daniel_reetz wrote:At some point in the near future I will start a thread where people can post samples for a sort of "diybookscanner standard image set". Perhaps we can make a sample book that the whole community can scan as a quality test.
i think this is a critical need. If we all had access to the same edition of some off-copyright book, we could u/l pics produced by our scanners for comparison. we could also compare stats of how long to scan, how many missed pages, and OCR error rates.
last summer, i bought a beautiful copy of Euclid's Elements from Blackwell's in Oxford. It is a nice mix of text and drawings, newly printed from a typeset done in the 1850s or so. It's off-copyright, but it's probably not widely enough distributed to be well suited to this task. Perhaps some government publication, that's in every library would be better suited to this task, perhaps some pages from the Federal Register.
Here's an idea. Dan, you should write a book on DIY Bookscanners. And publish it yourself. Add test patterns on certain pages. And sell it to guys wanting to tune their systems. i'd buy one. so yould most of the lurkers here. Especially after they've used the PDF free download to build their first unit that they want to calibrate. Since you would retain rights to the book, you could grant free license to scanner dudes to post imagery therefrom. The book sales wouldn't be in the same league with JKRowling, but it'd probably pay for some conferences and stuff.