How to Keep a Subject Centered When Cropping a Vertical Video
Make this article actionable
Send the article context into Vife Agent and turn it into a plan, checklist, or draft you can keep working on.
Cropping a wide horizontal shot into 9:16 is not a resize. It is a re-composition, and the subject you framed comfortably in a 16:9 shot now occupies a narrow slice of the original frame. If you set one crop position and leave it, anyone who walks, leans, or gestures out of that slice will drift toward the edge or get clipped.
Quick answer: pick your crop window based on the subject's movement path, not the first frame. Then keyframe the crop position at three points — start, middle, end — so the window travels with the subject, and keep the face inside the upper third with margins reserved for captions and platform UI.
Why the first frame is the wrong place to start
The instinct is to scrub to frame one, center the subject, and export. That works for a locked-off interview. It fails for anything with motion, because the first frame tells you where the subject is, not where they will be.
Consider a person walking left to right across a wide shot. In the original frame they might enter at the left third, cross the center, and exit near the right third. A crop window centered on their starting position will hold them for a second or two, then they slide out of frame while the crop stays put.
Before you touch a crop control, do this:
- Scrub the whole clip at speed. Watch the subject's path, not the shot. Note where they enter, where they pause, and where they exit.
- Mark the extremes. Find the leftmost and rightmost points the subject reaches. Those two positions define the total travel distance your crop window has to cover.
- Check vertical movement too. A subject who stands up, sits down, or leans in changes the vertical position of their face. Vertical drift is easier to miss than horizontal drift because it happens more slowly.
- Decide what the crop is protecting. Usually it is the face and hands. Sometimes it is a product the subject is holding. Write that down before you start keyframing, because it becomes your review criterion later.
If the subject's travel distance is wider than your 9:16 window can cover without the background looking like it is sliding, you have a creative decision to make: follow the subject, or let them move within a wider crop and accept more headroom. Both are valid. Following is usually better for talking-head and walking shots; a wider crop is better when the background carries meaning.
Turn the useful parts into next steps
Vife Agent can convert this guide into a prioritized workflow with tasks, risks, and reusable prompts.
Step 1: Choose the crop window
A 9:16 crop of a 16:9 source keeps the full height and roughly a third of the width. That means your horizontal freedom is limited and your vertical freedom is zero unless you scale the footage down inside the vertical frame.
Two practical approaches:
- Full-height crop, horizontal travel only. The crop window is the full height of the source and a narrow vertical slice. You move it left and right. This preserves maximum resolution and is the simplest to keyframe.
- Scaled crop with vertical freedom. You scale the source down slightly inside the 9:16 frame, which gives you room to move the window both horizontally and vertically. Useful when the subject's face needs to sit in the upper third and the source framing puts them low. The trade-off is a small resolution loss and more visible background movement.
For the walking example, start with a full-height crop. Place the window so the subject's face lands in the upper third of the vertical frame — roughly the band between one-fifth and one-third down from the top. That position leaves the lower two-thirds for the body, gestures, and captions.
A note on framing the face
The upper third is a convention, not a rule. It works because it leaves breathing room below the chin and keeps the eyes away from the very top edge, where platform UI often sits. If your subject is seated and gesturing low, you may want the face slightly higher. If they are holding something up, you may need to drop the face toward the middle to keep the object in frame.
Step 2: Keyframe the crop at three points
Three keyframes is the minimum that produces believable motion. More is fine, but three forces you to think about the path rather than nudging frame by frame.
Start keyframe. Position the crop window where the subject is at the beginning of the clip. If they enter from off-screen, place the window slightly ahead of them so the crop is already moving when they appear — a crop that starts moving at the same instant as the subject reads as more natural.
Middle keyframe. Go to the midpoint of the clip and reposition the window on the subject. This is the keyframe that prevents the "sliding background" look. If the subject pauses mid-clip, put the middle keyframe at the pause rather than the exact midpoint, so the crop settles when they do.
End keyframe. Reposition on the subject at the end of the clip. If they exit frame, decide whether the crop follows them out or holds on the background. Following out is usually cleaner for a short.
Then play the clip from the top and watch the background, not the subject. If the background appears to accelerate or stutter, your keyframe spacing is uneven relative to the subject's speed. Add an intermediate keyframe at the point where the motion changes.
Example input
A copyable brief you can adapt for your own clip:
Source: 3840x2160 horizontal, 24 fps, 18 seconds.
Subject: one person walking left to right, enters at left third, exits right third.
Target: 1080x1920 vertical.
Crop: full-height window, horizontal travel only.
Face position: upper third of the vertical frame, eyes roughly 28% down from the top.
Keyframes: t=0s (subject at left third), t=9s (subject at center), t=18s (subject at right third).
Bottom 22% of the vertical frame: reserved for captions, keep clear of the subject's hands.
