 |
 | Home | Contact Us | FAQ | Search & Site Map | Link to Us |
 | Sign In | Join | Other 45 Sites in Network |
| PhotoKB HomeDiscussion Groups Digital Photography Digital PhotoDSLR CamerasZLR CamerasPoint & Shoot CamerasFilm Photography 35 mmLarge FormatMedium formatDarkroomFilm and LabsOther EquipmentPhoto Technique Nature PhotographyPeople PhotographyTechnique GeneralGeneral Photo Topics General TopicsAustralian PhotographyUK PhotographyDirectoryPhoto Clubs |
Photo Forum / Digital Photography / DSLR Cameras / February 2006Hot pixels covered by warranty?
Should I bother asking Canon to change the sensor of my 20D due to the presence of a cluster of hot pixels? It really becomes noticeable at long exposure times (of the order of seconds) in night shots. It sounds like nothing, but it is annoying to find a red blob in a night shot. thanks. > Should I bother asking Canon to change the sensor of my 20D > due to the presence of a cluster of hot pixels? It really becomes > noticeable at long exposure times (of the order of seconds) in > night shots. It sounds like nothing, but it is annoying to find > a red blob in a night shot. Changing the sensor will just move the hot pixels around. Don't waste your time at Canon, do this instead: www.google.com: dark frame subtraction >>Should I bother asking Canon to change the sensor of my 20D >>due to the presence of a cluster of hot pixels? It really becomes [quoted text clipped - 6 lines] > > www.google.com: dark frame subtraction I guess that means that you subtract a "dark" image (obtained by taking a photo with the lens cap on) from the image you want to clean. The problem is that hot pixels become hotter with increased exposure time, so you have to match the exposure times before subtraction, which is a big pain in the a.s. Another problem is that it seems that it is not just an isolated hot pixel, but a cluster (although that could be an artifact--the result of transforming the raw image into jpeg). One isolated pixel I can accept, but not a blob! > I guess that means that you subtract a "dark" image > (obtained by taking a photo with the lens cap on) from the > image you want to clean. The problem is that hot pixels become > hotter with increased exposure time, so you have to match the > exposure times before subtraction, which is a big pain in the > a.s. Well, if you think it'll be easier to hassle Canon, then who am I to stop you? But if you want quality long-exposure images you gotta subtract properly collected and processed darks, even if you manage to get Canon to give you a new sensor. > Another problem is that it seems that it is not just an isolated > hot pixel, but a cluster (although that could be an artifact--the > result of transforming the raw image into jpeg). One isolated > pixel I can accept, but not a blob! When you do the dark-frame subtraction, the computer does the arithmetic per pixel. All 8 million of them, independently. It doesn't matter where the hot pixels are, how they are clustered, etc. > When you do the dark-frame subtraction, the computer does the > arithmetic per pixel. All 8 million of them, independently. It > doesn't matter where the hot pixels are, how they are clustered, etc. Dark frame subtraction will set the hot pixels to 0. With or without it, N scattered hot pixels affect the image less than N hot pixels grouped in a blob. > > When you do the dark-frame subtraction, the computer does the > > arithmetic per pixel. All 8 million of them, independently. It > > doesn't matter where the hot pixels are, how they are clustered, etc. > > Dark frame subtraction will set the hot pixels to 0. image(hot pixel value) = thermal noise + signal dark(hot pixel value) = thermal noise image(hot pixel value) - dark(hot pixel value) = signal You will only get a hole if the dark hot pixel is saturated, and you get less signal if the image hot pixel is saturated. I'd recommend you actually try the procedure, but you appear to be dead-set to hassling Canon. Carry on! > image(hot pixel value) = thermal noise + signal > dark(hot pixel value) = thermal noise > > image(hot pixel value) - dark(hot pixel value) = signal This linear relationship does not hold. Hot pixels are already saturated. I'll give it a try though, as I dont hold any hope in Canon. >This linear relationship does not hold. Hot pixels are already >saturated. Huh? The RAW value of a hot pixel grows first above the average noise, and then increases as the exposure time is increased (or the ISO is raised). It goes through the whole range of potential RAW values before clipping at 4095, and reaching 4095 usually requires a pretty long exposure. The phenomenon is relative to global noise, and to signal.  