Vertical video stopped being a niche export preset the moment short-form feeds became the default way people watch. Today a clip can be shot on a phone, generated by an AI model, edited in a browser tab, and published to three platforms in one afternoon. The problem is that each of those steps speaks a slightly different language about size: resolutions, aspect ratios, pixel dimensions, bitrates, codecs, and crop windows. When the numbers disagree, the result is a video that looks soft, gets its captions hidden behind a button, or stutters on the first loop.
This guide walks through the sizing decisions that matter most for Instagram Reels, from canvas setup to export, plus how to keep one master file that survives being reused elsewhere. It is written for creators, social teams, and anyone generating footage with AI tools who wants the output to look intentional rather than accidental.
Why sizing decisions shape performance more than people expect
Reels are distributed inside a full-screen, swipeable feed where the viewer's thumb is already moving before the video loads. That context creates three practical consequences for sizing.
First, the platform re-encodes almost everything you upload. A video that arrives with a low bitrate or an odd frame size gets compressed again, and second-generation compression is where mushiness appears. Uploading a clean, well-sized master gives the encoder less to fight with.
Second, interface elements sit on top of your pixels. Username, caption, audio label, action buttons, and the progress bar all occupy real screen space. If your subject or text lands in those zones, viewers see a shoulder where a face should be.
Third, retention is measured in the first seconds. A video that starts with a black letterbox bar or a blurry upscale reads as low effort, and the swipe happens before your hook arrives. Sizing is not a cosmetic detail; it is the first quality signal a viewer receives.
A useful mental model is to treat sizing as three layers: the canvas (aspect ratio and resolution), the composition (where important elements sit inside that canvas), and the container (frame rate, bitrate, codec, file size). Fixing them in that order prevents the common trap of endlessly re-exporting a clip whose composition was never right to begin with.
Canvas setup: resolution and aspect ratio that Instagram prefers
The default Reels canvas is a 9:16 vertical rectangle. In practice that means 1080 × 1920 pixels for a full-quality upload. Other vertical sizes will play, but the platform scales them, and scaling is where detail disappears.
A few rules of thumb are worth memorizing:
- 1080 × 1920 is the target for a native, full-screen vertical upload.
- 1080 × 1350 (4:5) works well for feed posts and can be reused as a Reel, but it will be pillarboxed or cropped in the full-screen player.
- 1920 × 1080 (16:9) landscape footage is the most common source of blurry Reels, because it must be blown up or heavily cropped to fill a vertical frame.
- Square 1080 × 1080 is acceptable for archive content but wastes roughly a third of the visible area.
What happens when you upload lower than 1080 wide
A 720 × 1280 export is not automatically bad — plenty of phones record at that size and it plays fine. The issue is headroom. Once the platform re-encodes and a viewer watches on a high-density display, a 720-wide master has no spare detail to give. Text edges soften first, then skin texture, then fine patterns like fabric or foliage.
If your source is genuinely lower resolution, upscaling with an AI video enhancement pass before export usually beats letting the platform do the stretching. Modern upscalers reconstruct plausible detail rather than smearing pixels, which keeps text legible and faces crisp. Do the upscale as a dedicated step, review the result at 100 percent zoom, and only then place it on the 1080 × 1920 timeline.
Handling footage that was not shot vertically
When you must use landscape or square source material, choose a strategy deliberately rather than letting the editor guess:
- Crop to the subject. Reframe so the subject fills the vertical frame. This preserves sharpness because you are using real pixels, not scaled ones. Check every shot — heavy crops can decapitate a subject or lose hand gestures that carry meaning.
- Blurred background fill. Place the original clip in the center with a blurred, zoomed copy behind it. This keeps the whole image visible but reads as a repost rather than native content. Use sparingly and never for a hook shot.
- Designed layout. Build a deliberate composition: landscape clip up top, captions or data below. This works well for tutorials, product comparisons, and before/after content where two visuals need to coexist.
- Split and stack. Cut a wide shot into two vertical panels and animate them in sequence. More editing effort, but it converts a static wide shot into movement that suits the format.
Whichever route you take, set the project canvas to 1080 × 1920 first and work inside it. Editing in landscape and converting at export is how off-center subjects end up half-hidden behind a caption.
Safe zones: composing around interface overlays
Reels has reserved screen regions. Ignoring them is the single most common reason good videos look badly framed on a phone.
Mapping the danger areas
The bottom of the frame is the busiest. Caption text, the account name, the audio ticker, and the action rail consume a meaningful band along the lower edge. The right side holds the vertical action rail — like, comment, share, and more — extending upward well into the frame. The top carries the progress indicator and sometimes a label.
