← AI Terminology
Inpainting
Inpainting fills in missing or masked regions of an image (or other signal) with plausible content, optionally guided by text or surrounding context.
A core generative editing operation.
A core generative editing operation.
Why It Matters in AI
Editing beats regenerating from scratch. Inpainting powers object removal, photo repair, and iterative design in diffusion tools — a primary creative and industrial vision feature.
Key Points
| Aspect | Description |
|---|---|
| Use | Remove objects, fix defects, extend content carefully |
| Input | Image + mask (+ optional text) |
| Models | Diffusion inpainters, older CNN/GAN inpainters |
| Product | Photoshop generative fill-class features |
| Quality | Seams, identity drift, prompt adherence |
| Related | Outpainting, image-to-image, ControlNet |
Simple Analogy
Restoring a torn photograph by imagining what belonged in the missing corner, matching the surrounding paper’s style.
Common Usage Examples
- Stable Diffusion inpaint workflows
- Mask product blemishes in e-commerce
- Batch restore damaged scans
- Combine with ControlNet edges
Summary
In short: Inpainting regenerates masked image regions coherently — the essential generative edit for repair and object removal.