Import Track

Export Track writes .platetrack.json files; Import Track is the other end - it reads one back into a TRACK_DATA object so every downstream node keeps working without re-running the tracker. Sounds boring. It's actually the node that turns tracks into reusable assets instead of one-shot computation.

How it works

It loads a TrackData from JSON - plate size, reference frame, the full per-frame homography chain, confidence, keyframes, all of it - and hands you the same object Planar Track produces. Because TRACK_DATA is a self-contained contract, the track doesn't care whether it was born this session or last week.

Inputs and outputs

• path - a .platetrack.json file, typically one written by Export Track. There's a file picker on the node.

Output: track_data.

What it's for

• Saving and reusing a good track across workflow versions without re-tracking, and without re-paying the compute.
• Sharing: hand someone the JSON and their copy of the pack can apply your track to the same footage.
• Hand-editing: the format is plain JSON, so if you're the sort who fixes tracks in a text editor rather than the widget, you can.

The node checks the schema version on load and raises a clear error if the file is newer than your installed pack supports - upgrade the pack and it reads again.

Installing it

Same pack - PlateTrack in Manager or clone + pip install -r requirements.txt. OpenCV only.

Where people get burned

• Feeding it a file that isn't a platetrack JSON. A Nuke .nk or an AE keyframe text won't parse - those are exports for compositors, not for the graph.
• Expecting it to need the source footage. It doesn't; the track is self-contained. But if you apply a saved track to different footage, that's on you - a track is tied to the plate it was made on.

It's a quiet utility node, and that's the point: tracks become files, and files are portable.