How to extend an image with AI

Cropped too tight, wrong aspect ratio, or just need more room? AI outpainting extends your photo beyond its original borders — generating new content that matches the existing scene seamlessly.

Last updated May 17, 2026

What AI outpainting does — Magic Eraser

AI outpainting (also called generative expand or uncrop) adds new content beyond the edges of an existing image. Unlike simple canvas extension (which adds blank space), AI outpainting generates contextually appropriate content — if the edge of your photo shows sky, the AI extends the sky with matching clouds and color. If the edge shows a room, the AI generates plausible room continuation. This is useful when a photo was cropped too tightly, when you need to change aspect ratios (portrait to landscape or vice versa), when a social media platform requires dimensions your original doesn't match, or when you simply want more breathing room around the subject. The AI analyzes the content at each edge and generates a natural extension that looks like the photo was originally taken with a wider field of view.

단계별 안내

1. 1

   Upload the image

   Open Magic Eraser on web, iOS, or Android. Upload the photo you want to extend. Any image type works — portraits, landscapes, product photos, social media content. JPEG, PNG, HEIC, and WebP supported.

2. 2

   Choose the expansion direction and size

   Select which edges to extend: top, bottom, left, right, or all sides. Set the expansion amount — how much new content to add. For aspect ratio conversion (portrait to landscape), extend the sides. For adding headroom above a tightly cropped portrait, extend the top.

3. 3

   Generate and review

   The AI generates new content at each extended edge, matching the existing scene's lighting, color palette, perspective, and content. Check the transition where original content meets generated content — it should be seamless. Review the generated content for accuracy and coherence. Export at the new, larger resolution.

추천 대상

• Converting portrait photos to landscape (or vice versa) for different platform requirements
• Adding breathing room around a tightly cropped subject for better composition
• Fitting photos to specific social media dimensions without cropping (Instagram square, Facebook cover, Twitter header)
• Extending backgrounds for banner or hero images that need more horizontal space
• Creating panoramic effects from standard photos by extending both sides

최상의 결과를 위한 팁

The AI produces the best extensions when the edge content is predictable — sky, grass, water, walls, and simple surfaces extend naturally because the pattern is consistent and the AI can generate matching content. Complex scenes at the edge (crowds, detailed architecture, text) are harder because the AI must invent plausible content. For best results on portraits, extend the background rather than the subject — extend the sky above or the room around the person, keeping the subject fully within the original image area. Small extensions (10-25% of original size) produce more convincing results than large ones (50%+) because the AI has more reference relative to what it needs to generate. For very large extensions, do multiple smaller passes.

자주 묻는 질문

Is outpainting the same as uncropping?

Yes. Outpainting, uncropping, and generative expand all refer to the same AI capability: adding new content beyond the original image borders. Magic Eraser calls it image extension.

Can I extend in just one direction?

Yes. Extend any combination of edges — just the top, just the sides, or all four edges. Choose based on what you need: more sky (top), wider background (sides), or more foreground (bottom).

Does the extended area match the original?

Yes. The AI generates content that matches the original image's lighting, color temperature, perspective, and content style. The transition between original and generated areas is seamless when the edge content is consistent.

Is image extension free?

Yes. AI image extension is available on the free tier with daily usage limits. Premium removes limits and allows larger extensions.