5d mark II strange black dots?

Discussion in 'Canon EOS' started by ivan_gunduli_, Dec 1, 2008.

  1. I have seen few of high ISO night shots with new 5D mark II from various sources and it seams to me that there some very strange picture quality issues. On night cityscape shots strange black dots appear around light sources. That thing would not bother me if the problem is specific for high ISO values but you can see those black dots even at 100 ISO. Can anyone confirm that issue. I think no matter what camera settings are, those kind of things should not happen. I have provided link bellow where you can see /glubsch.wordpress.com/>http://glubsch.wordpress.com/ (scroll down on night city shots)
     
  2. Maybe my eyes are failing, but I checked 100 & 1600 and do not see black dots. I do hope the fog had rolled in before you took those.
     
  3. I see what you're talking about: the black dots to the right of the lights running up and down the buildings. Wow. Much more noticeable in the higher ISO shots. This doesn't look like a user error to me. Did you shoot JPEG or RAW? If RAW, which RAW converter did you use?
     
  4. I can see in your blog that you used ACR 5.2 with PS-CS4. Did you also shoot JPEG? If so, do they have the same problem? Does conversion with the Canon raw converter utility show the same issue?
     
  5. Seems very obvious to me, totally bizarre, and totally unacceptable. I wouldn't expect that it would show up so much more prominently at higher ISOs if it was a RAW converter issue, but I guess it's possible. If the same thing showed up in camera JPEGs, I'd say it's got to be a hardware problem.
     
  6. If it's in the camera, hopefully it's a firmware problem, not a hardware problem... Since the black dots appear right after the bright areas I would guess that there is a bug in some compression/decompression code somewhere. Did you have highlight recovery turned on? It's conceivable that there is an interaction.
     
  7. "Ronald Moravec Maybe my eyes are failing, but I checked 100 & 1600 and do not see black dots. I do hope the fog had rolled in before you took those." Scroll down and look at night city shots. Blog is not mine I stumbled on it. I just noticed that error and have same informations as you guys.Thing is that I have seen few other shoots from other people (some of them on dpreview but I can't find them now) in similar shot conditions and they all show that kind of errors. I think it's some kind of in camera processing glitch but I'm not sure. Since I'm planing to by 5d mark II myself I was wandering has anyone else encountered same problems and maybe somebody knows why is that happening?
     
  8. You can see them in the bridge shots as well. Disappointing.
     
  9. At the risk of telling the Emperor he has no clothes on but where I come from they are called windows....... Now we all knew this stupidity would start but listen guys, you are pixel peeping other peoples stuff at web resolutions, for goodness sake at least talk about your own pictures and any issues you might have with them. This is just crazy. Take care, Scott.
     
  10. Normally I find that your opinions tend to gel with mine, but I'm afraid I disagree with you here.

    The small images on that page link to full sized versions - here, for instance - and the black dots along the edges of the buildings are clearly unacceptable...
     
  11. Nothing stupid or crazy about this one Scott. Take a look again at full resolution. A very real problem.
     
  12. Close up
    00RfCJ-93895584.jpg
     
  13. I don't see the problem in these night shots, which were JPEG encoded in the camera: http://flickr.com/photos/akihabaranews/2960657187/sizes/l/in/set-72157608231370176/ That leads me to suspect a problem in the RAW converter used in this example.
     
  14. Sorry Kieth, Ivan and Dan, The Emperor is clothed in his finest, I didn't link to the full sized images before commenting, (not unusual for my big mouth :) ) It does appear to be a strange issue and down to individual pixel level. Take care all, Scott.
     
  15. Funny how it only appears in a series of what seem to be city street lighting, mainly in the area of the provided crop. I wonder if there might be something physical about the lights that is being represented by that black pixel. I live in SF, have no real desire to go to work on time tomorrow, so will take the train downtown and check out the lights. I've scanned the photo at almost all ISO's and only found this in the presented crop, another area with similar lighting, and a few isolated places after that. I am sure everyone will be eagerly anticipating my report.
     
  16. Daniel- If you're talking about the crop above, you've got to take another look at the full pictures. It's all over the place, especially where the lights are running up the building edges. It's hard *not* to see it everywhere. And like I said earlier, it's also in the bridge photos, both ISO's (worse as ISO goes higher, but still there at 100). Look to the side of any of the lights- it's there.
     
  17. "Upon further examination", you're right. I see it in the bridge lights. Seems to be mainly on the right hand side. So, as a 30D portrait shooter, how much difference does this make? I wouldn't crop this far on shots this big without expecting something. BTW, I really did think it could have been something physically about the lights. The Embarcadero in SF has some pretty artsy architechture.
     
  18. "I don't see the problem in these night shots, which were JPEG encoded in the camera: (link) That leads me to suspect a problem in the RAW converter used in this example." They are all over that picture: on the left bottom corner you can see stripes of black pixels. They are harder to notice because of in camera noise reduction but they are still visible. On the lights in the restaurant (right from the entrance)
     
  19. You're right. Many are more subtle since they don't stand out so well against the generally darker buildings, but they're definitely there. If that's a camera JPEG and from a different camera, it's worrisome.
     
