WP3: Data capture in Artiste

Kirk Martinez - Soton-ECS-IAM

WP runs from Jan-Sept. 2000

In order to test the systems we will build a project data set using images from the partners in Artiste. These may be tailored to the type of research envisaged. We need to produce a rough list of numbers and types of images needed and if necessary agreements on their use.

Database capture formats

A tab or comma delimited text file is probably sufficient for textual data import. Images should as far as possible be in a format which has readers available (eg JPEG, TIFF, etc). One major exception will be SCOPYR images which will require a converter. High resolution is desirable, as that is one of the strengths of the project. Similarly calibrated colour or at least sRGB would be ideal to facilitate cross-collection colour comparisons.

My guess is that we should aim initially for 1000 items from each collection but this will depend on the items available (that is half the Nat. Gal. but a tiny sample of the V&A - although it would be enough for one type of image such as textiles) and their size. For storage reasons the very largest images from say the NG might not be stored in the demo system. Remote delivery is probably available for viewing them however. The multimedia server in ECS/Soton has around 90GB spare at the moment. Any images stored will be firewalled for non-project participants but can be made available to other sites as requested by the providers.

In the case of the V&A we have a fairly good idea of the areas of interest initially and I would suggest we aim for images of coins, textiles and pre-20C paintings (to overlap more with other partners). We have already received two CDs full of textile images.

Draft plan

mid-July: first list of items and specification of data requirements

end-July: first images and data sent to IT Innovation for importing

end Sept. main bulk of data sent and imported.

Images sent to ECS-Soton on CD will be saved and passed on to IT Innovation for databasing. There will probably be a requirement to gather updates to the test data because the nature of the queries to be solved is fairly open. These are the results of the initial requirements surveys:

VAM

textiles

engravings with corresponding images, tapestries

faxes of queries with appropriate digital image

Uffizi

pics with butterfly conserv. with colour front

conservation images later

some high-res digital images captured recently, especially with different cracquelure

Louvre

b/w vs colour

butterflys on back and colourpic

UV pics

scopyr format (Soton to convert) images of panel paintings with interesting cracks

National gallery

paintings with transferred faces etc???

vips IRs

colorimetric images - especially with interesting colour texture/conservation/cracks