How to make a vertical video from horizontal footage

Everything you have was shot wide. Everything you're publishing this week is vertical. So you crop it, and either a head leaves the frame or the subject ends up as a distant figure in a room.

The arithmetic is worth stating plainly, because it explains why this is hard rather than fiddly.

You are throwing away two thirds of the width

Crop a 16:9 frame to 9:16 at full height and you keep about 32% of the width. Not a trim — two thirds of the picture, gone.

That's the whole job: deciding which third survives, shot by shot. No automatic reframe can make that decision well in general, because it depends on what the shot means. A tool can find a face. It can't know that the point of the shot is the thing the face is looking at.

Set the canvas first

Do this before you cut anything: switch the project to 9:16, then edit.

Editing in 16:9 and cropping at the end is the classic own goal. You'll compose shots that don't survive the crop, time cuts against a frame you're not delivering, and place titles where the vertical frame has no room for them. Every framing decision you make is a lie until the canvas is right.

Same rule for generated shots: generate them at 9:16 rather than generating wide and cropping. The model composes for the frame you asked for, and a vertical generation puts the subject where a vertical frame wants it.

Three ways to fill a vertical frame

Choose per clip. Using one strategy for the whole video is how you get the decapitations.

Crop in (Fill)

Centre-crop to cover, then drag the crop to where the subject actually is. Best for a single person, close-ish, in a shot with a clear main subject.

Check every clip. Interview and podcast setups habitually place people off centre, so the default centre crop lands on an ear and a lamp. It takes two seconds to notice and two seconds to fix.

Letterbox over a blurred copy

The whole wide frame, shrunk to fit the width, floating over a blown-up blurred version of itself filling the rest. Blur BG, per clip.

This is the right answer whenever the width is the content: screen recordings, slides, wide landscapes, two people talking to each other, anything with text across the frame. It looks like a deliberate choice rather than a compromise — which it is — and it's the option people forget exists.

Stack two shots

Vertical frames have room for two things above each other. Speaker on top, screen share below; product on top, hand demo below. Two PiP layers, each cropped to its own box.

More work, and the most native-looking result of the three, because it uses the shape of the frame instead of apologising for it.

You can mix all three across one video. Cutting from a stacked layout to a tight crop is a perfectly good visual beat.

Where the crop should sit

Eyes on the upper third. A face centred vertically in a 9:16 frame looks like a passport photo. Eyeline high, headroom tight — this single adjustment separates competent vertical from cropped horizontal.

Follow the subject, don't chase it. If someone walks across the wide frame, don't hand-animate the crop to track them tightly; you'll get a nauseating pan. Pick the position where they spend the most time and let them move within the frame. If they truly cross the whole width, that's a cut, not a crop.

Keep the reason for the shot. Ask what the shot is for. A person pointing at a whiteboard, cropped to the person, is now a person pointing at nothing.

Leave the bottom fifth empty. Which brings us to the part that ruins otherwise good clips.

The UI eats your frame

Every vertical platform draws its own interface over your video: caption, handle, sound name, buttons, all in the bottom fifth and along the right edge. Exact dimensions vary by app and change without notice, so the practical rule is:

• Keep anything important out of the bottom ~20% and the right ~15%.
• Faces, product, action: the middle and upper half.
• Your own captions and titles: above that bottom zone, not at the classic subtitle position, or the platform will bury them behind its own.

Check on a phone before you publish. A frame that looks balanced on a monitor can have its subject sitting exactly where the Follow button goes.

Captions, positioned properly

Vertical video is watched muted by default, so captions carry the audio. Two things matter beyond having them: keep them clear of the UI zone, and keep the lines short — two or three words at a time reads far better in a narrow frame than a full sentence wrapping to three lines.

When no crop works

Some shots simply have no vertical version. A wide two-shot, a landscape establishing shot, a full-frame diagram. Your options, in order of preference:

1. Blur-fill it and move on. Ten seconds, always works, looks intentional.
2. Stack it with something else in the frame.
3. Split it into two shots — cut from one side of the wide frame to the other, each cropped tight. This is how the reframe becomes an edit rather than a compromise, and it's frequently better than the original.
4. Re-generate it vertically, if it's a generated shot. Cheaper than fighting a crop that can't work.
5. Cut it. If a shot needs two thirds of its width to make sense and you don't have two thirds, it may not belong in this version.

Export

Export at 1080×1920. Don't agonise over bitrate — every platform re-encodes on upload — but do export at high quality so their pass is the only lossy one, and don't render 1080p from 4K sources by accident.

Then watch it on a phone, muted, once. That's the format's actual viewing condition, and it's remarkable what it reveals that a desktop preview hides.

The order that works

1. Set the canvas to 9:16 before editing anything.
2. Generate new shots vertical, don't crop them later.
3. Choose per clip: crop in, blur-fill, or stack.
4. Eyes on the upper third; subject out of the bottom fifth and right edge.
5. Captions above the UI zone, short lines.
6. Export 1080×1920, high quality, once.
7. Watch it on a phone with the sound off before publishing.