Most video reframing tools give you one option: point the AI at your footage and hope it follows the right person. That works fine when there is only one face in the shot. The moment a second person walks into frame, you are rolling the dice. FaceStabilizer takes a different approach by giving you two distinct tracking modes. auto-tracking and face lock, so you always have the right tool for the scene in front of you.
Understanding the difference between these two modes is the fastest way to get cleaner reframes with less fiddling. This guide breaks down exactly how each one works, when to reach for it, and how FaceStabilizer's supporting features keep both modes rock-solid even in tricky footage.
Two approaches to face tracking
At the core of every reframe is a simple question: which face should the crop follow? FaceStabilizer answers that question using advanced AI face detection that handles the tough stuff: side profiles, low light, partial occlusion, even subjects at a distance. It just works.
Once faces are detected, you choose how the tracker behaves. Auto-tracking dynamically follows the most prominent face in every frame. Face lock pins the tracker to one specific person you select, ignoring everyone else. Both modes feed into the same stabilization pipeline and the same export engine. The only difference is how the target is chosen.
Think of it this way: auto-tracking is hands-off and adaptive, while face lock is deliberate and persistent. Neither is universally better. The right choice depends entirely on what is happening in your footage.
Auto-tracking: best for single-subject content
If your clip features one person talking to the camera (a vlog, a talking-head tutorial, a selfie-style Story), auto-tracking is the way to go. You drop your video in, hit track, and FaceStabilizer locks onto the dominant face immediately. There is no need to click, select, or configure anything. The tracker evaluates each frame and follows whichever face is largest and most centered, which in a single-subject clip is always your speaker.
Auto-tracking also handles movement beautifully in solo scenarios. If your subject walks across the room, leans toward the camera, or shifts from a standing desk to a whiteboard, the tracker adjusts the crop smoothly without losing them. Because there is no ambiguity about who to follow, the system can commit fully to one target and optimize the stabilization path around their movement.
This is the default mode for a reason. The vast majority of content that needs reframing (vertical cuts of podcast hosts, solo product demos, fitness walkthroughs, cooking videos) has exactly one face that matters. Auto-tracking nails these cases in a single pass with zero configuration. Most competing tools stop here, offering only this automatic approach and calling it a day. FaceStabilizer goes further.
Face lock: essential for multi-person scenes
The moment a second person appears in your footage, auto-tracking has to make a judgment call. Which face is more prominent? If your interview subject leans back while the host leans forward, the tracker might swap targets mid-sentence. If a waiter walks between the camera and your speaker at a restaurant, the crop could jump to the wrong person entirely. These are not edge cases. They are everyday shooting conditions.
Face lock eliminates the guessing. When you click "Select Face," FaceStabilizer scans the current frame and shows you a message like "3 face(s) found. Click to select," with clickable rectangles drawn over each detected person. You click the one you want, and the tracker commits to that individual for the entire clip. It does not matter if someone bigger walks in front of them, if the lighting shifts, or if five other people crowd the frame. Your locked subject stays centered.
This is indispensable for multi-person scenarios:
- Two-person interviews: lock onto the guest and create a dedicated vertical cut of their answers, then re-track with the host locked to get their side too.
- Events and conferences: the speaker is on stage but audience members, crew, and other panelists constantly enter and exit the frame. Face lock keeps the crop on the speaker.
- Weddings and ceremonies: dozens of faces, but you only care about one. Lock the bride, the groom, or the officiant and let the rest of the crowd blur past.
- Group fitness or dance classes: the instructor is surrounded by participants, but your vertical clip should feature only them.
Without face lock, you would need to manually keyframe the crop position every time the tracker drifts to the wrong person. That is tedious, error-prone work. With face lock, you make one click and the problem disappears.
It never loses your subject
Real-world footage is messy. People turn away from the camera. They duck behind a podium. Someone walks directly in front of them for a few frames. In any of these moments, other tools panic: snapping to a different face, freezing the crop in place, or throwing an error.
FaceStabilizer handles these interruptions gracefully. When the tracked face temporarily disappears, the app holds the last known crop position. The moment your subject's face reappears (even if they have moved to a different part of the frame, changed their head angle, or are partially occluded), it picks them back up and resumes smooth following.
This works in both modes. In auto-tracking, it means the system does not immediately jump to a different face when your subject briefly looks away. In face lock, it means your locked target is never permanently lost just because they turned to the side for a few seconds.
Think of it as a safety net stretched under a tightrope. You do not plan to fall, but knowing it is there lets you move with confidence.
This eliminates an enormous category of tracking failures that plague other reframing tools. No more jump cuts to the wrong person. No more frozen frames. The crop stays where it should be, and when your subject returns to visibility, the tracking resumes as if nothing happened.
Switching stabilization without re-tracking
Here is something that trips up users of other tools: changing the feel of your stabilization usually means re-running the entire tracking pass. If you tracked a five-minute clip and then decided the crop movement is too snappy, you would need to start over from scratch. That is a waste of time and compute.
FaceStabilizer separates the tracking data from the stabilization curve. Once your tracking pass is complete, you can switch between the three stabilization modes. Smooth, Balanced, and Responsive, without re-tracking. The tracking data stays the same; only the way the crop moves between tracked positions changes.
- Smooth: cinematic, gradual crop movement. Ideal for interviews, narrative content, and anything where a calm, steady frame feels right.
- Balanced: the default. A natural middle ground that tracks movement without feeling sluggish or jittery. Great for most everyday content.
- Responsive: fast, tight following. Built for action-heavy footage like sports, dance, or fitness where the subject moves quickly and the crop needs to keep up.
This means you can track once and then experiment freely. Try Smooth for your interview clip. If it feels too slow when the guest gestures enthusiastically, switch to Balanced and preview again instantly. No waiting for a new tracking pass. The change applies to your second pass (the stabilization pass), which recalculates in moments based on the existing tracking data.
One note: if you change which face you are tracking or adjust your trim points, you will want to hit the re-track button (the ↺ icon) to generate fresh tracking data for the new selection. Stabilization mode changes never require re-tracking, but face selection changes always do.
When to use which: a quick decision guide
Choosing between auto-tracking and face lock comes down to one question: how many faces appear in your footage? If the answer is one, auto-tracking is faster and simpler. If the answer is two or more, face lock gives you the control you need. Here is a practical breakdown:
Use auto-tracking when:
- Your clip has a single speaker or subject throughout.
- You are batch-processing solo content like vlogs, tutorials, or selfie-style recordings.
- Speed matters more than precision: you want to drop a file in and export with minimal clicks.
- The face in frame is always the one you want to follow.
Use face lock when:
- Two or more people are visible at any point in the clip.
- You need separate vertical cuts for different speakers in the same footage.
- Background people, crew, or passersby might distract the tracker.
- You are working with event footage, ceremonies, panels, or any crowded scene.
- Your subject is not always the most prominent face. They might be smaller in frame or off to one side.
And remember, you are never locked into your first choice. If you start with auto-tracking and realize the crop keeps drifting to the wrong person, switch to face lock, select your subject, and hit re-track. The whole process takes a few seconds. FaceStabilizer is built so that changing your mind costs you almost nothing.
Most reframing tools force you into a single tracking philosophy and hope it works. FaceStabilizer gives you both, plus automatic re-locking, plus stabilization flexibility, because great video deserves more than a guess.