This chapter deals with metadata for each of the media formats.
Still image metadata
Still image metadata is relatively well established. It has its basis in the analogue world of slide and photographic libraries and in image centric subject disciplines like art history. Still image metadata has adapted over time to the challenges of the digital age and the complexities sometimes evident in relating digital images with their real world subjects and other analogue surrogates such as slides and photographic prints.
Descriptions of images can be broken down into discrete types, such as: objective things, either generic (for example: 'person') or specific (for example: 'Mona Lisa'); and more subjective or abstract concepts (sometimes referred to iconological) for example: innocence, reserve, calm. Descriptions could also include biographical details of the image such as its provenance and history. Another good way of thinking about describing images is to ask (and answer) the basic questions: who, what, where, when, why and how.
Time-based media metadata
Time-based media - video, audio, performance, installation and any combination of these - is characterised by movement through time, whereas the still image is characterised as a 'snap-shot' of a given moment. As we have seen, still images can be complex to describe and have multiple relationships with other objects and time-based media are the same – only more so.
Indeed much of the work that has been done in developing metadata elements and schemas for time-based media have concerned themselves with addressing their complex structure and relationships with other resources.
This is particularly evident in digital audio files that are created as part of a process of digitisation of analogue media, such as tape. For example an audio file digitised from multi-track tape will contain multiple digital files which will need to be described in context with each other. Conversely, a given audio file may be an excerpt taken from a larger piece, in which case information on the structure of the related digital content may be required to show the file in as broad a context as possible.
This results in a ‘family-tree’ or hierarchy of related data that has to be described within the metadata schema. Moreover, information specific to audio files, such as location points and markers within the files can be used to identify and locate important content, which, again, needs to be captured in metadata.
Many of the same metadata principles apply in moving image as they do for audio. Both media types, being time-based, have the same issues with resource aggregation and sometimes complicated structure. For video, this is to some extent compounded by the fact that no single digital video ‘industry’ exists as different cultures exist for commercial video production, broadcast media, academic research and teaching and in cultural heritage. There is, therefore, no such thing as ‘industry standard’ means of managing moving image resources.
That being said, metadata schemas for moving image resources are emerging that have some traction across the various communities of users. The Public Broadcasting Core (PB Core) is one attempt that has emerged from the public broadcasting community, and another is MPEG21, which has come from commercial video production. MPEG21 in particular can describe very complex structural dependencies and if a given collection of time-based resources contain elements of multiple picture streams or complex audio MPEG21 would be a schema to consider.
Metadata for 3D objects
Metadata of a different sort is also required for three-dimensional resources. There are many domains where 3D resources are created (for example in architecture and building, archaeology, product and furniture design and geology) that produce geometric data that often does not fit in other standard metadata schemas.
The sort of data produced by 3D resources includes:
- Geospatial data which refers to the geographic location of a resource and could be expressed as latitude/longitude value, or other co-ordinate data which can be rendered as maps.
- Data to describe 3D models, for example in buildings or products produced by software applications such as CAD
As 3D data becomes more commonplace, repositories and other systems have been set up to store and provide access to it. Sketchup.com for example provides access to millions of 3D models, and the Europeana web-portal provides access to 3D resources created by European cultural and heritage organisations. Communities who work with 3D data therefore are looking at new ways to be able to share and use resources more efficiently. Standards are beginning to emerge to facilitate this.
A good example is the CARARE project which was a research project looking into ways that 3D data could be ingested in Europeana and has produced a metadata application profile to use for 3D data. 3D models require specific descriptors to facilitate a more compete description and management. This can include, for example, elements such as: the number of vertices (ie faces), the geometric primitive types (points, lines, triangles, quads etc) or textures that have been used.