What is Inpainting?
Inpainting transformed AI image generation from a 'roll the dice' tool into a usable editing workflow. Instead of regenerating a whole image to fix one wrong hand, you mask the hand and let the model re-render just that region with context from the rest of the image.
Modern inpainting goes beyond simple fixes: remove unwanted objects, swap clothing, change backgrounds, fix facial expressions, or extend images beyond their original borders (a variant called outpainting).
How it works
Mask the region
Draw a mask over the area you want to change. Tools like brush, lasso, or automatic detection (for objects or faces) all work.
Conditional re-generation
The diffusion model generates new content for the masked region while conditioning on the surrounding unmasked pixels. This preserves lighting, color, and perspective continuity.
Common use cases
- Removing unwanted objects (photobombers, watermarks, distracting elements)
- Fixing AI generation failures (deformed hands, extra fingers, weird faces)
- Swapping outfits, accessories, or backgrounds in product photography
- Retouching old photos to remove scratches or damage
- Extending images beyond their borders (outpainting)
