Classify Your Footage
Classify labels each file in the media bin with a short code that says what kind of footage it is. This page covers how it runs on import, how to run it by hand, and how to correct a label.
For video, Classify works out the shot type, the framing, and how many main subjects are in frame. It also picks the frame that makes the best thumbnail. For audio, it works out whether the file is speech, music, or sound effects. Other tools rely on these labels. Face-camera files come pre-checked in the Face Tracking and Find Characters dialogs, and Camera Selection shows the code on each camera lane.
The video codes are: FAC Face, SCN Scene, WRK Work, SBJ Subject, HND Hands, SCR Screen, BRD Board, POV POV, MIX Mixed, GFX Graphics, ANI Animation, and FLM Film. Audio files get SPEECH, MUSIC, or SFX.
Let Classify run on import
Classify normally runs by itself on every file you import. The switch lives in Preferences, in the Import section: Auto-classify imported media. The setting's own description explains the trade-off: "Classifies new clips by shot type or audio kind so the AI editor can use them. Costs roughly $0.001 per clip. When off, classify manually via right-click → Classify."
If a needed API key is missing, Classify is skipped for the whole import and you get one notice reading Auto-classify skipped, followed by which key to add. The notice has a button that opens Preferences → API Keys. You get one notice per import, not one per file.
Classify files by hand
- Save the project. Classifying an unsaved project shows the error "Save the project before classifying."
- Select the files in the media bin.
- Right-click the selection and open the Classify submenu.
- Click Auto-Classify.
There is no dialog. The run starts right away, and the status bar shows Classifying while it works. When it finishes there is no notification. The badge just appears on the card.
If Auto-Classify is greyed out, hover it for the reason: Select media first or Source file is offline. When some of the selected files are offline, the others still run and a note says Skipped N offline item(s).
Set a label yourself
The same right-click submenu lists the codes below Auto-Classify. The list matches the file you right-clicked: a video file offers the video codes, written as FAC — Face, SCR — Screen, and so on, and an audio file offers speech, music, and sfx. A check mark sits next to the file's current code. Click a code to set it yourself, with no AI run and no cost. This works on files that have never been classified, so you can label footage by hand and skip the AI entirely.
The code you pick applies to every selected file of that same kind. Anything else in the selection is left alone. A label you set by hand sticks. Automatic classification won't overwrite it, and only Rerun Auto-Classify replaces it.
When every selected file already has a label, the submenu is named Edit Classification and its first entry becomes Rerun Auto-Classify. It is the same menu either way.
When classification fails
Failed classifications collect behind a warning triangle in the status bar with an error count, for example 2 errors, rather than piling up messages on screen. Click it to open the Errors list. Each row names the file and the reason it failed. Dismiss a single row with its ×, or empty the list with Clear all.
Read the results
A classified video's bin thumbnail switches to the frame the AI picked. The card also gains a code badge, for example FAC on a face camera or SPEECH on a voice recording. Hover the badge for the details, such as "FAC, (medium, 1 subject)" or "speech, loud, 2 voices". A label you set by hand has no such details behind it, so its badge shows the code alone.
For the same details in text, click a card's meta row to open the Module Results popover. Its Classify row reads, for example, "Shot type FAC · medium." or "Audio kind SPEECH · 2 voices."