Vertical Video Export Guide: Avoid Black Bars in Reels

Vertical export should be the least dramatic step in your pipeline. The footage is already shot or generated, the story is cut, captions are burned in, and the audio is balanced. Then the upload finishes and the preview shows two thick black bars, or a presenter whose forehead has been sliced off, or a lower third that vanishes behind the platform's own caption area.

The cause is almost always one of a handful of configuration mismatches, and most of them can be prevented in a few minutes of setup. This blueprint covers what actually produces black bars and crop errors, the 9:16 spec that survives platform re-encoding, the safe zones worth designing around, and a repeatable workflow you can hand to another editor.

Why vertical export fails more often than horizontal ever did

For years the standard deliverable was a 16:9 file, and the mismatch risk was low. Your camera, your timeline, and your final upload all shared the same rectangle. Vertical broke that agreement.

A single short edit now routinely mixes phone footage shot at 9:16, a screen recording captured at 16:9, a stock clip at 4:5, and one or more AI-generated shots that came out square, landscape, or in whatever resolution the generator happened to use. Every one of those has to land inside the same 1080 × 1920 frame.

When they don't, the editing software makes a decision on your behalf. Some tools letterbox and add bars. Some crop from the center, which is rarely where the subject is. Some stretch. Some do different things for different clips in the same sequence. The result is a file that looks acceptable in the preview panel and wrong on a phone — and the phone is the only screen that matters.

Platform processing adds a second layer. Instagram, TikTok, and YouTube Shorts all re-encode uploads and reserve parts of the frame for captions, buttons, usernames, and navigation. A file that is technically 9:16 can still look broken in practice if the important content sits underneath the interface.

The fix is not one magic setting. It is a short sequence of decisions: match the frame, protect the center, encode predictably, and verify on a real device before you publish.

The four root causes behind black bars and cropping

Source resolution and orientation mismatches

Black bars appear the moment the export frame and the clip frame disagree in shape. The math is simple and worth internalizing. Drop a 1920 × 1080 clip into a 1080 × 1920 timeline and scale it to fit by width: the height becomes roughly 608 pixels, leaving about 1,312 pixels of empty space split between top and bottom. That is 656 pixels of bar above and 656 below — more than two thirds of the frame wasted.

Scale that same clip to fill the height instead and you need a width of roughly 3,413 pixels, which means you are only seeing about a third of the original image. If your subject was centered in a wide shot, you will crop into their shoulders or lose context entirely.

Neither outcome is wrong on its own. What is wrong is letting the software pick at export time without you noticing.

Aspect ratio math applied at the wrong end of the pipeline

Many editors delay the conform step until they export, then fight with scaling controls under time pressure. Conform instead at the start: decide the master frame, then normalize every asset into it before you cut.

Pay attention to metadata that survives import. Phone footage often carries a rotation flag, so a clip may report as 1080 × 1920 while the decoder displays it as landscape. Anamorphic footage can carry a non-square pixel aspect ratio, which makes circles into ovals after export. Some generators output dimensions that are close to but not exactly 16:9 or 1:1, which produces hairline bars that are easy to miss on a laptop screen and obvious on a phone.

One more trap: timelines with mixed pixel aspect ratios may preview correctly while rendering differently. Always check the final file, not just the viewer.

Frame rate and bitrate mismatches

Mixing 24, 30, and 60 fps clips is a creative choice, but exporting at a frame rate that doesn't match your timeline causes stutter or duplicated frames. Variable frame rate footage — common from screen recordings and phones — can produce audio drift, frame drops, and playback that stutters on some devices while looking fine on others. Conform everything to a constant frame rate before export.

Bitrate is the other silent killer. A very low export bitrate forces the platform encoder to work harder, and the second compression pass magnifies blockiness in gradients, skin tones, and motion. For 1080 × 1920 vertical video, H.264 at 10–20 Mbps is a safe, widely compatible range; higher for busy footage with foliage, confetti, or fast camera movement.

Mixed-source timelines in AI-assisted projects

Generative clips introduce a fourth problem: inconsistency. Different shots may arrive at 1024 × 1024, 1280 × 720, or 720 × 1280, with different levels of sharpness, grain, and color response. One shot might have crisp edges while the next looks soft.

If you scale each clip independently in the timeline, you get a sequence where faces change size between cuts and the grain pattern jumps. Normalizing every generated clip to one working resolution, one frame rate, and one color space before editing solves the majority of this. It costs ten minutes and saves an hour of patching.

