Convert Horizontal Video to Vertical: 9:16 Reframe

Converting horizontal video to vertical means deciding what to throw away. A 9:16 frame keeps under a third of a 16:9 frame's width, so every reframe is a crop decision: a fixed crop window, a crop window that follows the subject, or a padded frame that shrinks the whole shot. Choose per shot, not per video.

What actually happens when you convert horizontal video to vertical?

Nothing is "resized". A reframe slides a tall window across your original footage and discards everything outside it.

Run the arithmetic on real numbers. Google's video ad specs list "1080 x 1920px (vertical)" as the recommended resolution for the 9:16 format (About video ad specs, Google Ads Help). Take a 1920x1080 source and scale it until it fills the height of that canvas: the source becomes about 3413 pixels wide, and the 1080-pixel window you keep is roughly 32% of the original frame width. Two-thirds of the shot is gone before you make a single creative decision.

That is why a wide two-shot — host on the left, guest on the right, a metre of table between them — cannot survive one crop window. No tall rectangle holds both faces. The reframe is not broken; the shot was never vertical.

Static crop, tracked crop, or blurred bars: which one should you use?

Method	How it works	Best for	Where it falls apart
Static centre crop	One fixed window, usually dead centre	Locked-off single-speaker shots, centre-framed talking heads	Anyone framed left or right of centre; walking subjects; anything the camera pans to
Tracked crop window	The window moves to follow the subject through the shot	Presenters who move, handheld footage, single-speaker interviews	Two subjects far apart — the window either picks one or swings between them, which reads as a wobble
Padding (bars, blur, or a colour fill)	The whole 16:9 shot is scaled down and centred in a 9:16 canvas	Screen shares, whiteboards, maps, charts, gameplay where the whole frame is the point	Faces become small; the picture uses a fraction of the phone screen

Padding deserves a caveat. YouTube's formatting specifications say: "Videos should be in their native aspect ratios, and uploaded videos should never include letterboxing or pillarboxing bars" (Video and audio formatting specifications, YouTube Help). YouTube's own player already handles odd shapes — for 9:16 video on computer browsers it "may add more padding for optimal viewing", white by default and dark grey in Dark theme (Video resolution & aspect ratios, YouTube Help). Baking bars in duplicates work the player already does, and the choice is then permanent.

Padding also changes what your upload is. Videos uploaded on or after 15 October 2024 with "a square or vertical aspect ratio up to three minutes in length will be categorized as Shorts on YouTube" (Understand three-minute YouTube Shorts). Pad a 16:9 talk into a 9:16 canvas and you have not made a tall video — you have made a Short out of a wide picture. If you are still choosing a shape, our breakdown of the best aspect ratio for YouTube Shorts covers that decision on its own terms.

What are the safe-area rules for a 9:16 frame?

The player draws its interface on top of your picture. Google's Shorts asset guidance describes the right-hand column plainly: "The RHS panel includes like, dislike, comment, and share buttons" (YouTube Shorts ads: Asset specs and best practices).

Resist any blog that hands you one exact pixel margin for every app. TikTok's own in-feed ad documentation states that "The safe zone size is determined by the dimension (vertical, horizontal, or square), ad caption length, and any additional formats used", and ships downloadable safe-zone files rather than a single number (TikTok Auction In-Feed Ads). Pull the template from the platform; treat everything else as a working rule.

Working rules that hold up regardless of the exact pixels:

Zone	What lives there	What to keep out
Right column	Like, dislike, comment, share	Faces, product, burned captions, logos
Bottom band	Title, description, channel handle, sound line	Anything you need read — including lower thirds baked into the source
Top band	System status and player chrome	Hook text, scores, timers
Centre	The subject's eyes and mouth	Nothing — protect it

The common casualty is a lower third. A name card that sat elegantly in the bottom-left of the original 16:9 interview now sits behind the description, or gets cropped in half by the tall window. Reframe first, then re-add the name as a caption inside the safe centre.

When does a landscape-to-vertical reframe genuinely fail?

Four cases, and the honest answer for each is "do something else":

1. Two speakers wide apart. Cut the moment to whoever is speaking and reframe on them one at a time, or accept a padded frame that keeps both.
2. On-screen text you did not author. Slides, code, spreadsheets, comparison tables. A tall crop reads three words of a twelve-word line. Pad the frame, or re-key the text as a caption.
3. Gameplay HUDs. Health bars, minimaps and killfeeds live in the corners — the exact regions a vertical crop deletes first. Our guide to gaming clips without losing the action covers the trade-offs there.
4. Wide establishing shots. A drone reveal of a valley is horizontal because the meaning is horizontal. Reframing it produces a tall slice of grass.

How do you check a reframe before it ships?

Watch it on a phone, at arm's length, with sound off. That is the condition most viewers will meet it in, and it exposes more problems than any preview window on a monitor.

Then run this list:

• The subject's eyes stay in the upper-middle of the frame for the whole clip, not just the first second.
• No cut lands mid-swing on a tracked crop window — a fast pan across a cut looks like a mistake.
• Nothing you need read sits in the right column or the bottom band.
• Burned captions are inside the safe centre, and readable with the phone at arm's length.
• Any text that came from the source footage is either fully visible or replaced.
• Hands, props and the thing being demonstrated are still in frame when they matter.
• The first frame works as a still — that is what a paused feed shows.

If an item fails, the fix is a different crop for that shot, not a different platform.

What should the exported vertical file look like?

Export at 1080 x 1920 for 9:16 (Google Ads video ad specs). YouTube's recommended upload encoding settings ask for an MP4 container, H.264 video, and this rule that people break constantly: "Content should be encoded and uploaded in the same frame rate it was recorded" (YouTube recommended upload encoding settings).

That last one bites gaming creators hardest: a 60 fps capture pushed out at 30 fps loses the smoothness that made the clip worth cutting.

Everpop reframes to vertical and burns word-by-word captions on footage you upload or share through a Google Drive folder, with review-first approval so nothing posts until you say so. If a clip comes back cropped wrong — and on a two-shot it can — every clip comes with three free re-renders, and you can export FCPXML, EDL or SRT to finish the job in your own editor. Pasted YouTube links are refused, because the source file should be yours.

A reframe is a small act of editing, not a conversion setting. Treat it that way and the vertical cut earns its place instead of apologising for the shape it came in.