Signature<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<> John P Sheehy <JPS@no.komm> ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>>< >>This linear relationship does not hold. Hot pixels are already >>saturated. [quoted text clipped - 4 lines] > clipping at 4095, and reaching 4095 usually requires a pretty long > exposure. The phenomenon is relative to global noise, and to signal. Even if they're not saturated, I doubt that the relationship is very linear. >> Huh? The RAW value of a hot pixel grows first above the average noise, >> and then increases as the exposure time is increased (or the ISO is >> raised). It goes through the whole range of potential RAW values before >> clipping at 4095, and reaching 4095 usually requires a pretty long >> exposure. The phenomenon is relative to global noise, and to signal. >Even if they're not saturated, I doubt that the relationship is >very linear. Do you mean that you believe even without clipping, signal alone plus hot-effect alone is not equal to the combined effect (disregarding random noise, of course)? Signature<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<> John P Sheehy <JPS@no.komm> ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>>< > Do you mean that you believe even without clipping, signal alone plus > hot-effect alone is not equal to the combined effect (disregarding > random noise, of course)? Right; the value of a hot pixel in an image is not equal to the value of the hot pixel in a "dark" control image + the correct pixel intensity in the image. In other words hot pixels are deffective; they dont merely have a different "zero" than other pixels. >> Do you mean that you believe even without clipping, signal alone plus >> hot-effect alone is not equal to the combined effect (disregarding [quoted text clipped - 6 lines] >In other words hot pixels are deffective; they dont merely have a different >"zero" than other pixels. Well, that calls for some experiments, when I get the time. On the theoretical side, I would think that you should be able to profile all the pixels in a camera, at a given point in its life, and keep a record of blackpoint offsets (compared to average) for pixels and lines, and the individual sensitivities, and make a RAW bitmap at a higher bitdepth based on the *known* response of individual pixels. The only wild card would be the random elements. Signature<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<> John P Sheehy <JPS@no.komm> ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>>< Can we possible see an example of this "cluster" of hot pixels? > In other words hot pixels are deffective; they dont merely have a different > "zero" than other pixels. If the pixels work fine at brief exposures and grow more visible at longer exposures, they are "hot": charges are leaking into the well, and dark-frame subtraction should handle the matter fine (saturation notwithstanding). If the pixels are "defective", they just sit there at the same value no matter what the exposure or subject: the "erasure" is noted. A simple series of exposures will be informative: dark frames, start at 1s and go until about 2 minutes (or whatever your typical exposure is) in 1 stop increments. Look at the raw data for both your question pixels and what you deem to be ok pixels. I guess the other concern must be "Will the issue get any worse" - possibly leaving you in a situation with an unusable camera, and out of warranty. > Should I bother asking Canon to change the sensor of my 20D > due to the presence of a cluster of hot pixels? It really becomes > noticeable at long exposure times (of the order of seconds) in > night shots. It sounds like nothing, but it is annoying to find > a red blob in a night shot. First, manufacturers won't replace the sensor if it has hot pixels; they will simply reprogram the camera to ignore the pixels. Second, if hot pixels are showing up only on long time exposures, that's normal and there's nothing wrong to begin with. It's the nature of digital photography. > Second, if hot pixels are showing up only on long time exposures, that's > normal and there's nothing wrong to begin with. It's the nature of digital > photography. For me, the keyword was "cluster" of hot pixels - not a "number" of hot pixels. That bit's definately not normal. >Should I bother asking Canon to change the sensor of my 20D >due to the presence of a cluster of hot pixels? It really becomes >noticeable at long exposure times (of the order of seconds) in >night shots. It sounds like nothing, but it is annoying to find >a red blob in a night shot. Why not just call Canon and ask them? I've found there service techs to be very good. I would ask them. I would ask them about anything thought wasn't right. ********************************************************* "It looked like the sort of book described in library catalogues as "slightly foxed", although it would be more honest to admit that it looked as though it had been badgered, wolved and possibly beared as well." _Light Fantastic_ Terry Pratchett |
Reply to this Message |
 Sign In
 Join
 My Latest Posts My Monitored Threads My Blog My Photo Gallery My Profile My Homepage Start New Thread Enable EMail Alerts Rate this Thread
|
©2009 Advenet LLC Privacy Policy - Terms of Use
This website includes both content owned or controlled by Advenet as well as content owned or controlled by third parties.