Self-Adjusting Spine Support

[ateeq85]

all of the ideas for the spine support are great however there is a problem that is not taken into account. whatever solution is used to adjust the spine as the book moves to another end one camera gets more of a fuller shot of a page then the other camera does so it may result in the font on one page looking bigger then the font on the other. the link to the book scanner used earlier seems like it operates the same way a flat bed scanner operates and is not using cameras so that issue doesn't become a problem.

it seems one solution would somehow involve the camera moving with the position of the book to accommodate the plattens position on the spine so both cameras retain the same zoom on each page. maybe the cameras being connected to the sides of the cradle wings so that when it lowers the camera lowers.

better yet a way to have the camera connected to the platten like the book liberator scanner was
http://www.youtube.com/watch?v=XNfQb0NZjWY

http://bkrpr.org/doku.php?id=start

with this design it isn't the cradle that is adjusting but its the platten that is you can kind of see as it goes past the middle of the book it leans. what if the new build could in some way be redesigned or modded to have a sort of balance shift in it the same way a portable scanner like this works.

The builds Dan has designed so far seem to concentrate more on the cameras focusing on books position on the cradle however when a book goes off center especially in the new build which is made to slide the cradle goes off center if you let it slide to adjust with the platen or the spine is adjusted and one side is more in zoomed in then the other one.

with the book liberator design the platen camera and cradle act like one unit so as the book gets further from the center whether its in the beginning or towards the end the platen and camera adjust.

could the old platen design from previous models be used but with a bungee like in this one so it will still be light? the platten could be on sliding tracks in a track pattern that would be a v shape like the cradle outline and have bearings so it could slide easily to adjust with the book and the cameras could be attached to the platen so wherever you are scanning in the book when you push down on the handle the platen would slide with the track to the needed angle the books spine is at.


also here is a link of a video to the atiz diy build explaining how their platten and cradle work when a book goes off center
http://vimeo.com/4179744

[daniel_reetz]

Good observations, but one correction - in the new hackerspace build the relationship between the cameras and glass are fixed just like in the book liberator or similar designs. No matter what happens, the image of the page should remain the same size.

If you are experiencing otherwise, post some example images and we'll figure out what is going on!

[dpc]

From looking at the hackerspace scanner design, it appears that the thickness of the book could vary the position in which the page contacts the platen glass, but not in the manner described by ateeq85.

Since the cradle carriage side arms are attached to a lifting arm that travels in an arc, thicker books would contact the platen at a different "front-to-back" position against the glass than a thin book would.

From a quick look at the DXF files, it would seem that worst case would be less than 1/2" of difference from the scan of a 3" thick book, vs an empty cradle. Might be slightly more or less as I don't know the angle of the lifting arm when an empty cradle contacts the platen.

You shouldn't see this difference across the course of a scan if you've got your spine leveling sorted out. It's only between the scanning of books with different thicknesses (which can be easily compensated for by adjusting the camera position before starting a scan).

[ateeq85]

It seems my observations were off it wouldnt be the first time, however i am encountering a regular problem despite me checking everything I could think of both cameras are both zoomed in equally and i amusing identical cameras however the font seems to come out smaller on the right camera. When all the post processing is done it seems like its a smaller book next to a bigger one all the camera settings are equal everything is the same. What could be causing this is it a issue related to spine adjustment or the plattens,

I am thinking of just seeing if I can get a bent plexiglass to place to see if that helps and just have to not use the platten adjustment for different books that need it.

And I have scanned at least 1500 pages so far so this is a reacurring issue for me.

Its like the left pages have a bigger border and it isn't related to scan tailor it's origin is in the scan I think it may be related to the cradle or platten but can't figure out how.

Sometimes i see the pages may not be pressing against the glass and will result in curved pages. i think the a good solution for adjusting this is to somehow have something pushing the book up against the platten so when one side of the book has more pages then the other it gets pushed against the platten. maybe something pressing against the cradle wings pushing them towards the platten as they go up. Maybe this would aid in spine adjustment too.

[dpc]

ateeq85,

It sounds like the cameras are not set the same. Do you see this difference across the entire book?

Just as a test, have you tried swapping the cameras to see if the problem will still reproduce? Even easier, just put two pages of notebook paper that are roughly the same size as your book's pages on each side of the cradle without a book and see if the scanned images of the left and right pages look the same. That should quickly tell you if it's a camera problem or something to do with the way the book's pages contact the platen.

If you scan the pages at the exact center of the book (i.e. pages 200 & 201 of a 400 page book), the book spine should be flat/horizontal. This position is pretty easy to set up, so try to reproduce the problem you're seeing with this orientation first.

If you have the book spine supported at the correct angle and have solid contact with the platen you shouldn't see any curve on the pages except in the gutter between the pages (not on the text).

Posting the images of the left and right pages before any sort of post-processing is done would help to determine the cause of problem you're seeing.

