Eight slides aimed at how OLED actually fails: perfect black it should nail, the near-black region it finds hard, the grays where retention shows, and the primaries where wear hides. Best run in a dark room.
Run it
8 slides · ←→ or tap · 1–8 jumps · Esc exits
Turn the room lights off first — half of these slides only speak in the dark.
How to run it
Dark room, normal brightness — not maximum, which triggers brightness limiting and hides uniformity issues under it.
On true black, the screen should be indistinguishable from being off. Any lit pixel is a stuck subpixel; any glowing region is retention or a panel fault. (Glow that moves when you move your head is your own reflection — the classic false alarm.)
Give the near-black ramp and dim gray slides thirty seconds each; faint clouds and ghosts take a moment to see. Each ramp band should be flat, and the dim gray field is where logos and taskbars reappear.
On the primaries, sweep your eyes across the field looking for patches that differ from their surroundings. Compare across the three slides: wear is usually strongest in one channel.
What you're actually seeing
An OLED lights every pixel individually, which is why black is genuinely off and why the test begins there — a field an LCD physically cannot produce. The cost of per-pixel light comes at the two extremes. At the bottom, driving emitters at tiny currents evenly across millions of pixels is hard, so the near-black slides are where uniformity problems live. At the top and over time, emitters age with use: static content wears its pixels unevenly, and that wear reads as ghosts on uniform fields long before it's visible in normal content. The slides are ordered to walk exactly that story, dark to bright.
If black isn't black on your screen, you're probably on an LCD — the equivalent findings there are backlight bleed and IPS glow, and the gradient test plus the dead pixel test cover that ground. Found retention or a ghost here? The burn-in test is the follow-up, including what recovery honestly can and can't do.
Because its failure modes are inverted. An LCD struggles to make black (backlight bleed, glow) and is safe with static content; an OLED makes perfect black trivially but earns its faults elsewhere — blotchy near-black handling, uneven wear from static content, and brightness limiting on large white areas. A generic screen test spends its time where LCDs fail; this one spends it where OLEDs do.
Why does the white slide get dimmer after a moment?
That’s ABL — the automatic brightness limiter. An OLED panel has a power budget, and a full white screen at maximum brightness would exceed it, so the panel ramps the whole field down. It’s engineering, not a defect, and every OLED does it to some degree. What would be a finding on the white slide: a color tint, or patches that are darker than the rest and stay darker.
Is blotchiness in near-black normal on OLED?
A little, especially on phones at minimum brightness: the bottom few code values are the hardest region for an OLED to render evenly, and mild vignetting or faint cloud shapes in the 2–8 range are common even on healthy panels. Worry when the blotches are strongly colored, have hard edges, sit in the same place across slides, or are obvious at normal brightness — those readings point at the panel rather than the physics.
What are the red, green and blue slides for?
Wear detection, channel by channel. OLED subpixels age individually — blue fastest — and the wear from a static logo rarely hits all three equally. A ghost or dull patch that appears on one primary but not the others is uneven subpixel wear; the same patch on every slide is more likely dirt on the glass (worth ruling out first) or full-stack burn-in. The burn-in test is the follow-up if you find something.
Should I run my TV’s pixel refresh before or after this test?
After — and only if you found something. The panel-maintenance cycle (pixel refresh, panel refresh, depending on brand) measures emitter wear and compensates for it, and TVs already run the short version automatically on standby. Running this test first tells you whether there’s anything to fix; a manual refresh afterwards, then a retest, tells you whether compensation handled it. Repeated manual refreshes on a clean panel just spend panel life.