Think at the pixel level.
The pixels burn-in do not depend on each other. Just because one pixel is bright, doesn't affect wear-and-tear of the adjacent pixel.
It's simply the contrast in wear-and-tear between adjacent pixels that becomes visible to the human eye.
Metaphors may help.
One smartphone manufacturer (that I worked for) that uses an OLED screen in some of the phones, have used the candle metaphor to describe OLED burn. I thought it was a very accurate "Plain English(tm)" explanation of why burn-in becomes visible to human eye.
- Basically, each pixel is like a candle that lasts approximately 20,000 hours of average picture level (insert your favourite number here)*
- There are a few million candles on a 1080p OLED. (1920 times 1080, times subpixels, depending on RGB or pentile)
- The candle burns fastest when the candle is new.
- The candle goes gradually dimmer as it burns down.
- The candle burns fastest when the pixel is brightest.
- The candle burns slowest when the pixel is darkest.
- The burning speed of the candle gradually slows down as the candle burns down. Dimmer if the candle is old.
- "Visible burn in" now means "some candles (pixels) burned down far more than adjacent candles (pixels), to the point where you can easily see visible differences in brightness between adjacent candles"
- The first, say, 200 hours, the candle burns fastest and brightest, creating largest differentials between adjacent candles if you don't wear evenly. This shows up as noticeable burn in, as the newer brighter-burning candles contrast with the older slower-burning candles.
- Doing a factory burn allows the fast-burn phase to be done-over-with, so that the candles burn slower. So that the burn-time asymmetry between adjacent candles create less differences (and thus, prevents burn in from becoming noticeable after a short time period).
- Rated 20,000 hours is only based on average picture level. This can easily mean 5,000 hours of full-white, 10,000 hours of bright gray, 20,000 hours of medium gray, or 40,000 hours of dim gray, infinite hours of full black (candle is off), etc. Again, replace with your favorite numbers, but you get the approximate idea.
- The graph of burnspeed-versus-pixel-intensity is not linear. It can vary from tech to tech and display to display (e.g. plasma vs. OLED). It's a steep cliff that plateaus out.
- The more you burn, the differences between adjacent candles become smaller small because you go into the plateau. So you can erase burn in just by keeping using the display to cause the pixels to plateau out. The burn-in will become fainter over time.
- Sometimes there's an additional short-term component to the burn-in too. (e.g. when the display cools down or when you exercise the pixel, burn-in disappears quickly). So a new variable may be added: An overheated/static candle can burn dimmer (and still burn down fast), while a cooled-down/variable candle will burn brighter (and matching adjacent pixels better). This covers the temporary image-retention aspect.
So you don't have to use the same image to create burn in.
-- It can be black bars (letterbox burn in)
-- It can be semi-static images that orbits closely around each other, like an animated station logo (it creates a blurry burn-in)
-- It can be area-differences in average picture level. A splitscreen video (video touching each other directly, no borders, no black gaps) can still create burn in. This is when one video is on average brighter and the other is on average darker. If you run these looping for hundreds of hours, you will get burn in, even though you never ever had black boundaries or gaps. This is because the brighter video burnt those candles faster. This is a use case that very RARELY happens, though, but it does happen, and has happened before (e.g. advertising billboards)
With the candle metaphor, there is no concept of "different kinds of burn-in". Just "different burn-in use cases" and "different burn-in patterns, including blurry burn-in and sharp burn-in"
Edited by Mark Rejhon - 7/31/13 at 9:48pm