Polly Elkin
Victoria and Albert Museum, London

Securing the Object: The Digital Photo Survey Project at the Victoria and Albert Museum

Keywords: digital photography, photo survey, The Victoria & Albert Museum collections

The Photo Survey project took place in the Victoria and Albert Museum (V&A) from October 1999 until May 2000, acting as a case study in obtaining and using the digital image. As the first large-scale trial of a digital camera for object photography within the Museum, the performance of the digital image was closely monitored against analogue equivalents. Although the minimum core data standard developed for this project also acted as a trial for retrospective cataloguing across the V&A, the main issue that I want to address is the capture and the successive application of digital visual records within a specific context of museum collections management.

Why the Photo Survey project took place and why a digital camera was chosen

Relatively recent security problems experienced within the V&A have focused attention on the vulnerability of objects on permanent display, particularly in the ground floor galleries. Subsequently, the need to have a readily available image of all objects displayed in this area of the Museum, primarily for security purposes, was identified. To begin the project, a complete list of all objects located in the permanent display areas of the ground floor was printed off from the V&A's automated collections management database, the Collections Information System (CIS). This list was manually checked against V&A Picture Library records to establish which objects already had an existing analogue image in the form of a slide, photograph or colour transparency. If no analogue image was recorded, the basic object inventory record in CIS was added to a tag set which formed the master list for the project. The tag set could be reloaded into CIS, sorted by location and subsequently printed out on a gallery-by-gallery basis detailing the museum number, name and location of each object.

It was found that approximately one quarter of all objects on display in the ground floor area of the Museum required new photography – 2,250 at part number level. Obviously, this work could not be requisitioned in the normal way where every object would have had to be removed from its case by the collection responsible and taken to the V&A photographic studio for a publication quality analogue image to be obtained. Not only would this have been unfeasible in terms of the time scale and finance available for the project, but it would also have had serious implications for the disruption of displays. In terms of risk management, the exposure of the objects themselves to handling, movement and the constant opening of cases was unacceptable. ¹ As a result, the decision was made to photograph the objects in situ, irrespective of whether they were wall mounted, free standing or exhibited within a display case.

So why was a digital camera used? Firstly, we needed a digital end product since the most immediate way to access an image is to see it linked in to an object catalogue record on CIS. This makes it accessible over the network across departments, at outstations and in remote stores. However, bearing in mind that the most costly part of digitisation is in obtaining and then scanning analogue photography, ² this process seemed unfeasible given that the photo survey project was running under a fairly tight budget. Additionally, an unlooked for by product of this method would have been a series of photographs requiring cataloguing and archiving whilst being of insufficient quality to generate income for their upkeep from reproductions, unlike studio photography. Compared with the time required to develop and check an analogue camera film, the immediacy of processing images already in a digital format made it the obvious solution for end use within an automated collections management system. Given these considerations it was decided to make this project the first major trial of digital photography within the V&A. To facilitate this, V&A Picture Library provided a Fuji DC330 digital camera, which can store up to 48 images at a time on a Smartcard, plus a freelance professional photographer to take the images.

Image capture - the realities of working in the museum gallery space

In terms of the array of objects on display, the V&A collections are challenging. There is great variation not only in the materials to be photographed but also in the scale of objects. Examples ranged from large pieces of three dimensional furniture and sculpture to small two dimensional prints and medals. Many of the galleries on the ground floor also display light sensitive material such as textiles and works on paper: the susceptibility of such objects to deterioration from ultra-violet light necessitates light levels of 50 lux or below. The initial trials carried out with the digital camera on a moderately overcast day proved that it could not register objects in sufficient detail at this level without additional lighting. Flash was not an option because of the glare reflected from the cases, even when using a remote hand held external flash. The solution was to use a photographer's lamp. Although this had the disadvantages of being cumbersome to move, it provided suitable conditions for obtaining an image of sufficient quality for three dimensional free standing objects, two dimensional works such as prints, and any objects displayed behind glass. However, at busy times in the gallery setting up a digital camera on a tripod and lighting the shoot could be problematic as although the galleries are well supplied with electrical sockets, it did still necessitate running extension cables along and across spaces which were sometimes subject to dense visitor traffic.

