Logo Utrecht University

How to digitise slides. Recommendations and working lists for the reproduction of a very special artefact

Scanning diapositives, and personal comments by experts on scanners

As many archives keep huge amounts of historically interesting 24 x 36mm diaspositives on reversal film stock, a short excursion into this field should not be missing. Scanning specialist Josef Stark recommends to discuss and decide on several aspects before starting any activity: the objectives of the reproduction, the number of items to be scanned, the quality of the scans that is needed, and also their sustainability (which could lead to a division into high and low quality reproductions for sub-collections), the budget in money and time that is available for scanning. The cross-reflection could lead to a number of alternatives: to rent a “slide scanner” for some time or to test a scanning service, when small collections are to be reproduced; for big ones, he suggests buying a (used) magazine scanner which does the work without the presence of the operator in about three hours for a magazine of 50 diapositives (at 3 ½ minutes per scan but it can take much longer according to the chosen setting. This would require a place at a certain distance to other staff members, as the diapositive scanner is relatively loud. Stark’s alternative would be a flatbed scanner, which can scan several items as a batch, is suitable for glass diapositives, as their sharp edges can damage the magazine scanner, copes with medium formats such as 4,5 x 6 (41,5 x 56mm) or 6 x 6 (56 x 56mm) and can be used for very small or heavy frames which often trouble the automatic transportation mechanism of the magazine scanner (see Stark 2015, p. 18-24, 31-32, 35, 36-42, 46-47, 78).

Batch digitisation may be a satisfying solution for large amounts of diapositives. For a flatbed scanner, this can only be recommended if the slides are homogeneous in tonal values. Nevertheless, one should keep in mind that a dispositive scanner creates one file per item. A batch of 6, 8 or even more objects on a flatbed scanner result also in one file. Their separation happens in post production and involves algorithms which can influence the file’s sustainability. The archivists Shingo Ishikawa and Darren Weinert from the National Film and Sound Archive find a parallel scan of several slides at once quite time-consuming (Ishikawa, Weinert 2010).

Hints by a scanner operator:
As to the working power of the computer (if it is also to be used for post production), Stark (2015, p. 50-51) suggests high quality processors (e.g. Intel (Integrated Electronics) series core i5 and core i7 or AMD (Advanced Micro Devices) series Ryzen 5 or Ryzen 7), a hard disk of at least 4GB RAM (Random Access Memory) or better 8 GB and more, a SSD (Solid-State Drive) to upgrade the hard drive, some external hard disks with high memory capacities for storage and back-ups, a monitor with a broad colour gamut and a wide angle of view (aspect ratio 21:9), as it allows to see two images at the same time.

Stark’s recommendations can be helpful as to what to look for. Nevertheless, they aim at a high-speed computer, as needed for gaming, which can be expensive. Besides, the ultra-high speed Solid-State Drive can slow down after a while, when it is “used up”. The technique is based on “[…] a grid of electrical cells to quickly send and receive data. These grids are separated into sections called ‘pages’, and these pages are where the data is stored. Pages are clumped together to form ‘blocks’.” (http://www.makeuseof.com/tag/solidstate-drives-work-makeuseof-explains/) An SSD cannot write data down on whatever empty space there is, as a hard disk drive can; it needs an empty page. The more pages and blocks are filled, the more it has to look for unused ones, transfer content to its “flash memory” (a kind of RAM) before deleting entire blocks to create fresh pages, and afterwards has to fill pages with the saved content; all this is necessary before it can start to record new data. The other point is the fact that a flash memory can only be written on a finite number of times.

As to post production packages of well-known software producers, Stark (p. 255) advices to compare the results with an open source alternative offered for free on the internet (e.g. IrfanView) to be sure that the reproduction is acceptable. His final advice is not to install different scanners at the same time, as this could disturb the USB-drivers.

Personal comments by experts on diapositive and flatbed scanners
A publication on scanning diapositives mentions another problematic point: the image resolution of today’s scanners. According to producers’s leaflets a “CCD-sensor” in a flatbed scanner can produce up to 4.800 spi (samples (pixels) per inch), a CMOS just 2.000 spi per row. When the author André Kramer started a series of tests, he realized that both constructions produce reproductions of lesser quality than real diapositive scanners (often called “film scanners”). Those are no longer produced by the big (Japanese) photo companies (Kramer 2015, p. 127, 129), which, since the 2010s, specialised in flatbed scanners. Only a German and a Taiwanese producer are still continuing in this field (Wagner [2017]). Kramer states that the thickness of the glass pane and the “bad optic system” in flatbed scanners reduce the resolution promoted by the producer “to one third” (Kramer 2015, p. 126). Drum scanners in high end version were superior in resolution (8.000-14.000 spi) and colour depth (48-bit) as they work with a photo-multiplier tube (an extremely sensitive light detector) which have a large dynamic range (Kraus 1998, p. 15). (The “dynamic range” is the difference between the darkest and the brightest point on an object; the dynamic range of a scanner, i.e. its potential to reproduce details in the darkest and brightest areas of the image, is determined by the quality of the sensor and the speed of the lens). Yet, these scanners can only be used for flat, opaque and flexible objects (e.g. lecture texts), are almost obsolete today as faster, less expensive flatbed scanners (high-end version 4.800-5.400 spi) pushed them out of the market. André Kramer, authors of the German manual for the software “Photoshop”, insists on “film scanners” for diapositives. Wilhelm Scheuer and Karsten Kettermann (1999-2000, p. 341) also reject flatbed scanners for diapositives; but they accept them for what they call “Großformatdias” (large-scale diapositive, 9 x 12 cm) as they seem to prefer scanning to photographing. Josef Stark (2015, p. 20, 76), specialist in scanning diapositives, refuses to use photography, as his results with a SLR camera were inferior to “magazine scanners”. He appreciated the scanner’s infra-red based scratch and automatic dust recognition as repairing these flaws on a digital photograph would require time-consuming retouches in post production. The consortium of image specialists that wrote UPDIG states: “Scanning transparencies is more straightforward than digital capture [with a still-camera]. Use one of the excellent, available software solutions to characterize (profile) your scanner.” (http://www.updig.org/guidelines/ir_fine_art.html)