Twitter header maker

What this Twitter header maker guarantees

Files leave this page at 1500 pixels across and 500 down, or at double that. No custom canvas is offered, no height box, no proportion menu — each of those would end the page’s ability to guarantee anything at all. The pair comes straight from X, whose help center lists it as the recommended dimensions for a header image, checked on September 7, 2026. Divide one by the other and you get 3, a strip far wider relative to its depth than anything a camera hands you.

Choosing the doubled export gives 3000 by 1000: unchanged proportions, four times as many pixels. X shrinks oversized uploads itself, and its reduction onto a dense display generally beats rendering straight to the smaller count. Proportions are the one thing no later step recovers; resolution can always be discarded.

The 60 pixels, drawn because X named them

Almost every platform in this family publishes nothing about cropping. X is the exception, and it is worth quoting rather than paraphrasing: a header at the recommended size can still appear trimmed, they write, most likely because of how such images display across monitor sizes and browsers, and 60 pixels on the top and bottom could be cropped.

The overlay is on from the start, so those strips are marked out while you compose. Two qualifications belong with it. The 60 is X’s figure; the middle band it leaves — 1500 by 380 — is arithmetic done here. And the sentence hedges twice, with most likely and with could be, so it describes a tendency across screens rather than a line anyone has undertaken to hold. Build to it, do not align to it.

Everything else about placement is unpublished. X names no margin at the left or right and says nothing whatever about the portrait that sits over the lower left of a real profile. That is why the second box in the editor is adjustable and starts off: it is your own margin, taken from looking at your own profile, not a specification anyone can hand you. Neither box is ever encoded — the downloaded file holds your picture and your words and nothing else. The published figures, dated, are gathered in Twitter header size.

GIF is accepted, animation is not

X lists three formats for these images and then, in a separate sentence, rules out animated GIFs for profiles and headers. The two statements are easy to read apart, which is how people end up uploading an animation and receiving its opening frame. Since a still GIF also compresses worse than both alternatives at this job, it is simply absent from the format list here rather than offered as a trap.

PNG is the default when you start from a color, because that header is flat color and lettering. Put a photograph behind the text and the format switches to JPEG, which is normally several times lighter for a photo; either one is a single selection away. On file size, X publishes a 2 MB maximum for profile photos and nothing for headers, so the receipt names that distinction instead of pretending the number is a rule. Aiming at a byte figure deliberately is what compress image is for: it hunts for the quality that lands under a target while the dimensions stay untouched.

Fitting a photograph into 3:1

A 3:2 photograph has to surrender half its height to become a header, so this is the shape where the crop matters most. The picture is scaled until it reaches all four edges and the surplus discarded; which surplus is yours to choose by dragging inside the frame, nudging with arrow keys, or using the sliders. Bars down the sides would be the alternative and they are worse, since bars written into the file are bars X shows on every screen forever.

What you are looking at is not an approximation. One routine draws both the editor and the download, differing by a scale factor alone, so the frame rehearses the file exactly. That is also why the type control is a percentage of the header’s depth rather than a pixel count: switch export sizes and the lettering keeps its proportion instead of jumping.

What is not here

No designs to pick from, no stock photography, no sign-in. There is also no profile-photo mode: X wants 400 by 400 there, a square, and square image already crops to 1:1 without making you calculate anything. For artwork you already have, the image resizer crops and scales to typed numbers in one pass, and image dimensions tells you what a file measures as stored and as displayed — two different answers for anything shot on a rotated phone. The same kind of page for a different network is the LinkedIn banner maker, working to that platform’s own published figure. The YouTube banner maker is the third of them, and the one whose safe area comes with the awkward qualification that it was published about a smaller canvas than the one recommended.

Nothing you open is transmitted. Decoding, compositing and encoding all happen in this tab, and creatorvalet.com exposes no route that would accept an image. Treat that as testable rather than trustworthy: watch the network panel as you build one, or read how to verify local processing.