Thursday, July 20, 2006

Photo scanning tips

Several comments regarding the excellent write-up, "8 Blunders People Make When They Scan Photographs ... And How To Avoid Them" by Sally Jacobs "The Practical Archivist."

The goal when scanning should be to capture all the detail in the original image. Scanning at 300 dpi is not sufficient, as a general rule, although it may be just fine when scanning an enlargement. The technical requirement (Shannon's sampling theorem) is that the sampling rate be at least twice the highest spatial frequency (the Nyquist frequency) to avoid loss of information and prevent introducing artifacts. Scanning at 600 dpi is barely adequate to capture the detail in 19th century prints such as stereoviews or wood block engravings. Scanning at 300 dpi (at 100% of the original) will also not provide sufficient detail when printing even a moderately enlarged copy from a small picture (such as would be desired for printing a copy of a detail from a stereoview or CDV in a textbook), and would provide disappointing quality if used for an exhibit display print. Scanning engravings at 300 dpi will result in severe Moire pattern artifacts.

TIFF images may be uncompressed, but TIFF allows a variety of compression schemes. Some are lossless such as LZW (based on substring encoding), others such as jpeg (based on Fourier transforms and cosine waves) are lossy (yes a TIFF image can use jpeg compression).

Having multiple backups is not sufficient – to survive long term they must be in different locations that are not subject to a common disaster such as a fire, flood, earthquake, war, etc. Also, each time that a backup format (whether the file format, the physical medium, the electrical interface, the computer instruction set, or the operating system, etc.) is superseded by a new one, you likely will need to make lossless copies into the new format or lose the pictures. There is a high risk that even if the physical medium, such as a disk survives with the data intact, it may be impossible or impractical to find a working device that can read the long obsolete format. How would you go about reading a well preserved DECtape or 8" floppy disk, for example, now ... in a hundred years? Where would you plug in your SCSI drive or RS-232 cable, now that your computer doesn't use such interfaces, and how would you obtain software drivers that work with the current operating system? Will your great grandchildren even know that some unfamiliar object contains pictures, or remember the password? ... Will they realize that the postage stamp size SD flash memory card that they are about to toss in the trash has a thousand pictures on it?

Also see: 1, 2, 3.