One of the principal features defining traditional media is a fixed and linear narrative structure. In Database Narrative media however, the narrative is developed by selecting scenes from a given media database. This can either be an interactive process (whereby the audience selects scenes from a collection) or a generative process (whereby a computer selects scenes from a collection).

Lev Manovich first related the database to cinema and new media in an effort to understand the changing technologies of filmmaking and new media techniques (Database as Symbolic Form, Cambridge. MIT Press, 1998). This article is both seminal and highly controversial. According to Manovich, cinema privileged narrative as the key form of cultural expression in the modern age but the computer age introduces its correlate, the database: "As a cultural form, database represents the world as a list of items and it refuses to order this list. In contrast, a narrative creates a cause-and-effect trajectory of seemingly unordered items (events). Therefore, database and narrative are natural enemies. Competing for the same territory of human culture, each claims an exclusive right to make meaning out of the world."

Context-free grammars play a central role in the description and design of programming languages and compilers. They are also used for analyzing the syntax of natural languages. In an art context, Context-free Grammars are used to describe a series of generic rules that a program uses to construct images, text or video into a meaningful ‘natural’ or ‘organic’ order. The simplest way to imagine them is as a sort of ‘search and replace’ strategy for constructing ‘sentences’. For example, a context-free grammar such as:

 

Position PERSON ACTION POSITION PERSON who is ACTION
PERSON Matthew PERSON Willoh PERSON Josh PERSON Philippa	ACTION sitting ACTION talking ACTION sleeping	POSITION next to POSITION behind POSITION in front of

Could result in the following ‘sentences’:

Position [Matthew] [sitting] [next to] [Willoh] who is [talking]

Position [Josh] [sleeping] [behind] [Philippa] who is [sitting]

Position [Philippa] [talking] [in front of] [Matthew] who is [talking]

Metadata is "data about other data", of any sort, in any media. An item of metadata may describe an individual content item, or a collection of data including multiple content items and hierarchical levels. In data processing, metadata provides information about, or documentation of, other data managed within an application or environment. In the context of this project ‘metadata’ refers to a variety of categorizations and relationships within and between portraits. Every clip will be categorized by metadata fields, for example:

As can bee seen the metadata classifications can be quite descriptive. The more metadata description, the easier it becomes to make meaningful relationships between the clips.

The meta-script is a term used to describe a series of context-free grammars and metadata categories that are used by a computer to create interesting combinations and compositions of portraits. A ‘scene’ in this script would look like the context-free grammar scenario above. That is, it would describe a generic ‘sentence’ or series of variables that are then ‘filled’ with metadata categories from the portraits. This leads to being able to create quite complex, yet naturalistic relationships between the clips.

DNA or Genetic Programming is a computer science term for a process whereby DNA replication, selection and mutation are simulated inside the computer.

It is a more natural and organic way to get a computer to select sequences than a purely random generator. Random systems tend to be more predictable, repeat often and do not take into account the actual data they are selecting. DNA/Genetic algorithms on the other hand tend to ‘approach’ a more human, less predictable and organic selection due to the nature of ‘evolving’ toward a series of possible answers based on the data provided.

 

In the context of this project Audio-Visual Analysis refers to the process of analyzing the portrait clips to find out information about how long they are, how loud people are, how much they are moving, etc. 

This is a highly developed area in computer vision and hearing. Basically it is possible to find out a lot of information about movement, volume, color, etc using algorithmic methods. This makes it possible to apply visual and aural composition rules (such as an even visual distribution of people, no more than one person speaking loudly, etc).