How to Track Text to a Moving Object in Video (Motion Tracking)

What motion tracking does

Motion tracking pins a text element, label, or graphic to a specific point in your video. As that point moves through the frame, the text follows it perfectly. The result looks like the text is physically attached to the object or person in the scene.

You have seen this technique in sports broadcasts (player names following athletes), product videos (feature labels tracking with the product as it rotates), travel content (location names moving with landmarks), and social media (name tags following people).

Without motion tracking, text stays in a fixed position while the subject moves away from it. With motion tracking, text and subject move as one.

The traditional way, and the AI shortcut

For years, motion tracking meant real work in After Effects or DaVinci Resolve. You'd manually pick a tracking point — a high-contrast spot on your subject — let the software analyze that point's pixel movement frame by frame, then parent your text layer to the resulting tracking data. It works, but it assumes you understand tracking points, parent layers, and how to clean up the inevitable frames where the track slips. It's a skill, and a fiddly one.

AI tracking collapses all of that into a click. You click the object or person you want to follow, and the model identifies the subject, learns its visual features, and tracks it through the clip on its own — no point selection, no layer parenting, no cleanup pass. The browser-based depth text tool takes this approach to place and track text relative to the subjects in your footage, which is what makes the whole technique accessible to people who never learned a compositor.

In practice: open the depth text tool, upload your video (any footage works, though a clearly visible subject helps), and click directly on the thing you want the text to follow so the AI can start tracking it across every frame. Add your text — choosing font, size, color, and its position relative to the subject — and it holds that relationship as the subject moves. Play it back to confirm the track is smooth, nudge any frames where it drifts, and export with the tracked text rendered in.

What it's actually for

The technique earns its keep wherever a label needs to belong to a moving thing. In product videos, feature labels follow the product as it rotates in someone's hand — "12MP camera" stays on the camera module, "USB-C" stays on the port — which reads as clean and informative. In fitness and sports, form cues pin to body parts ("keep elbows in" riding the athlete's elbow), which is genuinely useful for coaching breakdowns. In travel content, location names stay attached to landmarks as the camera pans. Team introductions become more dynamic when name tags follow each person instead of sitting in a static lower third. Cooking videos can label ingredients as they move into the bowl. And before/after reveals can pin their labels to the specific areas that change, holding position through any camera move.

Getting a clean track

Tracking quality depends mostly on what you give the AI. Good contrast between subject and background helps most — a red ball on a green field tracks effortlessly, a gray object on a gray wall challenges any tracker. The subject needs to stay visible; if it leaves the frame or gets fully blocked by something, the track pauses or fails for those frames, so keep it in view. Normal movement tracks cleanly, but very fast pans or rapid subject motion can introduce lag. Stable footage helps without being required, since the AI tracks relative to the subject rather than the frame, though very shaky footage adds difficulty. And good lighting with clear edges always beats a subject buried in shadow.

The text itself should be styled for motion, which is different from styling a static caption. Keep it short — one to three words, because long sentences glued to a moving object feel chaotic. Use bold, readable sans-serif fonts, since movement reduces legibility. Add a semi-transparent background box so the label stays readable against whatever passes behind it. And match the aesthetic to its context, using brand fonts and colors for product labels so the text feels integrated rather than stuck on.

Tracked text layers naturally with the rest of your finishing: a film grade under the labels, auto captions handling dialogue while the tracked labels handle visual callouts, depth text for elements that sit behind the subject while the tracked text rides in front, and zoom animations for dynamic reveals. The fastest way to see what it does for production value is to open the depth text tool, drop in a clip with a clear subject, and place a label that moves with it.

Related: How to add text behind a person | Add text to video for free