A practical composition rule: keep essential content within roughly the central 1080 × 1420 band, with the top and bottom bands treated as risky. Anything you truly need read — a headline, a price, a call to action — should sit comfortably above the bottom caption area and left of the action rail.
A simple test: open a published Reel on a phone, screenshot it, and overlay your editing timeline's safe-zone guides. After two or three cycles you will internalize the boundaries and stop needing the overlay.
Captions, logos, and CTAs
On-screen text should be sized for a viewer holding a phone at arm's length. Two to four words per line, high contrast, and a solid or semi-transparent backing plate if the background is busy. Thin fonts and outlines vanish after re-encoding, especially over moving footage.
Place a persistent logo either in the upper third of the left side or near the midline of the top edge, small and low-contrast. A watermark that competes with the subject is a distraction tax on every second of the video.
Calls to action belong in the video and in the caption, and they serve different purposes. An on-screen prompt like "save this" works best in the middle third where it is fully visible. A caption prompt can be longer and more specific, because viewers who reach it are already engaged.
Frame rate, bitrate, and file size: the container settings
Canvas and composition are visual. Frame rate, bitrate, and file size are technical — and they decide whether the visual survives the upload.
Choosing 30 or 60 frames per second
30 fps is the default for talking-head content, tutorials, and anything with normal motion. It produces smaller files and compresses efficiently.
60 fps pays off for fast movement: sports, dance, action shots, and slow-motion sequences that will be retimed in the edit. It also feels smoother during quick camera moves. The trade-off is a larger file and, at a fixed bitrate, less data per frame.
24 fps is a stylistic choice borrowed from cinema. It reads as filmic and is a poor fit for screen recordings or interface demos, where it makes cursor movement look choppy.
Also avoid mixing frame rates on one timeline without thinking. Drop a 60 fps clip into a 30 fps project and the editor either drops frames or duplicates them; both produce judder that viewers read as "cheap."
Bitrate targets that survive re-encoding
Bitrate is how much data each second of video receives. Too low and you get blocking in gradients, banding in skies, and smeared motion. Too high and the file becomes slow to upload without visible benefit.
For 1080 × 1920 vertical video, sensible export targets are:
- 8–12 Mbps for typical talking-head and lifestyle content at 30 fps.
- 12–16 Mbps for high-motion footage or 60 fps.
- 16–20 Mbps when the frame contains fine detail that must stay crisp, such as text-heavy graphics or product macro shots.
Use H.264 in an MP4 container for the widest compatibility. Variable bitrate with a two-pass encode gives better quality per megabyte than a fixed rate. Keep audio at 128–192 kbps AAC, because poor audio drives viewers away faster than soft video.
File size limits and how to stay under them
Most upload paths cap files at a few gigabytes and duration at a few minutes for Reels. A well-encoded 60-second vertical clip at 12 Mbps lands around 90 MB, which is comfortably fine. Problems appear with long clips, high-bitrate 4K masters, or files with uncompressed audio.
When a file is too large, fix the cause rather than blindly lowering the bitrate:
- Trim dead air at the start and end — usually the biggest single saving.
- Export at 1080 × 1920 rather than 4K; the platform will downscale anyway.
- Drop to 30 fps if the motion does not need 60.
- Use a two-pass variable bitrate export instead of a fixed high rate.
- Split very long content into a series rather than a single heavy upload.
Export settings by editing tool
Different tools hide the same settings behind different labels. Here is how to approach the common categories.
Mobile editors
Phone-based editors usually offer a "Reels" or "9:16" preset that sets resolution and aspect ratio automatically. Verify three things before exporting: the canvas is 1080 × 1920, the frame rate matches your project intent, and captions are not sitting in the lower danger band. Mobile exports also tend to default to a lower bitrate than desktop tools, which is fine for simple footage but can soften detailed graphics.
Desktop editors
Desktop software gives you full control. In your export dialog, set format to MP4, codec to H.264, resolution to 1080 × 1920, frame rate to match your source, bitrate to a variable two-pass target in the range above, and audio to AAC. Save this as a preset so every project starts from a known-good configuration.
AI video generators and upscaling passes
AI video tools typically output at fixed sizes — often square or landscape, sometimes 1080 × 1920 already. Two habits make AI-generated footage look native in a vertical feed:
- Generate or upscale with the final aspect ratio in mind. If the tool supports a vertical output mode, use it; reframing after generation often crops out the composition the model carefully built.
