Red spots on images with Canon 5DM2

Discussion in 'Canon EOS' started by john_forney|1, Apr 28, 2012.

  1. I have a problem with my Canon 5D Mark II. I was doing some astrophotography and long exposure pictures of star trails. After downloading them I noticed that there were stationary red, blue, green, and white dots. The more noticeable red ones were in the same place no matter where I had the camera pointed in the sky.
    I went back and looked at previous pictures I took, including a 5 second exposure portrait of someone sitting in a room with natural light, and when I enlarged the image the red dots were also there in the same place. I ran some tests and the red dots become brighter and more noticeable the longer the exposure and are consistently in the same place. They are even faintly visible at 1/40 to 1/50 of a second. Cleaning the lens had no effect.
    What can I do about this and what may have caused it?
     
  2. They are "hot pixels". They are normal and not considered a "defect".
    They can be removed by dark image subtraction or there are software packages that can be used to remove them. Alternatively you can clone them out or better still, stop looking for them at 100% magnification on your monitor. It's very unlikely you would ever notice them in a print.
    They are caused by things like leakage current. They are defects in the sense that they are not supposed to do that, but all sensors have them. You don't get 20 million perfect pixels unless you are very, very lucky. All sensors have most of their more serious "problem pixels" mapped out by the camera maker so you never actually see them.
     
  3. I may be wrong but I believe you can program the camera to map them out. You'll have to check that though. I have the 5D2 and mine has some annoying hot pixels too, I just haven't been bothered to try and map them out.
    The only real problem with hot pixels is if you shoot serious video work. Then it can become a real pain as you can't clone them out.
     
  4. try PixelFixer (http://www.pixelfixer.org). Take a dark frame to map the pixels and then run it on your raw files.
     
  5. If it's still under warranty, Canon will remap the sensor for you free of charge. If not, I sure they'll do it for some cost.
    If you look back through your archives, can you tell when it started? Or has it been clearly present since you owned it, and simply didn't notice it until now?
     
  6. You might have either or both of two issues:
    1. "Hot pixels" can be found on many DSLR cameras. Your raw conversion software will often automatically map them out after you run a few exposures through the software and you'll never notice them again unless you open your raw files in a very different conversion application.
    2. You may be dealing with hot pixels that appear with longer exposures, such as the many-minute-long exposures that we often use to do star trail photographs. Here there are two solutions. The in-camera solution is to enable the "long exposure noise reduction" (a.k.a. "LENR") setting on your camera. With this enabled, each regular exposure will automatically be followed by a "dark frame" exposure of equal length during which the camera essentially creates a photo that contains only the noise and hot pixel data, which it then subtracts from the actual exposure to reduce/eliminate the hot spots. The downside is that each exposure now will require twice as much time - your 30 second image exposure will be followed by the automatic 30 second dark frame exposure. The upside is that for very long exposures your image quality will be great deal better. (There are some semi-work arounds that use a single dark frame exposure to map the hot pixels and then subtract them from multiple exposures in post.) The second solution is to create your star trails in an alternate way. Instead of making a single very long exposure, you make an automatic series of may shorter exposures and then use on of several possible methods of automatically assembling them to produce smooth start trails.
    Dan
     
  7. If they are true hot pixels, not just dark frame noise, they can be mapped out with a little known set of commands. This is from several sites but I believe this link is one of the first. It has been confirmed as working on many Canon DSLR's including the 5D MkII.
    1. Remove the lens and put the body cap in place on the camera
    2. Put the camera in to "manual sensor cleaning mode". Press the "Menu" button, then select the middle "yellow wrench" tab, then "Sensor Cleaning" then "Clean Manually". You will hear the click of the mirror coming up (so you *could* be using a swab on the sensor - but don't do that - leave the body cap in place).
    3. Leave the camera in this mode for 30-60 seconds.
    4. Power off the camera off - you should hear another click as the mirror drops down into place
    That was it. Test the camera and see if the dead pixels are gone.
     
  8. Thank you for the replies. I have had the camera since around March 2010. Went I went back to my archived photos I noticed that the most noticeable spot first seemed to appear about three weeks ago. I was not really looking for small defects before. If I send the camera to Canon how much do you think it would cost to remap the sensor and would it really make that much of a difference?
    When I use the long exposure noise reduction under the image command is it better to set it on ‘auto’ or ‘on’. Is this what Bob Atkins refers to as dark image subtraction?
    What is the best method to assemble pictures together to reduce noise (to make longer star trails) as suggested by Bob Mitchell?
    For the method suggested by Scott Ferris would this work if I left a lens on and kept the lens cap on and covered the viewfinder? Some of my things including my body cap have been packed by movers and I may not have access to them for a few weeks.
     
