Created attachment 7946 [details] input.pdf - sample PDF file The sample file has a null XObject resource, similar to /Resources <</XObject <</Obj1 null >> >> This is a correct PDF but Ghostscript expects an actual resource dictionary and fails with PS errors. The problem is easy to fix and a patch will be posted shortly.
A patch that checks XObjects for null has been committed as a rev. 63f2476edccc4860312737b61e86c929c340cb72 Unfortunately, there are rendering problems related to either to JBIG2Decode or masked images.
Created attachment 7950 [details] Simplified sample file. This sample file has only a masked image, null XObject has been removed. The file looks OK in AR9 but blank on gs.
Created attachment 7958 [details] A little more simplified file. This file has a masked image only. Further simplification is not possible. Both the base image and the mask are rendered OK by Ghostscript when they are not combined into a masked image. The problem appears to be a regression. V. 9.02 renders the simplified file correctly, but v. 9.04 doesn't.
Created attachment 7979 [details] sample.tar.bz2 - PS sample This is simple PS program that exercises a Type 3 masked image. Surprisingly, it doesn't work on any version of Ghostscript.
My previous comment is not correct. v. 9.01 and 9.02 render the file correctly. v. 8.64..9.00 and v 9.04..HEAD render the file incorrectly
The strange re-appearance pattern of this bug is caused by the choice of the default device in different versions. x11 device works in all recent versions, x11alpha doesn't. I'm reassigning this bug to Michael as an x11alpha problem.
The bug is still reproducible.
Here is the command line from the customer's original email: Gswin32.exe -q -dBATCH -dNOPAUSE -r144x144 -dTextAlphaBits=4 -dGraphicAlphaBits=4 -dFirstPage=1 -dLastPage=1 -dPrinted=false -sDEVICE=png16m -dPDFIgnoreAnnots=true -I -sOutputFile="C:\%d.png" "C:\BeforeImportTopdfDocs.pdf"
With that command line, the file looks fine to me. What exactly is the issue?
Assigning back to support for clarification as to the exact issue.
I've sent the customer an output png file to confirm that it is correct and the patch referenced in comment #1. Assuming the customer is satisfied I'll close this bug and reopen one for the remaining x11alpha issue.
The customer confirms that the patch from comment #1 fixes the problem. I'm closing this bug and will open a new one for the x11alpha issue.