Generative AI Beginner

Outpainting

Widening a picture's frame and drawing the new edges to match

Key points
  • Outpainting opens up new space around a picture and fills it in so the scene keeps going.
  • The only clue it has to work with is the original's edge. That makes it harder than fixing something on the inside.
  • Widening a lot in one go breaks the flow. A little at a time, over several passes, reads far more natural.
  • It's especially handy for fitting a tall photo into a wide frame, or opening up room for text.
  • The real scene that got cropped off doesn't come back. A plausible outside gets built from scratch.
Contents

1The analogy

Want a blanket a little bigger? You don't rip open the middle — you stitch fabric onto the edge. And you can't just grab any scrap; you match it to the pattern and color of the square it's joining, so the whole thing still reads as one blanket.

That's the stitching outpainting does. It opens new space around a photo and fills it so a wall cut off at the edge keeps running sideways, or the top branches of a tree that got cropped out reappear.

Stitch on a blanket-sized scrap in one go, though, and there's only one seam to match against, so the pattern drifts. Add a hand's width, check the result, add another — that reads far more natural.

2In detail

The edge is the only clue it has

Outpainting starts by setting the original photo down on a bigger, empty canvas. The spot the original sits on stays fixed, and only the empty space around it gets built. At every step, the original's spot gets pasted back over from the source, so the photo itself never drifts. The method is the same one used to fix something on the inside — the only difference is whether you're working inward or outward.

The hard part is that the clue only comes from one side. Fill something on the inside and you get pattern and color from every direction; fill the outside and there's just that one strip of original edge to build the rest from. A clear line running through it — a horizon, a wall corner, a road — stitches on well. A messy edge gives a messy result.

Widen a little at a time

Double the canvas in one pass and the outer half sits too far from the original to guess well, so it drifts into a scene with nothing to do with the source. Widen a hand's width at a time instead, and what you just built becomes the next pass's clue, so the flow keeps going.

Pick one direction at a time, too. Stretch left-right and up-down together and the four corners have no clear clue to follow, so they blur out. Widen left first, finish it, then widen up — now the corners have an already-built neighbor to follow.

Even done gradually, a slow drift creeps in over several passes: colors fade a little, the same pattern repeats, light shifts direction. Every few passes, it helps to look the whole thing over and fix whatever's drifted.

You can say what fills the new space

Leave the sentence field blank and the tool just keeps going with whatever the original was doing — an alley photo gets more alley, a field photo gets more field.

Write a sentence and you decide what shows up in the new space. Describe both the original scene and what you want to appear outside it — describe only the outside and you'll get a scene with a different feel stitched on next to the original, and the seam will show.

Where it's useful

The most common use is changing the aspect ratio. A photo shot tall that needs to fill a wide frame gets widened on the sides instead of cropped, so nothing from the original gets thrown away.

There's a place to be careful, too. Asking it to redraw the rest of a person or object that got cut off at the frame's edge tends to look off — limbs and posts are shapes your eye tracks closely, and the smallest mismatch stands out. It's usually faster to widen first and then repaint just the awkward part.

It's also handy for opening up space for text. Building a banner or a cover and need room up top? Widen upward only and let the sky extend. The same trick adds a bit of extra margin outside a print job's trim line.

3More precisely

Outpainting treats the whole of the new canvas outside the original's spot as one mask to fill. Many tools rebuild a thin strip of the original's edge along with it, overlapping slightly, to soften the seam. The result gets a bigger canvas, but the detail inside the original portion doesn't increase — that's a different job, going the other direction, from making a picture bigger, though the two often get used together.

The blanket comparison breaks down in places. A person picks the matching fabric scrap by eye; outpainting invents the whole outside from just one strip of edge, with nothing like a bolt of matching fabric sitting in reserve. The middle of a real blanket stays untouched once you stitch on a scrap, while outpainting redraws a bit near the border too, so the original edge can shift slightly. And a real patchwork scrap is real fabric, while what outpainting builds outside is an invented scene from start to finish — run the same photo through it twice and the newly opened space can come back looking different both times.

4Try it yourself

5Common misconceptions

  • It's easy to think the real cropped-off scene comes back, but actually a newly invented scene appears there, with no regard for whatever was actually there.

  • It's easy to think widening a lot in one pass is fine, but actually the farther it drifts from the original, the more the flow breaks and it starts to look like a different photo stitched on.

  • It's easy to think a bigger frame means better image quality, but actually the detail inside the original stays exactly as it was — only the new space around it is freshly built.

7One-line summary

In shortOutpainting opens new space around a picture and draws it to follow the flow of the original edge, and widening it a little at a time reads as one seamless frame.

Spotted an error or have a better analogy? Suggest an edit · Last updated2026-09-02