I got a bit carried away with doing this. Instead of just smacking together all the 360 dpi TIFFs I scanned seven years ago, I had to scan a new set at a higher resolution, then crop them, then fix the page numbers, add chapter marks, and make the table of contents a set of live links.
I’ve got out of the way of thinking in PostScript, so I spent some time looking for tools that would do things graphically. Bah! These things’d cost a fortune, so armed only with netpbm, libtiff, ghostscript, the pdfmark reference, Aquamacs, awk to add content based on the DSC, and gimp to work out the link zones on the contents page, I made it all go. Even I’m impressed.
One thing that didn’t impress me, though:
I used to edit multi-gigabyte files with emacs on Suns. They never used to complain like this. They just loaded (admittedly fairly slowly) and let me do my thing. Real emacs don’t give warning messages.