Bug 690928 - zero-width clip paths rendered differently from gs
Summary: zero-width clip paths rendered differently from gs
Status: CONFIRMED
Alias: None
Product: GhostPCL
Classification: Unclassified
Component: PCL raster (show other bugs)
Version: unspecified
Hardware: PC Linux
: P4 normal
Assignee: Hin-Tak Leung
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2009-11-17 18:22 UTC by Hin-Tak Leung
Modified: 2011-02-06 00:45 UTC (History)
2 users (show)

See Also:
Customer:
Word Size: ---


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Hin-Tak Leung 2009-11-17 18:22:27 UTC
Page 3 of /tests_private/pdf/PDFIA1.7_SUBSET/CATX0988.pdf rendered badly through
pxl then raster - compare how the pdf looks (which still preserves some details
in the tilted images, but is all gone in the png):

gs -dFirstPage=3 -sDEVICE=pxlcolor -o /tmp/a.pxl CATX0988.pdf

pspcl6 -r600 -sDEVICE=png16m -sOutputFile=/tmp/a.png /tmp/a.pxl

pspcl6 -sDEVICE=pdfwrite -sOutputFile=/tmp/a.pdf /tmp/a.pxl

It appears that the author of the pdf uses a lot of zero-width rectangles as
clip paths over the images. These are rendered with a minimal(1-pixel?) width by
gs; but conversion through pxlcolor then ghostpdl to ppm, the zero-width
rectangles are rendered as zero-width. It is a difference between how gs and
ghostpdl treats zero-width clip paths.

(See also background info from Bug 690868 and Bug 690912).

There is a similiar bug about zero-width stroke in bug 690877
Comment 1 Henry Stiles 2009-11-19 07:50:36 UTC
Reassigning path and fill problems to Robin Watts.
Comment 2 Henry Stiles 2009-11-19 10:26:06 UTC
Actually and xl issue and shouldn't have been reassigned to Robin.
Comment 3 Hin-Tak Leung 2009-11-22 18:44:22 UTC
gs5.98-dragon.pdf tests_private/comparefiles is another file having this problem
with lots of zero-width clips.
Comment 4 Henry Stiles 2010-05-12 21:17:50 UTC
(In reply to comment #0)
> Page 3 of /tests_private/pdf/PDFIA1.7_SUBSET/CATX0988.pdf rendered badly through
> pxl then raster - compare how the pdf looks (which still preserves some details
> in the tilted images, but is all gone in the png):
> 
> gs -dFirstPage=3 -sDEVICE=pxlcolor -o /tmp/a.pxl CATX0988.pdf
> 
> pspcl6 -r600 -sDEVICE=png16m -sOutputFile=/tmp/a.png /tmp/a.pxl
> 

The Artifex rendering of "a.pxl" fairly well matches the HP rendering.  There are horizontal scanlines artifacts in the tilted images in both rendering.  How well this corresponds to the original PDF I didn't investigate, but this isn't a pcl problem.