Demo · drag the slider to compare. Drop your own image below to remove its background.
Your image never leaves your device. The model runs in your browser using WebGPU (or WebAssembly as a fallback). No upload, no signup, unlimited use — even on huge files.
Free background remover that runs entirely in your browser
PocketWebTools' background remover uses an in-browser AI model to cut a subject out of its background and hand you back a transparent PNG. There's no upload, no signup, and no usage cap — every step happens on your device, powered by WebGPU when available and WebAssembly otherwise. If you've ever bumped into "free for 1 image then upgrade" paywalls on the big-name SaaS tools, this is the version that just works.
The model is ormbg, an open background-removal network (Apache 2.0). On its first use we download a small ONNX file into your browser cache so subsequent runs are instant; you can list and remove cached models at any time from the "Local models" chip at the top of this page. The model is general-purpose — it handles portraits, products, animals, food, and most everyday subjects, not just people.
How to use it
- Drop an image onto the drop zone, paste it from your clipboard, or pick a file from your computer.
- Click Remove background. The first time you run it, the model will download (~one minute on broadband); after that it's a few seconds.
- Use the slider on the result image to compare before / after.
- Hit Download PNG to save the cut-out with a transparent background.
What it does
The model predicts an alpha matte — a per-pixel score from 0 (background) to 255 (foreground) — then composites that matte onto your image as the alpha channel of a PNG. The result keeps the full resolution of your input file: a 4000×3000 photo comes out as a 4000×3000 PNG with transparent backdrop.
- General-purpose isnet-based segmentation — works on people, products, pets, food.
- Up to 25 MB images supported, capped only because larger files start to slow down the canvas pipeline.
- WebGPU acceleration on Chrome / Edge / Arc / Brave, with a WebAssembly fallback on Safari and older Firefox builds.
Common use cases
- Profile photos, LinkedIn headshots, team-page portraits.
- Marketplace product listings where you want a clean white or transparent backdrop.
- Slide decks where you need to drop a person onto a different scene.
- Social-post crop-outs for stickers, memes, and thumbnails.
- Quickly previewing a cut-out before doing a more careful edit in Photoshop or Affinity.
Why "local AI" matters
Most background removers upload your image to a server, run the model there, and send the result back. That's fine for casual use — but it means your photo touches their infrastructure (and their logs), they need a paid quota system to recover GPU costs, and you're locked out the moment your network drops. Running the model in your browser flips all three:
- Privacy — your image never leaves the device. We have no servers in the loop.
- Cost — there's no inference bill for us to pass on, so the tool is free and unlimited.
- Offline — once the model is cached, it works on a plane.
Frequently asked questions
- Does my image get uploaded anywhere?
- No. The image stays in your browser the entire time. The neural network runs locally with WebGPU (or WebAssembly on devices without WebGPU), so there's no server in the loop, no upload, and nothing for us to log.
- Why does it ask to download a model the first time?
- The AI model that does the cutting-out has to be on your device to run locally. It's a one-time download (under 50 MB) cached by your browser, so subsequent uses are instant. You can review or delete the cached model at any time using the 'Local models' chip in the page header.
- What kinds of images does it handle best?
- It's general-purpose: portraits, products, animals, food, even semi-transparent objects. The default model is ormbg (an isnet-based open background removal network released under Apache 2.0). Hair and fine edges are handled well; very low-contrast subjects against a busy backdrop are still the hardest case for any model.
- Are there file size or resolution limits?
- Files up to 25 MB are accepted. Internally the model resizes images to a fixed working resolution and then up-samples the alpha mask back to your original image's size, so output dimensions match your input. Very large images (10MP+) may take a few seconds longer on slower devices.
- Why is my output quality different from remove.bg?
- remove.bg uses proprietary cloud models trained on a much larger dataset; we use an open Apache-2.0-licensed model that runs locally. For most images the difference is small. The trade-off you get back: no upload limits, no signup, no per-image fees, and complete privacy.
- Can I use the result commercially?
- Yes — the output is yours, just like with any image-editing tool. The ormbg model itself is Apache 2.0 (permissive open-source). We don't claim any rights over images you process here.
- Does this work offline?
- Yes, after the first visit. Once the model is cached your browser doesn't need a network connection to run it again — you can use the tool on a plane.
- What's WebGPU and why should I care?
- WebGPU is a modern browser API that lets web pages run computation on your GPU. It makes the background removal a few times faster than the CPU-based fallback. We auto-detect support and pick the fastest path; the badge above the action button tells you which one you got.