Back to home

Edit Text in a Screenshot

Replace existing words in an authorized screenshot with AI. Describe the target label, review the prompt, and check nearby text, icons, borders, and spacing.

Last updated: 2026-10-31

Prepare your text edit

These fields build instructions; they do not detect text in your image. Upload your JPEG or PNG in the editor below (up to 10 MB). Use an image you own or have permission to edit.

Review and edit the prompt below, then choose Generate when ready. Applying a prompt does not generate an image or spend credits. Sign-in is required to generate; your applied prompt and selected image can be restored after sign-in in this tab. Images are not transferred when you follow a link to another editing page.

Credit cost: —Try it free — no card needed
0 / 2,000
…

Credit cost: —

Content and copyright rules

Do not submit NSFW, harmful, explicit, sexually suggestive, or rights-infringing prompts.

Change an existing label in a screenshot of your own product or an authorized demo UI. This is an AI image edit: you upload a picture and describe what to replace. It is not a live interface update or a text annotation placed over the screenshot. The model may redraw neighboring details, so inspect the full result before using it.

Start with a readable, authorized screenshot

Use the editor above to upload a JPEG or PNG up to 10 MB. Enter the original wording, the replacement, and an optional location, then apply the prompt. You can change the instructions in the editor before choosing Generate. The form does not scan the screenshot or detect text blocks.

Use a capture of your own product, a demo interface, or a screenshot you have permission to modify. Remove personal information before uploading. For an actual product change, update the interface and take a new screenshot whenever possible.

How to identify the right text

Quote the full label, including capitalization and punctuation. Describe its position in relation to a stable feature: “the primary button at the top right, beside the settings icon” is clearer than “the button.” If a phrase appears more than once, specify which instance to change and which to keep.

For example, you might prepare this instruction for your own demo app. This is a prompt example, not a demonstrated output:

In my uploaded screenshot, replace "Create project" with "New project". Target location: the top-right primary button above the project list. Match the surrounding typography as closely as possible. Preserve other text, icons, spacing, borders and layout. Do not add extra wording.

A short replacement that fits the existing button is easier to review than a sentence that requires a different layout. If the change needs a wider control or a new line, decide whether a fresh capture from the original UI would be more accurate.

Edit, compare, and download

  1. Keep the original. Use a sharp capture with legible text and retain an untouched copy for comparison.
  2. Prepare the replacement. Name one label and its location. Ask to preserve repeated labels elsewhere.
  3. Review the prompt and settings. Select an editing model and check the displayed credit estimate. Applying the prompt does not submit a generation. Sign in when prompted; the applied instructions and selected image are saved for return in the same browser tab.
  4. Generate deliberately. Choose Generate only after checking the wording. Each new generation uses the product’s current credit rules.
  5. Compare the output with the source. Read the replacement character by character. Check neighboring labels, icons, dividers, button corners, borders, alignment, and spacing. Inspect at full size and at the size you intend to share.
  6. Download the reviewed image. If the output changes an important detail, return to the original and refine the instruction rather than accepting the drift.

The editor can display your uploaded source alongside a generated result for review. An edit is not evidence that a real interface or record ever displayed the replacement text.

What can go wrong

Small UI labels, compressed captures, and dense tables can produce misspelled or blurred letters. A model can alter a nearby icon or border even when the instruction asks to keep it. There is no guarantee of pixel-level preservation. For precise product documentation, editing the UI source and capturing it again is the more reliable workflow.

Keep private conversations, payment confirmations, financial records, and other evidence out of illustrative edits that could mislead someone about what occurred. This workflow is intended for authorized product and demo imagery.

FAQ

Can I click a detected word to replace it?

This page builds a written instruction from the words you enter. It does not provide OCR text-block detection or automatic word selection. Use the location field to describe the target yourself.

Will all other pixels stay the same?

No. The AI can regenerate surrounding details. Compare adjacent text, icons, borders, and spacing with the untouched source. If every pixel must stay fixed, use the original interface or a controlled image editor.

What if the same label appears twice?

Name a location and a nearby visual landmark, and ask to keep the other instance. Review both occurrences afterward. If the model still changes the wrong label, start again from the source with a narrower instruction.

Does the replacement change my app?

No. The result is an edited image. It does not update code, interface strings, or application data.

Can I preserve the original font appearance?

You can request a close visual match, but font identity and exact letter shapes are not guaranteed. Use the same-font appearance workflow for a typography checklist, or the general image text editor for other image types.

Return to the screenshot editor to prepare one focused replacement.