Daniel Reetz' DIY Book Scanner Tech Talk at Google HQ

You've seen the very latest thing that I created

with my computer

So, uh, January 2009 I was a grad student

and I was searching Amazon for copies of textbooks

that I couldn't afford

and Amazon recommended that I look at some cameras

The cameras just so happened to be

a quarter the price of my textbook.

And so I did what I always do when I'm in trouble:

I jumped in a dumpster

and started building things.

Over three days I built the very first DIY book scanner

and showed it to my friend Aaron.

Aaron's a software wizard

and understood that I needed postprocessing software which I had written in a photoshop script

So he turned around and wrote software in exchange for his own scanner.

As I built the second scanner

I documented the thing obsessively

in total detail

and put the thing on Instructables

as a 79 step tutorial on

how to build your own book scanner.

So what happened after that kind of blew my mind.

First of all, it got over a 100,000 views in the first month.

It's now 204,000 views as of today.

Second of all, I won a laser cutter.

Third off, I won a qbuild for laser-cut art.

But what blew my mind wasn't actually winning things.

What blew my mind was the people that contacted me

after that.

Immediately after I put up the Instructable,

I got dozens & dozens of messages

from people all over the world

saying "I really want a book scanner like that,

I'm so glad you put these plans up."

And they started telling their stories

as to why they wanted them,

what they were going to do with them,

and they also started back improvements,

which in a way, isn't really all that difficult,

because mine was made of trash.

So, what I did was grab the first guy who showed up

his name's Rob. He's a mathematician.

We founded diybookscanner.org

diybookscanner.org today

Excuse me for flipping back and forth here

let me back up one more step

The first five people to sign up were

two mechanical engineers, two software engineers,

and an intellectual property lawyer.

Today, we have 600 members,

between 90 and 150 builds,

it's hard to say because they're not all complete

and it seems like we get a new build

almost every day.

Just showing you a selection of builds here,

giving you an idea of the enormous variety

of stuff people have constructed to do this job.

And this is only about half.

So what is the DIY book scanner?

What exactly are we doing?

DIY book scanning at its core

is nothing more than using cheap

compact cameras to digitize books and other materials.

So the essential insight that we have

is not only that books and cameras

cost about the same amount,

but that the quality of the point & shoots

has gotten good enough that we can do serious

digitization work with them

As you know, scanning books with cameras

has many, many advantages,

the first being that it's an order of magnitude faster

than scanning with a flatbed scanner.

Anybody who's tried to scan a book on a flatbed scanner

knows it's miserable.

You break the binding, it takes forever...

it's an exercise in frustration.

Where we go from there is kind of interesting.

When you start trying to digitize books

using compact cameras,

you run into some interesting problems.

There is not an equivalence

between compact cameras and SLR.

So many of our people, including myself,

showed up saying, oh, well, to get better sharpness

we'll stop down the aperture.

Well, that's actually a problem because

believe it or not, these cameras don't have aperatures.

They have a neutral density filter

that drops in the optical path

and degrades image quality.

So in the case of compact cameras

just as one example,

you leave the aperature as wide open as you possibly can.

And that's one of the big differences between DIY

book scanning and all of the commercial stuff

is that we treat compact cameras

for what they are

and optimize them for the best possible parameters.

Now I'll talk a little bit about the construction

of the DIY book scanner. This is in the

my old -- the garage I used to live in.

There you're looking at lighting.

And for DIY book scanning

I almost always recommend

tungsten or halogen lighting.

The reason for recommending tungsten

or halogen lighting is technical:

they have a pretty flat spectral output.

The sensors on a camera

have essentially red, green, and blue sensitivities

that peak some way.

A lot of people that come to our forums

want to use compact flourescents.

They say "oh they're eco-friendly, and

they're in a variety and you get them in

any color balance you want."

But the output is spiky.

And it might not match the spectral sensitivity of your sensor.

So generally we recommend those

because they work best with all platforms.

