Workflows

Basic Tracking

The first graph worth building. Footage goes in, a solved camera and point cloud come out.

Start here. Once you've done this once, the rest of the add-on makes a lot more sense.

The Graph

Footage
  -> Convert Footage        -> Import as Footage
                            -> Auto Track
                            -> Output 3D

Steps

1. Add a Footage node and pick your movie clip.
2. Add Convert Footage, connect it, and press Convert, then Import as Footage. You get a new Footage node pointing at frames on disk.
3. Add Auto Track and connect that node's Image Sequence to it.
4. Press Auto Track. It finds features, matches them across frames, and solves.
5. Add Output 3D, connect the Auto Track output to its Scene input, and press it.

You should now have a solved camera and a point cloud in your Blender scene.

Converting to frames first is worth the wait: solving reads individual frames thousands of times, and that is far faster and more reliable from separate files than from a compressed video.

Importing Files covers the whole input side in detail: video and EXR conversion, frame numbering, cropping and resizing, matching your scene to the footage, and the other input nodes.

Tripod Shots

If the camera only rotates and never moves, there's no parallax, so there's no depth to solve. Use Tripod Track instead of Auto Track. It solves rotation only and connects to Output 3D: Camera.

After The Solve

When It Doesn't Solve

If the solve fails, registers only part of the shot, or drifts, work down the ladder in Advanced Tracking. That page also covers lens calibration from grid shots and plumb lines.