Elhadi.
← All posts
5 min readAI-written · in Suhaib Elhadi's voice

Safe zones for vertical video: where the app covers your ad

This update was drafted on a schedule by the AI I build with, from real project notes — part of the vibecoding experiment this blog documents.

So here's a mistake that costs nothing to avoid and I still made it.

You render a vertical clip. It's 1080 by 1920, it looks great in the player, the hook text sits right there in the frame where you put it. Then you post it, open the app on your actual phone, and the username is sitting on top of your first word. The caption's eaten the bottom third. The like button is parked on the product.

The video is fine. The layout is wrong. And it's wrong in a way you literally cannot see until it's live, because the thing covering your work isn't in your file — it's the app's own interface, drawn on top after you hand it over.

This matters more for ClipForge than it does for one-off posting, because the whole point there is volume — the engine cuts clips and posts them across accounts on a schedule. A layout mistake doesn't ruin one post. It ruins every post, quietly, until somebody looks at a phone. Which is the exact failure shape I keep writing about: silent failures beat loud ones for damage, every time.

What a safe zone actually is

Your file is the full 1080×1920. The app shows all of it. Then it draws its own stuff on top:

  • Top: the account header, the sponsored tag, sometimes a "following / for you" bar.
  • Bottom: the caption, the audio track name, the handle. This is the big one, and it grows when the caption is long.
  • Right side: the action rail — like, comment, share, save, the spinning audio disc. Full height, roughly a thumb wide.
  • Left and the very edges: small margin, plus whatever the phone's own rounded corners and gesture bar take.

The safe zone is what's left in the middle. Everything that has to be read — hook text, captions, the logo, the price, the CTA, the face — lives inside that box. Everything else can run to the edges, and should, because the video itself should still be full-bleed.

The actual numbers, and why I don't fully trust any of them

This is where I have to be careful, because there's a lot of confidently-stated numbers floating around and they don't agree with each other.

For a 1080×1920 frame, the published guidance lands roughly here:

  • TikTok in-feed: around 13% off the top (~250px), around 37% off the bottom (~700px), with a right-side action column around 20–22% wide (~240px) running the full height. (spec roundups put it in that range)
  • Instagram Reels: around 14% off the top (~270px), around 35% off the bottom (~670px), with roughly 6% (~65px) on each side, and the action rail eating into the bottom-right well above that 35% line. (similar roundups land here)

Two honest caveats on those.

First, they're third-party roundups, not platform docs. I'm citing them because they're consistent with each other and with what I see on a phone, not because they're authoritative. Different guides give different pixel counts for the same thing.

Second, they change. The apps redesign. Captions are variable-length. A long caption with a "more" link takes more room than a short one. Anything you memorize today has a shelf life.

So I treat published numbers as a starting box, not a spec.

The layout I actually use

Rather than trying to hold two different platforms' pixel maps in my head, I design to one conservative box that survives both, and then verify.

Keep the readable stuff in the middle ~50% vertically. If I stay out of the top ~15% and the bottom ~37%, I'm inside both platforms' guidance with a little room spare. That leaves a band in the vertical middle that's safe basically everywhere.

Stay off the right ~22%. The action rail is the most consistent obstruction across every platform, and it's full-height, so there's no "it's fine up here" exception.

Put the hook high, but not at the very top. Hook text wants to be up where the eye lands first, and the top obstruction is much smaller than the bottom one. Just under the header is the sweet spot.

Never put anything that matters in the bottom third. That's caption country. It's the single biggest obstruction and the most variable one.

Full-bleed the video, box the text. The footage fills the frame edge to edge. Only the elements that have to be read get constrained. A clip with the whole image shrunk into a letterboxed middle looks like what it is — a repost.

Burn your own captions, don't rely on the platform's. Platform auto-captions land wherever the platform wants, which is often right in the obstructed zone. Your own captions, placed in the safe band, get read.

Verify on a phone. Actually do it.

Both Ads Manager tools have safe-zone overlays — Meta's ads preview has a safe-zone toggle, and TikTok's ad tooling previews the in-feed layout. Those are the closest thing to a source of truth, and they beat any blog post including this one. Use them.

But for organic posting, the real check is dumber and better: post one clip to a test account and look at it on your phone. Not the desktop web player, which has a different layout and will lie to you. The phone, in the app, with a realistic caption length.

For something like ClipForge, where the engine produces to a template, this is a one-time cost that protects everything downstream. You get the template right once, check it on a device, and every clip after that inherits it. That's the same argument I made for input floors — a boring constraint in the pipeline beats catching it per-item, especially when nothing's going to catch it per-item.

Where this connects to the AI-video stuff

If you're generating the footage rather than cutting it from a source, safe zones are another argument for keeping text out of the prompt. Text generated into the scene is wherever the model decided to put it. You can't move it. If the model's rendered headline lands under the caption bar, your options are re-roll and hope, or ship it broken.

Text as a layer in the edit moves wherever you need it. Including out of the way.

And when you're planning shots, it's worth composing for the box up front — prompt the product slightly left of center so the action rail doesn't sit on it, leave headroom at the top. Cheaper than fixing it after.

The short version

Design to the middle. Stay off the right rail. Treat the bottom third as gone. Verify on a real phone with a real caption, because every number in this post, including mine, has an expiry date.

This one's auto-drafted from my notes on a schedule. If a number isn't in the notes, it doesn't show up here — I'd rather leave a blank than make something up.