Note: If you see this text you use a browser which does not support usual Web-standards. Therefore the design of Media Art Net will not display correctly. Contents are nevertheless provided. For greatest possible comfort and full functionality you should use one of the recommended browsers.

Themesicon: navigation pathMapping and Texticon: navigation pathArchive/Map

icon: previous page

Mapping images

When it comes to displaying a text, the data are already given and unambiguously identifiable (usually on the basis of a word). But what is to be done if these data still have to be generated when dealing with images. How can images be recorded formally and descriptively when so far no descriptive systems for electronic images have been developed (despite art history)? How can moving images be indexed? If images are analysed as data packages, our text-based approach takes us as far as Google shows us we can go: searching for images by examining the text associated with them. The fundamental question is: why is it still at all necessary to continue with the «structuralisticlinguistic paradigm» (cf. Wolfgang Ernst, «Beyond the Archive: Bit Mapping« of subordinating the image to the word? We do not have to address this generally in order to ask the simple question of what happens if the image, the data package or the video has no text attached to it? Any media archaeology that does not already know what it is looking for deductively but constructs classifications and relations


through inductive cluster analysis and iterative processes has to face these questions.

In television archives, editors are already working with the first forms of automatic video item sequencing. These software solutions work with algorithmically manufactured storyboards, so it is simple for the editors to search for material that can be used again. This pictorial analysis is not applicable only to narrative, but makes it possible to analyse any amount of visual data in relation to structures, textures, colour values and other parameters. [28] Behind this already lies the concrete practice, not just the vision, of searching for images with visual references, not on the basis of text. Images look for other images, images recognize text, especially handwriting, videos are broken down into visual indexes. In all these cases, patterns and structures help comparative analysis. These image search engines carry out a change from simple meta-data to complex annotations and to the semantics of the image. By analogy with full text searches, this could be called a full image search. [29] But what Wolfgang Ernst and Harun Farocki now call the «Visual Archive of Cinematic

icon: next page