Edit

Metadata-driven editing

6 minute read

What is Metadata?

Metadata is extra information attached to your source files, which is used to describe those files in finer detail.

For example, have you ever opened a shot from your phone in your favorite image editor? If you head to the Properties or File Info dialog, you’ll see some interesting metadata. Your phone’s make and model, the shot’s exposure and shutter speed, even the location where the photos were taken – all of this extra info is seamlessly folded into every shot on your phone. That’s metadata.

But how does metadata affect postproduction? Let’s sift it into two categories:

  1. 1
    Useful Metadata for Systems
  2. 2
    Meaningful Metadata for Humans

Useful for Systems

Systems use metadata to process files in granular ways. For instance, operating systems use metadata to display common file attributes such as Type, Size, and Date Modified, which can be visually grouped and sorted with the click of a column.

NLE’s and VFX tools also use metadata to expose crucial details about your project’s assets. Is this clip a QuickTime MOV, or MP4? Is the codec a flavor of ProRes, or H.264? What’s the frame rate? Bitrate? Does it have Linear PCM 48 KHz stereo, or no sound at all? All of this metadata is baked into your files to help this system, the host application, do its job.

Meaningful for Humans

But not all metadata is for machine-use only. More and more postproduction software environments give you, the human, the tools necessary for applying and wielding metadata in a meaningful way.

Keywords

Keywords are specific terms or phrases you can attach to each clip. Once you’ve created your own keyword set, it can help you and your team get (and keep) your arms around your project. The good news? You don’t need an exhaustive set of keywords to see the payoff.

For example, keywords can help you stay organized. Some apps generate keywords based on your project’s folder structure when you import media. It’s a rudimentary method of asset tracking, but if you’re following your project’s file and folder naming convention, it’s another breadcrumb trail for your assets’ source of origin. This comes in handy when working with incoming assets from disparate sources.

Keywords also help you look at your footage as ideas, not just as clips in a bin. Names, places, emotions, moments – all of these ideas are candidates for keywords that can quickly become part of your team’s vocabulary.

Once those keywords are available, that metadata becomes the basis for locating any asset in your project. Type some keywords into your NLE or VFX app’s Search function to quickly narrow down the number of clips. Don’t know any of the keywords? Some apps allow you to filter results by selecting keywords with checkboxes and logical conditions. Others let you save those keyword searches or filters for later use. In other words, adding keywords lets you create a custom search engine for your project!

So do you have Interview shots of John Smith, but also shots of John in the Factory as B-Roll? Isolating those ideas, then translating them into keywords, will help you find what you need even faster during the edit.

Project-Specific Information

Project-specific metadata may also come from other sources like cameras or Script Supervisors, in the form of:

  • Reel
  • Scene
  • Camera make and model
  • Camera angle
  • Crew and personnel names
  • Circle takes (takes that the director liked on set)
  • Continuity notes

If available, these become another way to quickly find and filter project assets so you can keep that momentum going during the edit.

Metadata – useful for systems, but can be meaningful for you. All postproduction apps support system-grade metadata, but more and more are supporting meaningful metadata which allows you to organize and sift through hours of footage in mere seconds. When used effectively, it can help you make creative decisions more efficiently, collaborate successfully, and deliver on time.

Editing with Metadata

When you use an NLE that is very highly metadata-oriented, and you make heavy use of its metadata features, it can open up an entirely different way of organizing and selecting your footage.

Organizing clips and folders is one of the most important tasks in editing because, if you do it properly, it allows you to find the clips that you need much more quickly. Traditional NLEs have a whole suite of features that allow you to identify the right clips quickly.

  • Hierarchical Bins. Many editors organize their clips into a series of nested bins (a.k.a. folders). The level may be three bins, one for each act of the film. Inside each act is a bin for each scene, and inside each scene is a bin for each camera, and inside each of those is a group of clips.
  • Subclips. A subclip is a portion of a clip that is saved as its own clip. You may have a 3-minute-long take that’s mostly not great, but there’s one particular line that’s beautifully spoken. The editor might create a subclip of just that one line.
  • Stringouts. A stringout of a scene is just a sequence that includes every single shot from the scene laid out in a row. The editor can then watch through and choose the best ones.
  • Selects. A selects sequence is a collection of just the best clips from a scene. It’s common to start with a stringout sequence, make a copy, and then delete all of the not-as-good takes. What’s left are your selects.

Metadata-driven editing replaces all four of those techniques with two simple but powerful functions: tagging, and searching.

Tagging

Tagging happens at the beginning of the project, and it’s as simple as it sounds. The editor (or an assistant) tags each clip with all of the attributes that may be useful later. Some software calls these “keywords”, but the concept is the same.

A clip will be tagged with the obvious information like the act number, the scene number, the camera, etc. In fact, those may automatically be applied if the production crew has handled their metadata properly.

But there is also a lot of more interesting metadata to add. You might add the names of the characters in the shot, whether the director was happy with the performance, the type of shot (wide, medium, close-up), whether the shot is moving or not, whether a particular prop is visible in the shot, or whether or not the actor remembered to wear his vest.

You can also add metadata to a portion of a clip, instead of to the whole clip. In the above scenario where a traditional editor might make a subclip, you would mark that same section as “good.”

Some of this metadata can even be added automatically by computers using machine-learning technology to automatically detect characteristics of the shot.

Searching

Now that you’ve put in all of the work to tag your footage with metadata, it’s time to make use of it. “Collecting” is just a general term for how you work with metadata, and it’s as simple as typing a few words into a search bar.

Typing “4 Rachel” will instantly show you all of the shots of Rachel from scene 4. If you add “best”, then it will only show you the takes that you’ve also marked as the best ones. Perhaps Rachel begins the scene by sitting on the couch and then walks to the window. If you add “couch”, that show you the best shots that include both Rachel and the couch in scene 4. You did all of that just by typing “4 Rachel best couch”.

Now that you’ve collected the shots you wanted, you can save this view for later, but that is often not necessary, since it only takes a couple of seconds to type those same words again.

If you want a stringout of scene 4, all you do is type “4” into the search bar. If you want your selects of scene 4, you just type “4 best”.

As you can see, metadata-driven editing tends to be much more fluid and adaptable than traditional organizational methods, but it requires the editorial team to shift how they think about sorting and finding clips, which can be disruptive to existing workflows.

FCP X was the first NLE to provide advanced metadata features, but metadata-driven editing is gaining in popularity and many different tools are beginning to add some similar features.

Continue readingfor free

Unlock all 100,000 words of the Frame.io Workflow Guide and learn how the pros do workflow.

Workflow Guide

Read the full guide

Presented By:

Video collaboration solved.