So, in short: what we do is just re-distilling the PDF by using poppler to read the PDF and render it to a cairo surface with the PDF backend. I see no clear reason why bitmaps from the PDF shouldn't re-distill properly without loss of resolution. (There could be a slight loss of crispness as the images could be converted from whatever format they were in, eg JPEG2000, to the format in which the cairo PDF backend prints bitmaps -- likely PNG if I had to guess, so that should be lossless at the expense of an increase in file size).
In xournal, the code is in xo-print.c, print_to_pdf_cairo() is the main function that exports the open document to a PDF; print_page_to_cairo() is where individual pages get exported to a cairo surface (either for new-style export-to-PDF or for File -> Print to actual or virtual printers alike); print_background() is where the background gets printed to a cairo surface and as you'll see in the BG_PDF case it just scales the page and asks poppler to draw the PDF page. Denis On 06/17/2015 04:29 PM, Denis Auroux wrote: > The new export-to-PDF renders things on a PDF cairo surface, which I > imagine is more or less the same as using File -> Print -> Print to File > -> PDF with the main difference that we create the cairo context > ourselves and set the page sizes to match ours, rather than letting > gtkprint set up the cairo context and use the page sizes set in the > virtual printer. > > So, basically, we create a cairo PDF surface, and then ask poppler to > render the appropriate page of the existing PDF file to that cairo > surface (and then add the annotations on top). > > I am not aware of any resolution parameters that we could set in there. > Well, actually, there is one: > cairo_surface_set_fallback_resolution () > (see the documentation of cairo_surface_t) > but I don't see why it would matter since it's only for falling back to > rendering to a bitmap and putting that bitmap in the PDF when cairo is > asked to do something it can't directly do into a PDF, and presumably > whatever poppler tries to do to show the scanned image is just read the > PDF-embedded image and draw another embedded image into the other PDF, > which shouldn't require any fallback. > Besides, the fallback resolution is apparently 300 dpi by default > according to the cairo docs, so that should be plenty. > > Does export also drop the resolution of other bitmaps -- either those > inserted by the user into xournal (if so it means the cairo surface > we're writing to silently compresses bitmaps without asking us -- > xournal wouldn't change the resolution), or bitmaps contained in the PDF > file that we're annotating? > > Anyway: I've seen this happen as well, on PDFs where I thought the > background was actual vector stuff and not bitmap, but I didn't > double-check. It's definitely annoying and worth investigating. In > principle poppler is used both to render the PDF on-screen for xournal > and to export it to the new PDF, so if it can do one well it should also > be able to do the other well. > > Denis > > On 06/17/2015 04:09 PM, D M German wrote: >> >> Hi Denis, >> >> One question about the new exporter. >> >> I have PDFs that each page is a scan (images). When I use export to PDF >> it drops the resolution significantly. If use the old exporter, things >> look ok. >> >> Is this based on some variable that controls the resolution of the >> background PDF? >> >> >> thanks! >> >> >> -- >> Daniel M. German "My friends would think I was a nut, >> turning water into wine, >> Solsbury Hill, Peter Gabriel -> opening doors that seem to be shut." >> http://turingmachine.org/ >> http://silvernegative.com/ >> dmg (at) uvic (dot) ca >> replace (at) with @ and (dot) with . >> >> >> >> ------------------------------------------------------------------------------ >> >> _______________________________________________ >> Xournal-devel mailing list >> Xournal-devel@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/xournal-devel >> > -- Denis Auroux UC Berkeley, Department of Mathematics 817 Evans Hall, Berkeley CA 94720-3840, USA aur...@math.berkeley.edu ------------------------------------------------------------------------------ _______________________________________________ Xournal-devel mailing list Xournal-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xournal-devel
