How to remove a background
- Add your photo — drop it on the box, paste it with Ctrl+V (⌘+V on a Mac), or choose a file.
- Wait a few seconds while the AI finds the subject. The first time, your browser downloads the small fast model (about 4 MB) — it starts loading as soon as the page opens, and after that it’s saved, so later photos start right away.
- Want finer edges? Press Improve edges (HD). This uses a larger, more detailed AI model. It is a one-time download of 40–80 MB, so it only happens when you ask for it. The tool remembers your choice for next time.
- Touch up if needed. Use Erase to remove leftover background and Restore to bring back anything that was cut by mistake. Zoom to 2× or 4× for fine edges, and press Undo (or Ctrl+Z) to step back.
- Pick a background — transparent, white, a blurred version of the original, or any color.
- Download a PNG (keeps transparency) or a JPG (smaller, for solid backgrounds), at the full resolution of your photo.
Private by design
Most online background removers upload your photo to their servers. This one doesn’t. Both AI models (fast and HD) run inside your browser, using your device’s graphics chip (WebGPU) when it’s available, or its processor (WebAssembly) when it isn’t. Your browser only downloads the AI models and the code that runs them; nothing is uploaded. Your image never leaves your computer or phone, which makes it a good fit for ID photos, product shots before launch, or anything personal.
Getting the cleanest cut-out
- Contrast helps. A subject that stands out from the background in color or brightness gives the sharpest edges.
- Good light helps with hair and fur. Soft, even light around the edges gives the model the most detail to work with.
- Fix small mistakes with the brush rather than starting over — zoom in, lower the brush size, and trace the edge.
- Transparent or glass objects (bottles, glasses, smoke) are the hardest case for any AI model and may need more touching up.
What you can use it for
Product photos for an online shop, profile pictures, passport-style photos on a plain background, stickers, presentations, thumbnails, or swapping in a new background for a cleaner look. Make sure you have the right to edit and use any image you process.
Frequently asked questions
Is it really free?
Yes — there’s no limit, no watermark and no sign-up. Because the work runs on your device, it costs us nothing extra when you use it, so we can keep it free.
Why is the first run slower?
The first time, your browser downloads the fast AI model and the code that runs it — about 4 MB in all. That takes a few seconds on a normal connection, and up to about half a minute on a slow mobile connection. It starts loading as soon as the page opens, while you choose your photo, and it’s saved in your browser, so later visits start in a few seconds.
What is the difference between Fast and HD?
Fast uses a small AI model and then lines the edges up with the details of your photo. It works well for most people, products and pets. HD uses a much larger model that is better with hair, fur and busy backgrounds. HD needs a one-time download of about 40 MB (or about 80 MB on computers with a fast graphics chip), so it only runs when you ask for it. On mobile data, you may prefer to try HD on Wi-Fi.
Which browsers work?
Current versions of Chrome, Edge, Firefox and Safari on desktop and mobile. Browsers with WebGPU (most recent desktop browsers and many newer phones) are fastest; others use a compatibility mode that takes a little longer.
What resolution is the download?
The same as your photo, up to 4096 pixels on the longest side. Larger photos are scaled down to that size first.
Which AI models does it use?
Fast mode uses the open-source U²-Netp model (Apache 2.0 license), run with ONNX Runtime Web (MIT license); both are served from this website. HD mode uses the open-source ORMBG model (Apache 2.0 license), run with Transformers.js and downloaded from Hugging Face when you ask for it. Both use your graphics chip when available and fall back to the processor automatically.