3D camera tracking / matchmoving
Reconstructing the real camera's move from footage so added 3D objects inherit its perspective and stay planted in the scene.
See it
What it is
3D camera tracking, or matchmoving, analyzes many 2D features across a shot and reconstructs the camera that could have produced their motion. The solve estimates camera position, orientation, and often focal length, then places track points in a virtual 3D scene. Added objects can share the original shot's parallax instead of sliding over the image.
Reach for it when 3D text, set extensions, debris, creatures, or other CG elements must sit in live-action footage from a moving camera. It is a scene solve, not an object tracker: the useful points belong to the stationary environment, while moving actors and cars should be ignored or masked out. After Effects has a built-in 3D Camera Tracker; feature work usually goes to dedicated solvers such as SynthEyes, PFTrack, or 3DEqualizer.
Gotcha: the solver needs parallax. A locked shot, a pure tripod pan, heavy motion blur, rolling-shutter wobble, or a wall with no detail can leave depth ambiguous. Correct strong lens distortion first, inspect the ground plane and horizon, and test the solve with a simple cube before spending time on the final asset.
Ask AI for it
Solve this moving shot with Adobe After Effects 3D Camera Tracker. Mask out the moving people before analysis, use Fixed Angle of View unless the shot visibly zooms, and delete track points attached to moving objects. Select at least three coplanar points on the floor, set the ground plane and origin, then create a 3D Camera and a null. Parent a test cube to that null and verify that it stays planted with correct parallax across the entire shot before replacing it with the final 3D element.