Vertical Video Safe Zone — will your caption survive the interface?

Every vertical platform draws its own interface over your video: titles, buttons, usernames, descriptions, progress bars — all of it sitting between your content and the viewer, all of it placed by the platform's designers without asking yours. The result is the most common silent failure in short-form video: a caption, logo, or key detail burned into the frame exactly where an overlay covers it. The video is fine; the interface ate it. This tool makes the problem visible and solvable: upload a frame from your clip, overlay each platform's obstruction zones, and drag a caption to the positions where it stays readable everywhere. The rule the tool enforces is one number pair — captions belong between 18% and 68% of the frame height measured from the bottom — and everything on this page explains where those numbers come from and how to design around them.

The overlay stack, platform by platform

Think of each player as a stack of zones painted on top of your 9:16 frame. The tool models the three major platforms with the geometry its overlays use:

YouTube Shorts reserves the top of the frame — roughly the upper 12% — for the video title and the search/camera entry points, with its action buttons running down the right side. The bottom carries the channel identity and the title again in expanded views. The practical consequence: nothing you need seen should live in the top eighth, and right-edge content competes with the action rail.

TikTok is the most demanding. The bottom ~14% belongs to the description, music ticker, and your username; the right side — roughly the rightmost 18% of the width between those bottom controls and the top bar — is a vertical column of like, comment, share, and save buttons; and the top bar holds account and search entry. TikTok covers more of your frame than any other platform, which is why the tool's "all platforms" view is effectively the TikTok view plus margin.

Instagram Reels follows a similar shape — actions on the right, identity and description at the bottom, a gradient behind the bottom text to force readability — with the gradient itself acting as an obstruction zone: content behind it survives technically but drowns visually.

The zones are models of current interface behavior, measured against the platforms and stamped with a verification date; interfaces evolve, so treat the overlays as the authoritative snapshot of their date and re-check after major platform redesigns.

Where the 18–68% band comes from

The caption band is arithmetic on the zones above, measured from the bottom of the frame upward. Below about 14% from the bottom, content sits under navigation and description chrome on every platform — so the band starts at 18%, keeping a margin above the highest bottom zone. Above about 88% from the bottom (the top 12%), content sits under titles and top bars — so the band ends at 68%, leaving headroom below the title zone and away from the visual center where action rails intrude from the side. Between 18% and 68% inclusive, a burned-in caption clears every modeled obstruction on all three platforms at once.

Two properties of the band matter in practice. It is a floor and ceiling, not a target: anywhere inside it works, and the best position within it depends on the frame's own content. And it is inclusive at both edges — 18% exactly and 68% exactly both pass — but positions a single percent outside fail, which is why the tool reports safety as a binary badge rather than a gradient of mostly-fine.

Why "centered" is not automatically safe

The instinct that saves horizontal video — put the important thing in the middle — fails vertically in two specific ways. First, the true visual center of a 9:16 frame sits higher than people expect, and "comfortably centered" placements routinely land in the lower third… inside the bottom zone, under the description chrome. Second, center placement ignores the horizontal asymmetry: the right rail means the right half of even a "centered" composition carries less safe area than the left half, so text that is centered but wide can have its right edge under buttons while its left edge reads fine. The reliable definition of safe is not "middle of the frame" but "inside the band, clear of the right rail" — a zone, not a point — and it is exactly what the draggable caption in the tool tests.

The drag workflow

1. Upload a frame from your actual clip — not a generic placeholder, because the point is to see the zones over your composition. The tool accepts image and video files and renders them inside a 9:16 preview.
2. Choose a platform view or "all." All shows the union of obstructions; individual views show each platform's zones, which is useful when a clip is destined for one destination and you want its extra room back.
3. Drag the caption. The slider moves the sample caption between 6% and 84% from the bottom. Inside the band, the caption box turns green with the badge reading safe everywhere; outside, it turns red with the warning to drag toward the middle. The color flip is the entire decision, live.
4. Export the position. When the badge is green and the caption sits where your frame's composition supports it, note the percentage and set it in your caption preset. In WeaverClip's clip composer, the caption engine respects these bands automatically and lets you preview before rendering; elsewhere, the Y percentage transfers to any editor that positions text by frame coordinates.

The loop is fast enough to run per clip, and worth it per clip: the cost is ten seconds, and the failure it prevents — publishing a clip whose text is unreadable — is the kind of defect viewers attribute to amateurism rather than to geometry.

Designing the frame for the zone

Safe-zone discipline works best when it starts at shoot time rather than at caption time. Three habits carry most of the value. Frame with margins: when recording vertical, keep faces and key action out of the bottom fifth and the right edge, which buys freedom later no matter where captions land. Plan graphics in the band: lower thirds, name straps, and stat cards should be authored to live between 18% and 68% from the start; a graphic designed for the bottom of the frame is a graphic designed to be covered. Watch the conversion from 16:9: center-cutting horizontal footage to vertical throws away both sides — and whatever lived at the edges, including any burned-in watermark or caption, which must be re-positioned into the band on the new canvas rather than surviving the crop by luck.