  20. I didn't realize my photos would create such an interest. I signed up with Photo.net just to participate in this thread :) Let me answer some of the questions here, even if they were already answered in my blog or somewhere else. The shots were taken in RAW format only. These spots show up even at ISO 100, though not as black holes. I've turned off all in-camera post-processing operations (though I don't think they apply on RAW shots anyway). My picture settings were set to "neutral" and Highlight Recovery was turned off. I Used ACR 5.2 in PhotoShop CS4 (trial version). I can see the same black spots using DPP (version 3.5.1.0) that came along with the camera. Scott, I'm happy for others to pixel peep in my photos. That's why I posted them, albeit for demonstrating the camera's great noise performance, not to talk about the black dots :) I took a shot of the Golden Gate Bridge last Saturday during the golden hour. The photos were taken at ISO 100. Unfortunately the black holes are quite visible in that photo. It's early in the morning right now, but I hope to add another blog entry with an example later today. Since I set the camera to save RAW+JPG, I will be able to see if the spots occur there too. I wish I knew who to contact at Canon to make them aware of this camera behavior. Not just an email that will get absorbed in the bowls of the corporate monster, but really to someone who's a subject expert. So if you know the right person, post it here. If you are hesitant to leave the contact info here in this thread, leave me a comment on my blog at glubsch.wordpress.com. I won't publish it (I have to approve all comments before they become public).
     
  21. looks like something to do with the blownout highlights and instead camera is converting pure white to pure black as if the highlight meter was on and a screen capture was taken displaying the highlight warning. I'm getting my 5d2 tomorrow and hope this does not happen with mine or i'll be pissed
     
  22. Ivan, nice photos, looks like you had fun in SF, which RAW converter are you using? try Canon DPP and if you see the black dots there is a hardware issue with sensor readout, if so, send it back to dealer for a replacement.
     
  23. Ivan isn't the photographer; Stephan Hoerold above is, and he already said it also appears in DPP. And if you look at the example Ken linked to above, or some of the threads on dpreview, it doesn't appear to be an isolated case.
     
  24. Stephan, Sorry I didn't realize it was your photos, do you have the long exposure NR set to ON or auto by any chance (this is different from high ISO NR)? There might be a bug with dark frame subtraction...set this option to OFF and try taking a 30+ sec exposure with the lens cap on and see what happens.
     
  25. Can somebody with 5d mark II post some fresh samples in similar lighting conditions please. I really want to know is that kind of problem noticed on all new 5d mark II cameras or only on few of them.
     
  26. Same issue here - more noticible on the right edge of the frame: http://www.usa.canon.com/uploadedimages/FCK/Image/REVERIE%20Still%20Photos/ Vincent_Laforet_5D_MKII_MG_0403.jpg
     
  27. Well, what do you know. The 30D has the same problem. I never noticed it before, but if you look for it, there it is. Maybe not as much, but definitely visible at 100% magnification at the borders between blown-out bright areas and surrounding shadows.
     
  28. Ken, Can you post a sample image from 30D that has this artifact? I have never noticed this with either the 20D or the 40D in similar situations, there are two 30+ sec exposures of Golden Gate bridge taken with 20D and 40D in my gallery with plenty of blown up lights but they don't show this artifact. This looks like a hardware issue in which saturated pixels can't hold charge, charge leakes out and the pixels are reset and read as zero. My concern is that this turns out to be like the MKIII AF fiasco in which cameras need to be returned and so on....
     
  29. Here's my example. After comparing it with those from the 5D2, I'm thinking that they are not the same phenomenon. On these, you can see a black edge between blown out areas and adjacent dark areas, but they are not always to the right of the blown out areas as they are in the 5D2 examples. Also, the black edge is quite different from the block-like black spot, which looks bigger than a single pixel, but it sure is hard to tell.
    00Rh2Q-94841684.jpg
     
  30. Thanks Ken, to me, your sample only shows some noise, CA and oversharpened edges, I don't see any systematic artifact like the ones in 5DII images.
     
  31. Could this be part of the sharpening process, if these images were sharpened?
     
  32. "Here same problem. Hmmmmm!" Hmmmmm, where?
     
  33. ^^ On the bridge lights above the couple to the right.
     
  34. I think I see what you mean on this one, though it seems more subtle than the other examples...
     
  35. Gleaned from trolling various forums: it seems to be pretty common (maybe even universal?); the problem appears in the RAW data, so it has nothing to do with converters, sharpening, etc.; one person said that, based on a side-by-side shoot, his 1Dsiii and 5D do not exhibit the problem, but his 5Dii does.
     
  36. I'm with Thomas on that photo on the Canon site, it does not show the issue being reported. This is of concern though; I have a 5D mark II on order. So, I would be very interested in hearing from more users of the 5D mark II. That is first hand experience. However, this thread is the only place that I have seen any evidence of an issue. So, I'm reserving judgment at this point. If it is a problem with the camera, then I would expect Canon to fix it.
     
