Text Remover
Take text off an image cleanly. Pick a style, choose a model, and download the result at full resolution with no watermark.
Pick a style on the left, then hit Apply. Your original stays untouched — you always see both.
How to use Text Remover
Upload the photo
JPG, PNG or WEBP at the highest resolution you have. The fill is built from surrounding detail, so there needs to be surrounding detail to build from.
Say what to remove
Name the object and where it is: "the person on the right", "the watermark across the centre", "the bin in the bottom left". Location does most of the work.
Check the fill, then download
Look at the filled area at full size before using it. Full resolution out, no watermark added.
See it in action — real before & after
Why use Text Remover
What Text Remover does
Take text off an image cleanly.
Text and watermarks handled properly
Overlaid text sits on top of real detail, so removing it means rebuilding what was underneath rather than painting over it.
Edges and shadows go too
An object's shadow is part of the object. Leaving the shadow behind is the most common giveaway of a bad removal.
No manual selection
No lasso, no clone stamp, no patch tool. Name the thing and where it is.
Precise targeting
Because the instruction names a location, you can remove one person from a group of four without touching the other three.
Full resolution output
The removal is applied at the original size, not to a downscaled preview that then gets enlarged.
What Text Remover changes
Remove the text from this image and rebuild the surface beneath it, continuing the underlying pattern, texture and gradient so there is no trace of where it was.
Your face, bone structure and skin tone are held fixed on every generation — only what is described above is changed. Add your own brief in the Custom tab to take it further.
Text Remover for the things that ruin otherwise good photos
Most unwanted objects fall into a few recurring categories, and each has its own difficulty.
Objects and clutter indoors
Cables, chargers, laundry and anything on a surface that should be clear — the staple of property photography.
Reflections and glare
A flash bouncing off glass or a window reflecting the photographer, the hardest case because there are two images overlapped.
People in the background
Tourists, passers-by and anyone who walked into the shot at the wrong moment — the single most common request.
Watermarks and stamps
Logos, signatures and semi-transparent overlays that sit across real image detail.
Key highlights of Text Remover
Benefits
Faster than any manual method
A clone-stamp repair on a complex background is slow and skilled work. This is a sentence.
No editing skill needed
The skill in manual removal is entirely in the selection and the patching. Both are handled.
Cheap enough to iterate
A pack starts at $2, so removing one object, checking, and removing the next is a sensible workflow.
Works on any subject
Portraits, property, product, landscape and document photos all behave the same way.
Print-ready result
Full resolution with no watermark, ready for a listing or a print.
What people use it for
What it is mainly used for
Take text off an image cleanly. Pick a style, choose a model, and download the result at full resolution with no watermark.
Group photos with one person to remove
Common after a change in circumstances, and usually the only copy of that photo.
Documents and screenshots
Removing a stamp, a label or an overlay from an image of a document you have the rights to.
Event photos with signage
Sponsor boards, exit signs and temporary notices taken out of the background.
Vehicle listings
Licence plates and street clutter removed before a car or bike is listed publicly.
Prompt ideas
Name the object and its position, and remove one thing per run. A list of five removals in one instruction produces one mediocre result instead of five good ones.
Try the other model
If a prompt is being ignored, switching between Nano Banana and GPT Image often resolves it faster than rewording.
Describe the light
Direction, softness and colour temperature. Light does more to how a photograph feels than any other single thing you can specify.
Re-run before rewriting
Generation is not deterministic. The same prompt twice gives two different results, and the second is sometimes simply better.
Name the object and where it is
"Remove the person standing on the right edge" gives the model an unambiguous target.
Say what should be there instead
"Fill with the brick wall that continues behind" is far more reliable than leaving the fill open to interpretation.
Remember the shadow
"Remove the object and its shadow on the floor" — a leftover shadow is what makes a removal obvious.
More about Text Remover
Erasing is easy; filling is not. The difficulty of a removal is decided almost entirely by what was behind the thing being removed. A wall, a sky, grass or pavement fills invisibly because the pattern continues. A face, a sign or a unique object does not, and the fill becomes an educated guess.
Property photography is where this pays for itself fastest. Bins, cables, cars and clutter are what make a room read as lived-in rather than for sale, and removing them is most of what a property retoucher is actually paid to do.
Every photo has something in it that should not be there. A stranger walking through the frame, a bin at the edge, a watermark across the middle, a date stamp burned in by a camera in 2004. Text Remover takes it out and fills what was behind it.
Getting the best result from Text Remover
- Nothing you upload is published, shared or added to a gallery — but do not upload anything you would not want to leave your device at all.
- The free on-device tools — compress, convert, crop, resize, rotate — cost no credits, so use them for mechanical work.
- If a result is nearly right, say what is wrong with it in the next run rather than starting over.
- Remove one object at a time and check each fill at full size before the next.
- Objects on a simple, repeating background — wall, sky, grass, pavement — fill almost perfectly.
- An object overlapping the main subject is the hardest case, because the fill has to rebuild part of the subject too.
- Always ask for the shadow as well as the object.
- Say what should be behind it. The fill is much more accurate when it is told what it is completing.
It's free to try and runs online — nothing to install. Explore the rest of our AI Creative Apps, or open the full AI photo editor for more control.
