[tex-k] dvips produces postscript which bombs out

Reinhard Kotucha reinhard.kotucha at web.de
Sat May 24 10:10:56 CEST 2003


>>>>> "Volker" == Volker Kuhlmann <hidden at paradise.net.nz> writes:

    > dvips(k) 5.92b kpathsea version 3.4.5 Copyright (C) 2001 Radical
    > Eye Software.

    > tetex 2.0.1, SuSE Linux 8.2

    > ghostsript 7.05.6

    > I am trying to produce eps files of bitmap images which contain
    > no background, i.e. are transparent (with a depth of 1 bit), for
    > inclusion in latex documents. The images should not obscure
    > those page areas in which the image is transparent. The eps can
    > be produced with ImageMagick 5.4.7 as convert img.png
    > eps3:img.eps. It uses eps level 3 language
    > constructs. ghostscript can handle this eps file without errors
    > (display it, convert to eps or pdf). latex (with the graphics
    > package) and dvips produce a ps file which causes ghostscript to
    > produce an error and stop processing the file.

AFAIK pdftex's png driver doesn't support transparency.  This is a
known bug.  I don't know whether it's pdftex's fault or a bug in
libpng.

Ghostscript (ps2pdf14) obviously does not evaluate the "%%BoundingBox"
comment and assumes letter paper.  That's why the the second logo in   
logo-pdf.tex is too small.  You could steel the BoundingBox from the
eps file you send to ps2pdf14.

ImageMagick's png->pdf conversion obviously doesn't support transparency,
(at least it does't work for me) but I must admit that I have a too
old version. 

My suggestion is to send a bug report to the ImageMagick developers.

Regards,
  Reinhard

-- 
----------------------------------------------------------------------------
Reinhard Kotucha			             Phone: +49-511-27060390
Marschnerstr. 25
D-30167 Hannover	                      mailto:reinhard.kotucha at web.de
----------------------------------------------------------------------------
Microsoft isn't the answer. Microsoft is the question, and the answer is NO.
----------------------------------------------------------------------------




More information about the tex-k mailing list