Tessa Meijer and Oliver Vicars-Harris
Tate, London

TATE ACTION - ACcess To Images ON-line

Keywords: digitisation, online art catalogue, British Art Information Project (BAIP), Tate collections

Overview

In the light of its recent expansion activities and in order to take the lead in applying cutting edge technologies to provide greater access to its collection, Tate has, as one of its digital activities, created a project to digitise the British Collection. The project has been given the title of British Art Information Project (BAIP) and it went into full production in Spring 2000. The images that BAIP creates are fed to the Tate website once a week, and will be available as part of the Collection Catalogue, with a fully searchable subject index. This paper gives a background to the project and tells what is involved in getting an activity of this sort up and running; there is a 'walkthrough' tour of the production processes, an examination of the means for delivery to end users, as well as a discussion of how the project fits into the wider Information Systems initiatives at Tate. The paper concludes with a summary of the key points covered, ensuring that the activities of the British Art Information Project are made clear.

BAIP is essentially in place to digitally capture the entire British Collection now housed at Tate Britain, which is made up of some 50,000 works ranging from large paintings and sculptures to small pencil drawings and sketchbook pages by J.M.W. Turner. It provides very high quality imaging and indexing processes, to be applied to the collection. The project is expected to run for eighteen months. A new team of seven specialist staff has been recruited, providing ten person years of working input to achieve completion of the project at the end of the eighteen months. The overall cost of the project runs at just under £600,000 and the Heritage Lottery Fund supports the project.

The Collection

It would be useful at this point to provide a brief overview of the collection, which is being covered by BAIP. The collection holds paintings, sculptures, prints, watercolours and drawings. There are around three and half thousand paintings that comprise the national collection of British paintings dating from the sixteenth century to the present day. The sculptures make up about one thousand of the works in the collection, forming the nation's collection of modern British sculpture dating from the late nineteenth century to recent conceptual works. The print collection covers works of all periods, including the nation's foremost study collection of modern British prints and is made up of around eight thousand works. Other works on paper include six thousand drawings of all periods, including the recently acquired and magnificent Oppé Collection of historic English watercolours, as well as an extensive group of twentieth century works. The Turner Bequest provides the most substantial numbers of works, including six and a half thousand watercolours, representing an unrivalled collection by Britain's most popular artist. One of the most interesting aspects of the collection are the twenty five thousand studies, pencil drawings and sketchbook pages revealing an essential insight into the formation of Turner's most famous master works. Although the Heritage Lottery funding is focused on exclusively these fifty thousand British works, there are also around eight thousand foreign works that will also eventually be captured using the same infrastructure.

Therefore even a collection as seemingly well known as Tate's has vast areas of works that are known to few people outside scholarly circles. There are also parts of the collection that are either too fragile or too obscure to be considered suitable for display. This is where digitisation comes in. Tate has found itself in a position where it has substantial quantities of works and also a large number of people wanting to look at them, so one of the most significant aspects of a project such as BAIP is accessibility. This collection, and in particular the Turner Bequest, is difficult to explore and any increased use would put it under impossible strain in terms of excessive handling and display (works on paper suffer from the effects of environmental conditions and lighting). Much of the Bequest, however interesting, has only ever been thoroughly viewed by Turner scholars, registrars and curators. Digitisation will open up parts of the collection that have hitherto been inaccessible to the general public. Like many public institutions housing collections Tate sees digitisation as key to solving the central problem by improving access and continuing to preserve those most precious parts of the collection.

Project Set-up

Getting approval for these types of projects can be far from easy and institutions often find themselves looking for endorsement from people least qualified to understand their aims. Large-scale digitisation initiatives require significant resource investment and in most cases this implies the involvement of a funding body, such as the Heritage Lottery Fund in the case of BAIP. The BAIP team consists of a Project Manager, and Indexing Co-ordinator, an Imaging Co-ordinator, two Indexers, an Imaging Assistant and a Project Assistant. With funding coming from an outside body, the project manager has to be fully accountable for every aspect of the project. Projects require careful planning at the best of times, balancing the delicate triangle of scope, time and resources. While everyone working within an institution on a digitisation project may be working towards the same goal, digitisation projects can and do go badly wrong.

