How to frame archive footage vertically

A finished vertical documentary short renders at 1080x1920, per the product documentation dated 22 August 2026. This page covers how to get horizontal archive footage into that frame without destroying the shot.

The short answer

Find the subject in the original frame first, crop to the action rather than the center of the image, and choose deliberately between cropping (loses edges), padding (loses area), and zooming (loses resolution). Never upscale beyond what the source file actually holds, and check the result on a phone at full screen before you accept it. Every vertical framing of horizontal footage is a loss; the job is choosing which loss the shot can survive.

The steps

1. Find the subject in the frame before you crop

Archive footage was composed for a television set held horizontally, not for a phone held upright. The subject is frequently off-center: a speaker stands to one side of a podium, a ship enters from the left third, a crowd fills only part of the width. If you place a 9:16 crop box in the geometric center of the frame, you will often capture empty space and cut the person out entirely.

Scrub through the clip and mark where the action actually sits in each segment. A single fixed crop position rarely works across an entire clip; the subject moves. Note the timestamps where the framing needs to shift, or plan to re-crop per shot rather than per scene.

Output of this step: a list of crop positions keyed to timestamps, based on where the subject actually is.

2. Crop to the action, then check the whole shot again

Once you have set the crop around the subject, play the full clip with the crop applied and watch what leaves the frame. Vertical crops are narrow, so context disappears fast: the second person in a conversation, the sign that dates the footage, the explosion happening just outside the box. In documentary work that context often carries information the narration depends on.

Two checks matter. First, does any motion exit the frame mid-shot? A car driving right will leave a left-anchored crop within seconds. Second, does anything important sit on an edge? If yes, widen the crop, pan it, or accept a different method for that shot.

Output of this step: a crop setting per shot that survives playback without cutting off moving subjects or essential context.

3. Choose between crop, pad and zoom deliberately

The three methods lose different things. Cropping loses the edges of the frame permanently. Padding, meaning pillarboxing the horizontal image inside the vertical frame with bars or a blurred background, keeps the whole image but shrinks it and looks like repurposed content rather than native vertical video. Zooming into the source to fill the frame vertically keeps the composition but discards resolution, because you are magnifying fewer pixels.

Match the method to the shot. Talking-head interviews and speeches tolerate padding well, since the viewer watches a face anyway. Wide establishing shots and disaster footage usually need a crop, because the scale is the point. Soft, low-resolution sources should not be zoomed at all; they fall apart fastest under magnification.

Output of this step: a declared method per shot, chosen for what that specific shot can afford to lose.

4. Do not upscale past what the source holds

An archival transfer below the target resolution cannot become sharp 1080x1920 vertical footage by scaling. Upscaling adds pixels by interpolation; it makes edges soft and grain smeary, and on a phone screen held close, the blur is visible immediately. Grain structure also breaks: real film grain scales naturally, interpolated grain turns to mush.

Work out what the source can deliver after the crop. Compare the width your 9:16 crop will retain against the 1080-pixel width of the target frame. A high-definition horizontal source can come close; a standard-definition transfer falls far short. Frame within the source's real capacity, and let the letterbox or background treatment absorb the difference instead of inventing resolution.

Output of this step: a final render size that matches the true resolution of each source, not the target spec.

5. Watch the result on a phone before you accept it

The desktop preview lies about vertical footage. On a large monitor, a 9:16 frame appears small and distant; on a phone, it fills your vision, and every flaw grows with it. Captions that fit comfortably in the editor can collide with the platform UI overlay on device. Faces read differently when the frame is physically close to the eye.

Export one test render, put it on an actual phone, and watch it full screen with sound. Check three things: the subject stays framed throughout, captions sit inside the safe area away from interface elements, and no shot looks soft or unstable at close distance. Fix what fails, then re-check.

Output of this step: an approved vertical master verified on the device the audience will actually use.

What this will not fix

Vertical reframing cannot restore composition that was never there. A wide shot whose meaning lives across the full width, a two-person scene staged far apart, a landscape establishing frame: these lose something in any 9:16 conversion, and no tool recovers it. Padding preserves the image but reads as recycled horizontal content, which costs retention on platforms built for native vertical video.

The honest decision is per shot. Some shots convert cleanly with a crop. Some survive as padded inserts. Some should be shortened, replaced with a still photograph from the same archive, or covered with a map or graphic while the audio continues. Treating every clip as automatically convertible produces a video where half the shots feel wrong for reasons the viewer senses but cannot name. Budget time for those decisions; they matter more than the export settings.

Where to go next

Before framing comes acquisition, and /guides/how-to-source-archive-footage covers finding licensed material and keeping the provenance record that a media cue sheet requires. For the underlying geometry behind why 9:16 crops behave the way they do, /glossary/aspect-ratio defines the ratios and their history.

<!-- faq -->

Frequently asked questions

Should I add blurred bars or black bars when padding?

Blurred-fill bars, where the same footage is scaled up and blurred behind the sharp centered image, look more native than hard black bars on short-form platforms. Black bars signal repurposed television content and tend to depress watch time. Either way, keep the sharp image large enough to read; if the padded image occupies less than roughly half the frame height, crop instead.

What safe margins do I need for captions?

Platform interfaces cover the bottom of the frame and parts of the top and right edge, and the exact zones change with app updates. Keep captions in the middle band of the frame, well above the bottom fifth, and verify against a current screenshot of the platform UI rather than a saved template. Burned-in captions cannot be moved later.

Can I use different framings for different shots in one video?

Yes, and you should. A single global crop setting across an entire edit guarantees some shots fail. Interviews can be padded while disaster footage is cropped tight, in the same video, without visual inconsistency, because viewers read each shot on its own terms. Consistency matters in caption style and grade, not in crop geometry.

Does AI reframing solve this?

Automatic subject-tracking reframing handles static, centered subjects reasonably and fails on the shots that matter most in documentary work: fast motion, multiple subjects, and frames where context carries meaning. Treat automated tracking as a first pass that you review shot by shot, not as a decision you delegate. Tools that generate visuals instead of using the archive avoid the problem by avoiding the archive.

How do I know a source is high enough resolution before editing?

Check the pixel dimensions of the delivered file, not the label on the archive page. An HD listing sometimes delivers a lower-resolution transfer. Compute the width the 9:16 crop will retain from the source's height, and compare it to the 1080-pixel width of the target frame. If the retained width falls well short, plan for padding or a graphic insert from the start rather than discovering the softness at final review.