Webcam Exposure / White Balance
Measure whether your camera is exposing you properly — or exposing the window behind you.
How this test works
The camera runs for two seconds while frames are sampled and reduced to a few numbers: mean brightness using Rec. 601 luma weighting, the share of pixels clipped at pure white and pure black, and the mean of each colour channel. Only the second half of the samples is used, because auto-exposure hunts for about a second after a stream opens and a single grab during that hunt reports wherever it happened to land. Clipping is then checked before mean brightness, and that ordering is what makes the page worth having. A bright window behind a dark face averages out to a perfectly respectable mid-grey — so a tool that judges on mean brightness alone tells the person with the single most common webcam problem that their exposure is fine. Both ends clipping while the average sits in the middle is the signature of backlighting, and it is reported as such.
What the results mean
- Exposure
- A band derived from mean brightness and clipping together. Well exposed means the picture sits in a usable range with little clipped at either end.
- Blown highlights
- The share of the frame at pure white. Nothing can recover detail from these pixels afterwards — the information was never captured. Above about 10% something bright is dominating the shot.
- White balance
- Estimated from the ratio of red to blue. Warm means an orange cast, usually a tungsten bulb; cool means blue, usually daylight or an LED the camera is balancing against the wrong reference.
- Auto-exposure drift
- How much brightness moved across the sampling period. A large figure means the camera was still hunting, which is normal for the first second but a problem if it never settles.
Common problems and fixes
- You look like a silhouette on calls
- You are backlit, and this test will say so. Move so the window is in front of you rather than behind. It is the single most effective change available and it costs nothing — no camera setting compensates for a light source brighter than your face.
- The picture is dark and grainy
- Add light in front of you rather than raising a brightness slider. A brightness control amplifies what the sensor already captured, including its noise; more actual light lets the camera use less gain, which reduces noise and often raises the frame rate at the same time.
- Everything looks orange or blue
- Mixed lighting is the usual cause — daylight through a window plus a warm bulb. The camera can only balance for one of them. Using a single kind of light source gives it something consistent to work with.
- Brightness keeps shifting as you move
- Auto-exposure is reacting to something bright entering and leaving the frame. Remove the bright object from shot if you can. Some camera utilities allow exposure to be locked, though browsers cannot reach that control.
Frequently asked questions
Why not just read the camera's exposure setting?
Because it is almost never available. The MediaStream image capture controls exist in the specification, but `getCapabilities()` returns an empty object on most platforms and browsers, and where it returns anything the exposure fields are frequently missing or read-only. Measuring the picture works everywhere and describes what you actually look like, which is the question being asked.
Why does the average brightness look fine when I look terrible?
That is backlighting, and it is exactly why this page checks clipping first. A blown-out window at 100% and a face at 15% average to something perfectly reasonable. The average is a real number that describes nothing anyone cares about, and reporting it alone is how most tools miss the most common webcam fault there is.
Can it tell me if my webcam is faulty?
No, and it is worth being clear about that. Almost everything this measures is lighting rather than hardware. A camera producing a dark, noisy, colour-cast picture in a poorly lit room will produce a clean one by a window. Move the light before suspecting the camera.
Why is the white balance sometimes reported as unmeasurable?
Because a dark or very flat frame carries almost no colour information — sensor noise dominates the channel means, and any ratio computed from it would be arbitrary. Rather than report a cast that is really noise, the page says the frame cannot support the estimate.