Aspect Ratios for Instagram and TikTok: A Practical Guide

A vertical frame is not a cropped horizontal frame, and treating it that way is why so much good footage dies on a phone screen. The moment you accept that the shape comes first, platform work gets simpler: you stop resizing finished edits and start designing for the window your audience will actually look through. Reels and TikTok both live in a tall, hand-held viewport, and every other format you publish is a negotiation with that reality.

This guide is about the practical mechanics of that negotiation — which dimensions matter, where interface elements will cover your image, how to storyboard once and deliver everywhere, and how to keep an AI-assisted pipeline from producing clipped subjects and soft upscales.

The Frame Is a Creative Decision, Not an Export Setting

Most teams still work backwards. They shoot or generate in a wide format because that feels premium, edit a hero cut, then squeeze it into a vertical container as a final step. The result is predictable: heads near the top edge, captions buried under a username, a subject split down the middle by a crop that was never planned.

Flip the order. Decide the delivery shape before you write the first prompt or set the first camera move. A 9:16 frame changes what a scene can contain. Wide establishing shots lose their function, because there is no room for horizontal context. Close-ups gain power. Vertical motion — a fall, a rise, a slow push in — reads better than lateral movement. Once you know that, the creative itself changes for the better instead of being damaged later.

There is also a distribution reason to care. Recommendation systems reward completion and rewatches, and both are shaped by how quickly a viewer can parse a frame. A composition with one clear subject, readable within half a second on a small screen, holds attention. A composition that requires the viewer to hunt for the subject loses them before the second beat.

The Four Formats You Actually Need

You do not need a dozen exports. Four covers nearly everything a creator or brand publishes in a week.

9:16 — the default for Reels and TikTok

1080 × 1920 pixels is the standard vertical canvas. Both Reels and TikTok present full-bleed vertical video, so this is the format that gets the most real estate and the least friction. Anything you plan to push as short-form should be authored here, not converted here.

One detail that matters more than the pixel count: the difference between what the platform stores and what the viewer sees. Both apps re-encode uploads, and both overlay interface chrome on top of your picture. Designing for the full canvas without accounting for that chrome is the most common unforced error in vertical production.

4:5 — the Instagram feed workhorse

1080 × 1350 takes more vertical space in the feed than a square post while remaining a still-image and carousel-friendly shape. If your content strategy includes static posts, quote cards, or carousel storytelling that funnels toward a video, 4:5 is the shape to design in. It also crops gracefully from a 9:16 master, which is why planning a tall source pays off twice.

1:1 and 16:9 — where they still earn their place

Square remains useful for cross-posting to platforms that still favor it, for thumbnails, and for product grids. 16:9 remains the shape of embedded players, YouTube, websites, presentations, and anything a client will screen in a meeting room. Neither is obsolete; they are simply not the primary delivery target for short-form social.

A useful rule: author in the tallest shape you will publish, then derive the wider ones. Cropping inward is a controlled operation. Extending outward is guesswork.

Designing for 9:16 Without Losing the Cinematic Feel

Vertical video has a reputation for looking cheap, but that reputation comes from composition habits borrowed from cinema, not from the format itself. Cinema is horizontal because screens in rooms are horizontal. A phone held at arm's length is a different viewing condition with different physics.

Map the safe zones before you place anything

Treat the vertical canvas as three bands. The top roughly 10–15 percent is where usernames, captions, and platform labels appear. The bottom roughly 20–25 percent is where captions, music attribution, buttons, and comment prompts sit. Both bands shrink and shift slightly between platforms and app versions, so keep critical information out of them entirely.

The practical consequence: your subject's face, your product, and any on-screen text you add yourself should live in the middle band — roughly the central 1080 × 1300 region. Overlay text you burn into the video should sit above the platform's own caption area, not behind it.

Composition rules that survive a vertical crop

A few adjustments do most of the work:

• Replace horizontal context with vertical depth. Foreground, midground, background stacked in the frame reads as information. A left-to-right arrangement does not survive.
• Give headroom, not ceiling. In a tall frame, too much empty space above a subject feels accidental. Fill the upper third with environment, texture, or light rather than nothing.
• Use negative space as pacing. A vertical frame with one subject and a large calm area creates rhythm that a busy horizontal frame cannot.
• Keep faces above the midline. The eye lands in the upper half of a phone screen first. Feet and floors belong at the bottom, where interface elements will cover them anyway.
• Prefer vertical camera moves. Tilts, rises, drops, and slow pushes feel native. Fast lateral pans become a smear.

