Pictures in Print - Image capture

Scanning

Resolution of acquired images
The image is scanned at a resolution of 600dpi and saved (as a tiff - see image storage - file formats). If it is a topographical print it is scanned at a depth of 256 (grayscale in black and white, colour only if necessary). While there was some debate about this, it seemed that any higher depth or the use of colour for black and white images would detract from the original object created by the artist and throw too much emphasis on the state of the individual print, as discolouration such as stains or foxing and the ageing of the paper would be given more prominence. It also allows for a far smaller file size.

Maps tend to use colour to convey information, so unless no colour was visible in the map, colour scans were made. Also, to allow for the greater complexity of detail and the need to distinguish small symbols and text, the standard depth scanned was 24 bit. This is probably excessive, but ensures a good starting image (but a rather large file). The size of files generated by this method has tended to be excessivetoo large with the bigger maps. Although the pieces of map are stored on CD at 600dpi, they have been reduced to 300dpi before re-assembly. The computer was simply too slow manipulating such large images, and eventually the choice of zoomable image software placed a limit on the file size at 500MB so larger files proved superfluous.

Using the scanner
The scanner has a SCSI connection which it uses via a TWAIN interface. This is generally reliable (but ran into conflicts with the USB connection) but no matter what Windows claims about multitasking it is not desirable to run other programs while scanning, as the process takes over the lion's share of the resources available to the machine. As scans can take several minutes it is very useful to have another machine available to run additional tasks in the mean time. Having been run on Windows 95, 98 and NT4, the scanning process was found to be most stable on NT, and on less stable platforms this proves an important issue. When the scanner crashes, not only do you lose the current work, but if you allow the Scandisk program to run (as it does by default on 95/98 after a system crash - NT should not crash, just the application, but this is not always the case) this can add 5 - 10 minutes to the restart time of a machine with plenty of storage space. Current developments in operating systems and connectivity - such as the USB or Firewire protocols, may improve this situation.

Scanning large images
While an A3 scanner can manage most topographical prints, maps can be far larger. Two solutions have so far been used for dealing with large items: in an ideal world a high specification digital camera would be resolve this problem. For items up to about A1 a clear acetate folder backed with stiff board has been used. This has crosshairs marked on it dividing it into rectangles just under A3 in size. If the item is firmly held in the folder it can be moved around the scanner until all parts have been scanned. The image can then be reconstructed by adding the images together as layers which can be adjusted until the crosses (and any other parts of the image) match. By making the crosshairs fine, and if possible in a translucent colour not used in that part of the map, it is possible to remove the crosses afterwards.

Many maps were dissected when bought; cut into smaller pieces and mounted on a cloth backing to make folding them more easy. This is generally a good thing for scanning, as these divisions make for manageable sections. The entire map can then be re-assembled from the scanned pieces. While this makes scanning easier, it does raise some problems. Often the pieces have not been mounted square to each other, stray threads from the linen backing snake over the map like ephemeral roads, or the edges of each dissection have crumbled through repeated folding. Where stout boards have been stuck to one corner of the map to form the cover when folded, their weight can put considerable strain on the map as it is moved around the scanner.

Both of these processes rely to some extent on keeping the item square to the scanner (although the first is more forgiving). This is often difficult, as aside from the awkward shapes of some items, they are frequently not cropped squarely or even appear to have been printed at a skew.

The upside down scanner solution
One further, extremely basic solution was arrived at. An A4 scanner was turned upside down, without the lid, and laid upon the piece to be scanned (with a sheet of transparent Mylar to protect the paper if necessary). This relied upon several factors: the scanner had to present a smooth surface to the paper, so any protruberances had to be removed and joints protected with tape. The scanner had to work upside down, which it did - where scanners move the scan head along rails this is not a problem as it will still travel along a flat plane. In our case, the setup worked very well, but obviously there is the potential to wreck a scanner, so careful thought should be given to the internal mechanism. For our project, it solved two problems, coping with small prints in very large albums which would have caused injury to binding (and operator) if they had to be turned over and laid face down onto a flatbed. The other use was in dealing with larger maps: once a sheet of Mylar had been placed on the map and some parallel guides clamped in place, the upside down scanner could be moved over the whole sheet scanning it piecemeal.