The 9:16 export spec that platforms actually accept

Resolution and pixel grid

1080 × 1920 is the standard vertical frame and the safest upload target. It divides cleanly (both dimensions are divisible by 8), which helps encoders avoid chroma artifacts and keeps scaling mathematically tidy. 720 × 1280 is acceptable for drafts or data-constrained uploads but will look softer after re-encoding. Avoid odd pixel dimensions and avoid upscaling a 720p timeline to 4K just for the label — you gain nothing but file size and render time.

If your source is larger, such as 2160 × 3840, downscale to 1080 × 1920 on export. This sharpens detail and reduces the chance of visible compression artifacts after the platform's own pass.

Frame rate, codec, and bitrate

Export H.264 in a High profile with 4:2:0 chroma subsampling and 8-bit color depth for maximum compatibility. If your tool offers H.265 or HEVC and the target platform accepts it, you can lower the bitrate slightly for the same quality, but compatibility is less predictable, especially on older phones and desktop browsers.

Use a constant frame rate that matches your timeline — 30 fps for most social edits, 60 fps for fast motion or gameplay. Target 10–20 Mbps at 1080 × 1920 and go higher if the scene is dense. Modern platforms re-compress everything, so a slightly larger file is generally safer than a thin one.

Audio settings people forget

Export AAC audio at 44.1 kHz or 48 kHz, stereo, 128–320 kbps. Aim for an integrated loudness around −14 LUFS with a true peak near −1 dBTP so your edit sits comfortably next to everything else in a feed. Audio that clips or sits 6 dB quieter than the rest of the platform is a bigger retention problem than a slightly soft image.

Safe zones: design for the interface that covers your video

Mapping safe zones at 1080 × 1920

The vertical frame is taller than the visible content area on most devices. As a working rule, keep these regions clear of essential information:

Top 250–320 pixels: often covered by status bars and platform headers.
Bottom 420–500 pixels: captions, usernames, sound labels, and action buttons.
Left and right 60–120 pixels: edge gestures and side-mounted controls on some devices.

That leaves a central band of roughly 1080 × 1300 pixels as the truly reliable area. These numbers shift as platforms redesign their interfaces, so treat them as conservative guidance and leave more room than you think you need. If you want one habit that solves most of it: never place text or a logo in the top or bottom 20% of the frame unless it is decorative.

Captions and overlays

Burned-in captions are excellent for retention but terrible when they collide with platform captions. Place your text block higher than instinct suggests — around 25–30% of the frame height from the bottom — and keep line length short enough that it doesn't run into the side margins. If you rely on platform auto-captions instead, leave that lower band empty so the text has room.

Logos, subscribe prompts, and progress bars should live in the middle band or in a corner that stays clear of both edges. Check them against the platform's own preview, not just your editor.

Crop, pad, or reframe: choosing a deliberate fix

Scale to fill

Scaling to fill eliminates bars and makes the subject large and readable. It works well for close-ups, talking heads, and any footage where the important content sits near the center. Plan for a loss of 30–45% of the original width when converting 16:9 to 9:16. If the action is off-center, this option fails.

Pad or blur-fill

Padding keeps the whole image visible and adds a background — solid color, blurred version of the clip, or a designed graphic. It is the right call for screen recordings, diagrams, wide establishing shots, and interviews where framing matters. The tradeoff is that padded footage occupies less screen area, so it reads as smaller and less immediate. Many successful edits use a hybrid: pad wide shots and fill close-ups.

Smart reframe and subject tracking

Reframing tools track a face or subject and animate the crop across the clip, keeping the person centered as they move. This produces the most native-feeling result, and it is worth the render time on any clip where the subject moves horizontally. Review the tracking frame by frame at the start and end of each shot — automated tools often drift on fast motion or when a second person enters the frame.

A quick decision table

Source	Best default	Watch for
16:9 close-up, centered subject	Scale to fill	Softness after 2× upscale
16:9 wide shot or diagram	Pad or blur-fill	Small subject, weak hook
1:1 generated clip	Scale to fill with slight crop	Losing hands or props at edges
Vertical phone footage	Use natively	Rotation metadata
Screen recording	Pad on a branded background	Unreadable tiny UI text

A repeatable export workflow, start to finish

Set the master frame first. Create a 1080 × 1920, 30 fps, square-pixel timeline before importing anything. Never rely on an auto-created sequence.
Normalize assets on import. Note each clip's native resolution, orientation, and frame rate. Transcode variable-frame-rate footage to constant frame rate if your tool handles it poorly.
Standardize generated clips. Conform AI-generated shots to one working resolution and frame rate, then apply consistent grain or sharpening so cuts feel like one piece.
Design with safe zones visible. Add a guide overlay or a placeholder rectangle marking the reserved top and bottom bands. Keep it on until the final render.
Fix framing per clip, not per sequence. Decide crop, pad, or reframe shot by shot, and write the choice down in your notes.
Check faces, hands, and text at 100%. Zoom in on the phone-sized preview rather than the desktop viewer.
Export a short test file. Render 5–10 seconds of the busiest section at final settings before committing to a full-length export.
Upload as a draft or private post. Watch it on a real phone, in the app, with the platform interface visible.
Adjust once, then batch. Once settings are confirmed, apply the same preset to every remaining clip in the project.

