Looks incredibly serviceable and well engineered. I would expect reasonable and consistent results from the rig. The biggest question would come down to the cameras.
With these kind of rigs (two cameras, not computer controlled, no computer display) your big potential sources of error are either accidentally failing to trigger one camera or cameras losing focus on the page (especially if you are at something like the end of a chapter where there is often empty space in the middle of the page where the camera's auto-focus area is). His solution of using the IR remote should significantly reduce the issue of failing to capture on one camera. Cameras exist with manual focus settings, but they are often pricier or too old to reliably find one worth recommending to others. The CHDK alternative firmware for certain cheap Canon cameras generally adds a manual focus option for the less expensive cameras (though the individual features depend on who is making the firmware build you get).
Another option worth investigating is the newest Raspberry Pi camera modules with external lenses. Those should give you manual focus and the ability to build up an automated workflow you like around things like moving files around and any pre-processing you need. An ~9 mega pixel camera gets you 300dpi resolution on a full sheet of A4 paper, which is a lot more than most books.
Right, step 1 -> get page images, step 2 -> author images into book file. While OCR is obviously useful for search, a rotated phone screen will let you comfortably read a pdf book just fine unless you are talking about something like a textbook, in which case you probably wanted a tablet anyway.
I wrote up a guide on the authoring process using FOSS tools for some Digital Humanities folks a couple years ago: https://github.com/wikey/bookscan
It gives some background on the problem and covers a Scantailor (page crop, rotate, deskew), pdfbeads (compression, book metadata) authoring workflow, with pdftk for some general odds and ends.
For those that are digital subscribers and wish to suggest exactly this turning off of 3rd party tracking for logged-in users, this looks like the most relevant email address: membershipsupport@theguardian.com
Be sure to include your subscriber/membership number in that email, which you can find in your profile https://profile.theguardian.com/
I can confirm that. I specifically signed up for FirstTech for their Chip and Pin cards. Their Mastercard is quite good for traveling and has Chip and Pin.
It was great when I was traveling in Iceland where pretty much every transaction was made with my card. It's a regional credit union, but I only had to visit a branch in person one time to change the PIN on my card.
First Tech security is pretty top notch. For online trnsacations over a certain amount they would ask for password verification.
I just wished online transactions owned it up for optional two factor online transactions, pin card transactions and SMS verifications for all transactions above $500.
These steps are generally split between the image processing ones, the OCR one, and a final step of combining/compressing pages into a single "bound" pdf/djvu file, at least if you are looking to use FOSS software.
For image processing, take a look at Scantailor (https://github.com/scantailor/scantailor/wiki), which will handle all the image processing steps for you and output images that are ideal for OCR.
I have not done OCR on mixed language text but I will say that tesseract has been under active development for years and does continue to improve.
A lot of those tools have received little development in the past couple of years. They tend to do what they do well and reliably so don't let that put you off, though anyone interested in adding to the developer pool would certainly be welcome.
For more general information and especially background discussion, take a look through the DIY Book Scanner forum: https://forum.diybookscanner.org/
The trouble is that only the most engaged minority of users would be willing to pay for the service directly. What I'm interested to see is whether we can build services where the people who care most end up covering the costs for the rest of the users. So you pay $5 for a facebook-like service that then provides free services to everyone who knows you. You would have all the same funding problems as other public goods but there are plenty of them that make it work. Of course, even if something like a public radio or other public support model would fund a social network sustainably, you would still have the network effect problem in terms of getting people actually switched over.
Optical ballots provide a verifiable trail that can be used to audit the electronic counting machines. Seems that whatever system we use could benefit from some randomly selected mandatory audits/recounts at every level.
Many of the journals currently publishing these kind of Registered Reports only publish special issues, which might help with your concern about journals being given control over what work is done in a field, though it also limits how much of the literature can benefit from the new approach.
Another way of thinking about a journal's incentives is that, by moving the primary peer-review stage before a study has actually been conducted, you greatly expand the number of studies that a journal will interact with, which may provide a larger and more valuable role for entities like journals in world where actual publishing is a trivial matter.
The problem with using the blockchain for this is that there are legitimate reasons to withdraw such a registration ranging from misstyping (s/increase/decrease anyone?) to, depending on your kind of study, the inclusion of personally identifiable information. The solution currently being pursued is to expand pre-registration, a practice from clinical research, into other fields.
In pre-registration a trusted intermediary is used to store a read-only copy of your study design and materials, ideally including a pre-analysis plan that specifies the analysis you will run for hypothesis testing (so that we can cut P-hacking out of the picture at the same time). That intermediary can allow researchers to withdraw registrations while preserving a stub that shows everyone a registration used to be there and why it was removed.
That is how pre-registrations work on the Open Science Framework (https://osf.io), a cross-disciplinary FOSS web tool run by the Center for Open Science.
With these kind of rigs (two cameras, not computer controlled, no computer display) your big potential sources of error are either accidentally failing to trigger one camera or cameras losing focus on the page (especially if you are at something like the end of a chapter where there is often empty space in the middle of the page where the camera's auto-focus area is). His solution of using the IR remote should significantly reduce the issue of failing to capture on one camera. Cameras exist with manual focus settings, but they are often pricier or too old to reliably find one worth recommending to others. The CHDK alternative firmware for certain cheap Canon cameras generally adds a manual focus option for the less expensive cameras (though the individual features depend on who is making the firmware build you get).
Another option worth investigating is the newest Raspberry Pi camera modules with external lenses. Those should give you manual focus and the ability to build up an automated workflow you like around things like moving files around and any pre-processing you need. An ~9 mega pixel camera gets you 300dpi resolution on a full sheet of A4 paper, which is a lot more than most books.