AI photo editor for changes you can describe

Upload one or more images, describe the changes you want and generate an edited result with GPT Image 2.5 or Nano Banana. Review the subject and surrounding details before downloading. Uploads and generation run online; generating an edit requires an account with credits.

Browse 6 examples

AI-generated image ideas

See what your photos can become.

See more prompts

See the references, read the prompt, and try it with your own photos.

Reference images: 3

Three selfies. One great photo.

Bring three separate portraits together in one natural group photo.

Selfies & portraitsSocial media

Reference images: 3

Try the whole look

Combine a portrait, jacket, and bag into a new outfit.

Selfies & portraitsSocial mediaSmall business

Reference images: 2

From desk photo to campaign

Give your product a studio scene inspired by a second image.

Small businessWork

Reference images: 3

A room with your pieces

Preview a chair and lamp together in your room.

Home & livingWork

Reference images: 1

Your pet, collectible edition

Reimagine your pet as a tiny collectible with its own packaging.

Social mediaHome & livingArt & stories

Reference images: 2

Sketch to real-world concept

Turn a sketch and material references into a realistic storefront.

WorkSmall businessHome & living

See more prompts

Choose a model before preparing the references

Model	Maximum inputs	Prompt limit	Output formats
GPT Image 2.5	16 images	20,000 characters	PNG
Nano Banana	10 images	5,000 characters	PNG or JPEG

GPT Image 2.5 offers 1K, 2K and 4K for most ratios; 27:16, 16:27, 9:8 and 8:9 are limited to 1K. Nano Banana editing offers 1K only. Both have square, landscape and portrait options. Check the ratio before generating—the initial form is portrait 9:16. A higher resolution does not guarantee preservation of fine lettering or product geometry.

Describe the change, then inspect the whole image

For a ceramic cup photographed on a busy kitchen counter, try: “Replace the countertop and background with a pale linen surface and a softly lit neutral wall. Retain the cup’s blue glaze, handle shape and viewing angle. Remove the utensils behind it.” This gives the requested change a clear scope without assuming the model can lock those details.

Upload the relevant image, wait for its ready status and review the prompt and output settings. After generation, open the result in history and compare the handle, rim, glaze and contact shadow with the source. If several details drift, simplify the next request rather than adding more unrelated references.

Use a conventional image editor when you need an exact crop, pixel-level repair or editable text layers. Browse DojoClip applications for other generation workflows.

Start with one clear edit

A different background

Describe a new setting for a product or portrait, including the surface and light. Say which subject and composition you want to retain, then check whether the result follows those instructions.

Remove a visual distraction

Name the unwanted item and where it appears. Describe the surface or scene that should replace it, then inspect the surrounding edges and details.

Change color, light or style

Request a warmer scene, another clothing color or an illustrated treatment. Be specific about the intended change and review faces, textures and text afterward.

Use several visual references

Bring images into one request and explain what each contributes, such as the subject, setting or visual treatment. The tool does not assign these roles automatically.

Choose the next image workflow

Create a professional headshot

Use the dedicated portrait workflow when your aim is a profile or team photo.

Open

Create a product poster

Start a separate poster workflow with a product image and campaign message.

Open

Generate a new image from text

Use the text-to-image generator when you do not have an existing image to edit.

Open

Create a video from your result

Use a finished image as the starting frame in the separate video generator, then review its motion and settings.

Open

Prepare the inputs and compare the output

JPG, JPEG, PNG or WebP inputs, up to 10 MiB per file.

GPT Image 2.5 and Nano Banana, with different image-count and prompt limits.

Multiple input images support one editing request; they are not processed as separate batch edits.

Choose the available output aspect ratio, resolution and file format for the selected model.

Examples can prefill instructions or demo references for review before generation.

History shows source images and available results, with full-size previews, downloads and status refresh.

AI photo editing questions

When are my own photos uploaded?

If you are signed in, selecting your files starts their upload to your account. If you are signed out, selection opens sign-in and keeps the pending files locally; upload can continue after sign-in. Wait for uploads to finish before generating. This is not a browser-only editor.

Can I paint a mask or lock a face or logo?

This form uses images and written instructions; it has no brush, mask or pixel-lock control. Asking to preserve a face, logo or product is guidance, not a guarantee. Compare the whole result with the source, including areas you did not ask to change.

How should I use multiple images?

All selected images are sent together in their selection order. Explain each image’s purpose in the prompt; there is no primary/reference selector or drag-to-reorder control. Remove unwanted inputs before generating. More images do not mean more separate edited outputs.

What happens if I switch models?

The output menus update, but the existing prompt and inputs remain. GPT Image 2.5 accepts up to 16 images and 20,000 prompt characters; Nano Banana accepts 10 images and 5,000 characters. Remove excess inputs or shorten the prompt if needed, then check the output settings.

Does choosing an example upload files or generate an edit?

Using only its prompt keeps your current images. Loading the demo can replace them after confirmation when needed, but demo references remain local until you choose Generate. Neither action starts paid generation by itself. This differs from selecting your own files while signed in.

How do credits and further edits work?

Each generation uses account credits. Changing the prompt does not modify an existing result, and another generation is a new request. Refresh checks existing tasks. To edit a downloaded result again, select it as an input; the page does not automatically use the latest output as your next source.