1. Upload the photo
Open the generator, manually select GPT image 2 and upload the largest copy of the photo you have as a JPEG, PNG or WebP file up to 30 MB. Use a photo you have the rights to edit and keep the original file unchanged.
An AI image extender, also called uncrop or expand image AI, redraws a photo you upload at a wider or taller aspect ratio and fills the new edges with matching scenery. In the existing image tool, select GPT image 2, upload the photo, choose 1:1, 9:16, 16:9, 4:3 or 3:4 and describe what belongs in the added space; an edit with your upload costs 15 credits. It suits turning a landscape shot into a vertical story, giving a portrait room for a wide banner and adding back an edge that was cropped too tight. The added area is an AI redraw: edges and details may not match the original exactly, and the original part can be lightly redrawn as well.
Upload only photos you have the rights to edit, with consent from anyone pictured. This independent website is not affiliated with or endorsed by model vendors. Extended areas are invented by the model; do not present them as part of the real scene, for example in news or as evidence.
Choose an aspect ratio that differs from your photo, then say what should continue into the new space and what must stay exactly as it is.
Open the generator, manually select GPT image 2 and upload the largest copy of the photo you have as a JPEG, PNG or WebP file up to 30 MB. Use a photo you have the rights to edit and keep the original file unchanged.
Choose the aspect ratio in the generator settings: 1:1, 9:16, 16:9, 4:3 or 3:4. A vertical story needs 9:16, a wide banner or video thumbnail 16:9 and a square post 1:1. Then write one instruction such as: Extend this photo to fill the new frame, continue the sky above and the beach on both sides, and keep the people, their position and the colors unchanged.
Sign in, check the displayed quote of 15 credits and generate. Look at the line where your original photo ends and compare faces, text and edges with the original. A retry is another request.
These are untested prompt suggestions, not measured results. Change them to match your own photo and check every result against the original.
Extend this horizontal beach photo into a vertical frame. Add more sky above and more sand and water below, in the same light and colors. Keep the people, their size and position, and the horizon line unchanged.
Extend this tightly cropped portrait to a wide frame. Continue the plain studio wall on both sides with the same color and lighting, and leave empty space on the left for a heading. Do not change the face, hair, clothing or expression.
Extend this photo of my own handmade candle into a square frame. Continue the wooden table and the soft background on all sides. Keep the candle, its label area and the shadows exactly as they are, and do not add text or a logo.
Extend this photo downward and upward into a taller frame so the top of the building and the street below are visible. Match the brick, windows and pavement to the existing photo and keep everything already in the frame unchanged.
The model generates a new image at the ratio you chose. It does not know what was outside the frame, so check what it filled in.
Brick, tiles, crowds and text are the hardest to continue. Look for a visible seam, duplicated objects or warped lines in the added area and generate again with a more specific description if needed.
GPT image 2 returns a whole new image, not your untouched pixels plus a border. Faces, small text and fine patterns in the original area can shift slightly, so keep your original file and compare the two.
The tool offers 1:1, 9:16, 16:9, 4:3 or 3:4. It has no custom ratio, no pixel size setting and no larger output option for GPT image 2, so the extended photo is not a higher-resolution file.
Extending keeps your scene and adds more of it. If you want a different location behind the subject, use the background changer instead.
Replace the scene behind a person or object with a new one.
Reduce blur and sharpen an out-of-focus or grainy photo.
Erase stray people or clutter and fill in the background.
Learn how to guide any edit with an uploaded photo.
Review GPT image 2, the model used for these edit quotes.
Check credit options before submitting a photo edit.
Read the terms for uploaded and generated images.
Yes. Upload the photo with GPT image 2, choose a wider or taller aspect ratio and describe what should fill the new space. The model redraws the photo at that ratio; preview the result before you use it.
Yes. Uncrop, expand image and outpainting all describe adding new picture area around a photo. On this site you do it by choosing a new aspect ratio for an upload edit.
No. GPT image 2 offers 1:1, 9:16, 16:9, 4:3 or 3:4 in this tool, with no custom ratio or pixel size setting.
No. The model returns a whole new image, so the original area can be lightly redrawn. Keep your original file and compare faces, text and edges.
No. GPT image 2 has no output size setting in this tool, so extending changes the frame, not the resolution.
No. Extending continues the scene you already have into a wider or taller frame. The background changer replaces the scene behind the subject with a different one.
No. Each edit with GPT image 2 costs 15 credits. Sign in and check your balance and the displayed quote before submitting; a retry is another request.
No. The button opens the existing image generator and does not prefill a prompt, select a model, upload a photo or submit a job. Set up the edit yourself and check the quote before generating.