Let’s let the excellent folks at Moviola introduce this topic for a look at why metadata is useful pre-production through post and even on to distribution and archival.
In discussing organization, I mentioned how the future of organization seems to rely ever-more on metadata. Much of this can be input during production via an iPad and compatible camera app, or with robust tools like Lumberjack, or Prelude, depending on the NLE being used. In the “Movie Slate” section of the slate we look at a camera and NLE agnostic approach to inputting metadata via the iPad slate and syncing it with Resolve. These are fancy modern metadata input methods, but many times, it’s just a patient assistant editor manually adding metadata in post.
If you are lucky enough to have metadata, but it’s in a separate file, Resolve has a feature to assign metadata from a CSV file to clips based on timecode or file name.
If you have a professional sound crew, important metadata should be stored in your audio files and you’ll see it appear during sync. Surprisingly enough, this is a very common way to get scene and take metadata into the video file. As mentioned, video metadata standards are anything but standard, yet audio metadata embedded in a broadcast WAV file is pretty reliable.
Metadata generated by third party software is usually not held inside the media file. No NLE, well, except Premiere, generally dares touch the original media file as it’s not good practice. What this means is that much of your metadata will not transfer between NLEs. This can be a real pain if you’re working between multiple applications.
A variable is basically a placeholder value that represents some other dynamic value. Trigger these by typing % in Resolve and a helpful list will auto-populate. These can be used for clip names as well as for output file names.
All of this said, much of the monotony of the above work is likely headed to obsolescence for something wicked this way comes. The machines are taking over, and metadata is their specialty. Many entry-level jobs will hopefully be made obsolete by upcoming machine learning technologies. Computers will analyze clips and tag the people in them, easily recognizing things like shot framing, subject motion, etc. DaVinci Resolve 16 is already planting themselves firmly in this territory and in a day and age where you can download free Python libraries to generate rough roto mattes automatically, I’m excited to see the development of Fusion over the next several years.