However in cases such as this tungsten

isn't practical, and so we've developed

high-output lighting systems

using pre-LEDs (?)

and that's the folding scanner

that I produced with my laser cutter

right after I got it.

and this is (?).

So the next part of the book scanner

is camera support.

On this camera we just had

two columns here

and on this scanner which we call "The New Standard"

it's sort of a standard

build for beginners in our forum.

You can see there are two 2×4s.

And the reason for going for camera support

is simply that if your cameras move while you're digitizing

it makes post-processing really hard.

If your book moves while you're scanning

it makes things really hard.

The next part of the scanner is called the platten.

And the platten is a v-shaped piece of glass

or plexiglass, or in our case now

we're starting to use Gorilla Glass,

which mechanically flattens the page of the book.

We don't mess around with computational dewarping.

We don't mess around with structured light

and other methods of dewarping the page.

Unless we have to.

It's much better to get good input

and good output.

So, what goes under the platten is the cradle.

And the cradle is what holds the book

and the cradle also accommodates

the thickness of the spine.

So you'll notice that the gap

on the cradle that

John (?) is holding is

adjustable.

You have to put all that stuff somewhere

so we have a thing called the base.

Where we put everything.

And finally we have the electronics.

Now, one of the fundamental things

that enable the DIY book scanning technology

in the first place

was that there's a hack for these cameras

that allows them to be triggered simultaneously.

So there's extra firmware

authored by David Sykes (?)

called Stereo Datamaker

and when they boot up,

they're waiting for a signal to fire at the same time

I can demonstrate later.

So, all of that gets you photographs of the pages.

And you need to post-process them somehow.

And this first software that Aaron wrote

was great for our first software

-- Photoshop scripts worked great --

but we really needed something more serious.

And we hooked up with the author

of a package called ScanTailor

which is the closest thing we have to pure magic.

It can eliminate lighting issues

binarize (?) extremely difficult pages

It's amazing how it cleanly handles really poor quality input.

And many of our scanners do produce poor quality.

What's so amazing about it also

is that he spent a lot of time

making it really simple to use,

free to download and modify,

and it also works on all operating systems.

We have a forum specifically to support it

and we're continuously to support it

developing it and pushing it further,

including dewarping (?).

So, dewarping progress which was posted two days ago

-- this is an example

where we moved from the page on the left

to the page in the center, using some spline detection.

So, that's a lot of talk about technology.

What I'd really like to talk about are people.

This is Rob the co-founder of my forum.

And the reason I like to talk about people is

because it's what people do with this

is what's important, not some gadgets with cameras

or whatever.

Or garbage for that matter.

So, Rob is sort of the canonical user

of DIY book scanners.

His home is absolutely full of

Science Fiction books, technical books,

just bending every MDF bookshelf

in the house.

And Rob, like many of our forum users

doesn't find it acceptable

to repurchase books continuously

like we did with vinyl, 8-track, cassette, mp3,

and now whatever comes out of the iTunes store.

So Rob built a book scanner

which he did when he's not building a

rod-logic-based (?) mechanical computer

and digitized all of his books.

So, go from there

to Ben Verady (?) who's a graduate student

at Tulane University

who's working on a project called

The Durationator.

The Durationator (?) it's currently in process

right now, will be a database of

all database records.

Such that you can actually search for legal advice

on whether or not you can scan something.

And this is one way that DIY book scanning

technology.

A recent member of our forum

is Patrick Hall, he's a linguist.

And he works specifically with fileslips,

which are a linguistics tool.

They're slips of paper

on which are written grammatical fragments,

words, definitions, etc.

And these fileslips are a precursor to a linguistic dictionary.

And he's particularly interested Native American

languages, California Native American languages,

so there might be, in his case,

17,000 slips that never actually became a dictionary.

And he'd like to (?) them, and use them.

And he did exactly that, using the exactly

the same firmware (?) plus a cardboard box.

