Inpainting
What is Inpainting?
Inpainting lets you paint over a specific area of an image and have AI fill that area with new content that matches the rest of the image, making it easy to remove objects, fix details, or replace elements without touching the rest of the composition.
At a glance
- Also known as
- Generative fillAI-powered content-aware fillMasked generation
- Used for
- Removing unwanted objects from imagesReplacing specific elements within a compositionFixing generation artefacts such as distorted hands or facesAdding new subjects to existing scenes
- Common tools
- Adobe photoshop generative fillStable diffusion with inpainting modelsRunwayDALL-eMidjourney
- Related terms
- OutpaintingImage-to-imageMaskingGenerative fillArtefacts (visual)
Ready to create?
Direct scenes, design characters, and ship full films
All-in-one AI creative platform with simple, transparent pricing, no speed throttles, and an infinite Canvas for max creativity.
How it compares
Compared with related concepts
Inpainting is the complement to outpainting. Inpainting works within the bounds of the existing image, replacing or modifying content inside the original frame. Outpainting extends the image outward, generating new content beyond the original borders. Both techniques use the same AI understanding of context and style, but serve different editing purposes: inpainting for refinement and correction within, outpainting for expansion and reframing without.
Think of it like…
Inpainting is like digital correction fluid: you paint over the part you want to change, and the AI fills in a fresh version that matches everything around it, leaving the rest of the image completely untouched.
Pro tip
When using inpainting to fix small artefacts or details, mask generously around the problem area rather than selecting a tight region: giving the model more context about the surrounding image typically produces better blending and more convincing reconstructions.
Types and variations
- Inpainting can be performed with or without text prompts.
- Prompt-guided inpainting uses a text description to specify what should appear in the masked region, giving the creator direct control over the replacement content.
- Non-prompt inpainting asks the model to intelligently reconstruct the masked area based on surrounding context alone, useful for object removal and background reconstruction.
- Some workflows use inpainting at multiple rounds, iteratively refining regions until the overall image meets quality requirements.
Ready to make your first scene in Morphic?
Try MorphicCommon use cases
- Inpainting is used to remove unwanted objects or people from images, correct generation artefacts such as distorted hands or faces, replace background elements, add new subjects to existing compositions, adjust specific details without regenerating the whole image, adapt compositions for different aspect ratios, and fix technically imperfect areas in otherwise successful generations.
- It is particularly valuable in AI workflows where a single generation may be mostly successful but contain one or two areas requiring targeted correction.
Ready to create?
Direct scenes, design characters, and ship full films
All-in-one AI creative platform with simple, transparent pricing, no speed throttles, and an infinite Canvas for max creativity.
FAQs
Inpainting is a technique where a selected region of an image is masked and then filled with AI-generated content that blends naturally with the surrounding area. It enables targeted edits to specific parts of an image without regenerating the entire composition, making it ideal for corrections, object removal, and element replacement.
The user defines a mask over the area to be changed, which tells the AI model which pixels to replace. The model then analyses the surrounding image context ( colours, textures, lighting, and compositional logic ) and generates new content for the masked region that matches the style and context of the rest of the image, guided by any text prompt provided.
Not always. Inpainting can work without a text prompt, in which case the model attempts to reconstruct the masked area intelligently based on surrounding context: useful for removing objects and having the model fill in the background behind them. With a text prompt, you can direct what appears in the masked region, replacing one element with something specific.
Common uses include removing unwanted objects or people from images, correcting generation artefacts such as distorted hands or anatomical errors, replacing specific compositional elements, changing backgrounds in specific regions, and iteratively refining AI-generated images that are mostly successful but have one or two areas requiring targeted improvement.
Inpainting fills or replaces regions within the existing image boundaries, while outpainting extends the image beyond its original edges by generating new content that continues the existing scene. Both use AI's understanding of image context, but inpainting is used for internal correction and modification while outpainting is used for expansion and reframing.
Inpainting is available in numerous AI tools including Adobe Photoshop's Generative Fill feature, Stable Diffusion interfaces with inpainting model variants, Runway, DALL-E's editing mode, and many dedicated AI image editing platforms. The technique is widely supported across professional and consumer-facing AI tools.
Masking generously around the problem area rather than selecting a very tight region helps the model understand the surrounding context and produce better-integrated results. Clear, descriptive prompts specifying exactly what should appear in the masked area also improve outcomes. Running multiple inpainting attempts and selecting the best result is common practice when precision is important.
Yes. Inpainting works on any image, whether AI-generated or photographed. It is widely used in photography retouching, background removal and replacement, and commercial photo editing workflows. The AI models used for inpainting are typically trained on diverse image data and can adapt to the style and content of both photographic and generated source material.
