Screen dead pixel test
Fill the screen with five solid colors in turn — black, white, red, green and blue — and just look. This page judges nothing — you find the dead pixels, it does not compute them. All it is doing is giving you a clean, accurate backdrop with no room light to get in the way.
Now: black. Dim the room, stare at each corner of the screen for a few seconds, then switch to the next color.
How to look
- Wipe the screen first. Dust and fingerprints show up far more than dead pixels on a black screen, and they wipe right off — a quick clean saves most of the false alarms.
- Dim the room. Reflections will make you read room light as a screen fault. Dark patches and backlight bleed especially cannot be told apart unless it is dark.
- Click "Full screen", then hold your gaze on one area. When switching between colors do not look away — a dead pixel is defined by "this dot does not change under any color", and if your eye follows the color change you will never spot it.
- On black look for bright dots, on white for dark dots, on red/green/blue for a single dead sub-pixel. A screen usually has over a million pixels, so counting them one by one is not realistic — a few sweep passes are enough, and anything really wrong will jump out on its own.
Once you find something
First work out which kind it is, because the handling is completely different: bright and dark dots mean the pixel itself has failed, and within warranty you can usually claim on "more dead pixels than the standard allows" (the standard varies — some makers say 3, some more). Dark patches, backlight bleed and uneven brightness are a backlight or whole-panel problem; by dead-pixel count they often fall short, but they are usually easier to have accepted as a defect. Note which kind you saw, and under which color — you can describe it far more clearly when you send it in.
One thing worth saying: a screen will now and then have a stuck pixel — not dead, just held in one state by the signal. Sometimes it recovers on its own after a while, or with a dedicated "pixel fixer" flicker. This page does not do that: those tools are genuinely disputed as to whether they work, and they carry a seizure risk, so it is not for us to decide on someone's behalf that they should watch one.
Common questions
No. It produces no reading at all and makes no judgement. The one thing the page can do is flood the screen with an exact solid color. Dead pixels are found by you looking at the screen — fix your eyes on an area, cycle back and forth through the colors, and the dot that never changes with them is the dead one. Handing this to a program does not work: a web page cannot read the screen's pixels, and the screenshot API tends to hand back a different, processed image anyway.
It is not enough. There are two kinds of dead pixel: stuck-on dots stand out most on a black screen (they are lit, everything around them is black) and stuck-off dots stand out most on white (they are black, everything around them is white). And a dot with a dead sub-pixel (red and green fine, only blue out, say) is invisible on both black and white — it only shows up under solid red or solid blue. So red, green and blue are not padding: each one covers a sub-pixel.
You can look, but they are not the same thing as dead pixels, and this page likewise draws no conclusion. Flood the screen with pure black and look in a dark room: gray corners mean uneven backlight or bleed, a whole uneven field means a brightness-uniformity problem. All of these need the room dimmed and to be viewed head-on rather than at an angle — otherwise what you are seeing is really room light reflecting off the screen. Knowing what it is tells you whether it is worth repairing.
Safari on iOS does not support full screen for non-video elements — that is a browser limit the page cannot get around. It still works without full screen: just tap the color buttons and the patch fills the area on the page; turn the phone sideways and lean in — you just cannot cover the browser's address bar. Chrome on Android generally can go full screen.