  37. I have Photoshop CS4 Extended with the new ACR and I'm busting at the seams to work with a 5D2 RAW file. I'd like to see what artifacts and problems I could get to show up. I know I'm probably asking a lot, since that's basically an original file, I just have no 5D2 yet. It can be anything, a photo of the dumpster in your apartment complex, whatever, especially if a streetlight is nearby. I'd be happy to post all results and findings. I'd forward them onto Canon as well.

    Thanks,
    Paul
     
  38. dpreview forums have a useful discussion here: http://forums.dpreview.com/forums/read.asp?forum=1032&message=30222755 In particular, there is this screenshot of RAW data analysis at http://www.panopeeper.com/Demo/Canon5DMkII_BlackPixels.GIF
     
  39. .kaa, Interesting, is the RGB value of the "dead pixel" below or equal to the black level?
     
  40. the raw is about 30mb and my email only allows like 10mb. do you have a ftp site?
     
  41. Very interresting. I was curious to see if my 50D could have the same thing, with this JPEG file, it doesn't seem so.
    00Rhj8-95119684.jpg
     
  42. I have been using the 5DMark2 for well over a week now and have yet to see the black dot issue that's been surfacing recently. Mind you I have been shooting JPG as I didn't have the extra workflow time required to convert to DNG and have to upgrade to CS4... call it laziness to install DPP maybe. This issue however is making me curious.... Have been doing some extensive tests at night - running the full gambit of ISO and seriously have not seen anything remotely close to what's been appearing. I will go out tonight and try shooting some bright/dark scenes RAW to test it out however. This is the flickr link to all the test shots with links to view in full resolution as posted. http://www.flickr.com/gp/79642675@N00/U5x5vZ They're nothing special - just test night shots in a tow truck pound as well as some test video.
     
  43. Topher, I hate to say it but your images do actually show this artifact.
    00Rhwm-95219784.jpg
     
  44. well holy bat guano in july - i'd been looking 100% at the brighter spots, not the smaller ones between the trailers. take a look at the 400ISO posted shot i have - it's very prevalent. that's native JPG by the way, unprocessed straight from the DCIM folder. oh this is gonna be interesting....
     
  45. I repeat a request from Paul DeCesare: Can anyoane upload a Raw file with this issue?
     
  46. Here's a link to a RAW file showing the issue. http://www.jepsonphotography.co.uk/misc/20081129_220756_0073.CR2
     
  47. Thanks Nigel for posting the RAW file, here is a crop from DPP conversion with all parameters set to zero, the RGB value for the black pixels is below the black level meaning that the camera's ADC sees them as "zero"! Unfortunately this looks like a hardware issue so I am not sure how they can fix it but it is possible to add a routine to the converter to detect these false zeros and replace them with nearest neighbor values (albite this might be bad for people who like to do astrophotography!) Thanks for sharing and congradulations on your new camera.
    00RiFi-95347684.jpg
     
  48. and here's a screen shot with RGB reading
     
  49. and here's a screen shot with RGB reading
     
  50. looks like new photo.net interface is buggier than what it was before! I can't attach the photo because is larger than 700 pixels! any ways, the reading for the black spot is (2,4,16) also this photo was taken in portrait orientation so the black spots are now on top
     
  51. and here is a conversion with ACR.
    00RiFu-95349684.jpg
     
  52. Whilst at massive enlargements there is obviously an issue here, on a note of sanity, has anybody seen a picture at normal viewing magnifications where this is obvious and detracts from the image?
    Seems to me another storm in a tea cup so far. People seem happy with their camera and pictures until they are being told they aren't! Yes it would obviously be better if this didn't happen but come on, lets start seeing some pictures at normal viewing conditions where image quality is degraded by this, at the level being studied so far the black dots don't equate to much difference to individual grain particles when using film.
    As an academic pursuit to establish why this is happening then it has some merit, maybe, but so far, as an image quality issue there is nothing.
    Take care, Scott.
     
  53. Scott- I would say that architectural photographers will have a really rough time of it. Very detail-oriented, lots of point sources of light. I would guarantee that some of the folks i have worked for in that industry would return the camera immediately if this were a hardware issue and not a firmware fix.
    (of course, they're already shooting on the 1dsmkIII, but still. It wouldn't become a backup body.)
    As someone who is looking to the 5dmkII to replace my hasselblad, I'm really disappointed. I don't need a perfect camera or ueber resolution, but I do need no obvious defects and the ability to print 20x24, and these dots would mean i was constantly worrying about missing something in the retouching process. It also is a bit of a confidence-buster, making one worry about 'what else will go wrong', though that seems a little silly. (once you have a 4x5 lens misfire a couple times, you're always worrying about it, and will probably end up replacing it, even if it works 98% of the time).
    imho. anyway, it has me really grumpy and very hesitantly postponing my purchase until next fiscal year.
     

