Remove old words from images you want to use again.
Remove text from an image by selecting the words you want gone. Clear an old offer, date or caption, keep the surrounding picture, then compare and download a PNG.
Only upload images you own or have permission to edit. By uploading, you agree to our Terms and Privacy Policy.
Highlight every unwanted letter and its outline. Leave wanted text and picture details unselected.
Keep this page open.
Before saving, check that the selected words are gone, the background blends in, and wanted text is intact.
Each new repair uses 1 credit. Comparing and downloading are free.
Remove text from photos, screenshots and worksheets.
To remove text from an image, upload a JPG, PNG or WebP and highlight the words to erase. Brush over the unwanted words, including any outline or shadow, then select Remove selected text. ImgUnmark repairs that area and preserves the unselected pixels and dimensions of the image you edit. Compare the result before downloading PNG. Your first Google sign-in includes one credit; each new repair uses one credit. Review fine detail where the words used to be.
Select text to remove
Use this image text remover to select an old offer or review label while leaving the product name or report heading outside the selection. Brush, Rectangle and Lasso let you choose the repair area; Eraser helps keep wanted letters clear.
Remove words from a picture with text selection and keep the dimensions of the image you edit. A date stamp can go while the boat, horizon and full photo frame stay outside the selection, ready for your travel album.
After you remove captions from images, compare the filled background with the original. Look at color transitions, bubble outlines or small printed letters. Download a separate PNG when the result is suitable; comparing the current result uses no additional credits.
Use the image text remover to select unwanted words, inspect the repaired area and save a separate PNG.
Sign in with Google, then choose, drop or paste a JPG, PNG or WebP into the image text remover. Check that the caption or label is readable in the preview. Your first sign-in includes one free credit.
Choose the text type for guidance, then highlight the words with Brush, Rectangle or Lasso. Include every letter and outline; keep wanted text outside the highlight. Use Eraser or Undo to refine the selection, then click Remove selected text.
Compare the original and repaired picture after removing the text. Inspect where the old words crossed a background, a border or other lettering. Download your PNG when it is ready for reuse; a further repair uses another credit.
Select visible text baked into a still image. This is pixel repair, not editing a text layer or recovering the original covered detail.
| Feature | What ImgUnmark offers |
|---|---|
| Tool type | Dedicated selected-text removal. Plain backgrounds can use local repair; more complex selections use an explicit text-erasing AI instruction. |
| Input formats | One still JPG, JPEG, PNG or WebP at a time. Animated GIFs use the separate GIF tool. |
| File and dimensions | Up to 10 MiB; 4,096 px on the longest edge. |
| Select and protect | Brush, Rectangle or Lasso to select; Eraser, Undo and Redo to refine. Select less than 20% of the image. |
| Text selection | A dedicated text-removal workflow. Select words yourself with Brush, Rectangle or Lasso; this page does not automatically detect all text. |
| Text-removal output | A separate PNG at the pixel dimensions of the image you edit. Pixels outside the selection remain unchanged. Browser previews above 2,048 px are scaled for display. |
| Login and cost | Google sign-in is required. First sign-in includes one welcome credit; each new repair uses one credit. Comparing and downloading an existing result use none. |
| Image storage | AI-assisted repairs send image content to a processing service. Originals, selection masks and results remain in private storage without automatic expiry. See the Privacy Policy. Privacy Policy |
| Failure handling | Confirmed failed repairs return the credit. If task status is unclear, check the same repair before starting another. |
Choose the words that no longer belong in your image. These generated inputs show six text-removal tasks, with actual local repair results and specific details to keep.
The campaign is over, but the saved catalog image still says SALE 30% OFF. Select the offer with the image text remover while leaving the bottle and HAND WASH label unselected. This dedicated local repair clears the promotional words; inspect the filled backdrop before reusing the picture in an evergreen listing. Try this product offer.
Keep: Bottle and HAND WASH label. Use for: Evergreen product listing.
Generated demonstration input. Actual local text-removal repair output. Inspect the repaired area; these simple-background examples do not establish performance on complex textures.
A saved travel photo has a date over the sky. Use the image text remover to select the digits instead of cropping away the open space above the boat. Inspect the filled sky, then reuse the full-frame picture in an album or travel layout. Try this photo date.
Keep: Sailboat, coastline and horizon. Use for: Travel album or full-frame layout.
Generated demonstration input. Actual local text-removal repair output. Inspect the repaired area; these simple-background examples do not establish performance on complex textures.