There are many unavoidable big issues to be tackled before a major project can get underway. Firstly, the objectives must be defined. The organisation will need to decide what it needs and why, carrying out feasibility studies in order to ensure that its needs can be met within the scope of the proposed project. When this work has been done and the objectives are in place, computer systems will need to be sourced. This includes defining and seeking out the appropriate specifications for the type of project that will be undertaken and embarking on the procurement of these systems. The set-up of the infrastructure will necessarily go hand in hand with this step, searching for and fitting out premises for the placement of equipment and staff, and most importantly of all the team will need to be recruited. With all this in place, the project can get underway - whereupon the real problems almost certainly begin.

One of the key steps to the success of any project of this nature is putting effective information and workflow management systems in place. Tate have facilitated the work of BAIP by procuring the development of an Image Management System (IMS) which was supplied by a company experienced in dealing with the image needs of museums and galleries, called iBase Ltd. The system, called inVisage, is dedicated to supporting the specific activities of the BAIP team at its present stage although it is planned to be scaled up as a phase two develops to fulfil a wider range of cross-functional asset management needs. For now there is no plan to roll out this software for use by any one other than the BAIP team, the Photography department and copyright officers. The software has been developed with four different interfaces to support the four key activities of Indexing, the work of the Photography Department, Digital Imaging and Project Management. The concept is to optimise the use of computers for what they are good at and help towards freeing staff from more mundane and time consuming administrative tasks.

Production - Imaging

The production process is the heart of the project. Since delivery systems come and go, the main investment and emphasis of the BAIP is on capturing images and creating the text in a way that will be of long-term value. As will be shown, images are stored in standard formats that can be easily converted for use in other ways and archived for future use in new delivery systems or for conversion to new formats and so on. BAIP needs to capture fifty thousand images in a relatively short time-span, so the objective here is to do it once and get it right.

All the production processes for the project are carried out internally in order to keep down costs and to retain control and gain internal expertise that would be of value to Tate as it continues to expand in the area of New Media. The project has ambitious workflow goals of capturing 100 to 150 works each day. In order to achieve this a new specialist team has been recruited and a studio has been set up located near the works on paper store in the Clore wing of Tate Britain. The studio is comprised of two rooms, one set up for Imaging and the other for Indexing. The Imaging team is made of an Image Co-ordinator, an Imaging Assistant and there is room in the team for a casual worker to ensure that the camera continues to work when the other team members are unavailable. In order to accomplish the desired output it is essential to keep the camera running for as many hours in the day as possible.

The studio has been deliberately placed near the works on paper store, which facilitates the transportation of works, most often kept in solander boxes, to and from the studio. The registrars assist with the collecting of works and also with the verification of locations and when stray works (those on loan or display, or undergoing conservation treatment) will be available for imaging. One hundred to one hundred and fifty works pass through the imaging studio per day and careful track is kept of their progress. The team also includes staff who are experienced in handling works of art on paper, safety and security are not compromised by the demands imposed by the heavy through-put.

Figure 1 The LeafVolare digital camera in action.

The camera itself is comprised of a Leaf Volare digital back and a Nikon lens. It can capture at a resolution of two thousand by three thousand pixels and builds colour not through a colour chip and interpolation, but utilises a black and white chip and colour filters. The image passes through red, green and blue filters consecutively and the high quality black and white chip delivers crisp images with clean colours that have not been compromised by passing through extra processes within the camera. The file sizes that the camera produces are just over 37 megabytes in the TIFF format. Works are placed underneath the camera on a copy stand, and the camera can travel up and down to capture A0 sized works down to postage stamps.

Figure 2 Image capture on the Mac G4.