  54. Brad, I don't think there is too much to worry about on the retouching process. A little testing under photoshop shows that it's fairly easy to create an action which removes the black holes...
    1. Color select pure white with a 58 tolerance
    2. Move selection right (or up) by 4 pixels
    3. Remove light shades from selection
    4. Dust removal filter
    So if mine has it, I'll live with it.
     
  55. @ Scott
    I would say the images from the original post are that bad. The lit edges of the buildings in the center and left of center look terrible at what I would think is a normal viewing magnification. http://glubsch.files.wordpress.com/2008/11/_mg_0167.jpg
     
  56. i took the 5Dm2 and a 400 f/5.6 prime to the waterfront here in toronto to shoot the city's skyline to test this issue a bit further. being -10 degrees celcius on the lake, it was a little chilly to say the least but did manage to knock off a few tests both JPG and RAW. I've yet to examine the raw files in depth as I need to still install DPP for the 5dm2 or ACR update (on the list of things to do today) but from the JPG's I can determine that the problem starts surfacing at ISO400. It's "slightly" evident at ISO100 - not there at all ISO50....
    this is a super close crop from ISO400... trivial maybe but still to me indicates an issue.
    [​IMG]
    For the full shot at different ISOs go here:
    http://www.flickr.com/gp/79642675@N00/U5x5vZ

    Interestingly, its evident on one ISO3200 shot but the following ISO3200 with high ISO noise redux set to ON it was not evident. JPG's have been posted full res, untouched direct from the DCIM folder with links to download the full res shot. The link above will take you to a private set on flickr - hover your mouse over the skyline shots which will identify what ISO they are.

    also this astro photog has reproduced it here:
    http://www.flickr.com/photos/50686373@N00/3085719227/
    and more seriously with this shot on a car:
    http://www.flickr.com/photos/50686373@N00/3087076142/in/photostream/
     
  57. [[The lit edges of the buildings in the center and left of center look terrible at what I would think is a normal viewing magnification.]]
    Has anyone here actually printed one of these "problem" images? I see lots of wailing and gnashing of teeth but has anyone actually, you know, used these images in anything beyond 100% and 200% viewing on web pages.
     
  58. Wonder when someone will email Canon to see what they say.
     
  59. Bruce, the answer from Canon is here.
     
  60. Thanks Xavier, I hadn't seen their response.
    What do you guys think? Are you satisfied it is a simple firmware isssue?
    Guess we will see.
    I don't have the 5D MkII yet so will be watching this issue closely.
     
  61. Rob: I've printed some of the images posted, and on an A3 print the black dots are clearly visible. They would render certain images unusable, in my opinion.
     
  62. Brad,
    You have to be kidding, a $2,500 prosumer 35mm body to replace a Hasselblad system where you are working to individual pixel level, you say your not concerned with uber resolution but that is what this thread is all about. I think your expectations for the 5D MkII were way too high. From a pros point of view, as Xavier points out, there is a very simple auto action you can use to eliminate it and that takes virtualy no time. It certainly wouldn't stop me buying one if it fitted my buisness plan.
    Don't get me wrong, yes the camera needs a firmware update to eliminate this issue, but lets keep it in perspective, a few shots, in a few conditions, from a few cameras, render some pixels in need of an auto adjustment that is very easy to do.
    Mike and Micheal, run Xavier's action and tell me the picture is unusable. How many shots do you print that need no alterations or adjustments? So far I just don't see this as a big issue, slightly annoying, yes, but particularly when put in the context of a full blown, new, highly complicated electronic item launch, really not a big issue.
    Take care, Scott.
     
  63. Scott-
    Sorry, I should be more specific. I love the hasselblad, but there are several projects i'm working on that need things it just doesn't do (high ISO, more inconspicuous appearance, AF, AE, etc.).
    Yes, I want high resolution (I guess printing 20x24 is high resolution) but I meant that I'm not pretending this camera is something it ain't- I know it doesn't _have_ ueber resolution- it isn't a leaf back, and any image won't be MF/LF quality, but I still have the 4x5 for that, while I'm selling the 'blad. I don't have the camera, so I admit I haven't tried the fixes that are mentioned, and it sounds like they may fix this issue fairly painlessly for those of us with a workflow that involves a decent amount of post-production already, all of which is good.
    But I'm less disappointed by this specific flaw, and more worried that this is a fairly blatant mistake, and that I should really wait until they've had a bit of time to work through the v1.0 bugs. I would like to be more confident in the camera while I'm working with it, but this all makes me a bit nervous, and I'd rather put off the projects rather than begin with technology that I'm nervous about.
    (Of course, at this point, perhaps i'm just sour graping, since availability is so limited. grin.)
     