We have quite a number of disabled people

on the forums, and this is one of my favorite stories

This man's name is Tristan

he's a mechanical engineering major

who built his own scanner

because he has difficulty reading with eyes

He's a perfectly normal human being,

hears fine with his ears,

super intelligent,

just can't read with his eyes.

Built a scanner, his computer now reads

his books to him.

Probably the greatest story to come out of

our forum is also

one of, let's say the second person,

Surya Darnu (?)

Suyran Darna is a village official in Indonesia

who wrote me and said

"This is the first time I'm writing anybody

on the internet but

we have this problem.

We have village holy books

which are wet, being eaten by bugs,

and being destroyed by fires and floods.

These books are everywhere, nobody can scan

them fast enough, and scanning them

is destroying them.

I want to make a book scanner

like yours, but we can't afford the cameras."

So we took up a donation, we sent him cameras,

He built a scanner, and

started scanning.

And as far as I know, the work

is ongoing right now.

In many First Nations communities

in the far north of Canada,

it's extremely difficult to get resources

especially books.

The schools there have chronic shortages

of books.

And when they can bring in either books

or milk on a plane, they'll probably bring in milk.

So the University of Toronto started a pilot project

using DIY book scanner technology

to enable these people to copy

their own books.

It's called the On-Demand Book Service

and it's a DIY book scanner

and a bunch of other printing and binding equipment.

Including DIY binding equipment.

Now, there's a lot of people doing things

for themselves, but there's also a lot of institutional use.

Misty De Meo is the

Digitization Assistant at the County of Brant Public Library in Brant County, Ontario.

And she exemplifies the use of DIY technology in libraries.

She digitizes materials for her community:

maps, etc, books of all kinds.

For example militia handbooks.

Missy not only works to use DIY technology in archival fashion

not just an ad-hoc fashion,

but she also is I think at the forefront of digitization

digitizing things now.

Not just historical things,

not just old dictionaries but things we can use now (?).

She also has a great digitization blog.

So, what this comes down to is that

DIY technology sitting still.

We don't have any one scanner

We don't have one project.

We don't have any one software.

We're just continuously producing new designs

new software, new innovation.

And in this case I worked with a man named Dário de Moura

in Brazil, who helped me package

and license under the GPL the artwork for my

first laser-cut scanner that goes into carry-on luggage

and we've also produced now

ruggedized scanners that can be used

all over the world

and what has come up over and over and over again

and this is the really important point

is that there are all kinds of situations

where by social contract or circumstance

economic circumstances, you name it

you can't just bring in

they can't buy a perfect scanner for $10,000

these are book scanners that you can't buy

for people and situations where you can't afford them.

And that is where DIY Book Scanning technology fits

and that is absolutely what it does best.

And our goal as a community

in sharing these things absolutely wide open, no secrets,

is that we help people help themselves.

They allow people in situ to digitize things

share things with their community that they can't share elsewhere.

I know it may seem a little foreign from here to the Googleplex,

but there are many people who believe

that Google's of the class actions in Google Books search

compromises everyone else's fair use rights,

and I'm one of them.

And there are many people who feel that their data in general

can't be trusted in the cloud.

And just for some examples of that

there are personal journals,

I don't know, porn collections,

medical records, and other things which can't be shared widely.

Particularly religious books.

And in these situations the most important things we can do

is put technologies together that people can use to help themselves.

So going forward in a world which is irrevocably changed

by the diffusion and wide distribution

of e-reading technology by amazing, incredible services like Google Books,

probably the best thing we can do

is empower people to go on with reading

on their own terms, in their own place.

One of the ways we can do this is to iterate

on designs like this and share them freely

so that anybody with access to the machinery

will produce them.

This machinery needs help.

It's too complicated, it's too expensive,

and it's widely applicable just yet.

But we're just a few steps from each thing.

But the reason it is that way is, we were in a hurry.

So, I'm personally pursuing this next version of the Standard,

I'll be sharing it with the world.

I hope to see all of you at the DIY Book Scanner forum, where this works out.

Thank you.