vac@cs.cmu.edu (Vincent Cate) (06/29/91)
I want 1 bit on my bitmap to end up as one dot on a 300 DPI
postscript printer, but my printer seems to still mess with
the bits.
I have a pbm file that I am to printing on a 300 DPI printer.
pbmtops defaults to a times 4 scaling (one bit in the pbm file
goes to a 4 by 4 group of dots on the printer), and I want
1 for 1, so I use "pgmtops -scale 0.25". This does get my
1800 by 3000 bit pbm file to come out to 6 inches by 10 inches,
like it should. However, there are all sorts of artifacts
as if the printer had to scale the image. The printer is a
DEC lps40.
What am I doing wrong? Is there another way to get a bitmap to
a postscript printer? pgmtopbm -cluster3 seems to make the
printer happier but it is not as good an algorithm for converting
from grayscale to a bitmap.
The data that I am printing starts out as a 600 by 1200 gif image
with 6 bits of gray scale. I have tried things like this:
giftoppm input.gif | ppmtopgm | pgmnorm | pnmenlarge 3 | pgmtopbm |
pnmcut 0 300 1800 3000 | pgmtops -scale 0.25
If there is a better way to do this please let me know.
Thanks,
-- Vincejansteen@cwi.nl (Jan van der Steen) (06/29/91)
vac@cs.cmu.edu (Vincent Cate) writes: >I want 1 bit on my bitmap to end up as one dot on a 300 DPI >postscript printer, but my printer seems to still mess with >the bits. If you want to address the pixels of the output device directly you can change the CTM with: matrix setmatrix % = [1 0 0 1 0 0] setmatrix This sets up a 1 to 1 mapping between the user and device space. At this point the device work space can be determined using: clippath initclip pathbbox Leaving the llx, lly, urx and ury of the output device on the stack. Another approach is to determine the resolution of the output device. A simple way to do so is using the "dtransform" operator: 72 72 dtransform % (1,1) inch in device space abs /resy exch def abs /resx exch def Using "resx" and "resy" you can scale the bitmap to assure a 1 to 1 mapping. >I have a pbm file that I am to printing on a 300 DPI printer. >pbmtops defaults to a times 4 scaling (one bit in the pbm file >goes to a 4 by 4 group of dots on the printer), and I want >1 for 1, so I use "pgmtops -scale 0.25". This does get my >1800 by 3000 bit pbm file to come out to 6 inches by 10 inches, >like it should. However, there are all sorts of artifacts >as if the printer had to scale the image. The printer is a >DEC lps40. I looked at the PostScript file as generated by the command "pbm2ps 1" and noticed some things which might cause (your) problems: %! PBM to PostScript 0.24 0.24 scale /picstr 32 string def 1142 1372.5 translate 416 405 scale 416 405 1 [ 416 0 0 -405 0 405 ] { currentfile picstr readhexstring pop } image ffffc01ffffffffffffffffffff007ffffffff803ffffffffc01ffffffffe00f [bytes deleted...] showpage The idea of the ".24 .24 scale" is to create a 1 to 1 mapping for a 300 dpi output device (.24 * 300 / 72 = 1). An obvious drawback of this approach is that it doesn't make sense for non-300dpi devices. A simple fix would be to determine the resolution first (see above) and then scale according to: 72 resx div 72 resy div scale In this case the line containing the "translate" should be changed to: 1142 resx mul 300 div 1372.5 resy mul 300 div translate The second reason why this prologue may not give the best results is that the user space is not "snapped" on the device space. This can be fixed by "snapping" before translating: 1142 resx mul 300 div 1372.5 resy mul 300 div snap translate where snap is defined as: /snap { % % x y snap --> x' y' % transform round exch round exch itransform } def These changes should assure that the bitmap is printed without any distortion on any output device. An obvious drawback of the applied changes is that the image will have different sizes on devices with a different resolution but that can't be helped. Jan van der Steen -- -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Jan van der Steen jansteen@cwi.nl Centre for Mathematics and Computer Science (CWI) Kruislaan 413, 1098 SJ Amsterdam, The Netherlands