As a collaboration between V&A Picture Library and Records & Collections services, the information recorded about objects and images needed to be logged in a format applicable to both the Picture Library image arena and CIS. This meant manually compiling Comma Separated Value (CSV) files to correlate object museum numbers and the digital image reference as the photography was completed. At the end of each day, the manual log of image file name and number plus museum number was input into an Excel file. CSV files are structured to record:

• prefix - in this instance dcl for low resolution digital camera.
• major ID - date of image edit (date in yyyy/mm/dd format).
• minor ID (operator initials and Smartcard number run).
• individual image reference numbers in three digits together with the object museum number.

For example, the ninth Smartcard to be used was recorded as dcl19991103bn09. The CSV filename was also used for each folder of 48 images created in Photoshop.

Image load - the transfer of the digital image from digital camera to object record

Once the initial trials had defined the necessary technique to use the digital camera effectively in the galleries, progress was rapid: a total of 2,050 objects at part number level were photographed, downloaded into Photoshop and edited in 17 days, an average of 121 objects per day. Four Smartcards were available for use which easily enabled the maximum daily imaging total to be reached. The images were loaded from each Smartcard into Photoshop via a floppy disc adapter and stored initially on a PC hard disc. They were carefully checked against the manual CSV file to ensure firstly, that the correct object was shown and secondly, that the image quality was sufficient. Many of the images required cropping to single out specific objects exhibited amongst others within a case. To adjust colour and lighting, the auto-adjust function in Photoshop 5.0 was used for quickness, although the lack of consistent calibration across the project was reflected in the final results. As expected, the most significant variations occurred between objects captured in naturally lit galleries and those requiring the artificial light of the photographer's lamp.

When the images had been edited in Photoshop the image files were converted in Debabelizer into a database compatible format of 768 x 768 pixels on a black palette. Proprietary image loading routines were then run to create both the thumbnail and larger image, together with the metadata (known as a dot roll file) to enable them to be loaded into the Collections Information System. In this way the digitised files were consistently formatted and tagged to ensure that they could be located and retrieved for future use.³ The thumbnail file sizes ranged from 20 to 25K each, the full screen image 80 to 110K: both were 72 dpi JPEGS. By placing the images into an image arena rather than the Collections Information System alone, they could potentially be exported to different systems and used for different purposes such as gallery interactives. However, as low-resolution record shots and not studio quality master images, they are most suited for linking to Catalogue records.

Cataloguing

The cataloguing phase of the project followed once the edited digital images had been transferred from the Smartcard to the image arena and all anomalies such as any disparities between CIS Inventory numbers and those on gallery labels had been resolved. From a security point of view, good documentation is invaluable in the identification and recovery of stolen objects. ⁴ Records should ideally combine the use of non-technical terms to assist identification but contain sufficiently detailed information to allow them to be easily distinguished from other similar objects.⁵ A minimum number of fields for data entry were selected to form the V&A core, defined using V&A data standards and comparison with other examples such as Object ID, initiated by the Getty Information Institute in 1993 to develop an international documentation standard for the information needed to identify cultural objects. ⁶ In addition to an object image, many of the fields specified by Object ID were included, namely object type, materials and techniques, physical description (including subject and distinguishing features such as damage and repairs), title, date, and maker. For the Photo Survey it was agreed that accurate information should be obtained without needing to revisit each of the objects or involve physically handling them, so some fields recommended in Object ID, such as object dimensions, were excluded. ⁷ The fields were also chosen for their future potential use in multiple outputs such as gallery guides and label text, resulting in the following specification:

• Object name: single term or compound term best describing a particular whole object.
• Descriptive line: a brief free text description of the whole object sufficient to identify it (a précis of all the core record information)
• Title: either assigned at the time of origin or subsequently.
• Collection code: the curatorial department responsible for the object - for those working within one collection only, this is automatically inserted based on user ID.
• Artist / maker: individual or institution (responsible for object creation and the process involved)
• Date: either in textual or numeric form
• Attribution place: where the object was created or originated. (This information is to be entered into the Attribution tab of the Catalogue record screen in CIS, as in Figure 1.)
• Physical description: the general visual appearance of the object.
• Materials and techniques: a free text field to describe the basic materials and media of which the object is constructed and the technique used to make it. (See Figure 2).

