The short answer: upload images, pick a target ratio (12 common ones), and AI places the full original in the center of the new canvas, extending the background seamlessly; the subject is never cropped, people are never stretched, and texture, perspective and lighting continue outward. Landscape to square, square to banner, vertical to cover; up to 50 images share one ratio.
What image extension does
| It does | It doesn't |
|---|---|
| Place the full original centered in the target canvas | Crop the subject to fit a ratio |
| Extend outward with continuous texture, perspective and light | Stretch or distort people |
| Compute output from the ratio (short side ≥1024) | Apply a new style (use an editor for that) |
| Share one ratio across a 50-image batch | Set different ratios per image |
The direction is decided automatically by "target ratio + original aspect": a landscape extended to 9:16 grows up and down; a vertical extended to 16:9 grows sideways; you never compute the canvas yourself.
Choosing among the 12 ratios
| Ratio | Typical use |
|---|---|
| 1:1 | Social feeds, square e-commerce (default) |
| 16:9 | Site banners, newsletter headers, video covers |
| 9:16 | Short-video covers, mobile posters |
| 4:3 / 5:4 | Lookbooks, print layouts |
| 3:2 / 2:3 | Classic photo ratios |
| 2:1 / 21:9 | Ultra-wide banners and headers |
| 3:4 / 4:5 / 1:2 / 2:3 | Vertical social and posters |
One rule: pick the ratio of the place where the image will live: the original always stays centered; only the background extends.
How to extend step by step
- Upload images: one or many, up to 50; JPG / PNG / WebP, each up to 10 MB and 50 megapixels;
- Pick the target ratio: 12 options, 1:1 by default; the whole batch shares it;
- Optional: a note: describe the new areas, e.g. "extend into a soft city night, keep the reflections";
- Download: review each result, then download singles or the whole batch as a zip (files keep their names under the
eyesme-filledprefix); output is computed from the ratio with a minimum side of 1024.
Four scenes in practice
- Social square: a food shot extended to 1:1, original centered, sides filled naturally:
Landscape-to-square: the tabletop and background extend naturally to both sides
- Portrait to widescreen: a half-body portrait extended to 16:9, the field and distance continuing in perspective:
The person is never stretched; the wheat field and horizon extend in perspective
- Vertical cover: a full-body beach shot extended to 9:16, sky above and sand below:
The full body stays; sky and beach extend up and down for short-video covers
- Print ratio: an interior reading scene extended to 4:3 with the surroundings filled for layout flexibility:
Interior light and furniture continue, leaving more room for print crops
Complex backgrounds and edge subjects
- The more regular the texture and the simpler the background, the more natural the fill;
- Complex patterns, dense crowds or subjects close to the frame edge may show repeated texture or slight distortion; test one image first;
- Extension follows the original style; to grade the extended image afterwards, apply a film look with cinematic lighting.
vs manual cropping and general AI editors
| Manual crop / stretch | AI image extender | |
|---|---|---|
| Workflow | Resize canvas, crop, stretch or clone stamp | Pick a ratio and extend outward |
| No crop | Often crops the subject to fit | Full original kept centered |
| Seamless background | Visible patching or repeated texture | Follows the original texture, perspective and light |
| Batch | One image at a time | Up to 50 images with one ratio |
| General AI editors | AI image extender | |
|---|---|---|
| Getting started | Rewrite prompts and try ratios | Twelve ratios + custom, one click |
| Subject preserved | Can crop, stretch or reshape | Full subject kept, never cropped or stretched |
| Background continuity | Often needs multiple retries | Follows original texture, perspective and light |
| Batch consistency | Tune each image; canvas drifts | One target ratio for the whole batch |
In one line: manual extension wins on fine control, general AI on freedom, and the extender wins on "ready-made ratios, intact subjects, one canvas for the batch".
Frequently asked questions
How does extension work?
After you pick a ratio, AI places the full original inside the target canvas and fills in the new background. The original is never cropped or stretched.
Will it crop the subject?
No. The full original is kept in the center and new space is added around it.
Which ratios are supported?
1:1, 4:3, 3:4, 16:9, 9:16, 3:2, 2:3, 5:4, 4:5, plus 2:1, 1:2 and 21:9, twelve in total.
How does it handle complex backgrounds or subjects at the edge?
Regular textures and simple backgrounds fill most naturally; complex patterns, dense crowds or edge-hugging subjects may show repetition or slight distortion; test one image first.
Can I process multiple images?
Yes. Up to 50 images per batch, sharing the same target ratio.
What output size do I get?
Computed from the target ratio with a minimum side of 1024; the card shows the estimated size.
Does extension change the original style?
It follows the original texture, perspective and lighting rather than applying a new style; use another editing tool for a different look.
Which formats are supported?
JPG, PNG and WebP, up to 10 MB per image.
Features and details reflect the tool as of November 12, 2026 and may evolve; the before/after examples are from the tool's real demo assets, with the cover composited from the margined original and extended halves.
Ready to post in any ratio? Eyesme's AI Image Extender extends up to 50 images to one of twelve ratios with the subject intact; to swap the new background's scene entirely, pair it with AI background changing.




