Hacker News Reader

Comment by codetrotter | Hacker News Reader

> send a message to the owner and ask them to scan this particular book.

Scanning books take a lot of time. And making an actual high quality scan of a book is difficult even if you have a good scanner.

Physically sharing books could make it so that important books eventually get it into the hands of people with the right expertise and equipment to efficiently produce stellar quality digital scans of the book.

View on HN

You can scan a 500 page book in 20 minutes with you phone, non-destructively. Though it’s not fun and the quality will not be ideal, especially if there are graphics in the text also. For text-only books, quality matters little as long as the OCR can read the text.

I think that's ambitious. I've scanned a handful of books and although I don't know the page count, it's always been multi hour projects. Especially the editing afterwards to mark the text on the page and adjust warp tend to be time consuming. I even had a mounted tripod and remote to standardise.

Voice Dream Scanner on iOS recognizes pages and text and performs OCR. You just go snap snap snap.

For text that is sufficient. If you want to capture the layout also, or graphics, things get much more complicated. Even just getting headings requires post processing. So yeah, proper ebook vs just extracting the text are two different problems.

vFlat Scan [0] automates the unwarping.

[0] https://www.vflat.com/

I just tried this with a thumb-held warped page and Tesseract for OCR--worked pretty well! (Tesseract by itself struggles with warped pages.)

There's also https://scantailor.org/ (and a maintained fork at https://github.com/4lex4/scantailor-advanced ) which semi-automates unwarping and other corrective tasks in scanned books.

Graphics or any kind of symbols. I tried this with an out of print edition of an economics textbook from the 70s: the scans were fine but I never found any OCR that could even semi-faithfully reproduce the equations, nor anything in a table.

It occurs to me that a place where scans and first-pass OCRs of this stuff are shared with the intention of version controlling the resultant epub would be handy, such that small PRs could be applied cumulatively and the result be versioned (and therefore referenced). That's considerable work and legal risk, though.

That's not too far removed from what Distributed Proofreaders did/does.

You can use a device like www.czur.com to scan a book pretty fast and have a perfect pdf/epub version in under 30 minutes.

bookscan.us has a book scanning service for $1/300 pages.

There's a bit of friction though. It could be much simpler.

There's also the various https://diybookscanner.org variants. Not the quickest imaginable system, but it's much better than going <i>snap snap snap</i> with a cellphone.

Reply on news.ycombinator.com