[LA2]

also here is a link of a video to the atiz diy build explaining how their platten and cradle work when a book goes off center
http://vimeo.com/4179744

Here's a glimpse of how the Internet Archive's "Scribe" does the same thing, http://www.youtube.com/watch?v=qUbeWM4Kxmg

[sethg]

It's been a long time since my last post, but I have some new ideas. I completed my first build (I'll start a thread for that sometime soon). To save time I went with a pretty standard cradle design, but now that it's done I can see much more clearly what the issues are. I'm working on scanning a book that was printed in Wittenberg in 1664. Since the book is pretty old, it is showing signs of age. The whole book is warped, but the acid-free paper has held up great.

There is a problem though. With some books like this, there are margin notes that were printed with the book. Some of those notes go pretty much straight into the binding. I'll attach a couple pages to show what I mean.

This first one shows the margin notes. This page is earlier in the book, fairly close to the front cover (the book is about 1200 pages).

This second page is in the book a ways and you can see the notes are getting lost. I'm also getting an annoying line projected on the page as the glass focuses my light there. As I got farther into the book, the curve of the pages made it impossible to get the v-point of the platen into the crease.

So, what can I do? Well, I have one idea, but before I put any work into it, I'm hoping for your thoughts. I don't remember where, but somewhere there was a video of a robotic scanner that started with the book open wider than the platen, and then pushed the sides up against it. I'm wondering if we could do somthing like that in a simple fashion, perhaps using paint rollers on the sides and a small platform in the center. Each piece then could be suspended independently with springs. Here are some very basic illustrations:

1.png (29.82 KiB) Viewed 10366 times

4.png (121.58 KiB) Viewed 10366 times

5.png (107.95 KiB) Viewed 10366 times

What do you think? I'd appreciate any thoughts or suggestions.

[dpc]

The issue with the text in the gutter aside, how much of this problem you're seeing is related to this particular book being warped/lumpy? From the photos you've posted, the platen is clearly not going into the gutter fully and pressing the pages flat. This is evident from the white reflection line cast on the page from the edge of the glass. That should be a straight line (and should be down in the gutter instead of cast on the page).

It would really help figure out what is going on if you would describe (or post pictures!) of your cradle and platen setup. One of the first questions that I"d ask is do you see this same problem with any thick book, or is it only with warped books such as this one? Would your current cradle/platen setup prevent your platen from going fully into the book's gutter and pressing the pages flat with any book? If you held a book in your hands and pressed it up against your platen, can you get the platen apex to go fully into the gutter? If you can't, then there's no cradle design that's going to do that with an old book such as this. You might have to create a single-page scanner and concentrate your efforts on scanning a single page at a time.

The problem that I see with your proposed cradle setup design is that the spine of the book needs to change its angle and rotate through the platen angle throughout the scan. In your drawing, you have the book in the center position where the spine is horizontally oriented. When you begin a scan, in order to get the platen to go fully into the gutter, the book spine will be in the same orientation as the left side cover of the book (it lies flat against the left cradle side). Conversely, when scanning the last page of the book, the spine should lie flat against the right side of the cradle. With thinner books (< 1") you don't have to worry about this sort of thing that much and could probably get away with a simple v-shaped cradle, but for thicker books the spine needs to rotate from one cradle side to the other during the course of the scan. Is that spring loaded spine support in your design going to allow this?

I believe the video of the scanner that has a cradle that pinches the book into the platen is here: http://www.youtube.com/watch?v=Tm_e3AeoaU4 . They are using a bellows system to apply pressure to the bottom of the cradle sides and pinch the book.

[sixtysix]

I gather that that particular scanner costs $175,000. We soldier on.

[sethg]

Thanks for the suggestions. I'll have to try pushing the book against the platen by hand. I'm not sure if any method will get at all the notes. As I said before, when I have some time to organize my pictures I'll put together a thread to show off my build and share ideas, but for now, here's one picture of it:

The concept is the same as the hackerspace scanner, but since some of these books are huge it's a good deal bigger. Plus, I didn't have access to a cnc router, so it's all done by hand. My platen follows the suggested 100 degree angle, and I must say that I'm extremely happy with the quality of the images. This is intended for the large collection of rare books at Wisconsin Lutheran Seminary, so it needs to do a good job. I fear it may simply not be possible to get all the notes, since some of them go very deep into the gutter.

I'm using a pair of Canon T2i cameras. It took me 2hrs and 15 min of actual scanning time to do all 1600 pages of that Wittenberg book. That includes time spent making a lot of minor adjustments, but I stopped the timer every couple hundred pages to copy the files off my 8 GB SD cards. They fill up fast at 24 MB per raw image.

I realize my book pressing design posted above has problems, but I really like that concept. Think about it if you were placing a book on an upside-down platen. How would you do it? Would you open the book to the exact angle and then press it down? No, you would probably open it past 100 degrees, seat the gutter against the v-edge of the platen, and then press the sides down. Maybe there is some other way we could implement this concept. The only issue I can see is that for a very thick book you may not be able to press the sides all the way against the glass without destroying the binding (or breaking the glass).