  9. I have a 60D and the process described by Scott also worked without removing the lens.
     
  10. Since you had the same problem when NOT doing astrophotography, I wouldn't expect LENR to make a difference, I'd try Scott's suggestion, which, assuming it works (at resolving the problem manual clean works w/ or w/o the lens mounted), using the same method w/ the lens cap on should have the same result.
     
  11. My hunch is also that LENR is not the solution to the whole problem, though a photographer doing star trail photography with long individual exposures will still want to use LENR in that case.
    Dan
    (By the way, "astrophotography" is not the same thing as star trail photography in my experience. Unless I'm mistaken, "astrophotography" more typically refers to the sort of photography astronomers might do - e.g. shooting through big telescopes, etc. Star trail photography is more under the heading of "night photography.")
     
  12. LENR should not be used when doing star trails with long individual exposures. The reason is LENR takes as long as the exposure, so a 5min exposure takes 10 minutes in total, this means you end up with a locked up camera for 50% of the time and you need to realign each image for the gaps in the trails and is a huge waste of time. LENR works the same on any exposure of the same length though, so if you are using 5 min exposures take one five minute dark frame and subtract it from all the others, this allows you to expose for a much higher percentage of the time, over 99%, this makes post work much easier.
     
  13. I noticed that when I use LENR that the red dot is often replaced by a black dot which may be noticeable on a white background when the image is blown up. I will try Scott’s technique tonight and let you know how it works out.
     
  14. Scott, I think you could afford to be just a bit less absolute in your comments about LENR, or perhaps you misconstrued my "long individual exposures" description to mean a series of long exposures to be assembled in post. Let me clarify:
    If you are doing night photography and making single long exposures above some minimum threshold exposure time*, using LENR will effectively produce a very large reduction in hot pixels and certain types of noise. It most definitely will result in improved image quality. A downside of using the technique is that, indeed, as I pointed out, the camera makes a second "dark frame" exposure after each "real" exposure, thus doubling the time required for each exposure. When I make a photograph that requires a 6 minute exposure, if LENR is activated this exposure will be followed by a 6 minute dark frame exposure - and the total time will be 12 minutes. Whether or not the improvement in image quality is worth the wait is a decision that each photographer will have to make. For me, it is worth it. I console myself by recalling that if I had used film, reciprocity failure would have resulted in the need for an even longer exposure.
    If you are doing night photography - typically, but quite always, star trail photography - and you want to capture images over a long period of time by capturing a series of shorter component exposures and then combining them into a single "long" exposure in post, it is very unlikely that you'll want to use LENR. One of the primary reasons for using the "many short exposures" technique was, in fact, to avoid the need for using LENR in extremely long exposures where, for example, your two-hour shot might be followed by a two hour dark frame exposure! So for very long digital night exposures the combined exposure technique is a great one to use, and LENR is not necessary nor advisable.
    As to how long you can expose before you'll want to use LENR, there is no single objective answer. One factor is that more recent cameras have done a better job of controlling the hot pixel issues that plagued very early DSLRs. On one early 8MP cropped sensor body that I used, I wouldn't even think of trying a 30 second exposure without LENR. Today, some photographers feel that they can get away with a minute or two without LENR on full frame cameras and some are willing to push it further. So there is a subjective component to the choice as well. For me, I usually want the very best captured raw file I can produce, so I use LENR if the exposure will be 30 seconds or longer and sometimes I use it for somewhat shorter exposures. (The wait is not so annoying either with the short exposures.) When I do critical work, especially to produce large prints of night subjects, I do find the effect of hot pixels in my files still. On the other hand, for certain types of images and for less critical output this might matter less.
    The method of taking a single dark frame - which I also mentioned earlier in this thread as an alternative - has some validity, too, and is almost certainly better than not using LENR at all. Downsides are that you must deal with the subtraction in post using somewhat different techniques, and in order for the dark frame image to work correctly the sensor temperature and exposure time must be the same. (The sensor temperature will rise over the course of a series of long exposures.) As to the example of a series of five minute exposures, it seems to me that the odds that you would do this would be pretty low. If you are doing the series approach, in the vast majority of cases you can use much shorter individual exposures (that won't require LENR at all) or a single five minute exposure with a single dark frame.
    Take care,
    Dan
     
  15. I tried the technique suggested by Scott on the first page and it resolved my problem. I tried a few long exposure shots and the red dot was not there. Thank you.
    I also contacted Canon customer service before I tried this. They said it would not harm the camera but were not sure if it would help.
     

Share This Page