Remove Objects from Photos
Object & Text Remover erases people, objects and text you brush over and fills the gap with an AI model running in your browser, so the photo is never uploaded.
Paste a photo (Ctrl+V), drag and drop, or click to upload
Image file · processed on your device, never uploaded
Brush over what you want gone, then press Remove. The AI model runs in your browser: MI-GAN is 28 MB to download once, and your photo is never uploaded.
At a glance
- Accepts
- PNG, JPG, WebP, GIF and other images your browser decodes
- Outputs
- The photo at full resolution, in its original format
- Where it runs
- Entirely in your browser, on your device
- What leaves your device
- Nothing - the file never leaves your device
- Limits
- Up to 16.7 megapixels; first use downloads a 28 MB AI model (208 MB for Best quality); large areas, busy patterns and faces can look smudged
- Account
- Not required - there is no sign-up
- Price
- Free, with no watermark and no usage cap
- Made by
- Visrut
- Last updated
- 11 October 2026
To remove an object from a photo, brush over it and press Remove: an AI inpainting model fills the area with a plausible background, and you download the photo at its original resolution. Everything runs in your browser - the model (MI-GAN, MIT licence, or LaMa for best quality, Apache-2.0) is downloaded once and the photo is never uploaded. It works best on people, signs, bins, wires, blemishes and clutter in front of simple backgrounds.
How to use it
- Paste (Ctrl+V), drag and drop, or click to upload a photo.
- Brush over the object so it is fully red, a little past its edges, including its shadow.
- Press Remove. The first time, the AI model downloads (28 MB for Quick).
- Press Compare to check against the original. Brush over anything left and press Remove again, or Undo (Ctrl+Z) a pass you don't like.
- Download the result in the same format, or copy it.
Tips for a cleaner result
- Cover the whole object. Any edge left outside the brush gets copied back into the fill as a smudge.
- Do big things in pieces. Several smaller passes usually beat one huge one, because the model has more real photo around each hole to copy from.
- Try Best quality on texture. Fabric, leaves, gravel and brick are where Quick is weakest.
- Removing text? See remove text from an image for tips on captions, stamps and screenshots.
Watermarks and copyright
Use it on your own photos, or on images you have permission to edit. Removing someone else's watermark or copyright notice may breach their copyright. To hide something rather than replace it - a face or a number plate - use Blur Brush instead.
Frequently asked questions
How do I remove an object from a photo?
Upload the photo, brush over the object until it is covered in red (include its shadow), and press Remove. An AI inpainting model fills the area with what it thinks was behind it. If a trace is left, brush over it and press Remove again.
Is my photo uploaded to a server?
No. The AI model is downloaded to your browser and runs there, on your graphics card with WebGPU or on the CPU otherwise. Your photo never leaves this tab.
Why is there a download the first time?
The filling is done by an open AI model that has to be on your device to run locally. Quick uses MI-GAN (MIT licence, about 28 MB); Best quality uses LaMa (Apache-2.0 licence, about 208 MB). Each is downloaded once from Hugging Face and then cached in your browser.
Should I use Quick or Best quality?
Start with Quick: it is a small download and fills plain backgrounds - sky, walls, water, paper, grass - well in well under a second on most computers. Switch to Best quality when the fill looks smudged or has odd colours, which happens most on detailed textures like fabric, foliage or brickwork. Best quality is a bigger download and several times slower, but in our tests it was clearly more accurate.
Does it keep the full resolution of my photo?
Yes, up to about 16.7 megapixels. The model works on a 512 x 512 crop around what you brushed, and only the brushed pixels are replaced in your full-size photo - everything else stays exactly as it was. Photos over 16.7 megapixels are scaled down to that, because some browsers cannot draw a bigger canvas.
What doesn't it do well?
It invents what was behind the object from the pixels around it, so it cannot recover real detail. Large areas (more than about a quarter of the photo), busy or patterned backgrounds, straight lines that pass behind the object, and anything that needs a face or a body part rebuilt are where it goes wrong. Shadows and reflections are not removed unless you brush over them too.
Can I remove a watermark?
Technically it works on watermarks, logos and date stamps, but only remove them from images you own or have permission to edit. Removing someone else's watermark or copyright notice may breach their copyright and, in many countries, laws that protect copyright information.