The review is complete, but a flattened report image still says DRAFT COPY. Remove only that text from the screenshot and leave WEEKLY UPDATE, the review date and project rows outside your selection. Check the small lettering before using the image in a presentation. Try this draft label.
Keep: Report heading, date and project rows. Use for: Presentation-ready report image.
Generated demonstration input. Actual local text-removal repair output. Inspect the repaired area; these simple-background examples do not establish performance on complex textures.
A Monday joke dates an otherwise useful pet image. Erase the caption from the photo by selecting MONDAY AGAIN? and leaving the dog and its outline untouched. Review the mint backdrop before using the picture in a new post; add replacement wording in your design editor. Try this social caption.
Keep: Dog, fur and silhouette. Use for: New social post or meme template.
Generated demonstration input. Actual local text-removal repair output. Inspect the repaired area; these simple-background examples do not establish performance on complex textures.
A flattened comic panel needs an empty balloon for new lettering. Highlight WE MADE IT! inside the bubble, leaving its outline, tail and characters outside the selection. The dedicated local text repair clears the words on this white background. Inspect the fill before relettering the panel in a separate editor. Try this comic dialogue.
Keep: Speech balloon outline, tail and characters. Use for: Text-free panel for relettering.
Generated demonstration input. Actual local text-removal repair output. Inspect the repaired area; these simple-background examples do not establish performance on complex textures.
A worksheet image already contains the answer 12. Use the text selection tools to remove the handwritten text from the image while leaving 8 + 4 = and the printed instructions outside the selection. Check the equation before using the PNG in another practice activity; this is image cleanup, not PDF editing. Try this written answer.
Keep: Printed equations, title and instructions. Use for: Reusable practice activity.
Generated demonstration input. Actual local text-removal repair output. Inspect the repaired area; these simple-background examples do not establish performance on complex textures.
For occasional edits
20 credits · 20 edits
For regular creators
50 credits · 50 edits
Best value for studios
90 credits · 90 edits
Try text removal with 1 free credit on your first Google sign-in. Each new repair uses 1 credit. Prices are in USD. Credit-pack purchases are currently unavailable in this local preview; existing credits can still be used.
A careful selection protects the rest of the image. The area underneath the letters still needs to be reconstructed.
Text over a face, small print, a product edge or a repeating pattern may leave reconstructed detail that differs from the hidden scene. Inspect it before reuse.
Keep the selection close to the unwanted text. Large blocks need more background reconstruction; splitting work into repairs also uses additional credits.
Use images you own or have permission to edit. Removing a label does not provide a license or verify that an image or document is authentic.
What stays unchanged, how selection works, and what to check before saving.
Upload your image and use the text selection tools to mark the words you want gone. Include the letters’ border and shadow, then click Remove selected text. Compare the original and result before downloading PNG.
Yes. Select only the unwanted caption in this image text remover. Pixels outside the highlight stay unchanged in the image you edit. Use Eraser to keep wanted letters outside the selection. If unwanted and wanted text overlap, the hidden detail needs reconstruction and may not be accurate.
This tool uses your selection rather than automatically detecting every word. Choose a text type for helpful guidance, then highlight only the letters you want erased. Wanted words stay outside the selection. For general automatic watermark cleanup, use the homepage tool.
This text remover keeps unselected pixels unchanged. The background inside the selection must be filled in. Plain colors and smooth gradients can use local repair; textured areas may need AI reconstruction. Inspect the filled area because the original hidden pixels cannot be recovered exactly.
Yes. This text remover keeps the pixel dimensions of the image you edit and preserves pixels outside the selection. The browser preview may be smaller for display; the downloaded result is a separate PNG.
Your first Google sign-in includes one welcome credit. Each new repair uses one credit. Viewing, comparing and downloading an existing result use no additional credits. Confirmed failed repairs return the credit; check an uncertain task before submitting another.
This tool removes selected image pixels. It does not add replacement words, match fonts, extract text with OCR or edit original PDF and Word files. Export a page as an image for image-only cleanup; the output here is PNG.
You can select dialogue or handwriting as image pixels. The comic example clears words inside a white balloon while retaining the outline and characters. The worksheet example removes a written answer while retaining the printed exercise. Review your own result; this tool does not translate comics, create replacement lettering or edit a source document.
Compare the result at a readable zoom. Refine the selection around any remaining outline, or choose Use original image to return to the uploaded picture. Each new repair uses one credit, so inspect the current result before processing again.
AI-assisted repairs send image content to a processing service. ImgUnmark keeps originals, selection masks and results in private storage without automatic expiry. Read the Privacy Policy before uploading sensitive content.