[pdftex] pdftex & file size...

Thierry Bouche Thierry.Bouche at ujf-grenoble.fr
Mon Nov 12 13:54:09 CET 2001

Concernant « Re: [pdftex] pdftex & file size... », Hans Hagen écrit : «
» At 11:59 AM 11/12/2001 +0100, Thierry Bouche wrote:
» >Hi,
» >
» >I face the following problem, which was always there, but gets quite
» >annoying in this peculiar case...
» >
» >I have a PDF file (a small (150 p.) science book with 80+ images,
» >most of them vector arts, and hyperlinks).
» >
» >i make the PDFs for images with distiller, they use the same fonts as
» >the main text, from (AI, fontless) EPS of 50 K, I get PDFs of 300 K
» >(imbedding or not the fonts makes essentially no difference, and
» >obviously pdftex does the good job of factorizing the fonts in both
» >cases)
» why is the pdf so much bigger than the eps if there are no fonts in there?

I don't know... In fact, it is the second edition of something I did 3 
years ago. To be safer, I redid the PDFs, they were more or less of
the same size as the EPS at that time (acro 3, I guess) and now much
bigger (acro 4 or 5 is idem). the fact is that the pdftex's PDF is
more or less the sum of all these included PDFs (minus duplicated
fonts, + incredibly few bytes for the whole text...), I wondered whether 
this could have to do with the 'preserve editability' switch in
illsutrator, but I didn't generate the PDF direct as AI export, I made 
'AI EPS', then distill.

» >the output of pdftex is 23 MB
» >if I distill this to a PDF at 1200 dpi, I get a 2.5 MB file.
» what role does the 1200 play here if the graphics are vector?

It's the option set, should make no difference as the fonts are PS &
the graphics vectors (only 2 of them are scans). Many of these images
have patterns, that were relatively badly handled with AI 7, and may
not be vector.

» >Is there out there any solution (even a commercial plug-in) that would
» >allow the kind of optimisation that distiller does, _without_ killing
» >the hyperthings?
» what happens if you optimize the file with acrobat (exchange)?

grows by 100 K. I used PDF consultant, who only found some unused name 
destinations, this saves as much as 50 K...

I can't put an example now, because the thing is in another place, but 
I could tonight. I suspect there are lots of objects duplicated in
each PDF, that distiller strips off. But, as a general fact, pdftex
makes huge files compared to distilled ones. I wonder whether
something could be done about it, at least by post processing.


More information about the pdftex mailing list