Automatic Cinema
Automatic Cinema computes narratives from a media database (video, text, sound). It includes a system for capturing and tagging media, as well as for steering and controlling the performance. The aim is to free film from its static, once-and-for-all edited form and to enable a multitude of ways of viewing it. The concept is multimedia and multichannel, serving film just as well as spatial installation, theater, or art performance.
Geometric Storytelling
On a desktop, the important things are often found in the top right (Mac) or bottom left (Windows). Many users probably arrange their data according to their own logic: office stuff in the top left, the latest project in the middle, temporary data in the bottom right. One could therefore divide the desktop into regions and label them; the closer an icon is to the center of a region, the more strongly it possesses that region's quality.
Automatic Cinema stores the video data in a two-dimensional matrix. Regions of this matrix are defined by a keyword, so the video data acquire a stronger or weaker thematic association depending on their proximity to the keyword within the matrix. Keywords can be placed in causal-logical or relational connections using arrows or lines. The colors of the lines additionally define a positive or negative value. By applying style information, Automatic Cinema can compute a narrative path and use it to select media data.
Components
The infrastructure of Automatic Cinema consists of server software for storing the media and a command client for control. To play back the media data, several media players communicate with the controller in real time.
Controller (2012)
Displays the generated narrative as a timeline, sets the style parameters as a two-dimensional matrix, and activates tags to define the content of the narrative.
Heksler (2014)
Software for automatically fragmenting video sequences at hard cuts, tagging them, and storing them in the Automatic Cinema database.
Further reading:
- https://automatic-cinema.com
- https://github.com/urshofer/heksler