  64. Hi Scott. Yes, the action helps, but it didn't remove all the dots on some of the images I ran it on. On one of my test images, for example, it failed to remove about 50% of the dots. Not sure why; perhaps because some of the lights with adjacent dots were not entirely blown out (i.e. white), but retained some yellow.
    Of course for most images the dreaded dots won't be a problem. However, for some images they will.
    I hope that a firmware update will be possible, and that a recall won't be required, as it was with the 24 -105 f4 L lens flare issue.
    I really must stop being an early adopter!
     
  65. Scott, Is there a reason why you are so insistant that I feel the same way you do about this? If I bought a TV and found out it didn't have a power button, I wouldn't keep it just because I can still turn it on ond off with the remote. If you are ok taking the extra\alternate steps to work around the problem, I guess you're just more accepting than I might be...but I hope it's OK if I don't share your opinion. I don't own this camera but was thinking about upgrading to it and this kind of discussion is important to me as I'm not so willing to work around things like this if I can avoid it. I'll wait to see if it can be fixed with a firmware upgrade or if it's going to take a recall to resolve.
     
  66. Thanks Chris for taking the time to post these images, I may not be correct but it seems that the artifacts are more pronounced when the fine lights are in focus (e.x. your astro shot) when the lights blurry (e.x. your cityscape shots, I belive this was mirror vibration since you were using 400mm) the artifacts are much less visible.
     
  67. Mike, the photoshop action was made very quickly, it needs to be polished and tested on more than one image... My intent was just to show that it should be fairly easy to create one.
     
  68. Thanks for making it for us, Xavier. I'm not an expert in these things by any means, so if you are prepared to share your expertise with the rest of us, I for one would be very grateful. I don't do many night shots, but I'd like to know that I can if I want to!
    Of course, the availability of fixes in Photoshop does not absolve Canon of the duty to fix this issue....
     
  69. The artifact is commonly referred to as the 'Black Sun' effect in CMOS sensors. A description of the mechanism is shown in this article. It is doubful that this issue can be solved with a firmware since it typically require the addition of a circuit to provide a reference level that is not affected by the bright signals and a black sun detection circuit in the readout path.
     
  70. Interesting Berg, if there is leakage of charge from an overexposed pixel to the adjacent area I wonder why the black pixels appear only to the right of the bright pixels.
     
  71. Thanks Berg, very interesting...
    Mike, OK its sunday morning, I've tried to do better starting with the 6400 ISO image from http://glubsch.wordpress.com.
    1. Color Select pure white with a 58 tolerance
    2. Move selection right by 4 pixels
    3. Color Select Dark Shadows
    4. Dust Removal Filter with a radius of 13 pixels and a threshold of 19 levels
    It's better, but there is still some room for improvement though. I promise that once I have this camera, I'll put my optimized action on a "5D Mark II Tips and Tricks" page on my website.
     
  72. I don't get it. If Canon puts out a black-dot model, why didn't them label it as such on the package? .....
    :)

    Seriously, it seem to me as a very minor issue.

    Happy shooting,
    Yakim.
     
  73. Yakim,
    This is the super AF black-dot version ;)
     
  74. Thanks for this Xavier. Your new action is much better for that particular image. Unfortunately it doesn't work as well for all images; in fact I would doubt that it's possible to make an automated action which would work on all images. In particular, the number of pixels needed to nudge the white selection to the right might differ from image-to-image, and if the "black-spotted" highlights are not all the same colour as the colour chosen with Colour Select, some of the spots will be missed.
    I think any action would have to be tweaked on an image-by-image basis. Of course this is not the end of the world; as Scott suggested most images need some work done on them, but I'd rather not be forced to fix images due to intrinsic camera defects.
     
  75. Super AF? With 20D's AF system I am not so sure....
    Happy shooting,
    Yakim.
     
  76. Has anyone been able to eliminate the black dots - by changing the exposure automated options in the C.Fn II exposure settings?

    I just ordered one of these cameras and still have time to cancel.</
     
  77. Mike, you are right about the action not being universal. There is still some work to do regarding the color of the highlights, but I am less worried about the nudging that should be fairly standard. On the other hand, the selection should be feathered so that the "make-up" correction has smoother edges...
    And if it's not universal, then a little tweak on a per image basis is not a problem as I am already forced to fix my images for the equipment's intrinsic "defects": noise, white balance, contrast, distorsion, flare, vignetting, etc. :)
    I agree with Yakim, it is a minor issue.
     
  78. Having thought long and hard about what to write here, I'll cut to the chase, I have just cancelled my order based on what I've read above....I need a camera I can depend on in ALL situations (currently using 5D), not one that provides me with more work in Photoshop. Nikon will apparently announce the release of the D700x on December 20th, if this goes down as well as the D700 did....So, Canon, I know you read these forums, you need a solution to this issue before the D700x hits the streets I would suggest.
     
  79. I've just posted a new blog in response to the many comments found here and on other sites. In case you are interested (I compare it against an ISO 50 crop), you can read it here .
     
