In this cluster are Jake Berger and Stephen McConnachie, who brought to the group their experience and knowledge working with archives and large collections of moving images, from an institutional perspective in the UK. Joined by Mila Oiva, from a research perspective analysing large collections of Soviet newsreels.

These are some of the challenges they highlighted for us in terms of encoding cinematic time for computational analysis:

  • Encoding performative aspects of carrier technology, for production and exhibition. For example common rules of thumb to estimate the frame rate of a projectionist cranking a film reel by hand in early films (Stephen) or high frame rate such as in some of Ang Lee’s films.
  • What’s an acceptable levels of artifice: e.g. Peter Jackson’s frame interpolation and colorisation techniques which was very controversial in the archiving community.
  • Encoding relational historical categories, such as “Victorian” or “post-war”, and non-linear categories, such as cyclical time “weekly” (Mila). This came up again on the discussions on subjective time during the ideation and prototyping sessions.
  • Jake also made the point of defining use cases for this research, keeping in mind the users and uses of archives, as well as the wider purpose of these public collections. He argued that some of the organisations best positioned to use archives data would be the least likely to be trusted with this data (eg. FAANG’s).