dsmith@applga.aa.cad.slb.com (J. Daniel Smith) (10/02/90)
I'm trying to print 1 pixel wide lines on a 300dpi PS printer.
Following the techniques on page 62-63 of _Real World PostScript_,
I have the code shown below.
The problem is that the top line of the box is thicker than the other
sides. I tried printing this file on two different Apple PS printers,
and the result was the same on both of them. I also tried different
values for "setlinewidth" (0.24, 0.12 and 0.003). And I tried putting
the SnapToPixel before the "rmoveto". None of this worked! There
must be a way to do this properly (i.e. *not* doing everything in
1/300 inch units).
Thanks for any and all help.
Dan
P.S. Internet mail isn't working properly right now; use either
"uunet!sharkey!applga!dsmith" or "smithda@buster.cps.msu.edu" or just
post a followup article.
=========================================================================
J. Daniel Smith Internet: dsmith@applga.aa.cad.slb.com
Schlumberger CAD/CAM BITNET: smithdan@msuegr
Ann Arbor, Michigan Usenet: uunet!sharkey!applga!dsmith
SINet: AAACA1::SMITH
The two most common things in the universe are hydrogen and stupidity,
but not necessarily in that order.
=========================================================================
---------
%!PS-Adobe-2.0
%%EndComments
/bdf {bind def} bind def
/inch {72 mul} bdf
%
% Logo sizes
/boxy {.25 inch} bdf % box start this far up from division name
/boxlength {7.5 inch} bdf % length of entire logo box
/logoheight {.1875 inch} bdf
/SnapToPixel % snap X,Y coord to device pixels
{ transform
round .25 add
exch
round .25 add
exch
itransform
} bdf
/drawboxes
{gsave
.12 setlinewidth % 1/600 inch hairlines
newpath % long, narrow box
0 boxy SnapToPixel moveto
boxlength 0 rlineto
0 logoheight rlineto
boxlength neg 0 rlineto
closepath
stroke
grestore
} bdf
%%EndProlog
%%BeginSetup
0.5 inch 1 inch SnapToPixel translate % 0.5 inch margins
%%EndSetup
drawboxes
showpage
%%Trailerlambert@spectrum.cs.unsw.oz.au (Tim Lambert) (10/03/90)
>>>>> On 1 Oct 90 22:21:16 GMT, dsmith@applga.aa.cad.slb.com (J. Daniel Smith) said: > I'm trying to print 1 pixel wide lines on a 300dpi PS printer. Why not say "0 setlinewidth"? Tim
woody@chinacat.Unicom.COM (Woody Baker @ Eagle Signal) (10/04/90)
In article <LAMBERT.90Oct3110022@nankeen.spectrum.cs.unsw.oz.au>, lambert@spectrum.cs.unsw.oz.au (Tim Lambert) writes: > >>>>> On 1 Oct 90 22:21:16 GMT, dsmith@applga.aa.cad.slb.com (J. Daniel Smith) said: > > > I'm trying to print 1 pixel wide lines on a 300dpi PS printer. > > Why not say "0 setlinewidth"? Because as has been pointed out before, 0 setlinewidth is not guaranteed to produce a 1 pixel wide line. setting it to .12 is supposed to explicitly do that. A lot of times the problems with wide lines can be traced to rounding problems. As I understand it, you can round up or down to .25, 0 .5. Since any dot touched by a line gets filled in, in certain roungings, you will have problems. This seems to occur when you have rounded such that you are straddling the boundry between 2 ajacent rows of pixels. Some time back, the rules were posted to the net. I'm not sure just where they are. Cheers Woody
lambert@spectrum.cs.unsw.oz.au (Tim Lambert) (10/05/90)
>>>>> On 4 Oct 90 00:32:29 GMT, woody@chinacat.Unicom.COM (Woody Baker @ Eagle Signal) said: > In article <LAMBERT.90Oct3110022@nankeen.spectrum.cs.unsw.oz.au>, I wrote: >> >>>>> On 1 Oct 90 22:21:16 GMT, dsmith@applga.aa.cad.slb.com (J. Daniel Smith) said: >> > I'm trying to print 1 pixel wide lines on a 300dpi PS printer. >> Why not say "0 setlinewidth"? > Because as has been pointed out before, 0 setlinewidth is not guaranteed to > produce a 1 pixel wide line. setting it to .12 is supposed to explicitly > do that. Huh? The Red Book says that a line width of 0 produces a line one device pixel wide. If it doesn't then there is something wrong with the implementation of PostScript on your printer. I haven't tried it that much, but it has worked whenever I've tried it. You can draw a one pixel wide line easily and very quickly using Bresenham's algorithm so there aren't any excuses for an implementation that doesn't get it right. > A lot of times the problems with wide lines can be traced to > rounding problems. As I understand it, you can round up or down > to .25, 0 .5. Since any dot touched by a line gets filled in, in certain > roungings, you will have problems. This seems to occur when you have > rounded such that you are straddling the boundry between 2 ajacent > rows of pixels. Yes, I can see how this can happen. But if a line width of 0 gives you a line two pixels wide, I don't see how a line width of 0.12 could make it narrower. Tim
fryd@g.gp.cs.cmu.edu (Michael Fryd) (10/06/90)
> Newsgroups: comp.lang.postscript > Date: 1 Oct 90 22:21:16 GMT > Sender: dsmith@applga.aa.cad.slb.com (J. Daniel Smith) > Reply-To: smithda@buster.cps.msu.edu > > I'm trying to print 1 pixel wide lines on a 300dpi PS printer. > Following the techniques on page 62-63 of _Real World PostScript_, > I have the code shown below. > > The problem is that the top line of the box is thicker than the other > sides. I tried printing this file on two different Apple PS printers, > and the result was the same on both of them. I also tried different > values for "setlinewidth" (0.24, 0.12 and 0.003). And I tried putting > the SnapToPixel before the "rmoveto". None of this worked! There > must be a way to do this properly (i.e. *not* doing everything in > 1/300 inch units). The problem is that the endpoints for each line segment must be properly alligned with the device pixels. The Solution is to use SnapToPixel on each corner of the box. This requires a slight rewrite to the code because SnapToPixel only works with absolute coordinates, not relative moves. the solution is to change: 0 boxy SnapToPixel moveto boxlength 0 rlineto 0 logoheight rlineto boxlength neg 0 rlineto closepath to: 0 boxy SnapToPixel moveto boxlength boxy SnapToPixel lineto boxlength logoheight boxy add SnapToPixel lineto 0 logoheight boxy add SnapToPixel lineto closepath Michael Fryd President Voice: (412) 751-5557 MEFCO, Inc. Fax: (412) 751-8403 2401 Coulter Road Email: Michael.Fryd@CS.CMU.EDU McKeesport, PA 15131-4251