Skip to main content
Image to Image

OpenAI Image Editor

Edit source images from instructions using OpenAI GPT Image 2, curated and custom canvas sizes, and safe source-image preprocessing metadata.

View details
Try it in Ampere

Inputs

Loading input fields...
Execution Steps

Loading workflow structure...

Loading curated examples...

Overview

OpenAI Image Editor edits one or more source images from a natural-language instruction using OpenAI GPT Image 2. Use it when you already have product shots, brand visuals, mockups, or reference images and need controlled changes, masked edits, or visual variants.

Use cases

  • Change a product photo background while preserving the product.
  • Retouch a specific area with a mask instead of regenerating the whole image.
  • Create alternate social or ad visuals from existing campaign assets.
  • Use multiple reference images to guide a new edited composition.

Input tips

  • Provide HTTPS image URLs that are publicly reachable without login.
  • Describe exactly what should change and what should stay the same.
  • Use mask_image_url only when editing one source image and a specific area should change.
  • Choose size, quality, background, and output format when final handoff details matter.
  • Use exact GPT Image 2 sizes such as 1024x1024, 1536x1920, 2560x1440, or another valid WIDTHxHEIGHT value.
  • Use JPEG or WebP compression only when file size matters; PNG ignores compression.
  • Keep source images practical; large files may be resized, converted, or compressed before editing.

Expected output

The AI Tool returns edited images with downloadable URLs and optional content type, file name, file size, width, and height. It also returns model, source image count, mask status, source preprocessing metadata, background handling, output format, quality, size, token usage when available, and cost metadata. The output view renders a single image or gallery plus edit settings and privacy-safe source image details.

Caveats

  • Source images must be reachable HTTPS PNG, JPEG, or WebP files.
  • Masks only work with exactly one source image and must include an alpha channel.
  • Large source images or masks may be resized, converted, or compressed before provider handoff.
  • GPT Image 2 currently supports auto or opaque backgrounds; transparent backgrounds and input_fidelity controls are not supported.
  • Image edits can alter details unexpectedly; review output before publishing or using brand assets.
  • Generated text, logos, product details, and rights-sensitive content still need human review.

Related AI Tools