Skip to content

Learn

What Is Outpainting? How It Works

Outpainting is an AI image-editing technique that extends a picture beyond its original borders. The model enlarges the canvas, treats the empty margin as a mask and generates new pixels that continue the scene's content, lighting and perspective, so a tightly cropped photo becomes wider, taller or fully reframed.

Start on the free Starter plan.

Outpainting - FunAI
Outpainting - FunAI

Cropping in Reverse: Grow the Frame

Outpainting keeps every original pixel and generates only the new canvas around it, matching the light, perspective and texture at the edge so the extension reads as part of the same photograph.

Open the AI Image Extender

A mask outside the frame

The new margin is the masked area. The model fills it and leaves your original image untouched.

Context from the edges

Each denoising step looks at the visible border, so lines, horizons and textures continue instead of restarting.

A prompt for the new space

When the edge is ambiguous, a few words tell the model what belongs there: more sky, a longer table, an empty wall for text.

Outpainting models on FunAI

Each of these models extends an image in a different way. Open a model page to see its inputs and its price per run.

  • •
    Image Outpaint:Directional outpainting: choose which edges to expand, with an optional zoom-out that pulls the frame back. Only the expanded areas are generated.
  • •
    FLUX.2 Pro Outpaint:Adds new scenery around photos and artwork, with per-edge expansion settings.
  • •
    Bria Expand:Extends a photo to a target aspect ratio or canvas size, with an optional prompt and negative prompt.
  • •
    Luma Photon Reframe:Reframes and expands an image to a new aspect ratio, with control over where the original sits in the new frame.
  • •
    Ideogram Reframe:Ideogram V3's reframe feature, which expands an image while preserving the main content.
Open the AI Image Expander
Outpainting Models

Outpainting explained

Outpainting, also called image extension, image expansion or 'uncropping', is cropping in reverse. Instead of cutting pixels away, you add empty canvas to one or more sides and let a generative model fill it. It is the sibling of inpainting: inpainting regenerates a masked area inside the picture, while outpainting regenerates a masked area outside the original frame. The original pixels stay where they were; only the new margin is invented.

Most outpainting runs on a diffusion model. The tool places your image on a larger canvas and marks the new margin as the region to generate. That region starts as random noise, and over a series of denoising steps the model predicts pixels that agree with the visible edge: a horizon continues at the same height, shadows fall the same way, a brick wall keeps its pattern. Many models do this in a compressed latent space rather than on raw pixels and decode the result at the end. A short text prompt can steer what appears in the new area, for example 'more sand and a calm sea'.

Use outpainting when the subject and composition are right but the frame is wrong: a square product shot that has to become a wide banner, a portrait photo for a landscape slide, a vertical version of a horizontal image, extra headroom for a title, or more breathing space around a subject that was cropped too tightly. It is the wrong tool for placing a specific new object at a precise spot inside the picture; that is an inpainting job.

Tips for clean outpainting

  • Extend in moderate steps: adding a quarter to a half of the width usually blends better than tripling the frame in one pass.
  • Keep the subject away from the edge you extend, so the model continues background instead of inventing half a face or a hand.
  • Describe the new area in the prompt when the border is ambiguous, such as a cut-off sign, a doorway or a partly visible object.
  • Watch the seam for duplicated objects and repeating textures, and regenerate or switch models when they appear.
  • Use a reframe model when the subject's position, pose and perspective must stay exactly as they are.
  • Upscale after outpainting, not before, if the extended image is going to print or a large screen.

How to Outpaint an Image on FunAI

Feature visual preview 1
Feature visual preview 2
Feature visual preview 3
Feature visual preview 4
01

Open an extend or reframe Tool

Start in the AI Image Extender or AI Image Expander Tool. If your real goal is a new aspect ratio with the subject kept in place, the Image Reframe Tool is the closer fit.

02

Upload and set the new frame

Upload the image and choose how far to extend it. Image Outpaint expands chosen edges or all sides; Bria Expand takes a target aspect ratio or canvas size.

03

Describe the new area (optional)

If the edge does not make the continuation obvious, add a short prompt such as 'continue the wooden table and soft window light'. Leave it empty to let the model infer.

04

Compare and download

Check the seam and look for repeated patterns. If the extension drifts, run a second outpainting model on the same image and keep the version that reads as one photograph.

No public creations to show yet.

Everything you need to know