Skip to content
Runs entirely in your browser

Change your webcam background and record it, all in your browser

Replace your background with a studio, an office or your own image — or just blur the room behind you — and record the result to a file on your own disk. No account, no upload, no watermark, nothing to install.

Your camera, microphone and recordings stay on your device.

What it actually does

  • Backgrounds that change live

    Pick a backdrop, blur your room, or drop in your own image. The swap happens on the frames as they arrive, not in a render queue afterwards.

  • Recording that keeps the background

    The recorder is fed the same composited canvas you are watching, so the file you save looks exactly like the preview. No export step.

  • Nothing leaves the machine

    No account to create, no footage uploaded, no watermark burnt in. The recording sits in memory until you choose where to save it.

How it works

  1. Choose a background before the camera turns on

    The picker opens first, and it works without camera access. Eight built-in backdrops, a live blur, or an image from your own device.

  2. Grant camera access in the studio

    This is the first and only time anything asks for your camera. The microphone is requested at the same time and is optional — refuse it and you can still record silently.

  3. Record, and adjust while you do

    Change background, mute, or switch the camera off mid-take. The timer runs off the wall clock, so it stays honest even if your machine is busy.

  4. Review, then save to disk

    Play the result back before committing to it. Save opens a real file dialog where your browser supports one, and falls back to an ordinary download.

Why this exists

Tools in this space usually give you one half of the job. Three shapes are common, and each leaves something out.

Browser recorders capture your camera to a file quickly and without an account, which is exactly right — but they hand you your real room. Fixing the background afterwards means an editor, a mask, and a lot of time for a two-minute clip.

Cloud video editors can remove a background properly, and they can do far more besides. The cost is that your footage is uploaded to somebody else's server, usually behind a sign-up, and the free tier is often the one with the watermark.

Installed virtual-camera apps solve a different problem well: they register as a camera device, so Zoom and Teams see the processed feed. That takes a download, an operating system they support, and frequently a licence.

This does the useful half of the first two at once and on your own machine: the background is replaced as you record, and the recording never leaves the tab. What it deliberately is not is the third — a browser tab cannot present itself to your operating system as a webcam, so this records video rather than feeding a meeting.

What runs where

The claim that nothing is uploaded is worth only as much as whatever enforces it, so here is what does.

There is no server to upload to. The site is static HTML, CSS and JavaScript. It contains no API routes, no server actions and no database client, so there is no endpoint that could accept a video even if something tried to send one.

Your browser blocks the attempt anyway. Every response carries a Content-Security-Policy whose connect-src allows this origin and nothing else. Any request to another host — a fetch, a WebSocket, an analytics beacon — is refused by the browser, not by our good intentions.

Even the machine-learning model is first-party. The segmentation runtime and its weights are served from this domain rather than a model host, which is what makes that policy possible to keep so tight.

The privacy page goes through each permission and what happens to it, including the one third-party request the segmentation library tries to make and how it is stopped.

Browser support

Recording needs a handful of browser APIs, and the studio checks for each one at runtime rather than guessing from your user agent. If something is missing it names it.

Browser support for webcam recording and background replacement in Background Studio
BrowserStatusDetail
Chrome and Edge (desktop)Full supportRecords VP9 WebM. GPU edge refinement runs. The combination this is developed against.
Firefox (desktop)Full supportRecords VP8 or VP9 WebM depending on version.
Safari 16.4 and laterSupportedRecords H.264 MP4, because Safari cannot record WebM. Edge refinement runs.
Safari before 16.4Recording unavailableMissing canvas.captureStream. The studio says so instead of failing part-way through.
Browsers without WebGL2Works, softer edgesFalls back to 2D-canvas compositing: correct output, but a visibly softer boundary.
Phones and tabletsWorks, but demandingThe interface is responsive. Sustained realtime segmentation is heavy on mobile hardware and battery.

Frequently asked questions

Do I need an account?
No. There is nothing to sign up for, because there is no server to sign in to — the app is a static site with no backend, no database and no user records.
Is my video uploaded anywhere?
No. Frames go from your camera to a canvas to a file, all inside the tab. Three separate things enforce that rather than one: there is no server code to receive an upload, the site's Content-Security-Policy tells your browser to refuse any connection to another host, and the segmentation model is served from this domain instead of a third-party CDN. The privacy page spells out each layer.
Is there a watermark, a time limit, or a paid tier?
No watermark and no paid tier — there is no billing code in the app. There is no time limit either, though it is worth knowing that a recording is held in memory until you save it, so a very long take is bounded by your own machine rather than by a rule.
Can I use my own image as a background?
Yes. "Upload" is a misnomer here: the file is read into memory, kept in your browser's own IndexedDB storage, and drawn from there. It never travels. It also stays in that browser profile only — it will not follow you to another machine.
Can I use this as a virtual camera in Zoom, Teams or Google Meet?
No, and no web page can. Feeding a processed image into another application means registering as a camera device with the operating system, which requires an installed driver — something a browser tab has no way to do. What this is good for is recording a finished video. If you need the effect live inside a meeting, use that app's own background feature, which the background guide walks through.
What file do I end up with?
A WebM file using VP9 video and Opus audio in Chrome, Edge and Firefox. Safari cannot record WebM, so there you get an MP4 with H.264 and AAC. The app negotiates this with the browser rather than assuming, and the file plays without conversion in the browser that made it.
How good are the edges around hair and shoulders?
Better than a plain mask upscale, and honestly dependent on your room. The model produces a low-resolution mask, which is then re-snapped to the edges actually present in your camera frame — so it needs a visible edge to find. Even, front-facing light gives clean results; strong backlighting or a wall the same tone as your hair will still show artefacts. Why, and what to do.
Does it work on a phone?
The interface does, and recording works. Sustained realtime segmentation is genuinely heavy on phone hardware, so a laptop or desktop will give you a smoother recording and a warmer phone will give you a worse one.

Pick the tool you came for

Guides

Ready when you are

Choose a background, grant camera access once, and record. Nothing to install and nothing to sign up for.

Record a video