Remove text from an image
Paint over a caption, a camera date stamp, a label or handwriting in your own image, and AI fills the area with the background around it. Works best on plain or natural backgrounds. Your image never leaves your device.
How it works
- Choose or drop a photo you took or have the rights to edit.
- Paint over what should go with the brush. Paint a little past its edges; undo or clear if you slip.
- Press Erase. Repeat for anything else, then download the photo.
Questions
Which photos may I edit?
Only photos you own or have permission to edit. Do not use this tool to remove a photographer’s or owner’s name, logo or other identifying marks from images that belong to someone else.
What works best?
Small to medium objects against a natural background: a passer-by, a bin, a cable, a sign, a date stamp. The AI fills the painted area with what plausibly surrounds it; it does not know what was really behind the object, so very large areas or busy patterns can look smudged. Erase in a few smaller passes for the best result.
How large a photo can I edit?
Up to 17 MP on a computer and 13 MP on a phone, which covers photos from most phone cameras. The model fills the area around your brush strokes at up to 512 × 512 pixels and blends it back into the full-size photo; the rest of the photo is not touched.
How long does it take?
It depends on your device. Browsers with WebGPU (recent Chrome, Edge and other Chromium browsers on most computers) use the graphics card and are much faster; others run on the processor. A progress bar counts the image tiles as they finish, and you can cancel at any time.
Is my image uploaded?
No. Your image never leaves your device. The first time, your browser downloads the fill-in model (23 MB) and the AI engine (about 3.8 MB) from this site and keeps them for next time. After that the work happens on your computer or phone; there is no upload, no server-side copy, no account and no tracking script on this site.
Is it free?
Yes. No sign-up, no daily quota, and nothing is added to your result. Because your own device does the work, it costs us nothing per image.