I've been working on getting crunchjpgs.pl to work (going so far as to write my own version) and I hit the same stumbling block with crunchjpgs.pl and my own code. What I'm seeing is a halt at the top of parseObj with "Expected object open tag" followed by an unintelligible string that includes 'PDF1.4'...I believe it gets called via dereference which is called during call to 'clean' that occurs during 'cleanoutput' execution. When I change parseObj to return undef instead of die-ing, I get a PDF that is appropriately smaller than the original, but without images (acroread actually hangs, but kpdf starts but shows no images and reports: *Error (361875): XObject 'Im4' is wrong type*).

My input PDF is produced by Inkscape.

Any suggestions, ideas?

Thanks,

--Fred