How to use Remove background
- Choose a photo: a product, a person, a pet or a logo.
- The first time, the tool downloads its AI model (about 30 MB) once; after that it starts straight away, even offline.
- Wait about 10 to 30 seconds while the model finds the subject on your device.
- Pick a transparent, white or coloured background, then download the PNG.
A clean cut-out without uploading your photo
Product listings, profile pictures, slides, stickers and thumbnails often need the subject on its own: the shoe without the floor, the person without the office behind them. This tool finds the subject of a photo and removes everything else, leaving a PNG with a transparent background that you can place on any colour or design. If you need a solid background instead, choose white or any colour before you download.
Most background removers upload your photo to a server and send the result back. This one does the work in your browser. The photo is read, processed and saved on your device, so pictures of people, unreleased products and private documents never pass through anyone else's computer.
How it finds the subject
The tool uses IS-Net, a segmentation model from the open DIS project published under the Apache 2.0 licence, run by ONNX Runtime Web inside a background worker in this tab. The model looks at a 1024 by 1024 copy of your photo and decides, for every point, how likely it is to belong to the main subject. That likelihood becomes a soft mask rather than a hard outline, so edges blend smoothly instead of looking jagged.
The mask is then scaled up and applied to your original photo at its full resolution, so the cut-out keeps every pixel of detail the camera captured. Only very large images, above about 40 megapixels, are scaled down first to stay within what a browser tab can safely hold in memory, and the tool tells you when that happens.
What to expect
The first time you use it, the model downloads once, about 30 MB, from WeaveAI's own engine server, and your browser keeps it. Later runs start straight away and work offline. Processing takes about 10 to 30 seconds depending on your device, because the model runs on your own processor rather than in a data centre.
It works best on a clear subject that stands out from its surroundings: products, people, pets, cars, food and logos. Fine hair, fur and transparent objects such as glass come out softer, and a background with the same colours as the subject can blur the edge. For tricky photos, crop close to the subject first, which gives the model a better view.
Questions
- Is my photo uploaded?
- No. The AI model is downloaded to your browser, and the photo is processed on your device. The only download is the model itself, from WeaveAI's own engine server; your photo never leaves your device.
- Why does the first run take longer?
- The model, about 30 MB, downloads once and is then kept by your browser. Later runs skip the download and only need the 10 to 30 seconds the model takes to process the photo.
- Which photos work best?
- A clear subject that stands out from its background: products, people, pets, cars, logos. Fine hair and see-through objects such as glass come out softer, and a busy background can confuse the edges.
- Is the result full size?
- Yes. The model works on a 1024 by 1024 copy to find the subject, and the cut-out is then applied to your original photo at its full resolution.