jimomura@lsuc.uucp (Jim Omura) (03/23/88)
The following is a couple of letters from Lou Steinberg
to me and a couple of postings of mine to BIX. I've said
some other things as well, but these are about the most
important. I suppose I could say that the best way to follow
this it to join BIX, but that's obvious. For those who don't
know I'm the BIX moderator of the 'canada' conference -- *NOT*
the 'graphic.disp' (graphic display) conference. So I receive
not direct benefit from people posting to that conference on
BIX. My interest is really just that -- interest. Unless
of course you join the 'canada' conference while you're on
BIX. Then I get money! Yay!
:-)
Again, I do not read 'comp.graphics' regularly so please
respond by Net mail.
Cheers! -- Jim O.
From uucp Sun Mar 6 09:30:33 1988
>From uucp Sun Mar 6 00:51:34 1988 remote from utzoo
>From harvard!rutgers!aramis.rutgers.edu!lou Sat Mar 5 10:04:28 1988 remote from linus
Received: by linus.research (3.2/4.7)
id AA25186; Sat, 5 Mar 88 09:27:00 EST
Posted-Date: Thu, 3 Mar 88 12:34:37 EST
Received: by harvard.harvard.edu; Sat, 5 Mar 88 04:44:58 EST
Received: by rutgers.edu (5.54/1.15)
id AA17023; Fri, 4 Mar 88 12:13:16 EST
Received: by aramis.rutgers.edu (5.54/1.15)
id AA19273; Thu, 3 Mar 88 16:21:18 EST
Received: by rutgers.edu (5.54/1.15)
id AA28005; Thu, 3 Mar 88 12:47:53 EST
Received: by bearcat.rutgers.edu (5.54/1.15)
id AA16864; Thu, 3 Mar 88 12:34:37 EST
Date: Thu, 3 Mar 88 12:34:37 EST
From: linus!harvard!rutgers!aramis.rutgers.edu!lou (Lou Steinberg)
Message-Id: <8803031734.AA16864@bearcat.rutgers.edu>
Newsgroups: comp.graphics
To: harvard!linus!utzoo!lsuc!jimomura
Subject: Re: Grey Scale Slope Content Index
In-Reply-To: USENET article <1988Feb29.223957.20454@lsuc.uucp>
In article <1988Feb29.223957.20454@lsuc.uucp> jimomura@lsuc.uucp (Jim Omura) writes:
> I don't read 'comp.graphics' regularly so please comment via
> Net mail.
>
> [...] With the advances in display technology we can now consider
> using digital technology in normal consumer photographic
> applications, in effect, replacing the traditional family photo
> albums with computers. The current Amigas and Atari ST's are
> right on the border. We are faced with the problem of evaluating
> equipment on for their ability to represent continuous tone
> images with visually satisfying detail.
> [and] goes on to propose a measure of this ability]
One major problem with your proposed measure of quality is that it
does not measure the ability to render large areas of smoothly
changing brightness. For instance, you say
> Taking the ST again, and using the lower resolution 320 *
>200 * 16 colors, we find that horizontally, we get 4 pixels in
>1/100th the width (3 and a bit of the 4th showing). Of the 8
>grey levels, therefore, only 4 are useful in this small space.
This does reflect in some way the ability to show fine detail, but it
implies that a system with the same spatial resolution and only 4 grey
levels would be no worse at displaying pictures. If you imagine a
picture that has a large area shading gradually from white to black,
you will see that those second 4 grey levels are a real help. Of
course you can dither among the smaller number of grey levels, but
that effectively costs you spatial resolution.
In fact, it is hard to see how you could get one single measure that
applies equally to pictures with fine detail or texture and those with
large areas of smooth shading. Perhaps what you want is a graph,
rather than a single number. The horizontal axis might measure
spatial resolution and the vertical axis might give the brightness
resolution for regions of that size. E.g. for the ST described above,
for the horizontal direction, for a spatial resolution of 1/320 we get
a brightness range of 8, while for a spatial resolution of 4/320 we
get (8-1)*4+1 = 29 levels. (For grey level all that matters is the
total brightness of the 4 pixels, i.e. the number of different totals
we can get by adding up the brightness of each of the 4 pixels, where
each pixel can be considered to be in the range 0-7. In general for p
pixels each of which can have one of b brightnesses, we can get
(b-1)*p+1 different totals. This assumes the palette for each pixel
is the same. Hmm.. interesting idea: if alternate pixels, say, used
different palettes we could get a better brightness resolution for all
spatial resolutions above 1 pixel, using only a few more bits of
memory (to hold the second palette).)
Of course, only certain regions of this graph would be of interest for
any given application, depending on the picture to be displayed and
also on the visual angle subtended per pixel. (There is no point
in displaying more resolution than the human visual system is capable
of perceiving, but that depends on the visual angle a region subtends
and not the fraction of the full picture it takes up.)