Working with AI-generated clips in a vertical timeline

Normalize before you edit

Generators rarely output your target dimensions. A square 1024 × 1024 clip needs 1.78× horizontal scaling to fill a 1080 × 1920 frame, which is beyond what most sources can handle without visible softness. Where possible, request the tallest aspect ratio available at generation time, then do the final crop in the editor.

Upscale carefully

When you must enlarge a generated clip, keep it under roughly 130% of native size and add a light sharpening pass. Beyond that, artifacts become obvious once the platform encodes again. If a shot only exists in landscape, consider using it in a padded layout instead of cropping — a wide shot with a blurred background often looks more intentional than a mushy zoom.

Scene consistency and style drift

Because each shot is generated separately, small differences in lighting, lens character, and color temperature accumulate. Apply one consistent look — a color correction layer, a grain overlay, a subtle vignette — across the entire sequence. It is the single most effective trick for making generated clips feel like footage from one camera.

Pre-publish quality control checklist

[ ] Final file is exactly 1080 × 1920 with no embedded letterboxing.
[ ] Timeline and export frame rates match and are constant.
[ ] Bitrate is 10–20 Mbps for H.264, and audio is normal, not clipped or quiet.
[ ] All text sits inside the central safe band.
[ ] No face, hand, or product is cut by the frame edge.
[ ] The first frame works as a thumbnail in both square and vertical crops.
[ ] Playback tested on a real phone in the actual app.
[ ] File name and version are logged so you can re-export the same way next time.

Common mistakes and how to fix them

Symptom	Likely cause	Fix
Black bars top and bottom	Landscape clip scaled to fit width	Scale to fill, or pad deliberately with a design
Bars on the sides	Square or 4:5 clip in a vertical frame	Fill by height and let the crop handle the rest
Stretched faces	Non-uniform scaling or anamorphic pixels	Reset scale, check pixel aspect ratio
Cropped foreheads	Center crop on off-center framing	Use subject tracking or shift the crop manually
Stutter after upload	Variable frame rate source	Conform to constant frame rate before export
Blocky gradients and skin	Bitrate too low, or double compression	Export at 15–20 Mbps and avoid re-encoding intermediate files
Captions hidden behind UI	Text placed in the bottom 25%	Raise the text block into the central band
Blurry after upload	Upscaled source or 720p master	Keep master at 1080 × 1920, limit scaling

FAQ

Why does my vertical video still show bars after I set 9:16?

Setting the sequence frame does not change clip content. Each clip still has its own shape and must be scaled, cropped, or padded inside that frame. Check whether the bars are part of the sequence background or baked into the clip itself.

Should I export at 1080 × 1920 or 2160 × 3840?

1080 × 1920 is enough. Platforms re-encode anyway, and a well-encoded 1080p file often looks better than an oversized file that gets compressed harder. Use 4K only if you also deliver to a platform that genuinely preserves it.

Is 30 fps or 60 fps better for vertical short video?

30 fps matches most footage and keeps files smaller. 60 fps helps with fast motion, gameplay, and sports, but it roughly doubles file size and offers little benefit for talking-head content.

How much can I zoom into a landscape clip before it looks bad?

Aim to keep enlargement under about 1.3× of the native pixel dimensions. Past that, you are inventing detail, and the platform's encoder will make the softness obvious on a phone screen.

Do captions need to be burned into the video?

Burned-in captions give you full control over placement and timing and work everywhere. Platform auto-captions are convenient but can sit in exactly the band you left empty for other elements. Pick one approach and design the frame around it.

Why does my video look sharp on my computer but soft on my phone?

Desktop players are forgiving and often scale down. A phone shows the file at its native pixel size on a dense display, which exposes compression and upscaling. Always judge quality on the phone, in the app, after uploading.

Can I reuse one export for multiple platforms?

Yes, if the frame is 1080 × 1920 and the essential content stays inside the central safe band. Different apps cover different edges of the frame, so a conservative layout is the only reliable way to publish once and post everywhere.

What is the single most common cause of crop errors?

Editing at the wrong frame size and fixing it at export. Conform at the start, keep a safe-zone guide on the timeline, and check each shot on a phone-sized preview before rendering.