All the images of actual architecture for the Split project have been captured from 35mm slides or negatives via a S-VHS camera feeding into a 24-bit frame grabber, and thence to disk, where they are stored in JPEG format in spite of the fact that many of them are monochrome, or grey-scale. Because a 35mm slide has far more definition than can be captured in video resolution, my strategy has been to use a Sony PHV-A7E video camera for copying. The slides (or negatives) are placed in a holder, over constant white light, and the result projected onto a monitor in either VHS or S-VHS format. The machine offers colour balance controls, automatic focussing, white light balancing, and positive/negative modes - and the lens has 6x magnification. With suitable negatives/slides, therefore, three or more video images can be derived from parts of the area of one 35mm original, without much or any loss of resolution.
Note that because these images (whether a whole slide/negative, or just a detail) are presented to the framegrabber as video signals, albeit at S-VHS quality, none of these images can therefore be better than video resolution - but this is, in my opinion, quite sufficient for teaching and basic research purposes.
I have used a scanner only for capturing plans and the prints from the various out-of-copyright architectural manuals I have used, in an attempt to capture sufficient detail to make them readable (although, equally, I might have captured them using a video camera). Scanning positives is time-consuming, and storing the results is disk-consuming. Video resolution seems to me to give the best available balance between quality and efficiency - for both the digitizer and the end-user.
Frans van hoesel.