For a far more impressive scanner, developed by a Google engineer and including automated page turning see: http://www.theverge.com/2012/11/13/3639016/google-books-scan...

I believe there is a far more detailed breakdown of the project floating around somewhere.

It is possible, but not clear to me, that Google use this for their own book scanning project - Google Books.

Jason Scott says it destroys books:

http://ascii.textfiles.com/archives/4695

Ok... and is there some place where he explains why he believes that to be the case? The Google presentation explicitly mentions that a feature of the scanner is that it doesn't destroy books.

There's really nothing to go on except the small quip from the author, but it's not hard to imagine a scenario in which the machine has no way of adjusting for more delicate sources.

Watching the video from http://linearbookscanner.org on how the scanner works, I'm guessing the "doesn't destroy books" pitch is a response to other scanners where you're supposed to slice off the spine of the book and feed the cut pages into it. Given that it's just a vacuum pulling the pages through a slit in the apparatus, I can easily imagine much more delicate texts getting torn pretty easily using the device.

The FAQ on the Linear Book Scanner page itself admits that it does have restrictions on the type of pages and book dimensions it can work with, and that it has a relatively high damage rate (from the aforementioned website FAQ, "Out of 50 books tested, 45% had one or two of their pages either torn or folded.")

The LBS is probably fine for more modern texts that just haven't been digitized, but I too would be wary about automating any rare or delicate text.

Interesting product, but not really much of an improvement in speed if you have to do the page turns yourself.

The foot pedal for scanning saves a tiny bit of time, but it is still cumbersome to scan a whole book.

Also strange to emphasize MIPS CPU which should not be better than your average cheap smartphone ARM.

So, you are saving some time over regular scanner(been there done that), but it is still not practical for say 100 books.

If they had figured out a clever way to do the page turns this would have been a winner.

See http://www.diybookscanner.org for prior work and discussion of hardware/software in book scanners.

"Build your own digital library"?

For a single book, sure, makes sense, but this totally doesn't scale. Perhaps a nice gadget to have at a hackerspace, but then again you can probably just build this yourself.

This depends very much on your time/money tade-off, as well as the availability of texts, and the opportunities for sharing of scanned works.

In many cases,books are nonlonger in print, scanning is among the few ways to get a copy. Or you need a small number of works to fill out a library which is otherwise available online. Services such as Library Genesis make sharing works among individual scanners and scanner communities far easier.

For regions of the world with little access to printed materials, tools such as this are a godsend.

If you can scan a single book in 5 minutes, you could scan 8 books in a single episode of whatever you are currently watching on Netflix.

This will all come down to the software. the folks at http://www.diybookscanner.org/ have been using point and shoot digital cameras to make scanners for years so the hardware isn't the problem. The problem is the OCR and dealing with page curl. Without something to flatten the page, the quality of OCR goes down a lot. The video says they have solved it but I am skeptical. Good OCR software like Omnipage or Finereader usually sells for over $100 to start with and they don't do that well with curved pages so if this product really can then it would be worth the $200 to someone serious about book scanning. I would have to see some real proof that they can do high accuracy OCR before spending money on it though.

I have seen page-curl solved with two high-megapixel cameras; you can use parallax on the text to create a 3d-map of the page from which correcting page curl is quite easy to do. The problem is that you lose resolution in the highly-curled areas, so while a 1-2MP camera is plenty for perfectly flat pages, 16-20MP is better for non-flat scans.

here's their indiegogo page: https://www.indiegogo.com/projects/czur-scanner-build-your-o...

Previous techcrunch article posted here: https://news.ycombinator.com/item?id=10500915

another option that's roughly $1 for 100 pages:

http://1dollarscan.com/pricing.php

This is a good option depending on your needs (e.g., only select books, smaller collection, you don't want to do any of the work, etc.)

Hm - interesting. What stops students from sending in their textbooks to have scanned and shared with fellow students?

Not much besides the textbook gets destroyed, so no more margin notes or highlights.

It seems a modern camera-phone (my Samsung S4, and the S4 before it, takes crystal clear photos of text) and some clever mounting could easily replace the hardware setup: The software seems to be the interesting component here -- does it or something like it exist alone?

I write all my design notes, sketches etc on 5mm A4 square pads, this would be an incredible devices for me since I could scan them in really quickly once I'm done with them and I'd have a permanent record of all my stuff without a massive pile of pads to shlep through.

Could also attach metadata to them (project name etc).

I considered building something like this a while back as a side project but couldn't find a high enough resolution camera I could connect to a raspberry pi, this is 16MP, the rpi camera is only 5MP so that was a problem.

I was looking into similar products when I found http://booksorber.com/

I already have the DSLR, tripod and floodlight so it was the natural choice for me. I guess this one is good if you have none of that.

In the Cruz video there is a quick mention of a cloud service. Is the OCR'ing done in the cloud? Is the cloud necessary for the operation of the scanner? There is no need to spell out the privacy and security implications if that is the case.

Next time they probably shouldn't use The Sound and the Fury as an example book. It's got quite a few n-words scattered throughout the text that happen to show up in the video.