Bug 689773 - image generation does not produce same result as other programs
Summary: image generation does not produce same result as other programs
Status: NOTIFIED FIXED
Alias: None
Product: Ghostscript
Classification: Unclassified
Component: PDF Interpreter (show other bugs)
Version: 8.63
Hardware: PC Linux
: P2 normal
Assignee: Alex Cherepanov
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2008-03-28 05:37 UTC by Simon B
Modified: 2008-12-19 08:31 UTC (History)
0 users

See Also:
Customer:
Word Size: ---


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Simon B 2008-03-28 05:37:34 UTC
I'm not sure if this is a bug from your point of view, but here's the report..

If I view a pdf with, say, evince the pdf loads then a few seconds later the
quality of the pdf improves, particularly the text and some of the images. It's
like evince is doing some kind of optimisation for displaying the pdf.

With ghostscript, when I convert a pdf to an image, I get the same image as
evince first gets, but not the optimised version, i.e. not with the clearer text
and better quality images.
Comment 1 Ray Johnston 2008-03-28 07:58:16 UTC
Please attach a PDF that exhibits the described behavior.

When I compare Ghostscript to Adobe Acrobat 7, the screen view is the same. Note
that Acrobat defaults to 'smooth text', so you have to use -dTextAlphaBits=4 to
see the same quality from Ghostscript (we default to non-antialiased since it
is slightly quicker). Perhaps 'evince' is displaying non-antialiased, then it
repaints with anti-alias.

Without a sample file we can't do anything more on this issue.
Comment 2 Simon B 2008-03-28 09:13:04 UTC
Okay. I'll upload one on Monday.
Comment 3 Simon B 2008-03-30 23:08:31 UTC
Created attachment 3905 [details]
pdf with problem
Comment 4 Simon B 2008-03-30 23:09:57 UTC
Note also title "Office World" is/seems to be gone.

I am using:
gs -dPARANOIDSAFER -dBATCH -dNOPAUSE -dNOPROMPT -dMaxBitmap=500000000
-dAlignToPixels=0 -dGridFitTT=0 "-sDEVICE=png16m"  -dTextAlphaBits=4
-dGraphicsAlphaBits=4 "-r96x96" -dUseCIEColor  "-sOutputFile=OUT.png"
29-03-OE-2M1-OE_292M1036-01-4C.pdf
Comment 5 Ray Johnston 2008-03-31 09:22:24 UTC
I see the title with Ghostscript (rev 8609), but there is one substantial
difference compared to Acrobat and one minor difference.

Alone the bottom edge there with Ghostscript there is a blue bar. With Adobe
Acrobat 7 the bottom is white and the word "WERBUNG" in the lower left. This
is painted over by Ghostscript in a BX EX block in steps 9287 through 9301
(PDFSTEP values).

The minor difference is the white border painted around the image "I6" is wider
with Acrobat than with Ghostscript. This is painted by the "S" operator in step
15621.

I've attached screen shots that I get with each. My gs command line is:
  gs  -dDOINTERPOLATE -dTextAlphaBits=4 -dGraphicsAlphaBits=4 -dUseCIEColor -r82
bug_689773.pdf

(I chose r82 to match what I see on a full page view with Acrobat).
Comment 6 Ray Johnston 2008-03-31 09:23:39 UTC
Created attachment 3906 [details]
AR7.png
Comment 7 Ray Johnston 2008-03-31 09:24:47 UTC
Created attachment 3907 [details]
gs_r8609.png
Comment 8 Simon B 2008-04-02 00:47:30 UTC
Thanks for marking the attachments as private, but unfortunately bugzilla
doesn't let me see them! No matter though, I see what you are saying.

Using your command I also see no "office world" at the top, the background for
the page is blue instead of white, a missing background image on the page, and a
lower quality photo. Will attach screenshot.
Comment 9 Simon B 2008-04-02 00:48:02 UTC
Created attachment 3915 [details]
comparison (evince vs. gs)
Comment 10 Simon B 2008-04-02 00:49:22 UTC
I am using gs 8.62 with this patch:
http://ghostscript.com/pipermail/gs-cvs/2008-March/008183.html
Comment 11 Ray Johnston 2008-04-02 08:56:12 UTC
I did confirm that Ghostscript 8.60 displays this the same as Adobe (and evince).

I get different results with rev 8609 than was posted by the submitter as the
'comparison' image, but 8609 is still "wrong" (but closer to correct).

Initial assignment to Alex is reasonable, but this may not be in the PDF interp.
If not he will reassign.

Raising the priority since this is a regression from 8.60.
Comment 12 Simon B 2008-04-07 05:25:37 UTC
Is it worth changing status?
Comment 13 Alex Cherepanov 2008-11-25 21:51:13 UTC
All problems, mentioned in the bug report appear to be fixed in the current
development code (rev. 9239).