I’ve got a rather roundabout process (in LO 4.1 but used since 3.x) to have EPS images be preserved as vector images in a PDF:
Print to PS file and then convert to PDF using
ps2pdf -dAutoFilterColorImages=false -dColorImageFilter=/FlateEncode – the regular PDF-export cannot be convinced not to convert them to nicely anti-aliased raster images (whose resolution I have no control over) that of course look crappy in print.
So far so good, but now (with printing already underway) I realized that text in the PDFs (already in the PS files) has been turned into vector objects and can neither be marked and copied nor searched for. This is only true for some fonts (Bitstream Vera becomes vector drawings, Linux Biolinum and Helvetia don’t!). Sadly, I can’t change the font types now as that would mess up the layout.
=> Does anyone know a way to keep EPS images as vector objects, PNG images as uncompressed bitmaps and characters as characters? Or is there a way to add the lost information to the PDF after the fact? Some sort of OCR process that works on vector images instead of scanned bitmaps?