Just a quick question as i've never had a dead pixel issue, but think I have one now. I was going through photos I shot on our vacation and noticed a white fleck in one of them. I use Aperture for retouching. I used the loupe to hover over it and it's a perfectly square pixel. What's most disturbing is, since I can leave the loupe in the exact spot, but arrow through my photos, I notice the fleck/speck always remained...that is, in each photo, in the exact same spot, there's a tiny spot of white. What is this? Is this a dead pixel? I've owned a number of digital cams and never had this issue...then suddenely I drop $3.5K on one and get it...I think. Does this sound like a dead pixel to anyone who's dealt with this before? Not sure what I should do. I'm guessing this is a Canon service issue? Thanks for the help.
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
Dead pixel on 5D mark II?
- Thread starter c_smith
- Start date
G
Graham Meale
Guest
thenickdude
Senior Member
All good RAW processors will map out dead pixels completely automatically.
gareth hacon
Well-known member
I have the Nikon D3X and have noticed a small, tiny white spot. I believe that this is also a dead pixel, while so small I reckon it won't shoe on print I contacted Nikon and they have told me to send the unit back for the sensor to be re-mapped?? I believe that these are very unfortunate circumstances and i'm sure it is a one off but try to contact Canon and like Nikon i'm hopeful that they will also fix it for free.
Gareth Hacon
http://www.garethhacon.com
Gareth Hacon
http://www.garethhacon.com
glenn_chapman
New member
My research group has been studying defect development in many DSLR’s now for several years. We have never actually seen a true stuck high (white) pixel in all our cameras but I can probably explain what you are seeing. In all the studies Hot pixels appear to be the main source of sensor defects. But most people do not recognize that hot pixels can appear in short exposure shots as well a long ones because they come in two types.
Regular hot pixels: pixel accumulates signal even in dark exposures
Offset hot pixels: where the pixel has offset value which appears at even the shortest exposure, and an accumulation of value after that.
Currently we find that 30-40% of hot pixels have an offset value, and this offset can be quite high in some cases. We have seen a few offsets pixels with 40% of max (ie 102 out of 255 in 8 bits) in several cameras. Your white pixel at all exposures is probably of that type. You can see a plot of both hot pixel type behavior on our web page
http://isrl.nems.ca/index.php?option=com_content&task=view&id=29&Itemid=45
By the way keep in mind that the hot pixel value is added to by the regular exposure so you can reach saturation (your white condition) with those pixels in modest exposures in some cases. Demosaicing (which is the method to convert the Red, Green Blue pixel pattern into RGB for each pixel) spreads the hot pixel value to adjacent pixels
Now as to your concern about this being a defect in your camera when you purchased it, it is possible but I suspect that the defect developed after your purchase. Every camera we looked it developed hot pixels after a while. The important thing was how many of them were the offset type. Our studies strongly suggest that hot pixels are probably caused by cosmic rays and accumulate over time depending on the imager type (CCD or CMOS) and size. C sized CMOS imagers studied over several cameras seem to accumulate 2-3 hot pixels per year. (see a plot of defect development with time on our web page). We only have two full frame Canon 5DmkII in our study and only one showed some hot pixels. The 2.4 times larger area of the 5DmkII suggests it might experience a higher hot pixel growth rate. Our results are from camera images collected by volunteers so we would be happy to hear from 5DmkII owners willing to participate in the study. If you would like to participate we give you a set of test pictures which we will ana.lyze to tell you more about this defect.
--
Glenn Chapman
Regular hot pixels: pixel accumulates signal even in dark exposures
Offset hot pixels: where the pixel has offset value which appears at even the shortest exposure, and an accumulation of value after that.