- Do the enhancement pass before compositing. Upscale, interpolate frame rate if needed, then bring the clip into your editing timeline. Adding text and graphics after enhancement keeps them razor sharp, because the upscaler never sees them.
Be cautious with motion smoothing. Frame interpolation can make generated footage look fluid, but it also introduces warping around hands, hair, and text. Review at full speed, not frame by frame, and keep a non-interpolated version as a fallback.
Cross-platform consistency: one master, several crops
Most creators publish the same idea to more than one vertical feed. Fighting that reality with separate edits for each platform doubles the work and fragments your source files.
A more efficient pattern is to build a single master and derive variants:
- Master: 1080 × 1920, 30 or 60 fps, 12–16 Mbps, no baked-in platform-specific overlays, clean audio stem.
- Variant A: trimmed to the shortest platform's duration limit.
- Variant B: reframed slightly tighter for feeds that place UI differently.
- Variant C: a 1080 × 1350 crop for feed posts and profile grids.
Keep graphics in the master on separate layers when possible, export flattened only at the end. If your editing tool supports it, export the same timeline at two aspect ratios rather than rebuilding it. Ten minutes of setup saves hours across a month of posting.
A pre-publish quality control checklist
Run this before every upload. It takes about ninety seconds and catches nearly every sizing defect.
- Canvas: 1080 × 1920 confirmed in the export settings, not just the preview.
- Frame rate: consistent across all clips; no accidental mixed-rate timeline.
- Bitrate: inside the recommended band for the motion level.
- Safe zones: no text, faces, or logos in the bottom caption band or behind the right-side action rail.
- Hook: the first second is visually strong with no letterbox bars or title cards.
- Text legibility: captions readable at 50 percent zoom on a phone screen.
- Audio: normalized, no clipping, and levels consistent from start to finish.
- Loop: the final frame transitions acceptably back into the first.
- File: under the platform's size cap, correct container and codec.
- Thumbnail frame: pick a cover that is not a mid-blink or motion-blurred frame.
Common sizing mistakes and how to fix them
The blurry upscale. A landscape clip blown up to fill a vertical frame. Fix: crop to the subject at native pixel size, or upscale deliberately with an enhancement pass and inspect at 100 percent.
Text under the caption area. Headlines vanish behind the account name. Fix: pull all critical text up into the central band and re-check on a phone.
Banding in gradients. Skies and soft backgrounds show stripes. Fix: raise the bitrate and export with a two-pass variable bitrate setting.
Judder in panning shots. Mixed frame rates or too-low bitrate. Fix: conform all clips to one frame rate and slow the pan slightly.
A static first frame. The viewer sees a frozen image, assumes a photo, and swipes. Fix: start with motion — a hand entering frame, a cut, a zoom.
Over-cropped faces. Aggressive vertical crops cut off chins and foreheads. Fix: reframe manually with headroom, and check the crop on every clip rather than applying one setting globally.
Inconsistent look across a series. Each video has a different text size and position. Fix: build a template project with locked safe zones and reusable text styles.
FAQ
Does Instagram Reels require exactly 1080 × 1920?
No, but it is the size the platform handles best. Other vertical dimensions play, but they get scaled, which costs sharpness and can shift where your safe zones land.
Is 4K worth uploading?
Rarely for Reels. The file is much larger and the platform downscales it anyway. A well-encoded 1080 × 1920 export at a healthy bitrate usually looks better after re-encoding than a 4K file with a modest bitrate.
Should I use 60 fps for everything?
No. Use 60 fps when motion benefits from it, and 30 fps otherwise. Higher frame rates cost bitrate, and at a fixed bitrate that means less detail per frame.
How do I keep captions visible?
Compose within the central band of the frame, avoid the lower caption area, and keep text in the middle third where the interface does not overlap. Test on a real phone before publishing.
Can AI-generated footage look native in a vertical feed?
Yes, if you generate or upscale with the vertical aspect ratio in mind, refine the clip before adding graphics, and check the composition against safe zones. Text and overlays added after enhancement stay crisp.
What is the fastest way to fix an old landscape video?
Crop to the subject at native resolution, place it on a 1080 × 1920 timeline, add vertical-friendly captions in the central band, and export with a two-pass variable bitrate around 12 Mbps. That single pass fixes most quality complaints.
Sizing is not glamorous work, but it is the difference between a video that looks homemade and one that looks produced. Set the canvas once, respect the safe zones, tune the container settings, and reuse a master file across every feed. The technical layer then disappears from view, which is exactly when content starts to perform.
