Expand image with AI – extend edges, change format (outpainting)
Chat shortcut: /expand
Expand image instead of cropping: the AI extends the edge in the chosen aspect ratio and is meant to leave the subject unchanged. For photos in the wrong format for a story, banner, profile picture or print.
With Account
Cost: 9,000 tokens per run · Input and result are automatically deleted 24 hours after upload.
This tool can be used with a free account. Sign up for free
More image without reshooting
Outpainting extends an image beyond its edges. The AI reads what is visible at the edge and paints the scene outwards: the sky gets bigger, the table continues, the wall behind the person opens up in both directions. Light, perspective and grain are meant to be taken from the original; check the transition, touch-ups may be needed.
You need it when a good photo is in the wrong format: shot in portrait but needed as a banner or thumbnail; cropped square but wanted as a story; framed too tight, so the head or product is jammed against the edge. Instead of cropping, you choose the target format and get more image.
With the optional description you steer what appears at the edge, for example “calm meadow” or “more water in the foreground”. If you leave the field empty, the AI continues the existing scene.
How it works
- 1
Upload image
Choose a JPG, PNG or WebP up to 20 MB or drag it into the field.
- 2
Choose target format
16:9 for banners, 9:16 for stories, 4:5 for Instagram, 1:1 for profile pictures or one of the other aspect ratios.
- 3
Describe the edge (optional)
Briefly state what should appear in the new areas, for example “more sky” or “continue the wooden table”; up to 300 characters.
- 4
Check and save
Start the run, check the transitions at the edge and the subject, download the image or continue with “Image upscaler”.
Typical uses
Portrait to landscape
Use a phone photo as a thumbnail, website header or presentation slide.
Stories and reels
Turn landscape shots into 9:16 without cropping the subject.
Product image with room
More space around the product for text, price or uniform catalog tiles.
Shot too tight
Fill in cut-off heads, hands or edges with matching surroundings; the AI invents what's missing, it doesn't restore the original.
Good to know
- Input: JPG, PNG or WebP up to 20 MB, one image per run.
- Eight target formats: 16:9, 9:16, 1:1, 4:3, 3:4, 4:5, 3:2 and 2:3; the missing edges are filled to match the format.
- The image is regenerated on the target canvas; subject, light, texture and perspective are meant to be preserved, a pixel-exact carry-over of the original is not guaranteed.
- Description optional, in your language, up to 300 characters; leaving it empty means the scene is continued. Direction and subject position can't be chosen directly.
- Result in the canvas size of the format (around 1 to 1.6 megapixels), file format as delivered by the provider; for larger output, use “Image upscaler” afterwards.
- Cost: fixed token amount per run, shown before you start; account required, no subscription needed. The image goes to the AI provider (its terms apply); ChatX deletes input and result after 24 hours (hourly cleanup job).
Frequently asked questions
How does the AI decide what to add?
It continues the background, light, texture and perspective of your image. Without a description the scene is simply continued; with “more sky” or “empty studio background” you steer the result.
Is my image stretched or distorted?
Nothing is stretched: the AI regenerates the image on the larger canvas and is meant to keep the existing content; small deviations are possible. That's what distinguishes outpainting from scaling or stretching.
Which formats can I choose?
16:9 and 9:16 for landscape and story, 1:1 for square, 4:5 for Instagram, plus 4:3, 3:4, 3:2 and 2:3. You don't choose a direction directly, only via the description.
Can I open up a photo that was cropped too tight?
The AI generatively fills in what's missing: a cut-off head, too little room beside the product, a missing foreground; nothing from the original is recovered; the missing part is invented to match. Choose a format with more room in the right direction and describe what should appear there if needed.
What size is the result?
The fixed canvas of the format, for example 1365 × 768 pixels for 16:9 or 1536 × 1024 for 3:2. For print or large banners, use “Image upscaler” afterwards.
The result doesn't fit, what now?
Every run turns out a little different. Change the description or the target format and start again; each run costs tokens again.
Do I need an account?
Yes, this tool can only be used with an account. Signing up is free; each run is billed from your credit.