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

Scanning deficiencies, and scanning defective slides

Scanning can produce unwanted visual flaws. They are called “aliasing effects” and are generally produced by the computer (then called “artefacts”), but can also be a mixture of an optical phenomenon and the incapacity of the scanner to cope with it. Artefacts are often the consequence of an insufficient sampling (called “sub-sampling” or “underscanning”): the sensor registers only a part of the optical information instead of all and produces less electrical signals than necessary. In the following, some annoying flaws are treated. (For more information on artefacts such as “nonuniformity, dust and scratches, streak, colour misregistration, aliasing, and contouring/quantization” see also Williams 2000).


  1. Visible texture exterior to the original object – Moiré

Sometimes a scanned image shows a grid which may disappear according to Tom Ang (2006, p. 320-321) when the size of the inspected image is changed or the file is printed out in another format. The photographer explains the phenomenon as a conflict between the sensor’s rigid pixel structure and the grid of a printer or the raster of the computer screen. But it starts already with the digitisation of an artefact. When an object has a distinct pattern, it can “irritate” the sensor which sends a wrong signal (Sauter 2010, p. 50). The “conflict” between the motif’s strong regular structure (e.g. a jacket with a chequered pattern, a printed image with a visible grid) and the regular array of the image sensor in scanners or cameras results in the so-called “moiré”: the scanned images show diagonal wavelike structures in irregular width and length, instead of the original pattern with its horizontal and vertical arrangement. This phenomenon is the evidence that the signals of the two-line patterns interfere and superimpose.


As to slides, during their scanning project Shingo Ishikawa and Darren Weinert from the National Film and Sound Archive made the following discovery: “Some cinema slides were created using printed dot screens. When scanned, these screens can interfere with the scanner sampling frequency, creating unwanted textures, or moiré patterns in the duplicate […].” The archivists suggest: “To eliminate moiré patterns the slide is scanned at a slight angle to disrupt the relationship between the inherent image pattern and the scanner sampling frequency.” (Ishikawa, Weinert 2010) Tom Ang also advises to scan the image again, with the original slightly turned (c. 5°). If the moiré persists the object should be slightly turned again, or scanned with another resolution; the operator should look for an adjustment which makes the phenomenon disappear (Ang 2006, p. 320-321).


  1. Visible texture exterior to the original object – Newton rings

Another phenomenon can appear in a scanned image: one or more multi-coloured concentric or ovals rings. They resemble ripples caused by a stone when thrown into a pond, or an oil stain floating on a puddle (but less colourful). These are Newton Rings (also called Newton’s Rings or interference rings) and are described as follows: “Concentric bands of coloured light sometimes seen around the areas where two transparent surfaces are not quite in contact. The rings are the results of interference and occur when the separation between the surfaces is of the same order as the wavelength of light.” (Focal 1969) These rainbow-coloured rings are produced when air separates the slide or diapositive from the glass pane, or when a 24 x 36mm diapositive film image is framed between glass; the air pocket acts like a prism or a lens that refracts light. Newton rings are only visible in reproduction, not during capturing.


Hints by scanner operators:

Most photographic experts (e.g. Ang 2006, p. 320) as well as the team of the Canadian Museum of Civilization Corporation suggest to avoid that slides (and diapositives) touch the scanner bed. “Possible way to avoid them: a) use a mount to raise the negative [and the positive…]; b) insert an anti-Newton’s ring sheet between the negative and the scanner glass.” (Brosseau 2006, p. 22). TownsWeb Archiving proposes to use a mylar or acetate sheet to prevent any glass-on-glass contact (  Ang also asks to make sure that the glass is totally dry. The archivists Shingo Ishikawa and Darren Weinert give the following directive: “When the glass of the slide is in direct contact with the scanner glass, Newton’s rings are likely to form on the surface […]. Newton’s rings occur when light is reflected between the glass plate and scanner surface. The only way to eliminate this is by moving the two surfaces apart or by using anti-Newton’s ring glass on the scanner. To separate the two surfaces a simple thin card matte can be used. However, this may introduce further depth of field issues and a close inspection of the resulting scan is required to determine whether the slide image is in critical focus.” (Ishikawa, Weinert 2010) Photographer Tom Striewisch (2009, p. 394) recommends an antistatic cloth or to breathe on the negative or de-framed diapositive, as both are inclined to bend, and the humid air helps to reduce it.


