description: "After Effects Motion Tracking Basics — Back in After Effects, this module covers tracking — locking a layer (a graphic, text, or blur) onto motion that…"---

09 · After Effects Motion Tracking Basics¶

Back in After Effects, this module covers tracking — locking a layer (a graphic, text, or blur) onto motion that already exists in footage, such as a screen insert that needs to stick to a moving phone, or a logo that should ride along a moving object. This covers the Tracker panel's 2D point tracking and the simpler Track Camera for 3D scenes.

1. The Tracker panel and 2D Point Tracking¶

Select the footage layer with the motion to track, then open Window > Tracker. Click Track Motion to enter tracking mode — the layer opens in its own Layer panel with a track point overlay.
A track point has two boxes: the inner feature region (the exact visual pattern to follow — pick a high-contrast, distinct area like a corner or sticker) and the outer search region (how far AE looks for that pattern in the next frame — wider for fast/erratic motion).
Choose a Track Type in the Tracker panel: Transform (Position and/or Rotation/Scale from one point, or two points for Rotation/Scale), Perspective corner pin (four points, for a screen/sign insert that needs to follow a surface's perspective), or Raw (data only, for custom expression use).
Position the track point over the chosen feature at the layer's first frame, then click the Analyze forward button (the right-pointing play icon in the Tracker panel) to track frame-by-frame automatically.
Watch playback as it analyzes — if the track point visibly drifts off the intended feature partway through, stop, manually correct the point at that frame, and re-analyze forward from there rather than letting a bad track continue.

2. Applying the track¶

Once analysis finishes and looks clean scrubbing through, select the layer that should follow the track (e.g. a logo or text layer) and click Edit Target... in the Tracker panel to confirm it, then click Apply.
For a Transform track, AE writes Position (and Rotation/Scale, if tracked) keyframes directly onto the target layer, matching the tracked motion frame-for-frame.
For a Corner Pin track, AE applies the Corner Pin effect to the target layer with its four corners keyframed to the four tracked points — used for a screen replacement (e.g. compositing a new UI onto a phone/monitor visible in the shot).

3. Track Camera (3D scenes)¶

Track Camera (also in the Tracker panel, or Animation > Track Camera in Composition) analyzes an entire shot for camera movement and scene depth, generating a cloud of 3D track points across the footage rather than a single 2D point.
Once analysis completes, click on a well-tracked point in the scene (ideally one on a flat, stable surface) and choose Create Null and Camera (or Create Text/Solid) from the right-click menu on a selected track point — this creates a 3D-tracked null/camera that any new layer can be parented to for it to sit convincingly in the 3D scene, respecting perspective as the camera moves.
Track Camera works best on footage with plenty of fixed, high-contrast detail throughout the frame and a camera that moves with some parallax (not a purely static locked-off shot, which gives it no depth information to solve).

4. Stabilizing footage (the inverse use of tracking)¶

Warp Stabilizer VFX (Effect > Distort > Warp Stabilizer VFX, or drag from Effects & Presets) analyzes shaky handheld footage and smooths it automatically — no manual track point placement needed.
Key settings in Effect Controls: Result (Smooth Motion keeps some original camera move; No Motion locks it fully static), Smoothness (percentage of shake removed), and Method (Position, Position/Scale/Rotation, Perspective, or Subspace Warp for the most aggressive correction).
Warp Stabilizer needs to fully analyze the clip (shown as "Analyzing" then "Stabilizing" in the Effect Controls and directly on the layer) before playback reflects the final result — this can take real time on longer clips.

Worked example: a phone screen insert¶

Import a clip of a hand holding a phone that moves/tilts during the shot, with the phone's screen visible.
Apply a Perspective corner pin track: place the four track points at the phone screen's four corners on frame 1, and Analyze forward.
Import the new screen graphic as a solid/still, Edit Target it to the phone footage's track, and Apply — the Corner Pin effect keyframes now follow the phone's screen perspective throughout the shot.
If the source footage itself is a little shaky, apply Warp Stabilizer VFX to the phone footage first, before tracking, so the track has less erratic motion to follow.

Cheat sheet — motion tracking¶

Task	Where
Open Tracker panel	Window > Tracker
Start a 2D point track	Track Motion button
Choose Transform / Corner Pin / Raw	Track Type dropdown in Tracker panel
Run automatic tracking	Analyze forward/backward buttons
Apply tracked motion to another layer	Edit Target... then Apply
3D camera + scene solve	Track Camera
Attach a new layer to the 3D solve	Right-click a track point > Create Null and Camera
Smooth shaky footage	Effect > Distort > Warp Stabilizer VFX

How It Actually Works¶

2D point tracking is per-frame template matching, not motion prediction. At each step, After Effects takes the exact pixel pattern currently inside the feature region and searches the next frame's pixels within the search region for the location that best matches that pattern (typically via a correlation score, similar in principle to the cross-correlation used for audio sync in Level 2's multi-cam module, just computed on 2D pixel neighborhoods instead of a 1D waveform). Whichever location scores highest becomes that frame's tracked position, and the feature region's reference pattern is then updated (or kept, depending on settings) for the next comparison — which is exactly why picking a high-contrast, visually distinct feature matters: a flat, low-contrast area produces many similarly-scoring candidate matches nearby, giving the algorithm nothing to reliably lock onto, and why drift compounds once a bad match is accepted as the new reference pattern for the following frame.
A Corner Pin track's four independently-tracked points are converted into a perspective transform matrix applied to the target layer every frame. Rather than moving the whole layer as a rigid rectangle, Corner Pin computes the specific projective transformation that maps the target layer's four corners onto wherever the four tracked points currently sit — the same class of transform that lets a flat rectangle appear to tilt and recede in perspective, recalculated fresh at each frame from that frame's four tracked positions.
Applying a track writes literal keyframes onto the target layer's actual properties — it's a one-time data transfer, not a live link between the two layers. Clicking Apply reads the tracker's computed per-frame values and creates ordinary Position/Rotation/Scale (or Corner Pin) keyframes on the target, identical in kind to hand-placed keyframes from Module 2 — which is why re-running Analyze on the source footage afterward does not automatically update a graphic that was already applied; the target's keyframes are now independent data, disconnected from the tracker.
Track Camera solves for both camera motion and 3D scene geometry simultaneously using structure-from-motion: it finds a consistent set of 3D point positions and a camera path that would produce the observed 2D pixel motion across the whole clip. This is inherently an optimization problem with many candidate high-contrast features tracked in parallel across many frames, solved for the single camera-path-plus- 3D-points combination that best explains all of them together — which is exactly why it needs real parallax (points at different depths moving differently as the camera shifts) to be solvable: a purely static shot gives identical 2D motion for every point regardless of its 3D depth, so there's no depth information the solve can distinguish.
Warp Stabilizer works by tracking many points across the frame (much like a Track Camera pass) to estimate the unwanted camera motion, then applies an inverse transform to cancel it out, before subspace-warping the frame edges to fill the resulting gaps. Removing camera shake means each frame has to be repositioned/rescaled to compensate — which necessarily pushes some original frame content outside the visible boundary and requires either cropping in or synthetically extending edge pixels (the "Subspace Warp" method), which is the actual computational reason the analysis pass takes real time: it's solving the same kind of motion estimation problem as ordinary tracking, just densely across the whole frame rather than one feature.

Exercise¶

Track a moving element in a handheld or phone-screen clip using a Transform or Corner Pin track, apply it to a new graphic/text layer, and confirm the graphic sticks to the tracked feature throughout playback with no visible drift. If the source footage is shaky, apply Warp Stabilizer VFX before tracking and compare tracking accuracy with and without it.