Test your webcam

What the test tells you

Three numbers, and all three matter for different reasons.

• Resolution is what the camera is actually delivering right now, read from the video track’s own settings — not what the box claimed.
• Measured frame rate is counted frame by frame as they arrive in the page, using requestVideoFrameCallback where the browser provides it. This is the number that determines whether you look smooth or stuttery on a call, and it is frequently lower than the camera’s rating.
• Aspect ratio reveals whether the camera is cropping. A 4:3 sensor pretending to be 16:9 is doing it by cutting off the top and bottom.

Why the frame rate is lower than advertised

A camera rated for 30 fps rarely delivers 30 fps in practice, and the reasons are physical rather than faults.

Requested resolution versus what you get

The resolution dropdown asks with ideal rather than exact. That is deliberate: an exact constraint makes a camera that cannot manage the setting throw an error and give you nothing, while ideal makes it return its closest match. So asking for 1080p on a 720p sensor gets you a working 720p preview rather than a failure.

Two things surprise people here. Many built-in laptop cameras really are 720p sensors, whatever the marketing said. And some cameras will deliberately drop resolution to hold a higher frame rate, or the reverse, because the USB link cannot carry both.

Mirroring, and which version other people see

Video-call apps mirror your self-view because an unmirrored image of your own face reads as wrong — text on your shirt appears backwards to you, and you reach the wrong way when adjusting the camera. Everyone else on the call always sees the unmirrored image.

This page mirrors by default to match that expectation, and the toggle lets you see what others see. Snapshots are captured exactly as displayed, so turn mirroring off first if you are photographing a document, a whiteboard or anything with writing on it.

The snapshot

The snapshot draws the current video frame onto a canvas at full sensor resolution and encodes it as PNG — lossless, so it is a pixel-exact record of what the camera delivered. Because it is generated from the live stream rather than taken by a camera application, it carries no EXIF block at all: no timestamp, no location, no device serial, no lens data. It is not saved anywhere until you press download.

That makes it a genuinely useful way to check focus and framing at full resolution, which a scaled-down preview cannot show you.

Nothing is transmitted — the structural version

Every webcam-test site claims your video is private. Most of them are streaming it somewhere, because their architecture requires a server. Here is why this one cannot be:

• The site is static, with no backend and no API routes. There is no endpoint to receive a frame.
• There is no RTCPeerConnection anywhere in the page, so there is no WebRTC path out either.
• The site’s Content-Security-Policy lists no third-party origin in connect-src, so a request to anywhere else would be blocked by the browser itself.
• The preview is your own MediaStream attached to a video element. That is the entire pipeline.

Check it: open developer tools, watch the Network tab, and run the test. Nothing. Or turn off Wi-Fi and run it anyway — it works exactly the same.

Releasing the camera

Press "Stop and release the camera" and every track in the stream is stopped; the indicator light next to your camera goes out within a second. Closing the tab does the same thing. If the light stays on after both, another application or another browser tab still has the camera open — that is a good reason to check before joining a call.

Also worth testing

A camera that works and a microphone that does not is the more common failure on a video call. The microphone test covers levels, device selection and a record-and-playback loop, with the same guarantee that nothing leaves your machine. To record audio properly afterwards, there is the voice recorder.