DEEP FILM ACCESS Project (Digital Transforma4ons in the Arts and Humani4es: Big Data) February 2014 April 2015 Dr Sarah Atkinson (PI) s.a.atkinson@brighton.ac.uk
Interdisciplinary Principal Inves4gator: Dr Sarah Atkinson (School of Art, Design & Media) Co- Inves4gator: Dr Roger Evans (School of Compu4ng, Engineering & Maths) Partners:
Overview/Problem The film industry, academy and film archival ins4tu4ons currently face a significant challenge in evolving long- term preserva4on methods for digital film. Big film data is a daun4ng object for preserva4on and analysis but it could reveal many untapped opportuni4es. En4re data sets of individual feature film 4tles can exceed mul4ple terabytes.
Impera4ves Mul4ple interrelated streams of data are generated during the filmmaking process. These are currently stored separately, and their ini4al rich interrela4ons and links are lost. This project seeks to advance the integra4on of these dispersed materials to meet four key impera4ves: 1 Reducing the duplica4on of archival processes 2 Improving the integrity of data 3 Making the content of digital assets visible to search engines 4 Facilita4ng the mining of embedded or hidden data through deep searches.
Aims: To both improve ease of access to film and increase the depth and value of that access; To develop new forms of processing big and complex data sets generated by industrial digital film produc4on and evolve new methods by which to explore them; To improve and evolve discovery and research methods of big and complex data sets generated by industrial digital film produc4on; To contribute to the expansion of film and digital literacies within Arts and Humani4es research; To extend metadata protocols within film archival prac4ces; To contribute to the development of new ways in which the end users of film and media data can access and make use of archives and make links across different archives where data might be stored.
Outcomes A pilot / prototype integrated data model; A survey of metadata for all digital asset types associated with digital film produc4on; An established set of requirements for the analysis of the long- term management of digital film assets; An established mapping of exis4ng metadata to a more integrated view; The ini4a4on of a methodology for integra4ng assets with their associated metadata; The Ginger & Rosa structured and integrated dataset as an independent resource to be made openly accessible via www.sp- ark.org (in addi4on to being the proof- of- principle).
DEEP FILM ACCESS Will develop an integrated framework which will transform the big, dispersed and complex materials generated by industrial digital film produc4on into rich, integrated and malleable data sets, interweaving them into the DNA of the film itself.
Research Ques4ons 1. How can access to digital film as a big dataset be facilitated, expanded and deepened? 2. How can a system be devised that will link and layer the complex metadata that is generated during produc4on? What metadata schema will cover all the asset types in a consistent framework? To what extent can the crea4on of metadata be automated? To what extent will browsing and searching need to rely on basic text- based searches (e.g. manual tagging)? 3. What possibili4es for novel approaches to research will be facilitated through this integrated approach? For instance, how can access to the spa4al and temporal dimensions of digital film (through frame by frame analysis) reveal unique aspects hidden within a film sequence? Could advanced analysis of the assets captured during produc4on empower novel post- produc4on tools (e.g. in shot selec4on and sequencing)? 4. How can this integrated approach advance the field of digital film studies? For instance, if thousands of digital film sources could be analysed in parallel, what research challenges could become accessible, and could close qualita4ve examina4on and large scale analysis work together in this framework? How might this approach influence other disciplines? Others?
DEEP FILM ACCESS Project (Digital Transforma4ons in the Arts and Humani4es: Big Data) Dr Sarah Atkinson (PI) s.a.atkinson@brighton.ac.uk