On 8/29/13 8:08 PM, Steve Litt wrote:
On Thu, 29 Aug 2013 18:51:24 -0600
Ken Springer <snowsh...@q.com> wrote:
On 8/29/13 4:47 PM, Steve Litt wrote:
On Thu, 29 Aug 2013 10:19:34 -0400
Richard Heck <rgh...@lyx.org> wrote:
On 08/29/2013 03:59 AM, Liviu Andronic wrote:
On Thu, Aug 29, 2013 at 3:28 AM, Ken Springer <snowsh...@q.com>
wrote:
Much like the speakers in a sound system, it occurs to me the
actual quality of the printed output from a LyX document will
depend on the quality of the printer being used.
Right or wrong? If wrong, why?
Well, kind of. Of course the printing quality will depend on the
quality of the printer and the paper that you use. But once
exported to PDF, the typesetting quality of your document should
be rock-solid, whichever printer you use.
Obviously, a low resolution printer will give worse printed output.
This is especially true since the fonts used are (usually) vector
fonts.
That said, my two books /Frege's Theorem/ and /Reading Frege's
_Grundgesetze_/ were both printed, by the publisher, from a PDF I
provided. And they look great, if I do say so myself. I assume they
used very good printers!
Richard
Sort of on topic: After becoming dissatisfied with the 26 minute
print time, on my new Brother MFC-8810dw 40ppm printer, of my book
"Manager's Guide to Technical Troubleshooting
(http://www.troubleshooters.com/bookstore/mg.htm), I set the print
resolution down from 1200x1200dpi to 600x600dpi (11 minutes,
yeah!), my admittedly old eyes could find no appreciable difference
in the print, except for an almost imperceptable lightening at
600dpi. But the images were another matter: They looked better at
600 because 1200 showcased the mistakes and pixellations of the
artwork itself.
What would have happened if you applied stochastic screening to the
images? I'm assuming that would be retained in a PDF file.
Deeeuuuuud!
You're getting so tweak here I had to look up stochastic screening on
Wikipedia.
The graphics involved were:
1) A photo I took of a wrench
2) Various diagrams I did in Dia, Inkscape and Gimp.
I had to look up Dia, and thanks for that. I've added the downloading
of the Mac and Windows versions to my to do list.
Since I don't believe LaTeX can natively handle .svg, a lot of this
depends on the conversion LyX uses (which of course you can configure).
Keep in mind also that filesize goes way up with resolution, and that
makes it costly to email or download PDFs, and taxes printers in terms
of making the engine stop while loading the next big image.
.svg is a vector graphics file format. Stochastic screening is for
bitmapped files. It works best on lower resolution printers. It's been
years since I've use stochastic screening, so I don't know how it would
look on a 1200 dpi laser, for instance. But on a 300 dpi laser, you
would not believe how "real" a printed a photograph can look.
I've got a 20 year old computer in the back room with a desktop
publishing program on it that has a stochastic screening module.
I've known about stochastic screening for all of five minutes now
(http://en.wikipedia.org/wiki/Stochastic_screening) so let me dazzle
you with my opinions...
Any laserprinter I ever used used, as far as I know, a certain number
of dots per inch, each dot being a certain shade (I use
monochrome/grayscale) or color (if you want each page to cost you a
dime or a quarter or whatever). The shades are made from dot size,
with "dots" being a whole bunch of dots from the dpi measurement. As far
as I know, they cannot vary placement of their dots, only shade or
color via size (AM). So even if the PDF looked great on the screen, I'd
imagine the printer would have limited ability to reproduce the (FM)
stochastic screening.
You are correct on the dpi part, and contemporary printers can emulate
dpi output resolutions lower than what they are capable of.
However, you're wrong on each dot being a certain shade. The dot
produced by a B&W laser is either black or white. White meaning no
toner is applied to the paper. You can't change the color/shade of the
toner.
Where people go wrong is making the assumption that the dot produced by
the printer is a 1:1 ratio with the dot being printed. I.E., one
printer dot for each dot/pixel in the photo.
In reality, a number of printer dots are used to create the single
dot/pixel in a bitmapped graphic file. This is where the line screen
frequency, or LPI, comes into play.
