1) If I hooked it up to a timer or a switch with the right camera and good lighting, how good do you think the results would be?
If you have a camera with CHDK, my choice would be a short one-shot timer script that waited a second after you pressed the button to fire the shutter. This would get good throughput with the "pick it up and put it down" hardware, and should eliminate any vibration that might be caused by pushing the button.
If you don't have CHDK, I'd just go with pushing the button, though if you can get a remote release for your camera, I'd do so.
As for the results, it probably depends mostly on the lighting. You shouldn't depend on the camera's flash in this configuration, because if the material you're scanning is glossy, you'll get glare that will wash out content. Overhead lights may cause the camera itself to cast a shadow on the content. I'd take the shadow over the glare, but if you have some ideas for auxiliary lighting, that could make a big difference in the quality you can get.
o3h1p wrote:2) Does anyone know of any other portable projects people have done?
I don't. I've had some ideas for portable hardware, but I haven't gone into the field with anything yet.
o3h1p wrote:3) What is the drawback from this setup given that i don't have old books to scan.
The biggest drawback, other than the lighting, is that it's difficult to get consistency from one page to the next with a camera that's attached to moving parts. You might be able to improve this design with some kind of frame around the book, that will guide its placement from shot to shot.
One other drawback I can envision is that the size of the books you can scan may be limited to a certain range. A book may be too small for the stand to hold open, or so large that the stand blocks content.
o3h1p wrote:4) Would there be any problems if the pages are not lined up in the exact same position as the others? I assume the software would correct for that?
This is more of a problem with the "single command" kind of post-processing you propose than with something like Scan Tailor (which is what most people here use at some point) which handles most alignment problems transparently.
o3h1p wrote:5) Any camera recommendations? I'm assuming I should get something compatible with CHDK.
See the "camera recommendation" thread. I'm using CHDK-enabled 8 MP cameras, and they seem just fine to me.
o3h1p wrote:6) I am comfortable with both windows and linux. I assume that it should be easy to just throw all the images into a directory and a script can be easily designed that will rotate,crop,cut, and assemble with a single command. Has this been done? Maybe the final OCR can be done with AbbyReader or something I guess.
It has been done, but as I say, most people are using Scan Tailor because it does so many good things. Check out Misty's PDF builder too, or roll your own.
In general, most of us aren't doing single-command post processing at this point, but that may be more common in the future. Maybe you'll help to bring it to fruition.