Currently we find that 30-40% of hot pixels have an offset value, and this offset can be quite high in some cases. We have seen a few offsets pixels with 40% of max (ie 102 out of 255 in 8 bits) in several cameras. Your white pixel at all exposures is probably of that type. You can see a plot of both hot pixel type behavior on our web page
http://isrl.nems.ca/index.php?option=com_content&task=view&id=29&Itemid=45
By the way keep in mind that the hot pixel value is added to by the regular exposure so you can reach saturation (your white condition) with those pixels in modest exposures in some cases. Demosaicing (which is the method to convert the Red, Green Blue pixel pattern into RGB for each pixel) spreads the hot pixel value to adjacent pixels
Now as to your concern about this being a defect in your camera when you purchased it, it is possible but I suspect that the defect developed after your purchase. Every camera we looked it developed hot pixels after a while. The important thing was how many of them were the offset type. Our studies strongly suggest that hot pixels are probably caused by cosmic rays and accumulate over time depending on the imager type (CCD or CMOS) and size. C sized CMOS imagers studied over several cameras seem to accumulate 2-3 hot pixels per year. (see a plot of defect development with time on our web page). We only have two full frame Canon 5DmkII in our study and only one showed some hot pixels. The 2.4 times larger area of the 5DmkII suggests it might experience a higher hot pixel growth rate. Our results are from camera images collected by volunteers so we would be happy to hear from 5DmkII owners willing to participate in the study. If you would like to participate we give you a set of test pictures which we will ana.lyze to tell you more about this defect.
--
Glenn Chapman
Ah Pek
Leading Member
My 5D2 had a stuck pixel as well. However, there is a way to get the camera to remap it. You just switch to manual clean and then wait for 30s. Like magic the stuck pixel disappears.
--
'The majesticness of that duck is overwhelming!' - Bulbol
--
'The majesticness of that duck is overwhelming!' - Bulbol
pmevel
Member
I noticed one little blue spot yesterday on my 5D M2, but strangely, it was only visible on the LCD, not in LightRoom (Raw file).
Anyway, I did the manual clean trick, and it was gone on the next photo I took.
Every digital camera you have ever owned has had hundreds, possibly thousands, of dead pixels: it is just that the camera maintains a map so it knows where they are and conceals - even on RAW - them before you get a chance to see the image.
Fortunately some manufacturers have built in firmware to detect new dead pixels and update the map. Canon have implemented but, unlike some other manufacturers, not documented this feature on some cameras including both 5D models.
As someone else has mentioned, just select manual clean from the menu and switch the camera off soon after the mirror flips up - all of the hot pixel mapping is done with the shutter closed and the mirror down. No need to wait 30sec, however some have suggested they get better results if the camera is set to high-ish ISO, eg. 1600. I have no evidence that ISO makes any difference, but the fix is so simple it won't hurt to try it again if you don't get success the first time.
--
Its RKM
P
papparazzi
Guest
I have done that to my 5D(i) with no success.... :S
--
Nature and Landscape Photographer
http://www.pbase.com/jdf
Hi all-
Thank you for all the informative feedback. I did notice the manual clean on the sensor, but did not enable it as I wanted to read feedback before doing so. I'll make sure I do that tonight, and will then take a few test shots to confirm. Thanks again.
Thank you for all the informative feedback. I did notice the manual clean on the sensor, but did not enable it as I wanted to read feedback before doing so. I'll make sure I do that tonight, and will then take a few test shots to confirm. Thanks again.
Ear Mountain
Leading Member
I took delivery of my 5DMKII last May. The photos looked great and I saw no problems. Like everyone else though I was concerned about the "white dot" issue. Canon posted a firmware update which I installed according to the directions provided.
When looking at the photos from my next shoot I observed two red dots on each image in the same two places. In looking at 100% I could tell the red spots were one pixel wide each.
I found instructions to map the stuck or dead pixels (whichever they are). That was to remove the lens and install the body cap. Then turn the camera on and point the front down and initiate the manual clean sensor mode for a few seconds followed by turning the camera off. In the next and all subsequent photos taken there were no more red spots. They apparently have been mapped out.
I believe this feature was implemented in the 5DMKII but the original 5D did not have this feature built into its software.
When looking at the photos from my next shoot I observed two red dots on each image in the same two places. In looking at 100% I could tell the red spots were one pixel wide each.
I found instructions to map the stuck or dead pixels (whichever they are). That was to remove the lens and install the body cap. Then turn the camera on and point the front down and initiate the manual clean sensor mode for a few seconds followed by turning the camera off. In the next and all subsequent photos taken there were no more red spots. They apparently have been mapped out.
I believe this feature was implemented in the 5DMKII but the original 5D did not have this feature built into its software.
erikstefan
Leading Member
Lightroom and ACR remove hot pixels Automatically.
You have tested it on Lightroom now test it on ACR,
but in DPP it's still there.
