>John Doe wrote:
>> Nope, not on the 20D and as far as I am concerned that type of feature is
>> only a excuse for the manufacturer to do less quality control, testing and
>> for not buying better quality sensors.
>It is likely that a 0 defect policy for over dead/stuck sensor pixels in over 8M
>would result in such a low yield as to make the product unafordable. I may be
>wrong, and I bet that Canon would be quite tight lipped about it in any case.
>I don't know if they have a 'dead pixel' spec for how many dead sensor pixels
>are acceptable at production. In fact, it would not surprise me at all that
>they do have a threshold and that they do detect dead pixels in the factory
>tests. In turn have them mapped in memory and use that to correct every image
>taken. If so, it's likely a very small number of dead sensors, like 10, such
>that it would never affect an image enough to be noticeable after the correction
>Given that these are spatially seperate R,G,B sensors, the loss of one of them
>won't affect the combined RGB pixel when it is computed, using data from the
>pixels around it in the interpolation.
> From there, recognizing that sensor sites might die a year or 5 or 10 years
>later, there should be the means to map and correct these as they occur. There
>is a procedure in some Nikons that actually "fixes" some dead sensors (getting
>them out of the 'stuck' state. If this applies to the CMOS sensors in Canon's,
>I don't know.)
>[url]http://www.imaging-resource.com/NEWS/998087928.html[/url] is Olympus' response to the
>(Hard disks with dead sectors come to mind... in the 10 MB disk days we would
>would moan over a single dead sector in a hard drive. The cost/bit was too
>high. Of course this didn't affect the quality of a doent.)
>When film is returned it is not uncommon to have tiny scratches or bits of
>embedded dust in the emulsion. Not every frame, but 1 or 2 frames per roll.
>These are easilly corrected with ICE equipped scanners or a little handiwork in PS.