Camera height matters too. A slightly low angle makes a subject feel present and close, which suits the hand-held viewing context. A high, distant angle shrinks the subject and wastes the vertical dimension.

A Storyboard Method That Produces Every Ratio at Once

This is the workflow that saves the most time in practice, whether you are shooting, animating, or generating with AI.

Step one: draw the widest frame

Sketch or generate your scene in 16:9. This forces you to define the whole world: where the subject stands, what surrounds them, where the light comes from. Working wide first prevents the tunnel-vision composition that makes vertical-only planning look flat.

Step two: protect a center column

Inside that wide frame, mark a vertical column roughly one third of the width. Everything essential — the face, the product, the gesture that carries the story — must live inside that column. The outer thirds are permitted to be beautiful but expendable.

This single habit solves reframing for every platform at once. The 16:9 master is complete. The 9:16 crop is guaranteed to contain the story. The 4:5 and 1:1 crops are trivially derivable.

Step three: generate or export per-platform variants

Once the master exists, produce variants deliberately rather than in a batch-and-hope pass. Name files by ratio and destination (hero_9x16_reels, hero_4x5_feed, hero_16x9_site) so that nobody uploads the wrong container. If you work from a template system, lock the frame in the template and vary the content inside it — starting from a known-good vertical layout removes an entire class of mistakes. Browsing ready-made video templates is a fast way to see which layouts actually hold up in a tall frame before you commit your own.

Using AI to Reframe Instead of Crop

Cropping is lossy in a way that is easy to ignore until you see it on a phone. Generative tools let you extend the frame instead of cutting it, which changes the economics of multi-format delivery.

Outpainting, generative fill, and subject-aware reframing

Three techniques are worth knowing:

1. Outpainting takes an existing frame and invents plausible content beyond its edges. Feed a 16:9 shot into a vertical canvas and the model fills the sky above and the ground below. It works best with simple, continuous backgrounds — skies, walls, water, fog, gradients.
2. Generative fill handles smaller repairs: removing a boom mic that entered the vertical crop, extending a wall, continuing a table edge. This is where AI saves the most tedious hours.
3. Subject-aware reframing tracks the important element across a clip and moves a virtual camera window with it, instead of holding a fixed center crop. Essential for interviews, product spins, and any shot with movement.

A caveat worth internalizing: generated extensions are convincing at a glance and unreliable under scrutiny. Do not outpaint a face, text, or a logo. Keep invented content in peripheral, low-detail regions where the eye does not interrogate it.

Prompt patterns that read well in vertical

When generating footage directly in a tall frame with an AI video generator, describe the composition, not just the subject. Prompts that specify framing tend to produce usable results on the first attempt.

• Say the shape explicitly: "vertical 9:16 composition, subject centered, tall background."
• Name the camera behavior: "slow vertical rise," "static tripod, subject fills frame," "gentle push in."
• Describe what occupies the upper third: "hanging foliage above, soft daylight."
• Describe what occupies the lower third: "wet stone floor, shallow depth of field."
• Avoid dense horizontal staging: "crowded street stretching left to right" fights the canvas.

You can borrow structures from a prompt library and adapt the framing clauses rather than writing from scratch each time. A consistent prompt skeleton — subject, framing, camera, light, upper and lower detail — produces a consistent look across a series, which matters more than any single shot.

Resolution, Bitrate, and Other Details People Skip

Aspect ratio gets the attention; encoding gets the results.

• Export at native size. 1080 × 1920 for vertical, not an upscale of a 720-pixel source. Platforms re-compress aggressively; feeding them a soft master compounds the loss.
• Keep bitrate generous. For 1080p vertical, 10–16 Mbps is a sensible range for footage with motion and grain. Lower and fine detail turns to mush in dark scenes.
• Match frame rate to intent. 24–30 fps for cinematic pieces, 60 fps when you plan to slow footage down or when the motion is the point.
• Beware of letterboxing. If a horizontal clip must appear in a vertical video, design the surrounding space — blurred fill, a color field, a caption stack — rather than letting black bars carry the frame.
• Check your first and last frame. Many platforms loop short clips, so a jarring final frame reappears immediately.