DSPographer
Senior Member
To maximize the mapping out of pixels that are not so bad you can warm up the sensor a bit using live view. I had one pixel that was visible on long exposures only that didn't map out unless I used live view for two minutes just before the manual cleaning trick. Once I warmed up the sensor with live view the camera did map out the hot (or warm) pixel. When I updated the firmware the pixel came back and I had to repeat the live-view followed by manual clean trick to map it out again.
It certainly was implemented on the original 5D, having availed myself of it at least once and a search on this forum will find a large number of others who did the same.
However, the algorithm may have been improved in the MkII, because it was not 100% successful in the Mk1 and hot pixels which only appeared at high ISO or long exposure times were often missed. I expect that is why Canon don't document it, because even in the 5DII it still isn't a guaranteed fix.
--
Its RKM
Ear Mountain
Leading Member
Thanks for that correction. My bad!
I am, admittedly, a pixel-peeper and it serves my personal interests well. I'm only posting my observations here in the interest of further understanding of the phenomenon. The research project mentioned earlier is quite intriguing to me.
In this writing, the word "pixel" and "photosite" will likely be used incorrectly and I don't care much. If you're truely interested in the root cause, then you are probably smart enough to overcome my deficiencies.
My first one appeared today. It's purely coincidental that I caught it on the day it occured. I discovered it by examining a shot I took of my monitor displaying the Copper Hill test photo that they provide as a means to detect dust on the sensor. It presented as red and stuck out like a sore thumb on the light blue test image. I immediately confirmed its existence by taking other photos (solid painted walls and such and some lens-cap-on shots and whatever else I could point to from my recliner...I'm a lazy scientist).
The next thing I did was confirm that it did not exist in images I captured last night.
Then I started to examine the red dot more closely in PS CS3. CS3 does not seem to blend pixels together like some JPG/RAW viewers do when zooming in. At any rate, I zoomed in, in CS3 so I could determine how many pixels the "hot" one was bleeding into....it was far too large on the screen at 100% to be just a single pixel. I'm not sure of how images are rendered at various screen resolutions but, just for reference, I was using 1920x1200. I did not count pixels at other screen resolutions. Of course, it was most intense in the center and faded as it progressed("bled") to surrounding pixels. Interestingly, it radiated farther horizontally than vertically....about 16 pixels at its widest and 9 pixels at its tallest points....and practically identical in every shot....including the shot of the Copper Hill test image.
I was most interested in eleminating it and did so before I had the inclination to test for it in video. For the sake of gaining further insight into which and how many actual photosites are involved in video capture, this is regrettable. I'll have some more fun if another one turns up later.
The camera's Manual Cleaning mode completely eleminated(or masked) all traces of the troubled photosite in less than maybe 10 seconds. Of note, there are no visually detectable artifacts or other evidences at or near the location of the "hot" pixel. This lends evidence to the notion that it is likely confined to a single photosite(or connected circuitry) per instance and can be removed with no noticable effect....and the sites that were being bled on just go back to normal operation and performance. My hope is that more instances do not develop nearby, over time, in sufficient numbers to produce a noticable artifact after removal. I think the odds are fairly in our favor here.
I would love to know exactly what is the cause and what the camera's Manual Cleaning routine is doing to remedy the problem. Any pointers to technical facts will be greatly appreciated.
For those who are likewise interested in this phenomenon, I will update if additional pixels become unruly in the future.
I am indescribably thankful that these little over-anxious color makers are managable by the end users....in many, if not most cases. Every product of nanoscale has some latent issues and I'm thrilled that Canon had the foresight to help us easily deal with them....even if it exclusively serves their own interest. It's the rare game with two winners. And I bet I'm happer now than the poor schmuck that had to show up in the design review and say we have a problem.
It's an alarming experience the first time. Just try not to panic and then read as many related articles as you can.
Finally.....I'm sorry fellas'...and ladies. I'm practically never this chatty but when this happed, my nerd gland swelled up and was mashing on my talking gland and all this is what squirted out. I'll look for a nice ointment or a pill for next time.
Good luck...and have some fun along the way.
Oh yeah....the one question I wanted to ask is of you guys who already have some of this research under your belt. Is it even remotely possible that taking the picture of my very bright 24" LCD dispay could have triggered the hot dot?
In all seriousness, the picture I took of my montior(to check for dust no less) is the very first image that the red dot appered in.
