[Scribus] Setting linescreen when printing pdf file

LinuxLingam linuxlingam
Wed Jul 21 13:50:53 CEST 2004

On Wed, 2004-07-21 at 08:14, Louis Desjardins wrote:
> [snip] A file ripped at one single high 
> resolution is lots heavier than its counterpart at dual resolution. 
> The reason why we can treat images at this low 300 dpi resolution is 
> that images usually come at 300 dpi. So 300 dpi in, 300 dpi out. 

um,.... not quite correct. i have coined a term i have used to explain
this point in my workshops and seminars. 'pixels per inch' PPI. images
are *never* scanned at dpi DPI. Dots is what you get out of a printer.
photos are scanned in PIXELS.

so we have three units, each representing something DIFFERENT.
PPI (pixels per inch)
DPI (dots per inch)
LPI (lines per inch)

PPI is for photos scanned, DPI for vector resolution, and LPI is really

so if a photo is at 300PPI, it has to output at 150LPI, then ONE
halftone cell gets four PIXELS with which to compute its 'abstract color
data'. once the ripper does that, it then REPRESENTS that halftone color
using CMYK.
given that colors in real life are always a continuum {never discrete
values} and other issues of rgb to cmyk, this 4 pixels to a halftone
cell is calleld a quality factor.
you could output a 150PPI image to a 150LPI, it would work, but the
colors and the overall image may not be represented well enough.

you could even work with a quality factor of 1.5

2 is ideal.

therefore, please note that usually a 300ppi image is slight overkill
for a 133lpi. i have used 266ppi scans and got superb 133lpi output.

hope this clarifies the whole issue.

and yes, i am going to start writing a more thorough book on this and
related topix, the moment i finish some pending projects in the next two
or three months. god willing.

> As 
> for vectors, (type, vector graphics, logos, etc.), in order to 
> achieve a crisp job, we need a high resolution ripping at, say, 2400 
> dpi (it could be 1200 and still be OK). 

vectors are to do with LINEART. lineart needs the highest resolution
possible for output. so, in pre-press artworks, we have photos as
halftone images [grayscale, duotones, or cmyk-color, or cmyk+spot, or
hi-fi] and we have lineart, which is for text, for graphics and shapes.


> Linescreen is a necessity, it is not evil. ALL printed material is 
> screened. A 150-lines job can be of very high quality. Even newspaper 
> don't look so bad.

linescreening, or halftoning, is necessary because of its economy. this
morning, pick up your newspaper, look at all the pages, read thru it,
then look at its cover price. even assuming the impact of advertising,
it is a marvel how inexpensive it is to produce a newspaper, that is,
how little the per copy actually costs. thanks to halftoning. if we did
not have halftoning, people would have to probably paint every picture
by hand or mechanically, using tubs of colored inks....

the discussion on print-ink formulae [green at c80% y80%, etc] versus
c100%y100%, has to do with GCR [gray component replacement] and UCR
[under color removal]. google for more.

insight: colors in an original image are represented [quite poorly] in
printing processes, never reproduced.  represented colors are dependent
on the media, paper, and print-process. to sort of 'calibrate' for this,
GCR and UCR are important steps. have a special relationship to
something called 'dot gain' [google].

and am eagerly waiting for someone to start a discussion on spot
functions [dot shape: ellipse, round, diamond, custom...] frequency [x
LPI] and angle [minimize overlaps of translucent inks of cmyk]....


More information about the scribus mailing list