  80. I agree with Steve May- I've also lost interest in this camera until Canon owns up and fixes the problem. The response on the dpreview link that says the images from iso 100-400 look great is bunk. The guy needs to get his head, well, you know..... and look at the posts. I surely would not spend that money on a camera that's boasting 21 MP but not be able to take advantage of them while enlarging due to a camera flaw. A big benefit to this camera is its low light useability. Guess what- there are usually hotspots in those situations! I don't know enough, or care to know enough about PS to make scripts to solve a problem I shouldn't have to solve. And don't tell me I'm therfore not serious enough to own a camera like this then, as I'm sure some people will think. This camera is supposed to make your workflow easier, not add to it. I learned a long time ago- never buy v1.0 of anything..... I too want a power button on my TV!
     
  81. Well, I'm one of those who usually doesn't buy v1.0 either (particularly cars). But I really wanted to upgrade ASAP as I'm selling my photos online. I thought Canon learned from the 1D debacle (focusing issues) and wouldn't release another camera that has problems. But I'm quite certain this will be fixed in firmware, so I'm not particularly upset.
     
  82. I have faith it will be rectified in relatively short order, but I think I'll wait closer to 6-12 months though rather than jumping in next month like I was originally thinking. Unfortunately for you and others, there is more of a time crunch.
     
  83. I hope Canon realizes that their response to this problem is perhaps more important than the problem itself. All eyes are on THEM. If they don't fix it fast and try to whitewash it, it won't bode well. Nikon is providing some great alternatives. I have a 5d2 in the box at home(I'll get it tonight when I get there) and will check it out. If Canon acts like there is no problem or is very slow in a proper response, I'll have no qualms about selling all my gear and switching to Nikon.
    Shawn
     
  84. Michael, you were the one trying to convince me this was a major issue despite the fact that you don't own the camera, and have not seen a print from one. I think you are wrong, the original link printed out is not unusable, I did it today at work. I don't care if you buy a camera or not, I don't like it when people authoratively comment on stuff they just don't have or haven't even seen. But that is just me.
    Mike, I agree with you 100%, I hope this is fixed with a firmware upgrade, but having looked at Berg Na's link it would seem to be a bit more involved. This seems strange as the sensor is just a reworked 1DS MkIII sensor that has given no such problems for very critical photographers for some time. I wonder what the rework really did...
    Brad, my appologies this time, obviously you aren't daft. It is a great shame that this has shown up, and yes obviously anybody wanting the camera that will truly test its capabilities (20x30 inch prints before cropping counts! to get your 20x24 size you are looking at 20x30's) could be dissapointed. I do honestly think the vast majority of users will not worry, unless they are told they should, and most of the pictures taken by those users won't be affected anyway.
    That doesn't excuse the fault at all, I just tried to keep it in perspective, Canon have been stupid, on the trail of the 1D MkIII AF consumer cofidence debarcle and the fact that Nikon are now playing in the same league, they really should have made the testing work, what were all the Canon techs and photographers doing? It is amazing though that images we were looking at a few days ago as examples of how good the camera is are now being used to show how bad it is.
    Take care, Scott.
     
  85. Sorry Scott, just my opinion based on what I see in the shots, nothing authoritative meant by it.
     
  86. Looks like there is an easy workaround - turn off HTP, Lighting Optimizer and Noise Reduction - voila, no black dots. Sounds like an easy firmware tweak. Hope you didn't panic and cancel an order or return your camera.
     
  87. Ivan, you are correct. There is no easy workaround.
    I have tested my new 5D mkII all weekend.
    • It produces this problem from ISO 200 upwards.
    • With short and long exposures.
    • Whether the lights at distance are in focus or not.
    • Whether HTP, Lighting Optimizer and Noise Reduction are on or off.
    My 1Ds mkIII was tested against it. It does not exhibit the same problem the 5D mkII has
    There is a defect somewhere in the chain with the 5D mkII's image processing.
    Hopefully Canon will be able to address the problem quickly and efficiently via a firmware update.
    I take on board that I will still be able to use the camera outwith the shooting conditions that highlight this problem, but as a Pro who shoots night cityscapes and architecture frequently, this is an issue for me. The bottom line is; a new camera should not be defective in any way, shape or form. I will be contacting my CPN rep this morning to discuss the matter further.
     
  88. Jack,
    Can you post a few examples of the particular shots exhibitng this artifact?
     
  89. I am on a commercial shoot all day today. I'm sure you will appreciate needs dictate.
    Although I shoot RAW, I will try to post .jpg examples (or links to) asa I can manage.
    I have also discussed this with Canon directly. They are now aware of the situation.
    I will update you on progress as and when I can.
     