In this writing, the word "pixel" and "photosite" will likely be used incorrectly and I don't care much. If you're truely interested in the root cause, then you are probably smart enough to overcome my deficiencies.
My first one appeared today. It's purely coincidental that I caught it on the day it occured. I discovered it by examining a shot I took of my monitor displaying the Copper Hill test photo that they provide as a means to detect dust on the sensor. It presented as red and stuck out like a sore thumb on the light blue test image. I immediately confirmed its existence by taking other photos (solid painted walls and such and some lens-cap-on shots and whatever else I could point to from my recliner...I'm a lazy scientist).
The next thing I did was confirm that it did not exist in images I captured last night.
Then I started to examine the red dot more closely in PS CS3. CS3 does not seem to blend pixels together like some JPG/RAW viewers do when zooming in. At any rate, I zoomed in, in CS3 so I could determine how many pixels the "hot" one was bleeding into....it was far too large on the screen at 100% to be just a single pixel. I'm not sure of how images are rendered at various screen resolutions but, just for reference, I was using 1920x1200. I did not count pixels at other screen resolutions. Of course, it was most intense in the center and faded as it progressed("bled") to surrounding pixels. Interestingly, it radiated farther horizontally than vertically....about 16 pixels at its widest and 9 pixels at its tallest points....and practically identical in every shot....including the shot of the Copper Hill test image.
I was most interested in eleminating it and did so before I had the inclination to test for it in video. For the sake of gaining further insight into which and how many actual photosites are involved in video capture, this is regrettable. I'll have some more fun if another one turns up later.
The camera's Manual Cleaning mode completely eleminated(or masked) all traces of the troubled photosite in less than maybe 10 seconds. Of note, there are no visually detectable artifacts or other evidences at or near the location of the "hot" pixel. This lends evidence to the notion that it is likely confined to a single photosite(or connected circuitry) per instance and can be removed with no noticable effect....and the sites that were being bled on just go back to normal operation and performance. My hope is that more instances do not develop nearby, over time, in sufficient numbers to produce a noticable artifact after removal. I think the odds are fairly in our favor here.
I would love to know exactly what is the cause and what the camera's Manual Cleaning routine is doing to remedy the problem. Any pointers to technical facts will be greatly appreciated.
For those who are likewise interested in this phenomenon, I will update if additional pixels become unruly in the future.
I am indescribably thankful that these little over-anxious color makers are managable by the end users....in many, if not most cases. Every product of nanoscale has some latent issues and I'm thrilled that Canon had the foresight to help us easily deal with them....even if it exclusively serves their own interest. It's the rare game with two winners. And I bet I'm happer now than the poor schmuck that had to show up in the design review and say we have a problem.
It's an alarming experience the first time. Just try not to panic and then read as many related articles as you can.
Finally.....I'm sorry fellas'...and ladies. I'm practically never this chatty but when this happed, my nerd gland swelled up and was mashing on my talking gland and all this is what squirted out. I'll look for a nice ointment or a pill for next time.
Good luck...and have some fun along the way.
Oh yeah....the one question I wanted to ask is of you guys who already have some of this research under your belt. Is it even remotely possible that taking the picture of my very bright 24" LCD dispay could have triggered the hot dot?
In all seriousness, the picture I took of my montior(to check for dust no less) is the very first image that the red dot appered in.
Just an update regarding my own question:
I have since taken 11 more pictures of the Copper Hill test screen at various apertures and ISO's up to 6400. I think it was a pretty goofy idea that photographing the screen actually caused the red dot to show up. Pretty much just grasping at straws and now think it was purely coincidental.
No stuck/hot/red pixels after the additional test shots.
kdt
l-------------------------------------------------------
"Is it even remotely possible that taking the picture of my very bright 24" LCD dispay could have triggered the hot dot?"
I have since taken 11 more pictures of the Copper Hill test screen at various apertures and ISO's up to 6400. I think it was a pretty goofy idea that photographing the screen actually caused the red dot to show up. Pretty much just grasping at straws and now think it was purely coincidental.
No stuck/hot/red pixels after the additional test shots.
kdt
l-------------------------------------------------------
"Is it even remotely possible that taking the picture of my very bright 24" LCD dispay could have triggered the hot dot?"
Similar threads
Keyboard shortcuts
- f
- Forum