I did a 60 page magazine type manual from the 40's last night.
Scanned it in via flatbed scanner.
Ran Omnipage to OCR and make PDF
The only words that it seemed to stumble over were hyphenated words. It also picked up some of the line drawings as words, but the editor/processor prompted me for input.
All in all pretty impressed with the software. A little slow, but not unbearable.
Then I tried to OCR a few select pages (text only, text and graphics, and a full page table) with a digital camera.
The results were much worse. Even with the option to correct keystone and warping, It was 'choking' on significantly more words. I COMPLETELY blame my camera (only 8MP and marginal lighting) as I don't have a stand with lights etc. I hope to see better results with a better input image.
Omnipage is a bit pricey, but it handled mixed pages and tables quite well. Since I most of what I will be scanning are mixed pages, diagrams and drawings I need to preserve the layout and I was quite happy with the results.
I still have the images so I plan to try scantailer and see what happens. But I would say that if you can get a slightly older copy of Ominpage cheap (the older versions don't have PDF support, but the scan engine works fine... I have a copy of Omnipage 3 that came with my used scanner off e-bay, and other than not having PDF output (which can be gotten around) its just as good as the newest version.
