Page 1 of 1

One size doesn't fit all --or how to scan larger books

Posted: 21 Oct 2015, 07:57
by scann
Hi all, this is actually something that Dan Reetz wrote a while back ago in an email. And since it's a question that time over time people ask, I thought that it might be a good idea to post this in the forum. Since it's more than complicated to build a huge scanner -apart from being undeliverable, it might be impossible to handle- there are different solutions to overcome large books and large newspapers.

Below is Daniel's insight to this problem.


I am assuming that the books you are working on look something like the attached image. 24x36" per page. I'll tell you the bad news, and then the good news.
GreatRegister1-300x211.jpg (13.24 KiB) Viewed 14658 times
As far as I know, there is no V-shaped scanner in existence that can scan documents of that size, including the Hackerspace scanner & The Archivist. I have made a larger version of the Archivist, but it doesn't even scan half the area you need.

Some numbers:
at 24x36", to digitize at a minimum of 300PPI (Pixels Per Inch) you need

24*300=7200px tall
36*300=10800px wide

That's 77.1 megapixel resolution. So you're already well beyond the range of ordinary cameras and lenses. The very best lenses on consumer cameras resolve about 24mp, the best sensors available are about 36mp.

Then, regarding weight, those books are certainly very large and heavy, so any machine that lifts them toward the glass for photographing is probably not the best solution. Even with counterweights etc, it's just asking a lot of the operator and the machine.

You can likely still get this done, but it depends on your intentions. I'm going to make some assumptions.

0. You're not trying to capture pristine images of the page, you're trying to get information out of the page.
1. These books are handwritten. Therefore the writing should be bigger than printed text.
2. You're going to have humans reading them or transcribing the text.
3. You don't mind DIY in terms of setup.
5. The paper is not shiny.
6. Travel-ability is important to you.

Now, if you could stand copying the page at 150DPI, you could do 5400*3600, which is around 20 megapixels. There are quite a few cameras in the 24 megapixel range. A Nikon D5300 is a cheap one, and it can be mated with the inexpensive but excellent Micro Nikkor 40mm f2.8 lens, which is great for document imaging. A Nikon D5300 is about a $500 camera, plus a $60 power adapter (don't try to do this on batteries) plus a $300 lens is $860. Add $50 for a remote control for the camera if you're not going to control it by computer. Also a huge SD card.

Then, you need a copy stand. With books this large, you do not want to be trying to use an ordinary scanner. Rather, think of it as taking a picture of a poster or pattern. You want the page to be flat on the table, with one side of the book slightly propped up on a book wedge (you can DIY these, I've attached an example shape, foam inside cloth or a sandbag will work).
Your camera will be overhead, at some distance - 3-4 feet. To do that, you'll need a good tripod and a camera arm. I like the Manfrotto Magic Arm, but it can literally just be a board. Ignore that there are two cameras.
cameraarmside .jpg
cameraarmside .jpg (167.6 KiB) Viewed 14658 times
To light the book, you put a softbox on either side. That's something that looks like this: - those ones might not be ideal for travel because they are compact fluorescents, which can be brittle. You might look for an LED alternative. With those guys facing the page of interest, you should have bright, even lighting. Be sure to have the room lights off when you're shooting.

Finally, with this method you'll capture all the left pages, and then all right right pages, and combine them later. Focus your efforts on capturing the data while you have access to it, and worry about how it fits together later.

Basically, what you're doing here is creating a very flexible, large-scale copystand out of ordinary photographic equipment. I've attached a drawing which I hope makes things more clear.
2014-11-09 14.49.33.jpg

Re: One size doesn't fit all --or how to scan larger books

Posted: 21 Oct 2015, 08:59
by cday
Here's the above image rotated upright, if you would like to substitute it and delete this post:
2014-11-09 14.49_33_180.jpg
2014-11-09 14.49_33_180.jpg (86.74 KiB) Viewed 14646 times
Or in black and white, if you prefer:

2014-11-09 14.49_33_180_1-bit.jpg
2014-11-09 14.49_33_180_1-bit.jpg (88.42 KiB) Viewed 14646 times

Re: One size doesn't fit all --or how to scan larger books

Posted: 21 Oct 2015, 18:17
by russca
In my view the only two limits to DIY scanning of larger books are creativity and budget.

There are multiple ways to go.

For DIY bookscanning of rare or important larger-sized books, I would do this:
1) Buy a decent DSLR camera and a 50 mm prime lens. For example, Canon EOS 5DS has 50 megapixel sensor.
2) Buy a Photography Studio Softbox Continuous Lighting Kit
3) Build a wooden rig out of 2"x4" wood and plywood for an adjustable height cantilevered or ceiling/wall-hung camera mount.
4) Build a custom-sized table or just a top that can be moved or set to/on the exact X-and-Y positions while maintaining page straightness
5) Buy a custom-cut piece of glass or plexiglass for placing on top of pages
6) Set the camera to manual mode and make necessary camera adjustments
7) Section a book page into several quadrants and take a picture of each quadrant.
8) Stitch pictures automatically with Photoshop (File=>Automate=>Photomerge). Yes, Photoshop can stitch text scans.

So, 600 and up dpi scans are possible for books with pages larger than A1 size.

If you are tech-savvy, you can build an Arduino scanner set-up.. Please read this for ideas:
diy-negatives-scanner.jpg (36 KiB) Viewed 14619 times

Re: One size doesn't fit all --or how to scan larger books

Posted: 29 Oct 2015, 10:14
by scann
Awesome, russca! Thanks for your contribution.

Re: One size doesn't fit all --or how to scan larger books

Posted: 02 Nov 2015, 11:41
by davidlandin
You could also look at this video of a large version of my scanner which is being used in Pittston to scan town records

David Landin

Re: One size doesn't fit all --or how to scan larger books

Posted: 31 Dec 2015, 10:29
by hphistorical
Little late to the party on this post, I am doing this system now, building a wood copy stand to hold the DSLR in place. I am curious as to the angle needed for the other (uncopied) side of the book for the wedge. Any set size, or just improvise? I noticed a sandbag was mentioned, I'm guessing that a 45 degree angle will be enough, or maybe a little less?

cday wrote:Here's the above image rotated upright, if you would like to substitute it and delete this post:

Re: One size doesn't fit all --or how to scan larger books

Posted: 31 Dec 2015, 13:16
by rkomar
I wouldn't go less than 45 degrees, or you will see reflections from the opposite pages. Going more than 45 degrees cuts down on those reflections, but then you have to put the light source further away or you start to see its reflection in the images (especially for large books). If you don't want to put the light and cameras too far away, then 45 degrees is probably the best compromise.

P.S. I had David Landin's scanner in mind when I wrote this. It doesn't apply to the rig for shooting one page at a time.