Have you looked at old books in the google books archive? You quite frequently find pages where the page-turning robot has screwed up, with results like distortion, lack of focus, weird smearing effects as the page was still moving when the picture was taken, or plain thinks it can't deal with like fold-outs. I recently tried to find a 'plate' (illustration) in an engineering journal from ~1864; it was missing from both the google and archive.org scans; probably due to being a fold-out. If all the archives throw their paper copies away, quite a lot of information will be lost.
Fold-outs are the worst. I'm building a magazine encyclopedia and magazines are chock-full of weird inserts and fold-outs that never appear in any scans. I am trying to build an archive of all the physical magazines so that at some point people can go through and check them against the scans to see what's missing.
Scanning systems are set up generally to only scan "regular" media without any surprises.
