Scanning in your own books

[Ravensknight]

Originally Posted by harriska2

Some books are not available in e format. .

This is why I scanned about 10 books a couple of years ago and proofed them. Of course, now they're out in ebook form.
What I have learned from that is that all the books will eventually get the e-treatment. Even if not while I'm alive ;-)

[Katsunami]

Originally Posted by harriska2

Some books are not available in e format. And if you want to use it to highlight and write on it and annotate, e format is the easiest. Some of use refuse to have paper books....

In that case, I would just buy a cheap all-in-one, and copy the pages I would like to put annotations on, and bundle them with the book or put them into a binder. In case of a non-study book, I would never mark the book itself.

[markom]

Originally Posted by Katsunami

Woah... old thread. I'm still of the same opinion as I was a few years ago, even more so with lower prices, Kobo codes, and much more availability.

Scanning, OCR-ing and proofreading is way too much work. If a book costs €7. If it even takes only an hour (and it will take MUCH longer), it's not worth it. Would you want to work for €7, gross? I wouldn't, if I can help it.

There is really no need for proofreading, though, because every A5 format can be easily read on 6" reader in landscape if not in portraite.

Abbyy Finereader' ocr (pdf or epub) is pretty good for fiction (no need for 100% accuracy) and Abbyy' exact pdf image for scientific pdfs (100% accuracy).

So, in my case ocr-ed pdf is all it takes for reading, highlighting, scribbling, searching etc. thereon.

For an average 500 pages A5 book it takes me 1-1.5 hour for scanning (average flatbed scanner) and additional 1-1.5 hour for computer to create under 10 MB pdf using Acrobat' Clearscan ocr for exact pdf image.

[Wolfrott]

I do this! I am bed ridden and have mobility issues that stop me from holding hardcopies.

For years now, what I do is:

1. Either scan an intact book 2-pages at a time via a flatbed scanner, OR rip it up into sheets to feed through a duplex document scanner.

Time difference between the two are days; ease, speed, & quality of the duplex surpasses the flatbed. If the book is OOP, rare, sentimental, etc I recommend you don't rip it, scan it flattened on the flatbed.

2. OCR via Adobe DC Pro or through Microsoft Onenote. It depends on the scans; Adobe hates darker pages with fancier fonts and images like borders, chapter art breaks, but Adobe's speed at batch OCR can impress you to stick to it despite more time needed proofreading. OneNote takes longer as its manual OCR grabbing, but has nearly flawless rate of spelling errors in comparison to Adobe.

3. Use an online line break tool to fix paragraphs and indents for you. Lifesaver, I swear.

4. Proof read by comparing the text output file (pdf, word, txt, etc) to the scans. Most common mistakes are ff, tt, mm, rr, mistaken for each other.

6. Put together in Word, polish it off, Calibre, and you're done!

I can finish a 300 page book in a week following the above steps. Initially it took 2+ months.

[harriska2]

I use MS Word to fix line break issues. I don't know why but it just seems to work. Sometimes I have to be smart about it (first search and replace .^p with |, then replace all ^p with space, then finally replace all | with ^p) but I have lots of flexibility with a gui interface.

[alecE]

I scan perhaps 24 old paperbacks per year, partly to ease space restrictions, partly to provide durable copies of paperbacks that are falling apart. Standard procedure is:
- split book into 32-page signatures
- scan signatures using Canon P150 duplex sheet scanner
- ocr the output using Abbyy Fine Reader
- correct the obvious typos in Abbyy, mark where sentences run over from one page to the next, identify required and redundant hyphens at page breaks and then output to .txt file
- further editing in Notepad++ (systematic treatment of speech marks, ellipses, endashes, ligatures, apostrophes... specify the language as HTML and use entities...add markers to divide the text into chapters...use regex to tidy line breaks etc
- construct an empty e-book in Sigil using standard components (pre-canned title-page, front-matter, chapter styles, pre-defined css sheets)
- move text from Notepad++ to Sigil, add italics, additional breaks etc plus further styling
- create a "nice" front cover (usually scanned from the original cover) and add that
- add to e-reader via Calibre

Generally reckon from 4 to 10 hours work per e-book, depending on the complexity and the quality of the ocr process. Feel the work is justified as I gain digital copies of titles which are unlikely to digitised by the publisher, at least in my lifetime!