h a l f b a k e r y
Recalculations place it at 0.4999.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
or get an account
One of the disadvantages of the new display technologies is that they are in effect many thousands of single-pixel displays in parallel. While with CRT displays you generally either get a display or the thing is broken, with LCD displays it is common for single pixels to malfunction, being 'always on'
or 'always off'. This is frequently the case from manufacture, but such products are still considered servicable by the suppliers and are therefore not returnable.
I propose a device which sits in between the computer and the monitor, and generally is functionally invisible to both. However, it maintains a list of damaged pixels, and supplies a corrective processing to the surrounding pixels as they are displayed.
First, you set up the displays bug list in 'programming' mode. This involves marking the location and type of each defective pixel. Then the device is ready to use for that display.
As the information for each pixel is passed to the display, its location relative to each defect is determined. If it is adjacent to a defect, the output is modified so as to reduce the optical error of the defect. (There are several possible methods for doing this which are described below.) But essentially, the neighbours of a defective pixel are reduced or increased in intensity, dependent on what the defective pixel should have been.
The observant among you will have noticed two things.
1) This doesn't help at extremes of intensity. If, for example, all pixels should be fully on, then this won't fix the appearance of an 'always off' defect.
2) The device can't know the state of pixels further along the scan-line or in subsequent rasters. This must be solved in one of several ways:
2 a) Maintain a cache of the state of defective pixels from the previous frame. This is relatively easy, but may have slight issues with moving objects.
2 b) Maintain a cache of the previous scanline. Delay output by the same amount, effectively giving you a read-ahead of the amount required.
2 c) As 2b, but just cache and delay by 1 pixel. Ameliorate the defect over the two horizontal neigbours rather than all four nearest neighbours. If this works satisfactorally it is probably cheapest.
2 d) Ameleorate the defect in the direction of scanning, as Floyd-Steinberg dithering.
As mentioned, using fewer neighbouring pixels in the correction would probably lead to cheaper electronics. However, it would also reduce the effectiveness, both by increasing perceptablity in steady state, and increasing the borders of the intensity range where less improvement could be applied.
Unfortunately, this system may not be compatable with Vista's Digital wrongs management system. I'm not going to go in to that.
Please log in.
If you're not logged in,
you can see what this page
looks like, but you will
not be able to add anything.
Description (displayed with the short name and URL.)
||"Blend-in" bad pixels - So one bad pixel will be made to look like (up to) 9 bad pixels - and (just checking) that'll make the image look BETTER/less intolerable?
||No, you just dither around the bad pixel. (Dithering is a technique where you turn an area into a random-looking flurry of little colored pixels that, together, are perceived as the colors you wanted - rather than just picking the closest nearby color that you can display, and sticking with that.)
||So, if the broken pixel is blue, and you're trying to draw something teal, say, the pixels around the broken blue pixel would have to be bright green rather than teal, that is, less blueish and more greenish than in the underlying picture [thanks, marklar]. Looking at it from a distance, it would all sum up to green.
||<pedant>[Jutta] I think you meant slightly less yellow, but anyway we are dealing with light not paint so you have a green pixel (Red + Green = Yellow) </pedant>
||Subliminal Bob Marley, [BrauBeaton]?
||Shirley this would be far more complex, expensive, and prone to problems than just having a back-up screen.
In other words, completely halfbaked!
||I like the fact that you used the word ameliorate. I used it in a work report recently and had an argument with my manager about whether it was appropriate or not. Out came the dictionary, and I was proven right. Then he just said the customer wouldn't understand it and made me change the word anyway.
||As to the idea; bun for the word, the idea sounds overcomplicated, but neutrino has the right idea...
||[+] A good idea, and would seem
||I wonder also if, in some cases, the
whole image could be shifted slightly to
conceal the defect. For instance, if you
display a page of text, there's a good
chance that just shifting the window a
few pixels will place a chosen
(defective) pixel in either a black or a
white area, concealing the defect. This
doesn't solve multiple pixel failures, but
most displays should have only a very
few dud ones.
||If you can pack a battery, some wireless communication and a single pixel display into a device as big as a screen pixel and not much thicker than a sheet of paper, you'd have a stick-on repair pixel which you could stick over the dud pixel. There'd need to be some software setup to get it to display the contents of the right pixel, but after that, your screen would be as good as new, nearly.