This is something I know about, too, having once supervised this lab (in its previous incarnation as part of UVA's eText center): http://www.digitalcurationserv

This is something I know about, too, having once supervised this lab (in its previous incarnation as part of UVA's eText center): http://www.digitalcurationservices.org/ http://www.digitalcurationservices.org/. We used a pair of PhaseOne P40 scanning backs[1] with Hasselblad prime lenses, mounted aiming directly downward at a custom tabletop with adjustable book mounts. We used standard studio lighting (not strobes). Software-wise, for books and manuscripts we created batch jobs in Photoshop to minimally post-process the images (adjust levels & contrast, rename files, not much else). For 3D artifacts, everything was custom & manual. Files were scanned into a pair of Mac Pros for processing then burned in duplicate to archival CD for filing and shipping to the Indian company[2] that had been contracted to transcribe and encode (SGML at that time) the text. We chose manual transcription over OCR because it gave us roughly 99.9% accuracy versus 95% accuracy (at that time -- 1997-1999), and Apex only charged $.03/pg. That rate bought us two transcribers encoding each page and then compared against each other's work for error correction. In general we were quite happy with them. Not including the staff, I think there was about $75k in hardware and about $5k/yr in consumables (mostly repairing book stands, buying CDs and replacing CD burners we wore out). If anyone is curious, here are two projects I worked on: Early American Fiction (1789-1875): http://etext.lib.virginia.edu/eaf/ http://etext.lib.virginia.edu/eaf/ Walt Whitman Leaves of Grass archive (a whole bunch of versions): http://etext.lib.virginia.edu/whitman/ http://etext.lib.virginia.edu/whitman/ Note that these were both created 12-15yrs ago and offered both high quality scans and searchable text, and even basic comparison (split screen view of two texts: http://etext.lib.virginia.edu/whitman/whitframe2.html http://etext.lib.virginia.edu/whitman/whitframe2.html). Subsequently, UVA's library has joined TEI and I'm sure things are much more modern now, but I wanted to provide a little more flavor to what you posted, with some more examples. Obviously, manuscripts in any language are time consuming to transcribe. They are often in poor condition and handwriting can be downright illegible, and don't get me started on issues with accurately transcribing original authors' own grammatical and spelling mistakes! Argh! [1] http://www.phaseone.com/en/Camera-Systems/P-Series.aspx http://www.phaseone.com/en/Camera-Systems/P-Series.aspx [2] http://www.apexglobal.in/apextranscription/index.htm http://www.apexglobal.in/apextranscription/index.htm