I've been digitizing large swaths of my library for the past few months now. I use a big guillotine-type paper cutter to amputate the spines off of the books (usually manually cutting off the covers first so they don't get truncated on the side), then feed them through my ScanSnap at "Best" quality so an average hardcover-sized page comes out at about 1600 x 2600 pixels each. That's about 500KB if it's a page of text, and a book can run 50-150 MB. I save them as jpegs just so I have more flexibility in editing later, as I can always combine them into PDFs and OCR them at my leisure. Also with some books I want to run a batch transform on them in Photoshop to get rid of the page yellowing or some other repair work, and that's easier to do with individual pages.
But though I have lots of books scanned, I'm just beginning to move a few of them onto my Kindle, and I haven't yet figured out the correct PDF settings to make the files look good. PDFs that look good on screen (and are even readable on my iPod touch) often look wispy on the 16-shades-of-grey e-ink screen. Some look good, but not others. I'm assuming that adjusting the contrast will help, but haven't found the magic settings yet. Also I'm not sure if I want to shrink the file resolutions down to match the Kindle's resolution, or if I should let the Kindle resize them on the fly.
I've also been able to reduce e.g. a 50 MB book down to about 15 MB through PDF optimization settings, but without a satisfactory "before" result it's tough to decide on how far down I can go for the "after" version.
OCRing books into light .pdb files from html sources would be nicest, but that requires so much proofreading and correction that I've only done it with a handful of my favorite books so far. Joshua Tallent's "Kindle Formatting" has been very useful for those. Most books are destined to remain as images/PDFs, if I can just figure out how to display them properly on the Kindle.
My guillotine and pile of corpses:
One week's worth of spines:
