I believe the original images were the same as these (page 18-11 was missing too, I had to rescan it from my print version): http://www.scribd.com/doc/13252808/Feynman-Lectures-on-Physics-Volume-3
FYI, there is an annoying bug in AA9, which occurs on some pages of ClearScan OCRd pdfs when using the text touch up tool, preventing you to do corrections. It's described here, with a work around: http://forums.adobe.com/thread/514567 In short, one must use Acrobat 8 (other programs may work too) to do text touch up on Acrobat 9 produced Clearscan pdfs! Also, I have found that OCR artifacts are produced on some figures, sometimes to the extent that the figure is unreadable. That's very rare (one in several hundred in the Feynman books), but unless one is prepared to check every image in a large file (and then "declare" the mangled ones as images on the original file before re_OCRing, I believe that's possible in Acrobat) maybe it's safer to stick to the less glamorous but more faithful "searchable image" format for now. Michel 2010/4/11 Jed Rothwell <[email protected]>: > Michel Jullian wrote: >> >> Jed, it seems the way it performs depends on the original document's >> characteristics (resolution, fonts, multiplicity of fonts maybe?). > > Yes. There is remarkable variability between different kinds of documents > and different versions of Acrobat. I cannot figure out what all of the > controlling parameters are. It also depends a lot on the number and size of > the figures, and the amount of noise in the scan (extraneous dots). > >> >> In the case of the Feynman Lectures on Physics, volume 3 (quantum >> mechanics), of which I made a searchable backup of my print version >> from an image format pdf found on scribd : >> >> >> 1/ The ClearScan'd pdf file size was several times *smaller* than the >> original image-only pdf found on the web > > What is the URL of that file? I will run it through a variety of different > Acrobat programs. If ClearScan reduced the size I expect the original was > made a long time ago with an early version of Acrobat. > >> >> 3/ The OCR quality is much better than what you got with your EPRI >> document, without any "touching up" I got, for page 1-1 : > > That's probably a function of the quality of scan. A good quality scan of > cleanly printed text without a skew and without much noise will OCR far > better than an old one like the EPRI document. > - Jed >
