Spectacular Google Books failure

[daniel_reetz]

This caught my eye:

google_books_clip.jpg (74.42 KiB) Viewed 7948 times

It's from this book, which shows a few interesting things about the GB hardware. We've always known that they are platenless, but it appears that they use a heavy steel clip to hold the book in place. They also use visual fiducial markers - my guess is that those markers are in inch or half-inch increments, to auto-calculate DPI/book size. That's in addition to the infrared structured light they use to deduce the shape of the page.

Be sure to check out the reviews.

[daniel_reetz]

Also, given that they seem to be imaging the entire cradle area every time, it's safe to assume they're using a fixed focal length/zoom at a fixed position. That is the most bombproof way to scan, if you have enough resolution to work with.

[dingodog]

googlebooks scanning technique is awesome!!! (and it is free for libraries, while other *supposed* *specialists* ask for money and make a work many times worst than google) unfortunately managements of *public domain books* is not at same level (public domain book left for years in snippet view... and many other issues, but this is the hardware forum... the wrong place to talk about this problem)

[spamsickle]

I've started going platenless too on the "horizontal" format scanner, at least for books which don't have images that bleed all the way to the edge of the page -- art books, atlases, magazines, etc. I just have a bit of finger in the image holding the page down. It's not always as flat as an image with a platen, but it avoids reflections, and the scanning process itself goes faster.

I've also gotten away from post-processing of any kind. Since I was always saving the originals anyway, I decided to save the couple of hours I was spending per book changing the images into a PDF, and just read the originals. I don't suppose you can read them on a Kindle, but I'm still using the laptop as my reader, so it works for me.