A square grid of printed dots is used to create a single pixel from the
bitmapped graphic. Unless things have changed, the maximum number of
possible grey shades is 256. (Actually, it's 257, but that's not a
multiple of 2.) So the best possible greyscale picture that you can
print has a maximum of 256 shades of grey.
Example...
For ease of this post, let's say you use a 3X3 square of printer dots to
create the single pixel dot in the graphic. And, you want 50% grey.
For a single row of pixels from the file, the printer would print
something like the following:
xXx xXx xXx xXx xXx xXx xXx xXx xXx xXx xXx xXx xXx xXx xXx xXx
XxX XxX XxX XxX XxX XxX XxX XxX XxX XxX XxX XxX XxX XxX XxX XxX
xXx xXx xXx xXx xXx xXx xXx xXx xXx xXx xXx xXx xXx xXx xXx xXx
In the example above, the lower case "x" represents a place on the
printed line where the printer deposits no toner. The upper case "X" is
where toner is applied. That is as close to 50% grey you can have with
this example. You've just printed 16 50% grey pixels in a row from the
graphic using 144 printer dots.
I've added extra spaces to the typed line above to make the 3X3 square
more visible to the eye. In an actual print, those spaces would not exist.
Hopefully, you're using a monospaced font to read this. Otherwise, the
above example will really look horrible. Ideally, the em space in the
font is square, also.
You'll notice that each 3X3 square is identical in where the toner is
applied. On those old 300 dpi printers with ancient drivers, you can
see this pattern, if your eyes are good enough. Or use a magnifying glass.
In stochastic screening, the placement of the 4 printer dots varies,
possibly something like this:
xXx XxX xxX XXX xxx
XxX xXx XXx xxx XXX
xxX Xxx xXx xXx xXx
You'll notice each 3X3 square has printer dots in the different
locations, but still only 4 printer dots total for each 3X3 square. So
the eye sees 50% grey, but the apparent quality of the finished result
is way, way better than in the first example.
Indeed, the Wikipedia page talks of stochastic screening mainly in
terms of printing from plates (which I presume assumes print runs of at
least 100), and not from laser printers.
The results from laser printers really is unbelievable, given the lower
resolution. I've got a 20+ year old system in the back room that has a
300 dpi printer attached, and using stochastic screening I can print a
photograph with it that beats any print you can do with a 600 dpi laser
from the same time period that doesn't use a stochastic screen. In it's
day, there was no DTP software and hardware solution that could touch it
when it came to speed and quality of output. Nothing.
When doing stochastic screening, the ideal is to screen at the same dpi
as the final printing device. Next best is an even multiple. I.E.
screen at 300 dpi for printing on a 600 dpi printer. You also have to
decide on the finished physical size of the graphic before you start.
I used to use the regular graphic in all the drafts. When I was
satisfied with everything, then I applied the stochastic screen.
I just searched Inscape+"stochastic screening" and got a bunch of
useless stuff including an anti-Obama site (whaaaaat, how'd Google do
that?). Then I did the same thing for LaTeX, nothing ontopic for us.
Same thing with "Computer monitors". One site said most inkjet printers
use stochastic screening.
I suspect all home printers and laser printers now have some kind of
stochastic screening routines in their printer drivers. But, I've not
tested the idea.
I don't think stochastic screening will come into your life unless
you're using a very unusual printer, or long-run print at a print
house. It won't help me, AFAIK, because today's monitors can't use it.
Question... Who cares whether the monitor can use it? You're
interested in quality printed output, not monitor output. And at the
right (or wrong) zoom level, a stochastically screened image will look
like crap.
If filesize is no object, just go 1200dpi, and make sure your .svg
to .eps conversions are high resolution. If filesize *is* an issue, you
appear to know a lot more than I about how to minimize pixelization,
moire, and all that stuff.
I don't remember if I ever tried stochastically screening a vector
graphic. But my mind says it may turn out like crap. But I truly don't
know that for sure.
--
Ken
Mac OS X 10.8.4
Firefox 23.0
Thunderbird 17.0.8
LibreOffice 4.1.04