  90. Hi guys read this forum a lot but dont post very often as I dont know that much but had a 5d for 3 years and today got my mkii from Calumet, Ive got the flu and am in Scotland so no adventurous shots but I did one of an xmas tree with lights on but in the dark at H2 256K and blew it up to 500% and I cant see these black spots at all I dont know how to download an image if someone can tell me I will do so, also maybe my subject is ot relvant ? Regards Adrian
     
  91. Here is the picture
     
  92. Trying again
    00RjqL-96021584.jpg
     
  93. Nigel, more importantly, what's up with the horrid purple fringing around the lights? Even when viewed at 'fit to screen', the lights look purple! They weren't purple, were they.
    I had my eyes set on this camera too... :(
    -Rishi
     
  94. Yes, black dots are evident - especially in night shots
    http://www.6887.com/5dII/blackdots48.JPG
    And on my initial (first hour review) I did notice that turning off HTP did lessen the effect. However with the Christmas lights, nothing really made much of a difference, shooting jpgs/raws/CF options on/off and bracketing.
    I did print a sample (effectively a 16 x 24) and could distinguish the dots (just barely with reading glasses).
    I tried to produce dots in daylight - not very easy to do. Shooting the sun through tree leaves created a few dark spots (noticeable at 200%) but nothing as noticeable as the night shots.
    If any one wants a raw file, I'll upload one.
     
  95. ADMIN NOTE: Post deleted because of FTP errors from incorrectly formatted HTML. Sorry Kevin -Josh
     
  96. Kevin - You missed 1 important black dot in your image url, the one before 'jpg'.

    [​IMG]
     
  97. Kevin,
    This is the most accute form of this problem I have seen so far. In adition to the black pixels the lights are horribly square which is more distracting than the black pixels, I can't imagine the actual lamps or LEDs being square so looks like more trouble here...
     
  98. Kevin,
    Can you upload the RAW file please?
     
  99. Still waiting on official word back. In the meantime check the example linked to by HarryLally in this thread at DPR.
    Clearly visible at 100% and absolutely terrible looking!
    http://forums.dpreview.com/forums/readflat.asp?forum=1032&message=30298089
    There's definitely a serious issue here.
     
  100. Well the 200% crops do get a little rough looking, the actual image is breaking down on magnification.
    A 100% crop comparison [​IMG] [​IMG]
    I'll put up 2 raw files - ISO 200 and ISO 50. The 50 has less black dots.
    In camera JPGS are included.
    http://6887.com/5dII/_MG_9715.CR2
    http://6887.com/5dII/_MG_9716.CR2
    JPGS of same files (in camera)
    http://6887.com/5dII/_MG_9715.JPG
    http://6887.com/5dII/_MG_9716.JPG
    Gird (multiple specular highlights) crop. This could have been a disaster, but is reasonable (looks better than a 5d shot)
    http://6887.com/5dII/blackdots-grid2.jpg
     
  101. Whoa whoa whoa.
    The ISO 50 shot has less black dots. That sort of goes against this explanation for the black dots, doesn't it? Because if ISO 50 is just an implementation where an ISO 100 shot is taken overexposed by 1 EV, then there should be *more* of a chance for 'photon bucket spill' into the reference region, right? Which should then make this more of a problem at lower ISO.
    Does anyone see a flaw in that argument?
    What this might point to, then, is a problem with the noise reduction circuitry and/or algorithm, since that would be hiked up at higher ISOs... and perhaps hardly implemented at ISO 50. Just a guess.
    Rishi
     
  102. Kevin,
    Your RAW files demonstrate this phenomenon plus some strange artifacts, do I have permission to print these and show them to some people?
     
  103. Arash,
    No problem.
    K.
     
  104. Kevin did you check (Black dots problem ) sRaw mode with different iso in the same situation?
     
  105. Afshin - I'll try some sraw tonight.
    Still debating whether to keep or return the camera - that's why I'm working on the 'black dots' rather than enjoying it.
    The 5dII is a step up in terms of usability. The lcd, live view, remote shooting with live view, c1-3 user registers, quick menu, higher usable iso put this in a different class than the 5d. Feels more like a d300 for build quality. The resolution doesn't 'blow away' the 5d, but the improvement is there. With MF lenses, it is a huge improvement over the 5d (liveview focusing). Using the 5d vs 5dII back to back, the mkII makes the 5d feel antiquated (like jumping from a win xp computer to 98, or mac X to XP).
    The mkII does video too . . . . but I do prefer the rounded style of the 5d :)
     
  106. have not seen this on mine and I have done high iso night city shots
    Ross
     
  107. Thanks Kevin for taking the time :)
     
  108. PLEASE SOMEONE COMMENT ON THIS:
    The 'pixel buckets being overfilled' theory no longer makes any sense if ISO 50 doesn't show this black dot problem! Because ISO 50 is actually ISO 100 +1 stop... so the image taken at ISO 50 should run the risk of being overexposed much more than ISO 3200 (hence why ISO 50 has a limited dynamic range -- highlights blow out much more easily... if anyone wants an example of this, I can post it). Hence the ISO 50 should show more black dots if this theory were to be true.
    Which then says to me this has something to do with noise reduction (higher ISOs). Which then says to me perhaps there's a firmware/RAW converter fix? Or maybe that's just wishful thinking?
    Rishi
     
  109. Blah blah blah blah. Just write to Arash and Berg, better yet have a conference call, if you can get Bob's input then that is the one I'd listen to, the majority of people here are opinionated part time photographers with a healthy (mostly) interest in the gear, once you get down to the finer points of sensor physics you are way past most peoples knowledge and interest. You are building guesses on speculation, you can't work it out without testing only read about what other people think.