Review at: 0s, 4s, 9s, 14s, 18s.This is an example brief, not a measured result. Your keyframe times depend on where your subject actually is in your footage.
Step 3: Reserve safe margins for captions and platform UI
Vertical platforms overlay interface elements on top of your video. Captions, usernames, action buttons, and progress bars all compete with your subject for space. If you crop tight to the face, you will lose part of it to the UI.
A practical starting point:
| Zone | Approximate share of vertical frame | What to keep there |
|---|---|---|
Top | ~10% | Nothing important; UI and status elements |
Upper third | ~20–33% from top | Subject's face and eyes |
Middle | ~33–65% | Torso, gestures, held objects |
Lower third | ~65–85% | Captions, lower-thirds, key text |
Bottom | ~10–15% | Nothing important; UI and progress elements |
Treat these as starting points and verify against the platform you are publishing to, since overlay placement changes. The important habit is to leave the bottom band clear of the subject's hands and any object they are holding, because that is where captions usually land.
If your subject's hands are the point of the shot — a demo, a craft, a product — consider shifting the face higher and giving the hands the middle band, then placing captions in the lower third only if there is room.
Step 4: Review the crop at multiple timestamps
Do not review by watching the clip once at normal speed. Scrub to specific timestamps and check the same things each time.
Output review checklist:
- At 0s, is the subject fully inside the frame with margin on the leading edge?
- At the midpoint, is the face still in the upper third, or has it drifted?
- At the end, is the subject still inside, or clipped at the edge?
- At each timestamp, is any part of the face touching the top 10% of the frame?
- Is the bottom band clear of the subject's hands and held objects?
- Does the background move smoothly, or does it visibly slide or stutter between keyframes?
- Is the crop window ever wider than the source, showing black bars?
- If the subject pauses, does the crop also settle, or does it keep drifting?
- Does the crop motion feel intentional, or does it look like it is chasing the subject?
The last item is subjective. Two editors can watch the same clip and disagree about whether the crop motion feels natural. That is a judgment call, not a measurable property, and it is worth getting a second opinion before you publish.
Common mistakes
Keyframing on the subject's center of mass instead of the face. The body can move while the head stays relatively still, or vice versa. If your review criterion is "face in the upper third," keyframe on the face.
Using too many keyframes. A keyframe every few frames produces jittery, over-corrected motion. Three to five well-placed keyframes usually read better than thirty.
Ignoring the vertical axis. Horizontal drift is obvious. Vertical drift creeps in when a subject stands, sits, or leans, and it is easy to miss until you review the final export.
Cropping before you know the caption placement. If captions are added later, the crop may need to shift to accommodate them. Decide caption placement before you finalize the crop.
Assuming the crop will hold for the whole clip. A crop that works for the first five seconds is not validated for the rest. Review at multiple timestamps, always.
Forgetting that results depend on the source. A locked-off shot with a slow-moving subject is far easier to crop than a handheld shot with a subject who moves unpredictably. Handheld footage adds camera motion on top of subject motion, which means the crop window has to compensate for both. If the source is shaky, stabilizing before cropping usually produces a cleaner result than trying to keyframe around the shake.
Where this fits in a larger workflow
If you are producing vertical shorts at volume, the crop is one step in a longer chain: sourcing footage, cutting the segment, reframing, captioning, and exporting. Vife's AI image to video workspace is built for that kind of multi-step creation work, where you move between generating, editing, and assembling assets in one place. The crop technique above is tool-agnostic — it applies whether you are keyframing in a dedicated editor or working inside a broader workspace.
If you are weighing how much of this to do yourself versus automating, the pricing page covers what is available on each plan. Note that no crop workflow is fully automatic in a way that guarantees a centered subject; the review checklist above is what catches the cases where a crop drifts.
FAQ
Can I just center the subject and leave the crop static? Sometimes. If the subject barely moves and the shot is locked off, a static crop is the cleanest option. It fails as soon as the subject travels more than the crop window can hold.
How many keyframes do I need? Three is a good minimum for a simple left-to-right walk. Add one wherever the subject's speed or direction changes. More than five or six usually means you are correcting rather than composing.
What if the subject moves faster than I can keyframe? Slow the clip down, keyframe at the slower speed, then speed it back up. The crop motion will scale with the clip. Alternatively, widen the crop so the subject has more room to move before they reach the edge.
Should the crop follow the subject exactly? Not necessarily. A crop that tracks the subject perfectly can feel mechanical. Letting the subject drift slightly within the frame — as long as they stay inside your safe margins — often looks more natural.
Does this work for footage that is already vertical? If the source is already 9:16, you are not cropping for aspect ratio, but the same logic applies to reframing within the frame, such as when you want to shift a subject from center to the upper third.
How do I know if the crop is good enough? Run the review checklist at your chosen timestamps. If the face stays in the upper third, the subject never clips an edge, and the background motion reads smoothly, the crop is doing its job. Everything beyond that is taste.