Avoid dots in white spaces with custom halftone screensets - ghostscript

Question

I am trying to set up custom angle/frequency halftone screensets with ghostscript and ran into a peculiar problem. The resulting output contains black dots where there should be none.
As a reproducer I found this code:
%!PS-Adobe-2.0
% Sample code to explore the different screen and spot functions
% Written by Bruce Barnett
% Inspired by Michael Thorne
% PostScript Language Journal Vol 1, Number 4
%
%
% define some abbreviations
/l /lineto load def
/m /moveto load def
/rl /rlineto load def
/rm /rmoveto load def
/sg /setgray load def
/sh /show load def
/slw /setlinewidth load def
/st /stroke load def
/tr /translate load def
% and some places to store some information
/str 300 string def % define a string
/ss 50 def % defines square size
/fountstring 256 string def
% print a fountain
/PrintFountain {
% create a string containing values from 0 to 255
0 1 255 {fountstring exch dup put } for
% scale a 1 by 1 image to the size that will spread across the page
% first number is the width, second the height
600 45 scale
% construct/transform the image
256 1 8 [256 0 0 1 0 0] {fountstring} image
} bind def
% Print a tinted box
/TintBox { %define a tinted box, size (ss by ss)
tint 100 div sg
newpath
0 0 m
ss 0 l
ss ss l
0 ss l
closepath fill
} def
/LabelBox { % define a procedure to label a box
10 ss 2 div m % move to (10, ss/2)
lettercolor sg % select color
tint 3 string cvs sh
(% grey) sh
} def
/NextLine {-1 ss mul 10 mul -1 ss mul tr} def % goto next line
/NextLoc {ss 0 tr} def % goto next location (or place for a square)
/PrintMatrix { % define a procedure to print a row of squares
% if at the end of the row, go to the next line
% else - go to the next location
/tint exch def
TintBox % draw the box
LabelBox % add the label
/count count 1 add def % increase the count by one
count 11 lt { % move to next spot
NextLoc
} {
NextLine /count 0 def
} ifelse
} def
% show details of layout (Halftone, etc.)
/ShowDetails {
/str 300 string def
100 750 m
(Spot Procedure Name= ) show
/SpotFunctionName load show % => freq angle
currentscreen % => freq angle proc
pop % ignore procedure
( Angle= ) show
str cvs show % => freq
( Frequency= ) show
str cvs show
} bind def
% A procedure to set the screen angle
% and remember the name of the function
% so we can print it
/ScreenSet { % set screen function
% ang freq /spot_function ScreenSet
dup str cvs /SpotFunctionName exch def
load setscreen
} bind def
/PrintPage {
% also remember time to print page
/time_start usertime def
ShowDetails % prints the halftone screen
gsave % save the graphic state
ss ss 10 mul tr
0 0 m
/count 0 def
/lettercolor 1 def % select white letters
0 1 10 {PrintMatrix} for
10 1 20 {PrintMatrix} for
20 1 30 {PrintMatrix} for
30 1 40 {PrintMatrix} for
40 1 50 {PrintMatrix} for
/lettercolor 0 def % change to black letters
50 1 60 {PrintMatrix} for
60 1 70 {PrintMatrix} for
70 1 80 {PrintMatrix} for
80 1 90 {PrintMatrix} for
90 1 100 {PrintMatrix} for
grestore %restore graphic state
gsave % save it again, for the fountain
PrintFountain
grestore % restore
0 sg
% now print the elapsed time
100 775 m (elapsed time (milliseconds) = ) show
usertime time_start sub str cvs show
showpage % print the page
} bind def
% Here are the different spot functions
% These are suggested by Adobe
/spot_round { % simple round
dup mul exch dup mul add 1 exch sub
} def
% Inverted Round
/spot_iround {
dup mul exch dup mul add 1 sub
} def
% Euclidean Composite
/spot_euclid { % default on many new PS printers
abs exch abs 2 copy add 1 gt {
1 sub dup mul exch 1 sub dup mul add 1 sub
} {
dup mul exch dup mul add 1 exch sub
} ifelse
} def
% Rhomboid
/spot_rhomboid { % Rhomboid
abs exch abs .8 mul add 2 div
} def
% Line
/spot_line {
exch pop abs 1 exch sub
} def
% Diamond
/spot_diamond {
abs exch abs 2 copy add .75 le
{
dup mul exch dup mul add 1 exch sub
} {
2 copy add 1.25 le {
.85 mul add 1 exch sub
} {
1 sub dup mul exch 1 sub dup mul add 1 sub
} ifelse
} ifelse
} def
% Inverted Elliptical
/spot_iellipt {
dup mul .9 mul exch dup mul add 1 sub
} def
/spot_rb { % another from the "Red Book"
180 mul cos exch 180 mul cos add 2 div
} def
/spot_line { % simple line
pop
} def
/spot_line2 { % simple line going the other way
exch pop
} def
% End of definitions, now to print
% Use 8 point Helvetica
8 /Helvetica-Bold findfont exch scalefont setfont
% each "PrintPage" prints one test page
% Print 5 test pages or different screens
% using the same spot function
53 45 /spot_euclid ScreenSet PrintPage
75 0 /spot_euclid ScreenSet PrintPage
83 56 /spot_euclid ScreenSet PrintPage
106 45 /spot_euclid ScreenSet PrintPage
150 0 /spot_euclid ScreenSet PrintPage
% Now print 10 other spot functions, same screen
53 45 /spot_round ScreenSet PrintPage
53 45 /spot_iround ScreenSet PrintPage
53 45 /spot_euclid ScreenSet PrintPage
53 45 /spot_rhomboid ScreenSet PrintPage
53 45 /spot_line ScreenSet PrintPage
53 45 /spot_diamond ScreenSet PrintPage
53 45 /spot_iellipt ScreenSet PrintPage
53 45 /spot_rb ScreenSet PrintPage
53 45 /spot_linea ScreenSet PrintPage
53 45 /spot_lineb ScreenSet PrintPage
% I think you get the idea.....
% end of file
with gs -r1440 -sOutputFile=test%d.tif -sDEVICE=tiffsep1 -f example1.gs I get in test5(black).tif (spot_euclid freq=150) following patterns in progression from white to (near) black:
Where are these dots coming from and how to avoid them? I noticed that the dots within the white are gone at 720 but I need 1440 for my application.
I already looked around and found people with similar problems but no solution. This comment assumes that the white is not clean but that does not explain why the non white parts contain those dots.
EDIT: Smaller reproducer
I noticed that this behavior is also reproducible with a smaller example and stock ghostscript spot function:
<< /PageSize [10 10] >> setpagedevice
0.75 0.5 0 0 setcmykcolor
newpath 0 0 moveto 10 0 lineto 10 10 lineto 0 10 lineto closepath fill
showpage
quit
Running this with gs -r300 -dDITHERPPI=20 -sOutputFile=test%d.tif -sDEVICE=tiffsep1 -f test.gs I get the following output tifs for CMYK channels:
Cyan:
Magenta:
Yellow:
Black:
And for 600 DPI I also get dots in the empty channels:
Cyan:
Magenta:
Yellow:
Black:
I used low dither PPI to make the problem more visible but even in default settings, the points are there.

Answer 1

As noted by #KenS this was a bug in ghostscript that has been resolved.