5 minute read
You’ve almost certainly heard the term metadata and probably have some idea that it means information about frame rate and codecs. That’s true, but metadata is much more. If you break the word apart into “meta” and “data”. Meta literally means “about the thing” and “data” is the thing.
In the world of production, the “thing” is referred to as the “essence,” i.e. the media itself. Everything else is metadata. Everything we know about the media is metadata, whether it’s technical information or information about the content, everything we know is metadata.
Metadata is not new; it’s just a new name for something you’re probably already using. Instead of calling it metadata you’re much more likely to be calling it “notes”, “log notes”, “script notes” or “logging”.
Notes about the script, location notes, script notes, tape or card labels, log notes, and even transcriptions are additional descriptions relating to the shoot and edit that aren’t the media itself. They are all metadata.
This isn’t anything new, but the digital world lets us do so much more with metadata than we could with paper notes. That they are searchable alone make a huge difference.
In our media-centric world, we have technical metadata that describes the technical aspects of the file, such as frame rate, image size, codec, etc. Technical metadata is easy to derive because it’s inherent in the characteristics of the file.
Technical metadata applies to the entire media file (and all clips derived from it) because from the technical perspective, every frame is the same. Technical metadata tells us nothing about the content of the file.
Content Metadata describes the contents of the film or video image by applying a description of the content to all, or part, of the file. Without the Content Metadata, we know nothing about the content of the file. Content Metadata can apply to only part of a media file because the description will change as the content changes. An example of a piece of Content Metadata might be “Close up”, but if the camera moves, it might become a “Medium shot” within the same clip.
Metadata doesn’t just apply to our media world. Metadata is being gathered as we watch videos on YouTube and rate them, as we make purchases online, and as we make phone calls. You will find more on the connection between metadata and the uses discussed on the news in Metadata and the Wider World.
In the words of filmmaker and archivist Jason Scott’s, “Metadata is a love note to the future.”
As we alluded to above, in post-production there are two primary types of metadata: technical and content. There is a third use of metadata in the media and entertainment industries, and that’s distribution metadata covering the content and rights associated with a finished program or movie.
Technical Metadata covers the range of information we get from the camera (generally). Not surprisingly, it is the technical information about the media file. This includes the obvious technical information on Image size, Frame rate, Codec, etc. But technical metadata goes beyond the obvious and – depending on the camera or source – can include focus, white balance, GPS, camera and lens serial numbers, and a lot more metadata than you would think necessary.
Technical Metadata is used to manage media and to automate operations.
Technical metadata comes with the file from the camera because it is dependent on the camera and the type of file(s) it produces. Some technical metadata is in the file itself, but in other cases, it resides in a separate “sidecar” file in the same directory as the media file. Sidecar files are usually a XML files with relevant information in them. Each type of XML file is specific to the camera that created it.
These sidecar files are why it is important to always maintain the entire card structure before importing to an NLE.
Content Metadata is the part of the metadata spectrum that is most useful in production. This is information about what is in the media file. Once it was obvious (though slow) to find out what was in a frame of film by holding the film up to the light, but now metadata about the content of the media file is how we understand what we have.
Content Metadata can be as concise as a Keyword or as verbose as a transcript. Content metadata is generally – for the moment at least – added by a human somewhere before editorial starts. We say “mostly” because increasingly indexing speech and images for searching can be done by trained machines, sometimes inaccurately referred to as “Artificial Intelligence.”
Machine-based logging and metadata generation has its limits because, at the current state of the art, there is no discrimination about what is useful metadata and what is simply cluttering up the index!
Maximum discoverability makes sense in an asset library, where vast amounts of video are stored, but the flood of metadata out of machine-driven indexes often becomes too overwhelming to be useful in production. Content Metadata for production needs to be focused on the needs of the production and, aside from basic image search and transcripts, likely to be too broad to be useful.
Content Metadata is what we use for organizing projects and getting a head start on editing.Content Metadata is usually applied and managed within an NLE. Content Metadata generally does NOT travel with the file, but rather in the NLE’s Project file(s).
Content Metadata does not need to apply to an entire media file. It is more valuable when it is applied to content-based ‘selects’ or defined time ranges.
Distribution Metadata solves a problem most of us don’t have: tracking the distribution of finished programs through their many territories and variations. The industry solution is EIDR: Entertainment IDentifier Registry. It is a unique ID for every variation of every film and television episode ever made.
Because it is unique, the EIDR can be used by third parties – most notably ROVI – to add additional value for program guides by linking the program/episode ID to other data sources.
Video collaboration solved.