AI
Remove text from image
Words gone, picture kept.
Find the text in a photo automatically, then erase it and let the AI fill in what was behind.
Finds text in any language or script, including stamps and signs, and covers it. Nothing needs to be read, only located.
Finds where the text is, in any language.
Downloads once on first use, then runs offline. Your images are never uploaded.
Brush size
Cover the whole object, plus a little of what surrounds it. The model fills the painted area from the pixels around it.
Fills the brushed area from the surrounding pixels.
Downloads once on first use, then runs offline. Your images are never uploaded.
How did this tool do?
Good to know
Finding text in a photo, and filling in behind it
Removing text takes two steps: finding exactly where the letters are, then painting in what should be behind them. Find text uses a small detection model trained on text in real scenes, not on pages, so it spots captions on a wall or a date in a corner. It only has to locate the text, not read it, so it works for any language or script, and it covers each line with a little extra margin, since leftover edges of letters are what give a removal away.
The covered area is only a starting point. Detection is best on clear lettering: captions, date stamps from old cameras, labels and signs. Very small, faint or heavily stylised text can be missed, so brush over anything left, and Undo takes back a whole Find text step if it covered something it shouldn't.
The fill comes from an inpainting model that looks at the pixels around each covered area and continues them inward. It works best where the background is plain or evenly textured, such as sky, walls, paper or fabric. Text over a face, a pattern or a busy scene is much harder to fill convincingly, and running Remove a second time on the result often cleans up a faint trace.
Only remove text from images you own or have permission to edit: taking a watermark off someone else's photo doesn't make it yours to use. Both models download once and then run on your device, so the photo never leaves it, and the result is saved as a PNG at the full size of the original.
Where you'll meet it
Date stamps
Clear the orange date an old camera printed in the corner.
Captions and labels
Take text off a screenshot, meme template or product photo.
Signs in the background
Remove a distracting sign or number from your own photo.
Guide
How to use it
- 01Drop a photo, or paste one with Ctrl+V.
- 02Tap Find text. The words it finds are painted over; brush over anything it missed, or undo what it shouldn't touch.
- 03Tap Remove, then download. Run it again on the result if a faint trace is left.
Good to know
Questions, answered
Why didn't it find all the text?
Detection works best on clear lettering. Very small, faint or heavily stylised text can be missed, so brush over anything left, which works the same way.
Does it work for other languages?
Yes. It only has to find where text is, not read it, so Hindi, Arabic, Chinese and any other script are found the same way as English.
What does it work well on?
Captions, date stamps, labels and signs on plain or softly textured backgrounds. Text over a busy scene or a face is harder to fill convincingly.
Can I remove a watermark?
Technically it often can, but only do that on images you own or have permission to edit.
Is my image sent to a server?
No. The AI model itself is downloaded to your browser the first time you use the tool, then everything runs on your device — the image never leaves it. After that first download the tool works offline.