One more: preview on an actual phone, in the actual app, before publishing. Desktop previews hide overlay collisions that are obvious on a handset.

One Idea, Four Deliverables: A Worked Example

Say you are producing a short piece for a fictional cold-brew brand. The idea: condensation forming on a glass in morning light, a hand lifting it into frame.

Master, 16:9. Wide shot of a table, window light from the left, glass in the center-right. The hand enters from the right. This cut serves the website and a YouTube pre-roll.

9:16. The glass sits in the protected center column. Upper third filled with window light and a curtain edge. Lower third: table grain, a dark surface. The hand enters from below instead of the side — a vertical-native gesture. Text like "brewed overnight" sits around 62 percent down the frame, safely above the platform caption area.

4:5. Cropped from the same protected column with a little more table on either side. Useful as a carousel cover and as the thumbnail for the vertical cut.

1:1. The tightest crop: glass and condensation only, no hand. It reads instantly at grid size, which is all it needs to do.

The lesson: nothing was reshot, and nothing essential was lost. The vertical version feels authored because the gesture, the text placement, and the upper-third content were decided for that frame.

Mistakes That Quietly Cost You Reach

These are the recurring ones, roughly in order of damage:

• Designing for the full canvas. Text and faces land under the interface. Fix: stay inside the middle band.
• Center-cropping movement. A subject walking laterally leaves the frame mid-shot. Fix: reframe dynamically or change the blocking.
• Upscaling instead of re-exporting. Softness reads as low quality even when the content is strong. Fix: always export at native resolution from the master.
• Inconsistent typography across ratios. The brand looks different on each platform. Fix: define one type scale with ratio-specific positions.
• Burning in captions the platform will also display. Duplicate text is visual noise. Fix: turn off platform auto-captions when you burn your own, or leave space and skip the burn-in.
• Ignoring the first 1.5 seconds. The frame must communicate subject and stakes before any motion resolves. Fix: start on the most legible composition in the sequence.

FAQ

Does uploading a 16:9 video to Reels or TikTok hurt reach?

Not directly — platforms do not penalize a container. What hurts is the viewing experience: black bars, subjects cropped by the player, and text that becomes unreadable at the smaller effective size. A well-designed 16:9 with intentional framing can perform, but a native 9:16 almost always reads better on a phone.

What is the safest area for on-screen text in a vertical video?

Keep it inside the central region, roughly 1080 × 1300 in a 1080 × 1920 canvas, and leave clear margins above and below. Treat the top and bottom bands as reserved for the platform, and preview in the app before publishing, since overlay sizes shift with app updates and device sizes.

Can AI reframe my existing footage for vertical without re-editing?

Yes, within limits. Subject-aware reframing and generative outpainting handle static backgrounds, skies, walls, and continuous surfaces well. Shots with complex foregrounds, hands interacting with objects, or readable text in frame usually need a manual pass or a reshoot. Budget a review step for every generated extension.

Should I create each ratio as a separate generation or crop from one master?

Crop from a master when the framing already works; generate separately when the vertical composition needs to be genuinely different — a different gesture, a different camera height, a different amount of environment. The test is whether the story still makes sense in the tall frame.

How many exports does one campaign really need?

For most short-form work: 9:16 for Reels and TikTok, 4:5 for feed statics and carousels, 1:1 for grid thumbnails and cross-posting, and 16:9 for the site, presentations, and long-form. Four files, one idea, no reshoots.

Put the Frame First and the Rest Follows

Aspect ratio is the cheapest creative decision available to you and one of the most visible. Get it right and everything downstream — composition, captions, motion, encoding — becomes easier to reason about. Get it wrong and no amount of polish rescues a scene that was framed for a screen nobody is looking at.

If you want to build the vertical version first instead of rescuing it later, start in the Orelon AI video generator with the frame already set to 9:16, sketch your scene inside the protected center column, and let the wide and square cuts fall out of the master. Browse the Orelon blog for more workflow breakdowns, or spin up still frames in the AI image generator to test compositions before committing to motion. One idea, every frame, no reshoots.