Inpainting and outpainting
Inpainting is filling a masked region of an image with new generated content that fits the rest of the picture. Outpainting is the same idea applied outside the edges, extending the image beyond its frame.
Also called: inpaint, outpaint, outpainting, generative fill, generative expand
By Divya Sharma · Checked Nov 15, 2026 · How we review
You paint over an area, such as a person in the background or an empty sky, and optionally describe what should go there. The model regenerates only the masked area so it blends with the unmasked pixels around it. Photo editors sell this as generative fill.
How it works
With a diffusion model, the masked area starts as noise while the rest of the image is held fixed. The RePaint paper showed this can work even with a model never trained for inpainting: at each denoising step, the known pixels are taken from the original image and only the masked pixels come from the model. Dedicated inpainting models are also trained with the mask as an extra input, which usually gives cleaner edges.
Outpainting
Outpainting treats the new canvas area beyond the border as the mask. It is used to change aspect ratio, for example turning a square image into a wide banner, or to show more of a scene. The model has to invent everything in the new area, so the results depend heavily on the prompt.
Related terms
Inpainting is a targeted form of image-to-image editing. Tools such as ComfyUI and InvokeAI support both, and Adobe Firefly offers it in a hosted editor.
Limits
Seams, lighting mismatches and repeated textures show up on large masks. Removing or adding people convincingly is also an easy path to a misleading image.
Related terms
- Image-to-image
- Image-to-image is the task of generating a new image from an existing one plus a prompt, keeping some of the original's structure while changing style, detail or content.
- Diffusion model
- A diffusion model is a generative model that learns to turn random noise into data, such as an image, by removing a little noise at a time over many steps.
- Text-to-image
- Text-to-image is the task of generating a new picture from a written description, called a prompt. Most current text-to-image models are diffusion models conditioned on the text.
- ControlNet
- ControlNet is a neural network add-on that lets a text-to-image diffusion model follow a structural guide, such as an edge map, depth map or human pose, in addition to the text prompt.
- AI image upscaling
- AI image upscaling, or super-resolution, uses a neural network to enlarge an image and fill in plausible detail, so the result looks sharper than simple resizing would make it.