An Apple Mac G4, running the Leaf software, drives the camera and the IMS is run on a PC. The predominant platform at Tate is PC and so there is a part of the network dedicated to the BAIP's Mac to PC needs so that data can be exchanged through the two platforms without hitting any problems. A live video feed on the Mac allows each shot to be lined up and focused manually and accurately and demonstrates the action of the colour filters. The camera software also allows for adjustment to exposure and provides a toning curve to control the tone balance. If the exposure is correct, there is little need to adjust the curve. Too much fiddling interferes with the capture rate and when compared to the original the RGB colours are very true. In fact, BAIP does not use any ICC colour profiles at capture as the colours produced directly by the camera are more than satisfactory both for on-screen display and print.

Figure 3 inVisage screenshot, showing several fields.

inVisage contains a numerous amount of data fields including windows for viewing thumbnails and larger views of selected images. It is also possible to batch works, for instance by solander box number, which provides a valuable tool in helping the operator to ensure that everything in that box has been caught. There are also several data forms and data sheets that give additional information, fed through from the Tate collections database. Two versions of the captured image are exported from the Mac to the network before the import into inVisage is actioned. One file is saved as full data from the camera; no lossy compression is applied to this image and it is not cropped. This is a full 37megabyte file and is automatically archived and written to CD-ROM. At this size, it is only possible to store seventeen images per CD. A smaller 8bit uncompressed image at a reduced file size of around 16 megabytes is imported to inVisage where a further four files sizes are created using a set compression ratio of ten or twelve to one. Three of these files are JPEGs, at increasingly larger sizes and one of the files is a PNG.

The PNG file is there for future-proofing the on-line image pack. As it is unwise to further compress a JPEG to create a smaller and more web friendly JPEG because of the loss of quality incurred, the loss-less PNG format means that BAIP can make further image packs without incurring quality penalties. PNGs provide better compression than TIF and are readable by web browsers. All the web images created by BAIP are made from the PNG and this file can be used over again to form more JPEG image sizes. At the import into inVisage stage and before the image pack is created, there is an opportunity to record metadata about the image capture including the name of the tone curve used, exposure, and a space for comments. Few of these fields change as the aim is for consistency and a fast capture rate, which can be achieved by keeping the additional work to a minimum.

Conventional Photography and Scanning

The photography department at Tate is also involved in the project as they have specialist knowledge in the photography of paintings and sculptures, which is very different from copy-stand photography. Therefore conventional methods are used to capture paintings and three-dimensional objects, principally a ten by eight camera that makes five by four transparencies. A high specification transparency scanner is used to convert the five by four transparencies to digital format and the transparencies are scanned in at a resolution as close to that of the digital camera as possible. After the scanning process, the transparencies are imported to inVisage and treated in much the same way as the camera images.

Some images are also being imported from Photo-CD. These are images that were captured by the photography department before the start of this project and that were placed onto Photo-CD by a specialist laboratory. These images tend to be of a poorer quality than those produced by the camera, however they can be improved by the careful use of some image manipulation. This sort of treatment can often bring them closer to the original and more in line with the bulk of the project's production. The scanned transparencies also undergo some changes with the software allowing for adjustments to tone and colour. Once scanned, the images are compared with the original as the digital images produced by the scanner are now twice removed from the artwork itself and will have suffered changes en-route. Again, manipulation can bring these images closer to the original.

The iBase software contains basic image processing tools and this is quicker than routing the images through Adobe PhotoShop before import. Once in the database, adjustments can be made to the colour, tone, orientation and overall look of the images. The images from the camera need little if any changes made, other than some cropping. The camera is very objective and tends to produce images so true to the original that it would be almost unethical to tamper with them in any way. The scanned images and those from Photo-CD often need colour and contrast balancing as conventional media tends to be more subjective.

