Create a mosaic and build the corpus

Name it, drop the clips you want the matcher to draw from, then start writing

A Mosaic is a script-driven assembly. Before you can write the script, the matcher needs a pile of footage to pull from. That pile is the corpus, and you build it by dragging clips onto the workspace.

Make the Mosaic

Right-click the Mosaic section in the sidebar, choose New Mosaic, give it a name, hit Create. If the workspace is empty, the big "Create new Mosaic" button does the same thing.

Name is required. Frame rate inherits from your global Settings default and stays linked unless you turn off "use default" on this Mosaic.

Drop in the clips

Drag clips from the library onto the workspace. Each one you drop gets enrolled in this Mosaic's corpus. The right side of the workspace shows a chip strip with every enrolled clip, thumbnail, color, and the order you dropped them in.

Drag the same clip in twice and nothing happens. The corpus dedupes silently. Right-click a chip and choose Remove from corpus to take a clip out.

The matcher only sees clips in this corpus. Clips sitting in the library that aren't on the strip are invisible to it.

Peek at a clip's transcript

Click any chip in the strip. The right pane loads that clip's full transcript with word-level playback. Click a word, the playhead jumps. This is how we scan a clip to remember whether the line we want actually lives in there.

Once we Submit later and the workspace switches to attribution view, focusing a clip from a splice will highlight the matched word range inside that transcript automatically.

A clip without a transcript

Drag in a clip that hasn't been transcribed yet and the matcher just skips it. No error, no warning, it's silently absent. If you expect matches from a clip and aren't getting any, check whether it's been transcribed.

We learned this one the hard way more than once. Transcribe before you submit, save yourself the head-scratch.