Making It Say SomethingLesson 51 of 64

Contents

Making It Say Something

Lesson 51 of 64. Sign in to keep your place.

Module 7 · The surface and the system carrying it

Something is always sitting on top of your picture

9 min read

The crop chain, and the box that survives itShapes one shot must carry9:16 full screen, 1080 by 19204:5 in a feed, 1080 by 13501:1 square for grids16:9 for a site or a slideA square thumbnail from the centreWhat survives all fiveA tall box near the centreNothing in the bottom fifthNothing in the right-hand stripText kept in the middle thirdA wide master in the archiveAutomatic reframing tracks a face or a saliency estimate and has no idea which part of the framecarries the meaning. Composing for the intersection is what makes a new shape an export rather than areshoot.
Automatic reframing tracks a face or a saliency estimate and has no idea which part of the frame carries the meaning. Composing for the intersection is what makes a new shape an export rather than a reshoot.

Figure

The frame you export is not the frame anyone sees

You export a 1080 by 1920 vertical video with your subject beautifully centred. On the platform, a username, a caption of three lines, a follow button, a music credit and a column of icons are drawn over it. The bottom quarter and the right-hand strip are gone. Your subject’s chin is behind a share button.

Nothing has gone wrong technically. The file is correct. The safe area (the part of the frame that is reliably visible) is much smaller than the frame, and it was never in your editing preview.

The numbers, held loosely

As a working starting point for full-screen vertical video on the current crop of short-video apps: keep anything that matters out of roughly the bottom 20 per cent and the right-hand 10 to 15 per cent, and leave a margin at the top for the status bar and any header.

These figures move. Platforms change their interfaces without announcement, they differ between iOS and Android, and they differ between phone sizes: a tall phone with a notch and a short one with a home button crop differently from the same file. Treat published specifications as approximate and verify the only way that works: post it privately, open it on a phone, screenshot it, and look. Two minutes, and it catches the failure that no amount of reading catches.

Title safe is an old idea that is still right

Broadcast television has had safe areas for seventy years. Cathode-ray televisions overscanned (they deliberately drew the picture slightly larger than the visible screen so that ageing components never revealed a black edge) and the amount varied by set. Producers therefore worked to two nested boxes: action safe at about 93 per cent of the frame and title safe at about 90 per cent, with text inside the inner one.

CRTs are gone and the convention survived, because the underlying problem did not: the frame is still cropped by something you do not control, it is just an app’s chrome now rather than a television’s bezel. Every editor (DaVinci Resolve, Kdenlive, Shotcut, CapCut, Premiere) can draw these guides, and turning them on is a single setting.

The crop chain is the harder problem

A single shot commonly has to survive several different crops of itself:

  • 9:16 full-screen vertical, 1080 × 1920.
  • 4:5 in a feed, 1080 × 1350: the tallest shape most feeds allow, and therefore the one that occupies the most screen.
  • 1:1 square, still used for grids and some previews.
  • 16:9 landscape for a website, a slide or a television.
  • A thumbnail crop, often square and often taken from the centre.

Automatic reframing tools will do this for you and will do it badly, because they track a face or a saliency estimate and have no idea which part of the frame carries the meaning.

The robust approach is to compose for the intersection: the region that survives every crop you will need is a tall-ish central box, and anything load-bearing goes in it. Shoot slightly wider than the tightest crop so there is material to give away, and decide the crops before you shoot rather than discovering them at export.

Text is the first casualty

Text placed near an edge is the thing that gets cut, and cut text looks like incompetence rather than like a platform quirk. Two habits:

  1. Keep on-screen text in the middle third vertically, not at the bottom where captions and interface live.
  2. Set it larger than you think. A phone held at arm’s length on a bus is not a monitor at 60 centimetres. If you cannot read it on your own phone at a glance, it is too small, and your eyes know the words already.

Doing it free

DaVinci Resolve’s free version, Kdenlive and Shotcut all handle multiple aspect ratios, safe-area guides and reframing on a modest laptop. CapCut runs on a phone and does vertical work competently. For stills, GIMP, Krita and the browser-based Photopea all let you set up guides once and save them as a template, which is the real time-saver, because the setup is the cost and you only pay it once.

The limitation

Everything above is a moving target that you do not control and cannot appeal. A platform can change its interface overnight and turn a year of correctly composed work into work with a button on its subject. There is no specification you can hold them to.

The defence is structural rather than technical: keep a master at the widest sensible framing, so a new crop is an export rather than a reshoot. Archive the project file, not only the finished video. The people who get hurt by a layout change are the ones who only kept the delivered file.

Every lesson here is open. Your progress is recorded once you pass the Making Things with AI exam.

Check your understanding

Why does automatic reframing from 16:9 to 9:16 so often cut the wrong part of a shot?

Pick the one you would defend. Your answer stays private.

Next lesson · 8 minThe still is what recruits the play

Stuck on something?Ask Meher, your AI tutor

How to cite: Addaly (2026) ‘Something is always sitting on top of your picture’, in Making It Say Something. Edited by M. Mun. Addaly. Available at: https://addaly.com/learn/visual-storytelling/platform-shapes-and-safe-areas (Accessed: 13 October 2026).