Each image is examined carefully for quality and for any possible faults such as incorrect orientation, strange effects or incorrect data for the image. If the metadata suggests that any changes should be made to an image these will also be done at this stage. If the image passes, it will be signed off ready to be picked up by the server and fed daily to the Intranet and weekly to the website. Any images with faults logged against them are withheld; it is the sign-off status that is the trigger for their importation on-line. Faulted image pass through a double-checking system and in most cases faults can be rectified relatively quickly. Sometimes curators are consulted or images need to be re-loaded from the archive CDs, however despite the fast capture rate there are relatively few faults and they can be dealt with very quickly. Tate has future ambitions to move beyond static two-dimensional imaging.

There are plans for special treatments to allow users to interact with three-dimensional works and explore installations by means of specially built applications providing walk-through tours with links and accompanying text. Tate has made a bid for funds from the New Opportunities Fund (NOF), which is in its second stage of application at the moment. This would be seen as an expansion of Tate's activities in the field of digital imaging and would be designed to run directly on the website as well as the Intranet in much the same way as BAIP are publishing their images.

Production - Indexing

There would be little gain in digitising all these images unless they are indexed for easy retrieval. Therefore, the other half of the BAIP team concentrate on developing a detailed subject index linked to the images. This is the most time-consuming production activity, with a team of three indexing around one hundred and fifty items per day. The team works very closely together and the IMS also enables the communication of notes across the team as they work at their desks. This is essentially the second production stage, since the indexing is done from screen images rather then the originals and there is an emphasis on authority sources, for example core data migration from the Tate Collection database. This is not a cataloguing exercise; one of the most important aims for the indexers is to facilitate the retrieval of works with an emphasis on drawing out the visual references.

Figure 4 inVisage screenshot showing indexing fields.

The indexers are building a simple in-house subject classification system within the IMS, with the advantage of being highly relevant to the collection as well as having speed of application. This consists of three levels with authority control over the structure and individual indexers can insert only new keywords. The subject index includes classification terms such as 'Animals', and 'Movement' and so on, with deeper levels beyond. Users will be able to search for images using these sorts of visual references. The IMS automatically gathers metadata concerning record status and assists with photographic records and copyright management. There are also 'project management' reporting tools in place to track and analyse workflow progress such as how many records are made at each stage, by each indexer, tracking against targets and identifying blockages and backlogs.

User Systems

Figure 5. Diagram showing user systems

The schematic in Figure 5 gives a concept overview of the main data components behind and the user systems at the front. The top half of the diagram describes the delivery systems for BAIP, in other words the ways in which the images and subject index will provide a service to its end-users. The aim is to develop on-site digital access points to provide three distinct services. A major part of these services will also be realised in simpler form for web delivery.

The future ambitions for the current website include a wish for much more intuitive search tools, in other words trying to get away from typing in boxes, visual browsing that is image rather text oriented and to encourage serendipity through index links, i.e. giving an alternative to the 'dead end' currently reached at individual want level. There are three key systems in place that are involved in the feed of information to the website. The Collections Management System and the IMS are primary repositories, whereas the Collections Information System (CIS) is a copy platform for publishing. CIS transformed the website from static pages to being database driven in spring 2000, as well as providing some important new tools for the user. The website is updated weekly, and the Intranet daily, with both collections information and new images, approximately 15,000 to date. This is where the project is now, together with a number of other key building blocks in place towards the future vision for management and delivery of collections information at Tate.

Tate Digital Initiatives

Back-end > management...	Front-end > delivery...
Collections Management System	TateNet (staff intranet)
Image Management System	Website (public internet)
Collections Information Service	New art on the web
e-commerce applications	Commercial art/design portals

Conclusions

This has been a reasonably comprehensive look at the digital activities within the collections at Tate, and chiefly the BAIP. Key points covered were:

• A brief background to the project, looking at the collection and why digitisation is important.
• A look at getting a project of this kind and running.
• A walk-through of the production processes.
• An examination of the means for 'delivery' to end-users.

With the website being updated on a weekly basis as a result of these activities, it is well worth visiting from time to time at http://www.tate.org.uk. Look out for further website developments over the coming year.

September 2000