🚀 go-pugleaf

In <1565@eagle.ukc.ac.uk> sc7@ukc.ac.uk (S.Chan) writes:

>I have printed some decompressed images on hardcopy and they don't
>look quite the same as that is shown on a monitor, specifically
>the effects of noise, sharpness and blocking effects. (a 2400dpi
>printer was used to print the 256-level grey scale images)

>I suppose there is a great difference between viewing pictures on
>screen and on paper. Does anyone have any experience in resolving
>this kind of problem? I'm also interested to read materials
>that have been published on this problem.

>Thanks.

>Chan

The problem  you are experiencing are due to the  imperfect science of
gamut  mapping.  This  is  a hot topic in computer graphics  and falls
under color theory.  There are lots of  papers to  read on  this.  For
some introductory  reading you could look at the Foley-Vandam graphics
book.

Monitors use the RGB color space and are capable of displaying a whole
range of colors within a certain 'gamut'.  Printers use CMY(k) and can
also display a  whole  range of colors from a different looking gamut.
As you might  guess, it's easy to display  red  on a monitor, but it's
not so easy to display red on printer using Cyan, Magenta and Yellow.

Anyway, this a little of what I recall from my computer graphics
class.  As far as I can remember, color theory doesn't involve much of
compression research so maybe ask the folks in comp.graphics.*

Dan
--
-- -- -- --- --- --- ---- ----  ___  ---- ----- ----- ----- ------ ------ -----
Dan Mack (mack@cray.com)       / | \   	  "We're not hitchiking anymore... ..
Sun/SGI Resource Group         \/_\/           We're riding!"  -Commander Hoek
Cray Research, Eagan MN   ----- ----- ----- ---- ---- ---- --- --- --- -- -- --

In article <mack.715549997@deci> mack@deci.cs.umn.edu (Dan Mack) writes:
>In <1565@eagle.ukc.ac.uk> sc7@ukc.ac.uk (S.Chan) writes:
>
>>I have printed some decompressed images on hardcopy and they don't
>>look quite the same as that is shown on a monitor, specifically
>>the effects of noise, sharpness and blocking effects. (a 2400dpi
>>printer was used to print the 256-level grey scale images)
>
>The problem  you are experiencing are due to the  imperfect science of
>gamut  mapping.  This  is  a hot topic in computer graphics  and falls
>under color theory.  There are lots of  papers to  read on  this.  For
>some introductory  reading you could look at the Foley-Vandam graphics
>book.

>[remainder of discussion on color deleted]

Notice that the original poster was referring to gray-scale images, so
all this color discussion is irrelevant.  I've noticed the same difference
between displaying and printing *any* gray-scale images (not just lossy-
compressed).  I think the effect springs from the fact that a typical
monitor inherently does a small amount of lowpass filtering, so that the
pixel boundaries are less noticeable on a monitor.  There may be other
causes as well, but this is probably one of them.



--
Stan Reeves
Auburn University, Department of Electrical Engineering, Auburn, AL  36849
INTERNET: sjreeves@eng.auburn.edu

In article <sjreeves.920903165024@fourier.ee.eng.auburn.edu> sjreeves@eng.auburn.edu (Stan Reeves) writes:
>Notice that the original poster was referring to gray-scale images, so
>all this color discussion is irrelevant.  I've noticed the same difference
>between displaying and printing *any* gray-scale images (not just lossy-
>compressed).  I think the effect springs from the fact that a typical
>monitor inherently does a small amount of lowpass filtering, so that the
>pixel boundaries are less noticeable on a monitor.  There may be other
>causes as well, but this is probably one of them.

The major problem is that the monitor is able to display gray levels,
while the printer does only black or white. It makes a marvelous attempt
at pretending to do gray by varying the numbers and positions of the
black dots it puts on the paper, but it is still only an approximation.

The standard PostScript printer has 300 dpi, but only reproduces 53
lines per inch and 33 levels of gray, by default. If you want full
resolution from the printer (300 dpi, you get only two shades: black and
white. There are several other "halftone screens" available, but the
best number of grays is, as I recall, 39.

For more information, look in this month's Radio Electronic's, the
Hardware Hacker column.

A 2400 dpi printer should do much better than this, but it is still only
writing an approximation to gray.

In article <mack.715549997@deci> mack@deci.cs.umn.edu (Dan Mack)
writes:

>In <1565@eagle.ukc.ac.uk> sc7@ukc.ac.uk (S.Chan) writes:

>>I have printed some decompressed images on hardcopy and they don't
>>look quite the same as that is shown on a monitor, specifically
>>the effects of noise, sharpness and blocking effects. (a 2400dpi
>>printer was used to print the 256-level grey scale images)
                                ^^^^^^^^^^^^^^^^^^^^
>>I suppose there is a great difference between viewing pictures on
>>screen and on paper.

>The problem  you are experiencing are due to the  imperfect science of
>gamut  mapping.

[summary of gamut mapping deleted]

While what Dan writes is all true, it is not the problem in this
particul;ar case as the images are greyscale.

The problem is that the printer is not greyscale, and while 2400 dpi
is a lot of resolution, you have to trade some for halftoning to
simulate greyscales. If the matrix used is regular, you will get
patterning which the eye is rather too good at picking up !

You could try messing with the halftone screen frequency or the screen
angle.

Or you could do a randomised error-diffused dither yourself, and print
that (the image will get bigger, though)

--
Chris Lilley
---------------------------------------------------------------------------
Technical Author, ITTI Computer Graphics and Visualisation Training Project
Computer Graphics Unit, Manchester Computing Centre, Manchester, UK
Internet:   lilley@cgu.mcc.ac.uk        Janet:   lilley@uk.ac.mcc.cgu
Voice:      +44 (0)61 275 6095          Fax:     +44 (0)61 275 6040
---------------------------------------------------------------------------