Figure 1: The attribution tab of a CIS Catalogue record, containing seven of the core Catalogue record fields. Victoria & Albert Museum; © V&A Picture Library

Figure 2: The Physical tab of a CIS Catalogue record detailing the remaining core fields completed as part of the Photo Survey project. Victoria & Albert Museum; © V&A Picture Library

This also complies with the spectrum data requirements necessary to achieve Phase II of Registration. ⁸ Compliance with elements of other metadata standards such as Object ID and Dublin Core enables the V&A core record to act as a Crosswalk, a metadata standard that facilitates a shared objective.⁹

There are many structured data fields in CIS controlled by authority files. In the case of the core record, these are: attribution place, attribution maker, and their attribution role. The Getty Art and Architecture Thesaurus was loaded into CIS but the import or creation of other authority data is also possible, for example using information sourced from the Getty Thesaurus of Place names. ¹⁰ Figure 3 shows an example of an entry in the Terminology client module in CIS where authority files are created, structured and maintained. The authority files have a fundamental role not only as an aid to consistent data entry but also through their availability to help the user needing to search the database.

Figure 3: An entry in the place names section of the Terminology Client on CIS. Victoria & Albert Museum; © V&A Picture Library

Each Collection was sent a copy of the core record and asked to nominate sources of information to satisfy as many of the core criteria as possible for the team of three cataloguers to work from. Most of the Collections nominated label texts (some available in electronic form that could be cut and pasted directly into the relevant field on CIS), supplemented if necessary by Registered Descriptions which were created when an object was accessioned by the museum (until being replaced by the Catalogue function in CIS from 1998).

Once the core information had been entered onto the system, the digital image could then be permanently linked in. Both cataloguing and image capabilities were added to inventory and acquisition record functionality in the second phase of the development in CIS. The Index+ database and search engine allows searching from many access points, but searching for an object through a large number of images is time consuming and impractical.¹¹ The only efficient way to access images in the image arena through the catalogue record is if you know its CSV filename. Image loads can be run in CIS using the metadata generated by the loading routines to identify the correct image for each object catalogue record, but it was found by the cataloguers that it was more accurate to manually link in the correct object image by searching for the CSV reference via the image tab. Although the image can be seen as a thumbnail in the top right hand corner of all tabs in the catalogue record, the dedicated image tab shows a larger image which can also be enlarged to a full screen image, as shown in Figure 4.

Figure 4: The expanded image view within the Catalogue record on CIS. Victoria & Albert Museum; © V&A Picture Library

It is clear that together with the digital image the core catalogue record is a crucial source of information particularly in the context of object identification. The relationship is a symbiotic one - the low resolution of the images necessitates the detail recorded in the physical description field. Conversely, the image has a huge impact on the cognitive potential of the record when compared to the text entry alone. The information recorded acts as the descriptive metadata for the image itself, supplementing the technical metadata such as file format and resolution, and the administrative metadata such as when and by whom the image was created and downloaded.

Conclusion:

Museum collections (and their associated information) are dynamic with constantly changing displays, the rotation of light sensitive objects, and inclusion of new acquisitions. Periodic checks will need to take place in the future to maintain the photographic record of objects on display on the ground floor area. However, these displays are relatively static, particularly so given that the project did not encompass temporary display and exhibition areas.

