Dave,
I'd like to add some to your discussion.
In traditional halftoning (round, square, elliptical,...) dots are indeed
variable in size and not in spacing, hence the name AM screening. Because
our software packages generally work with 8 bits (so does PostScript,
which is the native language of most imagesetters), 256 levels of gray
can be represented, this means your supercell should be 16 by 16 dots.
This gives you a very simple formula:
linescreen (lpi) x 16 = imagesetter resolution (dpi) to acheive these 256
graylevels.
i.e. 150 lpi x 16 = 2400 dpi
The obtainable 256 graylevels are theoretical. There's always some loss,
but you should not try to output film at any lower resolution than the
one calculated like this, especially for digital negatives since our
drastic curves already throw out a lot of the available information. I
suspect we're limited to about 100 graylevels if we do everything
according to the book.
For your scan, the resolution should be 1.4 times the screen ruling at
least for best results. i.e. 200 lpi, scan resolution = 280 dpi at image
output size. So for scanning, the formula is as follows: Scanresolution =
Screenruling x enlargement factor x 1.4
Stochastic screening: here it's the spacing between the dots that varies,
hence the name FM screening. Take into account that with FM screening the
dot size is equally important. The problem is that very small dots don't
copy well (mostly due to film undercutting; ligth scattering through the
film still images somewhat under the black dots). On the other hand, if
you could use very high frequencies, you could get a continuous effect in
printing: testing does the job!
Some processes don't need or even don't like very fine or certain shapes
of dots! It could mean a great difference to generate round dots vs.
square dots, even at the same frequencies.
As for me, I've settled for AM screening (Pt), but at 600 lpi; imaged at
10160 dpi 1 film takes about 1 hour, making this a very interesting but
economically not very sane process if it's done just for that purpose.
Goed luck,
Carl Fransman
______________________
SALTO
Printing & Lithography
Groenplaats 1
3621 Rekem-Lanaken
BELGIUM
tel +32 89 71 87 90
fax +32 89 73 18 30
salto@salto.be
______________________
This archive was generated by hypermail 2b29 : Tue Jan 11 2000 - 12:10:48