The Evolution of Content Metadata

If we go back 30 years, content metadata was very simple: scene, shot and take information that applied to the entire ‘media file’ although the ‘media file’ was more likely a piece of film negative, or a section of a tape.

At that time, most content produced was based on a  script and it made sense to log metadata against the script. Scene and Shot are really shortcut descriptions (keywords if you like) of a section of script.

While scripted production continues at about the same pace, non-scripted production of all kinds has exploded over the last 20-30 years, and logging to a script made no sense: non-scripted production has no script by definition. What made sense is logging the content that was shot.

All content is tied to a time range in media but unfortunately the NLEs of the day were still focused on Scene, Shot and Take. Media Composer, classic Final Cut Pro and earlier iterations of Premiere Pro had all their logging focused on scene, shot and take. These metadata fields were also focused on adding metadata to the entire clip – the full duration of the media file.

As long takes became common for interviews and long form coverage of events, the inability to apply Content Metadata to only a range of time within a media file was a limitation. The only ways of applying Content Metadata to only a range of media required either:

  • Using Markers to span time ranges, or
  • Create Subclips for the media time ranges, and sorting the Subclips by Bins. Bin names are effectively Content Metadata for the clips in that Bin.

With Final Cut Pro X and Premiere Pro CC, we have other options. In Final Cut Pro X we can apply Keywords to a time range in a media file, which automatically sort into Keyword Collections. In Premiere Pro we can apply Temporal XMP to time ranges (as implied by the name), but we lack the self-organizing features of Keyword Collections.