCD running at its native resolution, or an easy fraction of it.
(2) CRT running at any resolution
(3) LCD running at resolution unrelated to the source material.


Andy.

On Fri, 29 Aug 2008 21:55:40 +0100, Andy Champ 
wrote:

[skip more of same]

> Your photos don't point to a generic conclusion.  The experiment we need 
> involves two computer-quality

No, my experiments were primarily about TVs!

> screens of each technology, comparable 
> size

Which they were!

> and two different resolutions.

As SD TVs they can only display the one, unless perhaps I could obtain
an NTSC signal, but I'm not sure that I can.

I admit on the web-page that 'care should be taken generalising from
this result'.  But I also go on to point out that it is at least a
result.  In fact, it's the only one we've got.  The correct scientific
approach is therefore cautious acceptance.

[skip more of same]