Burned-in captions versus platform captions

The band governs burned-in text — captions rendered into the video pixels before upload — because the platform cannot move what is already in the picture. Platform-provided captions are a different system: the player positions them itself, typically above the bottom chrome, and viewers can toggle them. Two practical consequences. Burned-in captions are the tool's subject and the safe-zone problem's real stakes: they are permanent, always on, and exactly as placed as you placed them. Platform captions relieve you of placement but hand you the trade of depending on auto-generated accuracy, which for specialized vocabulary, names, and accents needs correction wherever the platform allows edits. Many producers ship both: burned-in key text inside the band for the sound-off scroll, and corrected platform captions for accessibility — the two systems overlapping in purpose but not in mechanics.

Worked example: one frame, one fix

Note for vertical-video-safe-zone: The clip below is a hypothetical example — an invented frame used to demonstrate the workflow, not a real published video.

A fictional talking-head clip carries its burned-in caption at 12% from the bottom — the classic placement, chosen because it feels natural near the speaker's chin. Uploaded to the tool with the all-platforms view, the caption box lands squarely in the red: below the 14% bottom-chrome line, covered by description text and navigation on every destination. The fix is a drag, not a redesign: the slider moves the caption to 24%, the badge flips green, and the frame still reads naturally because 24% sits just below center — close enough to the original feel that viewers perceive no change, high enough that TikTok's description and Shorts' bottom chrome both pass underneath it. The alternative fix — re-editing the video to shift the speaker upward — would cost an hour for the same result the ten-second drag achieved, which is the entire argument for checking zones before publishing rather than after a comment points out the covered text.

One clip, three platforms: the common-denominator strategy

The band exists precisely so a single master works everywhere, and the strategy that follows is simple: design to the union of all three obstructions — the tool's all-platforms view — and every destination inherits a readable result. The cost of the common denominator is real but small: a few percent of extra margin at the band's edges, and accepting that a TikTok-only clip could run its caption slightly lower than 18% if you are willing to maintain separate masters. For almost every workflow the separate-master savings are not worth the version-management risk — two masters means one of them will eventually be uploaded to the wrong platform. The exception is channel-native series produced for exactly one destination, where per-platform tuning is a legitimate optimization; even there, the band remains the safe fallback when a clip gets cross-posted later, which clips always do.

Mobile-first realities the zones do not cover

Overlay geometry is necessary but not sufficient, because the viewing conditions of short-form video impose their own constraints. The audience is on phones, often small ones, in variable light, holding the device one-handed. Three design responses follow. Text must be sized for a five-inch screen at arm's length: caption lines that read fine on your editing monitor routinely dissolve on a phone, so the practical minimum is large — roughly the size that remains readable when you hold your phone at arm's length and squint slightly. Contrast must survive both bright daylight and dark rooms, which is why the professionals put a background behind caption text — a solid plate or a strong shadow — rather than relying on the frame's own contrast. And thumb reach matters more than composition theory: the lower third is where thumbs live and where accidental touches happen, another quiet argument for keeping critical content above the band's floor.

Long captions and multi-line text

The band is measured in frame height, and text grows downward from its anchor — so a caption that fits at one line can fail at three. The practical limits inside the band: two lines of caption-sized text fit comfortably anywhere in it; three lines start to threaten the ceiling when anchored near 68%, because the text extends below the anchor point toward the obstruction. The remedies, in order: shorten the caption (captions may compress speech without distorting it — a skill shared with subtitle work everywhere), split the moment across two caption beats timed to the speech, or anchor lower in the band with room for the text's downward reach. The same arithmetic governs stacked elements — a caption plus a logo plus a call-to-action: each needs its own clearance from the zones, and the stack's total height decides where its anchor can live. When the stack does not fit inside the band with margins, something leaves the frame; choosing what leaves is an editorial decision the tool surfaces but deliberately does not make.

What the zones are — and what they are not

The obstruction zones are models: measurements of where current platform interfaces draw their chrome, taken on a stated date, expressed as percentages of the 9:16 frame. They are accurate to the interface versions they were measured against, and they are not guarantees — platforms redesign, A/B test interface variants, and render differently across device sizes and app versions. Treat the tool as the conservative envelope: the band it enforces carries margin by construction, which is what makes it robust against small interface drift. The honest statement of scope: a caption inside the band has an excellent chance of surviving every modeled platform today, and the discipline of checking — not any individual snapshot — is what keeps it surviving as interfaces evolve. Re-verify after any platform redesign announcement, and re-run your own frames rather than trusting memory; zones are cheap to check and expensive to get wrong.

