PDF from ghostscript appears blank in acrobat reader

Question

I have (3 pretty large) PDF files which I obtain using ghostscript from a single source pdf generated with xelatex (both under ubuntu linux). The 3 files select different regions from the source, like so

gs -o out.pdf -sDEVICE=pdfwrite -dDEVICEWIDTHPOINTS=26562 -dDEVICEHEIGHTPOINTS=4252 -dFIXEDMEDIA -dCompatibilityLevel=1.4 -dEmbedAllFonts=true -c "<</PageOffset [-7447 0]>> setpagedevice" -f in.pdf

Only the PageOffset and the dimensions differ for the three files.

The last two files, which do not start at the origin, however do not display correctly in adobe acrobat reader DC (tested under windows). They do display in all other PDF-viewers I tried, but I need to submit them to a print service which does not seem to be able to deal with the problem.

What can I try to fix the files (I tried pdf fix utilities under Ubuntu, but that did not work)?

Answer 1

This problem can circumvented by reducing the media size of the document. Adobe Acrobat seems not to be able to display documents if the dimensions are too large (while other viewers might be).

Try running

gs -o out_a.pdf -sDEVICE=pdfwrite -dDEVICEWIDTHPOINTS=[half-width] -dDEVICEHEIGHTPOINTS=[half-height] -dPDFFitPage -dCompatibilityLevel=1.4 -dEmbedAllFonts=true -dDownsampleColorImages=false -dDownsampleGrayImages=false -dDownsampleMonoImages=false -f out.pdf

with [half-height] and [half-width] half the previous values and out.pdf the output of your initial ghostscript command.

If it still does not work, you may want to try even smaller values.

Answer 2

I just had this problem where Acrobat shows a blank PDF accompanied with the error "Insufficient data for an image". The PDFs in question were produced by libtiff. This solution will only work for libtiff-produced PDFs.

In this case libtiff used a 24-bit colour profile but said the image was only 1-bit (monochrome), and both Acrobat and Ghostscript preferred the colour profile's bit depth (so couldn't open the file), while other PDF readers preferred the file's overridden bit depth (and would show the file just fine).

In my case the fix was to change the PDF file to use a different colour profile, which meant doing a binary replace of this text:

/ColorSpace [/ICCBased 7 0 R]

with this text:

/ColorSpace /DeviceGray %0 R]

The % starts a comment ignoring the rest of the data on that line, which is required so the offsets of subsequent data does not change. As I am on Linux, I made the change with sed:

sed -i 's|/ColorSpace \[/ICCBased 7 0 R\]|/ColorSpace /DeviceGray %0 R]|' file.pdf

This fix was suggested by the fine folks over at the Ghostscript project and got all my broken PDFs working again.