How Our Screen Tests Work
Every test on this site is a visual inspection aid. It puts a known pattern on your screen so that your eyes can do the judging. That distinction shapes everything else on this page.
What a browser can help you observe
- Visual pixel defects
- Dark, bright and stuck dots become visible when the whole screen is one solid color.
- Uniformity patterns
- Flat gray fields expose bands, blotches, clouding and tint shifts that normal content hides.
- Gradients
- A smooth ramp makes step boundaries visible, which points at bit depth or the video pipeline.
- Near-black behaviour
- Single-level gray steps show where a panel starts crushing detail or banding.
- Motion patterns
- Moving shapes let you see trails, smearing and overshoot with your own eyes.
- Observed frame timing
- The browser can report how often it was able to present a frame.
- Touch coverage
- Drawing across a grid shows which areas of a digitizer respond.
What a browser cannot measure
- Physical luminance
- Nits require a light meter pointed at the panel. A web page has no access to one.
- Delta E and color accuracy
- Measuring color error needs a colorimeter and a reference, neither of which exists in a browser.
- True gray-to-gray response time
- Real GtG needs a photodiode and an oscilloscope sampling the panel directly.
- Input lag
- End-to-end latency has to be measured against an external clock, not from inside the page.
- HDR peak brightness
- Peak nits and sustained brightness are physical measurements of the panel.
- Color gamut
- Gamut coverage is a measurement of what the panel can physically reproduce.
- Contrast ratio
- Contrast is the ratio of two measured luminance values — again, hardware only.
Why there is no screen health score
A number like "Screen health: 87/100" would be easy to build and completely made up. It would combine things we cannot measure with things only you can judge, and it would give a false sense of precision at exactly the moment when precision matters — when you are deciding whether to return a device. So we do not produce one. The screen test report counts what actually happened instead: which tests you completed, how many issues you marked and where they were.
Why results say "possible" and "likely"
The diagnosis wizard works from your description of what you see. That is real evidence, but it is not proof: dust looks like a dead pixel, viewing angle looks like backlight bleed, and content artifacts look like panel faults. Every outcome is therefore phrased as a possibility with a confirmation test attached, never as a verdict. If a tool tells you it has "confirmed a dead pixel" from inside a web page, it is overstating what it knows.
What we do not recommend
Some stuck-pixel advice circulating online suggests pressing, rubbing or tapping the panel, or applying heat. We do not recommend any of it. The realistic upside is small and the downside — spreading a defect, damaging the polarizer, or voiding the warranty you were about to use — is not. Color cycling is the one software approach worth trying, and even that is explicitly not guaranteed.
About viewing conditions
Your results depend on the room, the brightness setting, the viewing angle and your own eyesight — three people can look at the same panel and disagree. Test in the conditions you actually use the screen in, view straight on from your normal distance, and re-check anything you find at a second brightness level before drawing a conclusion.
Privacy
Everything runs locally. Patterns are drawn with CSS, marked positions and observations are kept in your browser's local storage, and reports are generated on your device. No images of your screen are uploaded, because none are ever captured.
What this test checks
- What browser-based display tests can genuinely show
- Which display specifications need real measurement hardware
- Why every result here is phrased as an observation, not a verdict
Related tests
Not sure what you are seeing?
Run diagnosisSave your observations
Create a report