Cuneiform Digital Library Notes
2014:008        «              »
Digitizing in the round*

Klaus Wagensonner
University of Oxford

Approaching cylinder seals depends on the research questions that are being asked. Seals can be studied from quite different research angles. The basis for the study of sealing practices and seal use, for instance, is not the physical seal itself, but ancient seal impressions on clay tablets, locks, door and jar sealings etc. For clay tablets, in particular, the way of impressing a seal onto the surface of the tablet and the particular location of a sealing were not arbitrary and varied not just throughout the millennia and from place to place, but depended also on the type of document. The study of the sealing practice on clay tablets is hampered (1) by the lack of a proper photographic documentation of sealed surfaces and (2) by the omission of sealings on published handcopies for the sake of better readability. This lack of information makes a proper study of the sealing practices, for instance, on documents dating to the Ur III and Old Babylonian periods a quite difficult endeavour. Many handcopies reduce the “seal data” to the textual aspect of the seals and therefore only provide drawings of seal legends. The respective seal’s iconogrpahy is often omitted or accounted for in separate studies. The same applies to the way a seal is impressed onto the tablet’s surface. An essay on copying sealed documents is forthcoming by the author (Wagensonner [forthcoming]).

In many respects physical seals pose less issues. Printed catalogues and studies of larger groups of seals frequently put their focus solely on the iconography of the seals. Data on the raw materials of the seals is frequently of less importance and data on certain aspects as, for instance, the grain of a stone, is omitted entirely. But it is the latter, which sometimes might have influenced the choice of a specific stone, not solely for its color, but also for its texture and translucency.

Representations of (cylinder) seals in printed seal catalogues are based on modern seal impressions. Therefore, the afore-mentioned material aspect of the seals is completely lost. In a few cases the seal cylinders themselves are photographed and printed. But these representations merely give a clue to the appearance of a certain seal.

In January 2014 CDLI staff at the University of Oxford started to digitise the collection of cylinder seals in the Ashmolean Museum of the University of Oxford and preparing meta-data for inclusion into the CDLI database. The Ashmolean Museum houses a rather substantial and varied collection of seals, which amounts to approximately 1,050 individual seals. These were catalogued by B. Buchanan in 1966; seals that were acquired by the museum in the 1960s and 1970s were catalogued and discussed in a joint article by P. R. S. Moorey and O. R. Gurney in 1978. As indicated before, representations of the individual seals are accessible through black and white photographs of modern impressions. Information about the colour and type of stone is just given in the descriptions.

In the subsequent paragraphs I aim at discussing a method which combines both the iconographical and/or epigraphic aspect as well as a proper representation of the material a seal was cut from. This note contains a progress report on the imaging technique that is used.

CDLI has developed an imaging technique for seals, which is based on a technique introduced by the West Semitic Research Project. However, CDLI tried to follow a less cost-intensive approach for the capture of cylinder seals, in order to make this method also interesting for smaller collections, which are not willing to invest in expensive equipment. The result is a digital roll-out of the seal, which, in contrast to photographs of seal impressions, provides data on the iconography and the grain of the used raw material.

Method 1

One method uses a turn-table that spins the cylinder seal around while being captured by a stationary camera. The set-up for imaging the cylinder seals is rather simple (Fig. 1):

  • a turn-table (either operated by a computer or electronically)
  • a tripod, on which a camera is mounted
  • a light-source
  • a DSLR camera with a macro lens or any camera with the capability of producing macro images
  • a software that remotely controls the camera and is capable of doing time-lapse images (e.g., Smartshooter).

Fig. 1: Working station with turn-table and a DLSR camera using remote control

For this method it is necessary to mark the centre of the turn-table in order to place the cylinder seal accurately. Otherwise, the camera might lose its focus on the object. Due to the varying thickness of cylinder seals it is advantageous to attach a sheet of paper with concentric circles in the centre of the turn-table (Fig. 2).

Fig. 2: Turn-table with concentric mark-up and seal Buchanan 1984: No. 563 in the centre

The remote software allows not just for various camera settings as shutter speed or exposure, but, in particular, for time-lapse capture. There is a wide range of commercial software available in order to fulfil this task. For the digitisation of the cylinder seals in the Ashmolean Museum the program Smart Shooter was used (Fig. 3). The turn-table (custom-built by Kirk Martinez at the University of Southampton) needs about one minute for a full circle. While turning, the camera takes 40 images of the cylinder seal (Fig. 4). This amount of photos is sufficient to fully capture the seal. In order to ensure an even interval between the captures, the images are stored on the memory card in the camera and hence not directly transferred to the connected computer.

Fig. 3: Camera remote control via software ‘Smart Shooter’

The processing of the raw data to a digital roll-out poses difficulties at first (Fig. 4). Hardly any panorama or stitching software is capable of dealing efficiently with the differences in lighting between the various shots. A rather useful software is the stitching program PTGui, which allows for plenty of settings and adjustments.

Fig. 4: Raw data of seal Buchanan 1966: No. 563

Method 2

One of the main issues using a turn-table is the danger of losing the focus while the seal is turning. In order to avoid this, the turn-table can be moved in steps (for instance by 5 degrees) before checking the focus. So far, this can be achieved manually, and hopefully can be automatised in future. But there are other possibilities, which do not necessarily result in any increase in time and effort in order to reach the final result.

Using a soft support and mounted a DSLR camera directly above the seal can be carefully turned by hand and photographed. Good results can be gained from turning it by approximately 15-18 % between each shot.

But this method also allows for further adjustments. Since the seal is turned manually, the capture quality can be increased. A worthwhile is method is to produce High Dimension Range (HDR) photos in each position. In simple terms, HDR photos combine several (mostly three: a short, normal, and long) exposures in order to gain more information in shadows. This has also an advantageous effect on the representation of the color of the stone (Fig. 5b).

The final stage of producing a complete representation of an individual cylinder seal is to add the top and bottom sides of the cylinder. These are photographed separately. Interestingly, not many catalogues provide any representations on those sides. They do not only show the cross-section of a cylinder seal, but furthermore provide data on the central drill-hole, and so forth. For instance, there are plenty of examples, which show that the drill-holes are worn out by being hung on a cord.

Fig. 5a-b: Two processed examples: Buchanan 1966: No. 563 (left) and No. 633 (right)


Last but not least, using either a turn-table or rotating the object manually, this method can also be used for cone-shaped objects such as foundation pegs. Many collections of cuneiform artifacts house such objects. Up to now the representation of such objects in the CDLI database follows the usual “fat-cross” representation by juxtaposing 6-8 images of the cone's outer side next to top and bottom (Fig. 6a). However, juxtaposing does not present the text in a “fluid” manner. Using a sufficient amount of images and the afore-mentioned method it is possible to gather a complete rendering of the inscription (Fig. 6b).

Fig. 6: The cone AN 1929-777 in usual "fat-cross" representation (left) and stitched result (right)


SQLSTATE[42S02]: Base table or view not found: 1146 Table 'cdlndb.abbr' doesn't exist
ISSN 1546-6566    © Cuneiform Digital Library Initiative | Archival: 2014-04-15