Using a digital camera can be an effective procedure valuable to good collections management, particularly as in this example when it makes use of existing in-house technology. The initial expense of purchasing a digital camera can be offset against the cost of buying and developing large quantities of analogue film, making digital photography a viable method for museums of all sizes. The digital camera offers a quicker route to the digital surrogate than scanning, even considering the post capture adjustments necessary such as cropping and colour balance. ¹²

The management process to be developed and followed for digital images needs to be stricter, more detailed and more consistent than that which is sufficient for analogue images. This is because you are managing non-obvious filenames as opposed to photographic material that can be physically handled and more easily recognised by sight. The three kinds of metadata generated in the management of images in this project alone (technical, administrative and descriptive) signify the complexity of the task. Although it is recognised that it can be best to create the initial digital image in a format that allows it to be reused in a number of different ways, from high resolution images for publishing to thumbnail records,¹³ it would have been counter productive in this project since the aims of object security and safety would not allow the removal of large numbers of objects from cases to obtain a publication quality image. Furthermore, this method avoids the cost of creating and maintaining either analogue or high resolution digital archives in the appropriate storage medium. The small file sizes are important for transmission speeds across the network, particularly given that the number of objects that could potentially have an image linked to the catalogue record in CIS is about 1.5 million and growing. The low resolution of these images is also suitable given their potential use: the records will be viewed on-line rather than being printed out which would necessitate a higher resolution. ¹⁴

The feedback from the collections about the ability of the images to enhance the catalogue record had been very positive and the digital camera will be used again in the future to complete the photographic record of all objects on permanent display within the museum. The data entry itself provides a basic, starter record that can be expanded in future by curatorial staff. As such, it will be used as the standard for retrospective cataloguing (i.e. creating records at a point subsequent to the object's accessioning by the museum) across collections in the V&A. This is vital if the museum is to address documentation backlogs in a consistent and methodical way, and fulfil its role as a provider of accessible object information.

What I hope I have explored are some of the advantages and limitations of using low-resolution digital photography for a diverse range of museum objects, particularly within the specific and demanding context of public gallery spaces. The administrative care of objects such as inventory control and identification oriented cataloguing is an essential addition to the provision of good physical conditions for objects.¹⁵ An image surrogate allows the object to be identified remotely over the museum network and without unnecessary handling. Additionally, the existence of an image can be enormously helpful when trying to locate a missing object.¹⁶ However, future passes through the Catalogue record will add more detail to the core record beyond the security led focus, which will be vital given that memory institutions are increasingly using digital information resources for in-house applications and to contribute to broader cultural heritage networks. The benefits that can be gained from linking even a low resolution image to a basic catalogue record can contribute greatly to good practice in collections management beyond addressing the immediate need for security orientated object information.

September 2000

Notes

1. Fahy, A. (1995) (Ed.), Collections Management, p. 253, New York and London: Routledge.

2. Keene, S. (1998), Digital Collections: Museums and the Information Age, p.39, Oxford: Butterworth- Heinemann.

3. Besser, H. & Trant, J. (1995), Introduction to Imaging: Issues in Constructing an Image Database, p.10, Santa Monica, California: Getty Art History Information Program.

4. http://www.object-id.com/why.html

5. Fahy, A., Collections Management, p.255.

6. http://www.object-id.com/about.html

7. http://www.object-id.com/checklist/check_eng.html

8. http://www.mdocassn.demon.co.uk/phase2.htm

9. Zorich, D.M. (1999), "Museum Information: Reforming the Beast", Spectra - a publication of the Museums Computer Network, p.27, Issue 2, Volume 26.

10. http://shiva.pub.getty.edu/tgn_browser/

11. Besser, H. & Trant, J., Introduction to Imaging: Issues in Constructing an Image Database, p.17.

12. Besser, H. & Trant, J., Introduction to Imaging: Issues in Constructing an Image Database, p.21.

13. Keene, S., Digital Collections: Museums and the Information Age, p.39.

14. Besser, H. & Trant, J., Introduction to Imaging: Issues in Constructing an Image Database., p.13.

15. Fahy, A., Collections Management, p.251.

16. http://www.object-id.com/checklist/check_eng.html