--
Lou Steinberg
uucp: {pretty much any major site}!rutgers!aramis.rutgers.edu!lou
arpa: lou@aramis.rutgers.edu
From uucp Wed Mar 9 21:29:01 1988
>From uucp Wed Mar 9 19:04:34 1988 remote from utzoo
>From harvard!rutgers!aramis.rutgers.edu!lou Wed Mar 9 12:17:08 1988 remote from linus
Received: by linus.research (3.2/4.7)
id AA07297; Wed, 9 Mar 88 11:20:59 EST
Posted-Date: Tue, 8 Mar 88 17:22:40 EST
Received: by harvard.harvard.edu; Tue, 8 Mar 88 22:55:01 EST
Received: by rutgers.edu (5.54/1.15)
id AA01221; Tue, 8 Mar 88 18:07:26 EST
Received: by bearcat.rutgers.edu (5.54/1.15)
id AA00460; Tue, 8 Mar 88 17:22:40 EST
Date: Tue, 8 Mar 88 17:22:40 EST
From: linus!harvard!rutgers!aramis.rutgers.edu!lou (Lou Steinberg)
Message-Id: <8803082222.AA00460@bearcat.rutgers.edu>
To: harvard!linus!utzoo!lsuc!jimomura
In-Reply-To: <8803080917.AA22512@linus.research> (jimomura@lsuc.uucp)
Subject: Re: Grey Scale Slope Content Index
Reply-To: harvard!rutgers!aramis.rutgers.edu!lou
Cc: rutgers.edu!lou@aramis
You say:
Actually,
the whole point is that once you decide the "pseudo-pixel" size
for dithering based on the finest detail that you need, you can
theoretically re-dither on large spaces anyway, but if you
chose to base your dither on the large gradual grey, there's
nothing you can do about the fine detail except fake it with
"edging" algorythms.
Modern dithering algorithms (my own in particular) are adaptive: they
can give sharp spatial resolution in one part of the picture and smooth
shading (i.e. good brightness resolution) in another part of the picture.
They do not depend on any "pseudo-pixel".
I'm not sure if that answers your point. My feeling
is that you *could* use the general theory behind my
index system against the whole width or height of an image.
The only differences should be that 1. you get a greater
differentiation between systems at the low end.
Yes, you do. My point is that for certain classes of pictures and assumptions
about viewing distance, etc, that differentiation is meaningful
2. you
had a heck of a time trying to calculate the dither
permutations
Look at my message again - I give the formula. It's not that hard to
figure out.
and 3. all the index numbers become larger.
Yes, but since you should only compare the large-area number for one display
with the large-area number for another display, and not with any small-area
numbers, this does not really matter.
==========
graphic.disp/disc.bench #139, from jimomura, 1385 chars, Tue Mar 22 20:51:54 1988
Comment to 135. More refs to 135.
----------
I still haven't actually done any calculations based on Lou's
formula, but looking at it, I don't expect that my method will
produce any singificantly different results. In general, his
formula is faster. It *may* result in distinctions that my
method will tend to blur. But the blurring in my method was
deliberate. I worry sometimes when statistics imply accuracy
beyond their capability and thus distinctions among practical
equivalents. But there's no question that systems that would
show up as better using my approach will still show up as better
in Lou's approach. Also the proportion of the improvement looks
to be to be about right.
I think Lou may be right about the problem of fine detail
indexing. It may not be possible to integrate it with this
particular grey scale index directly. I think that for now,
I'll calculate "resolution" as cycles black to white across
the dimension being evaluated and keep it as a separate figure.
In most cases that'll be 1/2 the pixel resolution, but not
necessarily in cases where a direct transition from black to
white or the reverse is not allowed. I don't know of any
system currently having such a restriction, but it's certainly
possible.
I think that right now, what we need is to get some
money and hire a survey company to do some testing to see
whether the index corresponds to real human judgement.
==========
graphic.disp/disc.bench #140, from jimomura, 859 chars, Tue Mar 22 21:07:34 1988
Comment to 135.
----------
Some calculations:
Lou's equation is:
(( brightness.levels - 1) * pixels) + 1
For the Atari ST with 320 res. and 8 levels that works out to
(( 8 - 1 ) * 320 + 1
which is 2241.
For the Amiga at 320 res. and 16 levels it's
(( 16 - 1) * 320 + 1
which is 4801.
For the Color Computer 3 with 320 res. and 4 grey levels it's
(( 4 - 1 ) * 320 + 1
which is 961.
For the Atari at 640 res. and 4 levels it's
(( 4 - 1 ) * 640 + 1
which is 1921.
For the Color Computer 3 with 640 res. and 4 grey levels it's
Hmm. Same thing.
Note that The 2 Atari calculations show a slight distinction
between the 2 modes which did not show up in my original method.
It would be interesting to see if people would agree to this
ranking if they saw the pictures.
Also, we haven't gotten into color yet. But that's going
to be a pain. Uh, I mean a challenge. :-)
--
Jim Omura, 2A King George's Drive, Toronto, (416) 652-3880
ihnp4!utzoo!lsuc!jimomura
Byte Information eXchange: jimomuralou@aramis.rutgers.edu (Lou Steinberg) (03/24/88)
In article <1988Mar22.213059.17809@lsuc.uucp> jimomura@lsuc.uucp (Jim Omura) writes: > The following is a couple of letters from Lou Steinberg > to me and a couple of postings of mine to BIX. > [regarding the a discussion we've been having about calculating an > index, based on pixels and brightness levels/pixel, for how good a > display is at showing pictures] I'm not sure, but I think you may be missing my main point: no SINGLE number can do the job. Given two displays, say 256 x 4 bits vs 512 x 2 bits, one may be better for a picture with large areas of smoothly changing brightness and subtle contrasts while the other may be better for pictures with lots of fine detail. No single number can capture something like this. Compare this, for instance, with the way we describe speakers for good stereo systems: we show a graph that shows how the speaker responds over a range of sound frequencies, since one speaker may be better at high frequencies and another at low ones. Similarly, I was suggesting a graph of how well a display handles different spatial frequencies, i.e. different sizes of detail. Whether you use a single measure based on pixels across the whole screen or across 1/100 of the screen, or whatever, that one number must be misleading for certain classes of pictures and certain pairs of displays. -- Lou Steinberg uucp: {pretty much any major site}!rutgers!aramis.rutgers.edu!lou arpa: lou@aramis.rutgers.edu