  1. Visible texture exterior to the original object – staircasing (or stairstepping)

When diagonal or round lines or edges in an image look jagged like stairs, when the area between two colours seem lacerated, when the square form of the pixels can be remarked which makes details vanish, the effect is called “staircasing”. It is an “edge artefact” and appears on the edge of an area which is no longer straight but deformed. Tom Ang (2006, p. 320) sees the cause in too small a number of pixels, insufficient to fully cover the given information. He and Kraus (1989, p. 74-76) recommend to choose a higher resolution for the scan. As an “antidote”, a higher resolution is better than to eliminate hard contrasts in post production, as software would use mathematical “anti-aliasing” and “calculate” the edges smooth by producing a slight blur.


  1. Visible texture exterior to the original object – noise

A disturbing net of many white or light dots in darker parts of an image is called “white noise”. Kraus (1989, p. 168) compares this to “arbitrarily littered confetti”. Each dot on the reproduction is a pixel that is sending wrong information on its light value. This concerns especially small sensors when they receive little light (see Striewisch 2009, p. 397). Many cameras have a function called “noise reduction” that suppresses the effect, but noise can also be eliminated in post production. (For more information on noise see also Williams (2000).)


  1. Visible texture exterior to the original object – out of focus

If the reproduction seems overall slightly fuzzy, is missing details and its colours look washed out, the scanner was probably not focussing correctly as Tom Ang (2006, p. 320) suggests. The photographer advises to redo the scan, this time with a manually adjusted focus. If this happens with a diapositive, its frame should be changed to make sure that it lays totally flat on the pane.


  1. Visible texture exterior to the original object – jitter

Tom Ang (2005, p. 320) draws attention to “jitter”, an artefact which can be seen when the reproduction is strongly enlarged. An image with “jitter” shows jagged lines in the wrong colour, their details seem to be in the wrong position. He explains this as the result of a “scanner head” which is moved irregularly. His advice is to flip the object (let it make a 180°-turn) and to scan it again. If the phenomenon does not disappear, the scanner may be defect (and should be repaired) or performs poorly and should be replaced by one of higher effectiveness.


  1. Visible texture exterior to the original object – posterisation

If a reproduction shows hard transitions between colour areas, if the visible difference between the tones does not look smooth but stepped, if the colour density suddenly changes, then this can have two causes according to Tom Ang (2006, p. 321): due to a low resolution, there is not enough information to fill in gaps left in the tonal values, or an “exceeding” manipulation in post production has created the loss of data. He suggests to scan again with maximum resolution and to be careful when manipulating brightness, colour balance and contrast.


  1. Visible texture exterior to the original object – blockiness, macro-blocking

This artefact is produced when images are compressed too much. When enlarged, the reproduction shows many blocks formed by pixel groups of the same colour. Tom Ang (2006, p. 320) calls them “JPEG-Kernals” and recommends choosing another compression that doesn’t show blocking.


Scanning defective slides


Digitising broken slides can be difficult, not only when handling the object that has to be protected from further damage (if it can’t be repaired and stabilized before), but also for its reproduction.


As Shingo Ishikawa and Darren Weinert from the National Film and Sound Archive observe: “[…] a crack in the glass may scan better length ways than across the scanning head. Determining the best scanning angle can be a matter of trial and error. Angled scans can be corrected to a normal viewing alignment in the final file by rotating the image during post-processing.” (Ishikawa, Weinert 2010) They also write in their report: “Glass slides are digitised with any defects that cannot be easily removed in the conservation process, such as crystal patterns that form as deterioration by-products of the slide. This digital copy can be used at a later date to ascertain whether the slide is in a stable condition or any further damage has occurred. Scans are also made when further conservation is undertaken, to document the ‘before and after’ condition.”


The archive has to take the decision whether to scan such a slide with the series it belongs to, as it is practical (same scanner settings, almost identical file name, series completely done and therefore digital accessible in its totality); or to keep it outside the workflow to be treated later, because its scanning is more time-consuming, interrupts the otherwise smooth procedure and puts the broken item at a high risk.