Make a Pinterest Pin Image

How it works

Your reference image goes to an image-editing model with an instruction describing a tall Pin composition: a clear visual hook, strong vertical hierarchy, and room for a headline added later, with the subject preserved and no text generated. A second instruction rules out distorted subjects, unreadable text, watermarks, bad crops, and clutter.

The picture is generated at 1000 by 1504, because 1500 is not a multiple of eight and the generator would quietly round it down. That result is then scaled to exactly 1000 by 1500. Four pixels of reduction is invisible; four pixels of shortfall would have been rescaled by the platform on upload.

What this actually produces

It is worth being precise about the deliverable, because it is not a finished Pin.

Most Pins that work are two layers. Underneath is an image doing the job of attracting a glance in a crowded grid. On top is a headline stating what the Pin is for, because a picture alone rarely answers the question a searcher typed.

This page makes the layer underneath. The instruction asks for a hook, a clear top-to-bottom structure, and space where type can go — and asks for no lettering, because generated text is unreliable and because the headline is the thing you will want to rewrite without regenerating everything else.

Asking is not the same as guaranteeing, so check the frame. A review on 13 August 2026 returned a plate with INTER EC set across the top in a serif that looked entirely deliberate. The model cannot spell, and instructions not to write are the kind a diffusion model follows unevenly. Anything letter-shaped in the output is something to crop or paint out.

So the right way to use it is as a plate. Generate, look at the whole frame, open it in whatever you set type in, and add the words yourself.

Designing for a long life

The other platforms on this site reward being current. This one punishes it.

A Story disappears. A post falls down a feed and stops being served within a day or two. A Pin is indexed against what it appears to be about, shown against searches, saved into other people's boards, and reshown from those boards indefinitely. A large share of the views a Pin ever gets can arrive long after the day it was published.

That makes datedness the main risk. Anything anchored to a specific moment — a season, a sale, a reference that made sense at the time — stops being useful while the Pin is still being distributed, and it will be distributed by people who have no idea it expired.

The design consequence is to prefer the general to the topical. An image about a thing rather than about an occasion. Nothing in the picture that answers the question when.

The grid is the real viewing size

Pins are found in a masonry grid of other Pins, at a fraction of their full size, next to competitors for the same attention.

That is the size to judge yours at. Shrink the file to roughly the width of a column in a phone browser and look at it among others. What you are checking is whether it is distinguishable — whether the subject separates from its background, whether the tonal structure survives, whether it is obvious within a moment what kind of thing this is.

Full size only matters after somebody has already decided to tap.

This is also why the vertical shape matters commercially rather than aesthetically. A two by three Pin occupies more of the column than a square one does, so it simply takes up more of the screen while a person is scrolling past. The shape is not a style choice. It is space.

Publication gate

This page ships once the workflow has been run against an overhead scene, a single object, and a portrait, with each output measured at exactly 1000 by 1500 and viewed at column width in a live grid alongside other Pins.