FAQ

Why measure from the bottom instead of the top? Because the most contested space on every platform is the bottom — descriptions, navigation, and identity chrome all live there — so bottom-anchored percentages make the dangerous zone the zero point. The band 18–68% from the bottom equals 32–82% from the top; the numbers are the same rule expressed in whichever direction your editor prefers.

Does the safe band apply to faces and action, or only text? The zones are obstructions for anything that must be seen — faces, products, text alike. The band is framed around captions because burned-in text is the most common casualty, but the same geometry protects any element you cannot afford to have covered.

My caption is at 18% exactly. Is that safe? Yes — the band is inclusive at both edges, so 18% and 68% pass while 17% and 69% fail. In practice, give yourself a percent or two of margin beyond the edge unless the frame forces the boundary; the edge exists for the check, not as a target.

Can I upload video instead of a still frame? Yes — the tool accepts video files for the preview. The zones are frame geometry, so any representative frame of the clip shows the same picture; choose a frame with the composition you actually care about protecting.

Why does the TikTok view show a right-side column the others don't emphasize? Because TikTok's action rail is wider and more central to its layout than the equivalent rails on the other platforms — roughly the rightmost fifth of the frame. Content that must stay visible should keep clear of that column in the TikTok-specific view; the all-platforms union already accounts for it.

Do landscape or square videos have safe zones too? Different ones — the overlay geometry is a property of each platform's player for each aspect ratio, and this tool models the 9:16 case that short-form platforms are built around. For square or landscape destinations, test in the platform's own preview before trusting any generalized numbers.

Covered captions are an accessibility failure, not just a visual one

The safe-zone conversation usually runs in terms of aesthetics — an eaten logo looks amateurish. The deeper stake is access. Burned-in captions are how deaf and hard-of-hearing viewers get dialogue in environments where platform captions are absent, disabled, or wrong; when an overlay covers those captions, the clip is not slightly degraded for those viewers but effectively silent. The same is true for sound-off viewers in general, who make up a large share of any feed: a covered caption converts your clip from content into a guessing game. This is why the zone check belongs in the accessibility review alongside alt text and caption accuracy rather than in a purely visual polish pass. The ten-second drag test is, in practice, one of the cheapest accessibility fixes available in short-form production — and the red/green badge makes it a check that nobody has to remember to perform, because the failure is visible the moment the frame loads.

Cover frames and thumbnails have zones too

A clip's cover frame — the still that represents it on the channel grid and in feeds — sits under the same interface family: duration badges, selection UI, and grid chrome all draw over specific corners and edges depending on the platform. The transferable discipline: keep the cover frame's essential content — faces, text, the visual hook — inside the same conservative middle area you protect for captions, and preview the frame inside the platform's own picker before committing. Producers who design covers in full-frame and discover the crop in the grid are in the same failure family as covered captions: geometry assumed instead of checked. The tool's frame-upload workflow applies directly: load the candidate cover, see what the overlays touch, and adjust the composition until nothing essential is contested.

The caption-to-render pipeline

Safe-zone checking slots into the finishing pipeline at one specific place: after caption text is final and before the render commits. The order that prevents rework: write and time the captions first (with speech-accurate content), position the block inside the band using the tool's drag workflow on a representative frame, then render. Rendering before the zone check is the expensive direction — every covered caption found after render costs a full re-render, and batches make the cost multiply. In WeaverClip's clip composer the caption engine keeps text inside these bands by default and previews the result before the render runs, which is exactly the pipeline order encoded as product behavior; in any other editor, reproduce the order manually: content, then position, then pixels. One last pipeline habit with outsized payoff: check one frame per caption style, not one frame per clip — styles repeat across clips, so a style verified once stays verified, and per-clip checks become spot-checks rather than full passes.

FAQ — a few more

What about videos with hard-coded platform watermarks from other tools? They live where the other tool put them — frequently exactly where an overlay will cover them. Re-exporting without the watermark is the clean fix; where that is not possible, treat the watermark like any other contested element and check it against the zones before publishing.

Do the zones change between phones and tablets or desktop players? Interfaces scale and rearrange across device classes, and desktop players generally cover less of the frame than phones do. The conservative move is designing for the worst case — the phone player — because every viewer who sees the phone version sees the maximum obstruction, and desktop viewers simply get extra margin.

How often should I re-check these zones? After any major platform redesign and at least on a seasonal cadence if short-form is a core channel for you. The band carries margin by design, so small interface drift rarely breaks it — but the tool's value is precisely that checking costs ten seconds, which is cheaper than any assumption about yesterday's geometry.

Can the band be different for different languages or caption lengths? The band is frame geometry, so it is language-independent; what changes with language is how much text fits inside it. Longer-word languages hit the multi-line limits described above sooner, which is a sizing problem solved inside the same safe area, not a different safe area.

Does this replace testing on the real platforms? It replaces the surprise, not the final check. The zones tell you where obstruction lives before you publish, which is where the fix is cheap; a quick scroll through the live clip on one real device remains the last word, because it shows you the interface as your audience's app version actually renders it.