optimizing large PDFs

[sarah_pnix]

I work for an academic publisher and I'm trying to produce PDF "ebooks" of large technical textbooks. For example, the print book may be 700 pages with images (some vector-based, some pixel-based) on roughly 80% of the pages. I've become pretty familiar with the different Acrobat optimization settings and can get the PDFs down to the 20-40MB range, which I consider pretty reasonable for all that content. The problem is we're having some performance issues. The PDFs work pretty well in Adobe Digital Editions—some page turns are a little slow but not bad. However in the ebook reader app on the iPad, some of them are REALLY slow and even completely freeze up the app. Some large ones are fine, so it's not just a matter of file size alone.

Could be an app issue, could be an iPad issue, could be a PDF issue. Or all of the above. For my part, I'm trying to figure out what all I can do to make the PDF as streamlined as possible. Of course, the images still need to be high enough quality to see (we're not talking about pretty photos here, but technical drawings, charts, maps, etc.). There's a bookmark for each chapter start (average of 20-30 chapters), but I tried to not to muck it up with too many bookmarks/hyperlinks.

Any thoughts? I've pored over the line items in the "Audit Space Usage" window, but can't seem to find the silver bullet there. Has anyone encountered anything besides image resolution that can make the PDF more efficient?

I appreciate any ideas,
Sarah

[DSpider]

If the "Audit Space Usage" window says graphics take up the most space, then you should vectorize as much content as possible and simplify those vectors - meaning you should layer them instead of leaving them edge-to-edge, reduce nr. of nods where too many do the same job, etc.

So instead of having rasterized content (not "pixel-based"), try to incorporate as much vectorization as possible. They scale really well when zoomed (useful for technical plans) and layering them, reducing nods, etc., can drastically improve performance, especially for underpowered devices such as e-readers (though I wouldn't call the iPad "underpowered"... try GoodReader instead of iBooks and see how that goes).

I would recommend Vector Magic and editing the result with Inkscape (or Adobe Illustrator, whatever).

Don't bother vectorizing photos or something with depth perception in them, unless you know what you're doing. Because it could actually increase the filesize while reducing quality. Instead, you could reduce the number of colours (by indexing them to 256, maybe less depending on the content - for something predominantly blue you could go as low as 8 colours and still look gorgeous), use dot matrix to achieve a sort of "printing dots" effect, which will compress really well using TIFF. In fact, I did this quite recently (check my signiture for the note on printing dots). I turned a grayscale image of about 880 KB, indexed it to 1 bit (black and white) and kind of interpolated it, using lines instead of dots:



This is just a sample of a corner, of an item. But once the PDF is resized down to 10" or so (even on a computer screen) they both look virtually the same. Except one took up 880 KB while the other only 91 KB, 9 times less! Do this for a 90 MB PDF full of similar content (raster images) and you may get a 10 MB PDF. Give it a shot.

Either compress it beforehand as TIFF or let Acrobat compress it while it's reducing everything else. I usually let Acrobat compress it. It's not that I don't trust Gimp or anything... it's just that Acrobat is too expensive for what it can do. It better have some of that Photoshop joojoo, compression-wise! Am I wrong here?