DIY book scanner, 150 pages/min

[BeccaPrice]

With 1dollarscan, the pdf is OCR'd. What I did was save the PDF as a TXT file (I"ve got Acrobat Pro, so I can do that), and then had a file I could edit for scanning errors.

[kevinp]

Originally Posted by rkomar

As I said before, 150 pages per minute would a pair of pictures every 0.8 seconds. That's not possible manually. 15 pages per minute (900 per hour) would be 8 seconds per shot, which sounds more reasonable. So far, the two numbers put forth (150/min and 150/hr) both seem wrong compared to what people say on the DIY website.

Check out this video. Start around 5:00. It looks possible to me.

[HarryT]

Originally Posted by kevinp

Check out this video. Start around 5:00. It looks possible to me.

In that video, he scanned 8 double pages in 24s, so that's 16 pages in 24s, or 1.5s per page, which is 40 pages per minute. It's fast, certainly, but only approximately a quarter of the claimed 150 pages/min. I really don't see how he could have operated it 4x faster! It's very impressive even so.

[yekim54]

Originally Posted by rkomar

I have the cheapest version (Opticbook 3600), and I can get about 120 pages per minute at 600 dpi (black/white) for large books.

Are you saying you can scan 2 pages per second with the Opticbook 3600? According to the Opticbook sales literature: "The OpticBook 3600 plus also has an amazing scanning speed of 9 seconds per page that really shortens the time that need to be scanning a book."

[rkomar]

Originally Posted by yekim54

Are you saying you can scan 2 pages per second with the Opticbook 3600? According to the Opticbook sales literature: "The OpticBook 3600 plus also has an amazing scanning speed of 9 seconds per page that really shortens the time that need to be scanning a book."

Doh! I was complaining about bad numbers earlier, and did the same thing myself. 120 pages per hour, is what I should have said. The mass-market paperback number of 200-300 pages per hour was correct.

The reason the second is faster is partly because the scan is shorter, but mostly because the resolution was halved. It seems that the device is limited more by the digital side of things rather than the physical moving lamp-plus-sensors side. Switching to grayscale, or even worse colour, images slows things down significantly.

It's definitely slower than the DIY camera-based systems. The only advantage I can think of for it is that the images won't be distorted in the way that they will be from the camera. If you're going to OCR the results, then that won't matter much. However, if you want to save the images themselves to the ebook, then the ones from the scanner would probably look a bit better. Personally, as long as the distortion wasn't too bad, I'd probably pick the faster system since scanning a book for hours is a slog.

[HarryT]

No, the camera images aren't distorted. The cameras are mounted at an angle such that each is parallel to the page it's photographing.

[rkomar]

Originally Posted by HarryT

No, the camera images aren't distorted. The cameras are mounted at an angle such that each is parallel to the page it's photographing.

I meant that straight lines on the page won't be straight in the image. Good cameras at the proper distance would make those distortions small, but I expect that the cheap point'n'click cameras that are mounted closer to the book will show significant distortions.

[afv011]

Originally Posted by rkomar

I meant that straight lines on the page won't be straight in the image. Good cameras at the proper distance would make those distortions small, but I expect that the cheap point'n'click cameras that are mounted closer to the book will show significant distortions.

Not necessarily, although the optics will indeed have distortion, the software is likely to compensate for those.

[danielreetz]

Hi all,

The design of the scanner is such that each press of the button captures two pages simultaneously. 100-some pages per hour - as quoted above- is ridiculous. It does not take one minute to turn a page. Even if it did, that would be 200 pages an hour. . Personally, for me, it takes on average 1.5 to 2 seconds per page turn. You can do the math.

People on my forum regularly report 600-800 pages per hour on the scanner, 400 page books on a speed run take, on average, 15-20 minutes end to end. Some people on my forum report 1200PPH, though I believe that to be near an absolute maximum. People gotta pee, eat, and drink. People make mistakes. Occasionally a page needs to be captured twice.

Now, of course the photo-images of pages require some post-processing which takes time. But the capture stage is what requires your attention.

In general, IF it is possible to destructively scan a book, then that is the most economical option. However economics aren't everything. Many of us do not wish to destroy our books, and would like control of the ebook creation process, as well as solid backup of the page images, so that as new technologies come along, we can improve our ebook libraries and shift into new formats.

Over 3000 people have worked with me on this project, improving the hardware and software, and it's in use all over the world, and in several commercial and government scanning operations, and I've dedicated quite a bit of my life to offering kits to get things to the next stage of development. So I also disagree with HarryT that it is of no real commercial practicality, even if it is by definition concerned with all the cases not addressed by the big players.

My goal is to put one of these things in every hacker-and-maker-space in the world, because, as people noted here, most people don't need to own one (and it's not economical to buy one) to digitize the small number of important books they already own. When it comes to book scanning equipment, access is more important than ownership.

However for people with bigger libraries and/or dreams of helping bring small communities, hospitals, and makerspaces into the digital age, it is just the thing.

Cheers all,
Daniel Reetz
www.diybookscanner.org

[Turtle91]

Hey Daniel!
I was wondering when your ears would start twitching and you would notice this thread!

It's good to hear some actual numbers instead of relying on my old memory.

I hope I got most of the info right.

Cheers,

[cromag]

Originally Posted by wizwor

Back in the 80s, I had an Atari computer and wanted to scan images for a newsletter. There were no affordable scanners at that time and none whatsoever for the Atari, so I did something similar.

I bought this device called ComputerEyes, attached it to a video camera, and digitized images taped to a nearby wall. No OCR back then, but I'm not sure why this impresses. For a fraction of the cost of this device with no effort, I can buy a scanner with an ADF that comes with free OCR software.

Sorry, I noticed this late.

Back in the 90s I was putting out a newsletter on my Amiga. I had a hand-held scanner -- apparently, nothing like it is currently available. It was wider than a bar code scanner and had an RS-232 interface. A shame, because it would probably have done a good job at this.

[Kumabjorn]

Is it possible to get good working results with reference works? Something that is indexed and makes it possible to look up things fast and conveniently. I have a coupe of works that i would love to have in e-Book format, but alas the Publishers don't provide it.

[Turtle91]

Originally Posted by Kumabjorn

Is it possible to get good working results with reference works? Something that is indexed and makes it possible to look up things fast and conveniently. I have a coupe of works that i would love to have in e-Book format, but alas the Publishers don't provide it.

No. It will scan the text and OCR, but computers aren't smart enough to know what it MEANS...at least that I've seen.

If you wanted active hyperlinks in an ePub/html you would need to manually insert the links to the proper locations. If all you want is a digital copy you could save it in PDF.

[Ralph Sir Edward]

Truthfully, the scanning is the easy part. (I use and Optiscan 3600, with Finereader 5 and 9. Some books work better with 5, most with 9. It varies on the paper/font.)

The real time consumer is the proofing. And you never seem to get it perfect...

[HarryT]

Originally Posted by Ralph Sir Edward

The real time consumer is the proofing. And you never seem to get it perfect...

Very true indeed. I can only properly proof (ie do it at the "check every punctuation mark" level) about 10 pages an hour.