Words of wisdom

[Karyudo]

From a geometry point of view, I really like the idea of the cameras being mounted rigidly wrt (with respect to) the platen.

Right now, it seems like people are using camera adjustments that can't separate aiming point (i.e. centre of the page) and image size explicitly, and without introducing some skew. To get the sort of acceptable aim/angle/size wanted, people are looking at 6DOF arms. Obviously this produces good results when set up right, but it seems fiddly to get the camera aimed properly, and it doesn't seem very repeatable. Also, any skew has to be removed by software.

To have the best possible scans (ignoring reflections of the camera itself for the moment), I would expect that you'd want to position the camera exactly square to the centre of the page, so that any spherical aberration (barrel distortion?) affects the edges about the same all around. Also, it's got to be best to move the camera as far away from the page as possible, and then zoom in—optical zoom only, of course—to get the view with the squarest possible edges (i.e. smallest amount of spherical aberration). This is what I'm pursuing by tabulating view angles and heights for the A590.

To put this into practice and get fine adjustment on the size of the page, you'd really want the camera to travel along an axis perpendicular to the centre (or pretty close to centre) of the book. To have the ability to accommodate different size books, you'd also have to be able to move the axis closer to or further away from the bottom of the platen V. If I'm counting right, that's only 2DOF, and it implies a 2D orthogonal coordinate system that's 45° to what everybody has been using thus far.

I'm guessing it would be somewhat more difficult to implement my alternative take on this, but I have to imagine it will pay dividends down the road, in terms of setup time and repeatable scan quality.

In my purely conceptual design thus far, I'd have a linearly-moving platen cantilevered with drawer slides from a central light-and-platen support at the back, and adjustable camera mounts following my rotated coordinate system fixed rigidly to the sides of the platen. Whaddya think?

(I realize I'll have no credibility at all until I can either draw or build something to illustrate my points, and gain some cred as a builder...)

[spamsickle]

In theory, I expect I'll want the "film plane" of the camera to parallel the "page plane" of the book it's photographing, so I'm currently planning on using a couple of angle brackets to give me the same 45-degree pitch the sides of the platen will have.

bracket.jpg

(52.66 KiB) Downloaded 5809 times

But your "fixed so they don't move" comment made me think -- I probably don't need that screw between the brackets. There isn't any need to adjust this angle that I can see. Six degrees of awesomeness? Pfffft. I'll take no degrees of awesomeness, if it will give me consistent, reliable, repeatable rectilinear page pics. If I had a welder, I could just tack those brackets together, but since I don't, there's probably some kind of epoxy or bondo that will do as well.

I"ll probably go with the screw initially (because there's likely something I'm overlooking), but this is my plan right now.

[spamsickle]

I really need to email Ian from the BKRPR group again. A book scanner with cameras mounted on the platen would represent a kind of hardware middleground for our designs.

Yeah, it's too bad they don't have a forum there, the lack of one severely limits collaboration.

[daniel_reetz]

They have a Google Group but it's pretty quiet. I think (and I suppose this is now obvious) that a forum is a much better approach.

[jradi]

I almost forgot about that laser thing. Hmm.... Now I want to think of cool laser projects...

[jradi]

My design for the arms used two ultrapods (http://pedcopods.com/) that attached to vertical arms. I've had great luck with these as pocket tripods and I used to love them. That is, until this project.

My cameras mount is located off to the side, which makes it want to tilt down against the base and unscrew itself. No matter what I do, I can't get these things to hold firmly enough for the course of 10-15 minutes. I think they just weren't designed for that. They're great little pocket tripods for a quick shot, but not so great over a long period of time. Hopefully, I'm going to try that double L-bracket idea (shown above) a shot. If that doesn't work, I'll just have to redesign my arms before I do any more scanning... (6DOAA?)