    The 5D MkII is a great camera, it has an issue that affects some specific types of lighting, not an uncommon thing for a v1 but frustrating for owners and interesting to prospective owners. It is a camera, take some pictures and enjoy it, or choose not to buy it because it is crippled by its horrendous image quality, your choice, but endless untested speculation, even for those with an academic interest, is pointless.

    Seasons greetings all, Scott.
     
  110. I noticed this report from the first day it was posted, but hesitated to reply. Being an amateur astronomer for many years and also an electronics engineer by profession I may be useful on this issue. It is not something new in cameras and is found on most cameras used on demanding applications with ultra high contrast as star fields on an almost dark background. Oversaturated pixels might lead to neighbour pixels with abnormal values in the direction that pixels are sampled. The first time I faced this issue was on a special monochrome camera made for astronomical photos back in 1998. Trying to use DSLRs for astro imaging I have noticed this behaviour even at the time of D10.
    It is a normal behaviour for a ccd or cmos camera that has to overcome saturated pixels while sampling sequential pixels. Above all emotional thoughts and expectations, these marvels of technology cannot escape basic troubles of electronics. I will try to be as simple as I can but not too much as not to miss the meaning. The photons captured by a single pixel (any kind of photo sensitive device used) result to a charge stored on this insulated device. This charge is proportional to the amount of photons that have struck the sensitive area until the limit of storage (known as ‘well capacity’ normally measured in electrons), where voltage has the max value and result in a saturated pixel reading. Today's pixels are insulated to preserve the charge until the processing circuitry is able to read the value. Older versions of image sensors had leakage of charge to their neighbour pixels when oversaturated, usually in the row that they were wired and created streaks on the image with saturated pixels, an effect known as blooming.
    The stored signal (el. charge) has to be read by an analogue to digital converter but at first has to be buffered and amplified. These circuits take care of the conversion of the stored charge to a signal at levels usable as the input of A/D converters, but also take care of the required amplification as advised by user's ISO setting. Canon sensors are advertised to have amplifiers on each pixel (perhaps buffering for low noise operation) but also final row amplifiers acting as conditioners of the sampled signal. Sampling subsequent pixels connects these amplifiers to each pixel. Saturated pixels will lead these amplifiers to their physical limits and moving from a saturated (overcharged) pixel to a contrasty dark pixel (with very low charge) does not mean that the amplifier will be able to respond linearly as a perfect follower. It will overshoot, meaning it might read smaller values than the real ones, depending on the speed that we try to sample the pixels. There is always a trade-off between speed and accuracy of an amplifier, something decided at design time of the circuit. When seeking high accuracy and linear operation this amplifier must have the time to return to normal operating range and respond, and also feedback is chosen accordingly. Perhaps the "need for speed" is today’s must but it does not come without sacrifice.
    This is the problem with these black pixels, and if sampling and timing values are incorporated in the firmware files, something can be done through a new firmware that distinguishes movie sampling mode and high quality photo sampling settings. Until then, a user in demanding photos should use as low amplification as can be used (low ISO setting). This will also lead to high dynamic range and smoother tones. Higher amplification (higher ISO) will create more pronounced black pixels.
    Noise reduction is the technique of subtracting a dark (unexposed) image of the same duration from the light (exposed) frame. In this case the black pixel effect could be even worse because of errors in mathematics of an exposed pixel being darker than the corresponding dark frame's pixel. Avoid noise reduction when shooting ultra high contrast shots and really want to avoid black pixels.
    I did try to explain it as simple as i could. Hope this helped.
    Yiannis
     
  111. The truth is, CANON needs to wake up! They are pissing off a lot of people, (me included) professionals and prosumers with their lame product releases. The only thing more offensive is the response Canon provides, by attempting to put the responsibility on the camera owner, or pass posted images as having "no way to verify the image is from a Canon 5D II". What a crock of crap.
    I have an EIGHT THOUSAND DOLLAR six month old Canon 1Ds Mark III, (a replacement to the two month old original who's sensor failed). My camera just took a five week tour of the NJ, NY and CA service centers in hopes of a repair for a white ghost pixel issue. I was getting a dozen or so blank white pixels on my images. Canon' resolution is to basically say that they can't find the problem or re-create it.
    There's three things that need to happen here.
    1. If you have not laid out $3,000 for the 5D Mark II, DO NOT GIVE CANON YOUR MONEY in hopes that you might get a "good one" until THEIR problem is solved.
    2. If you did buy a bad camera, phone your retail place of purchase and DEMAND your money back.
    3. If you can not return your camera, bombard Canon until you get a solid answer on a solution.
    The only thing Canon is going to listen to ultimately is $$$$ Cash being swiped out of their hand. So be it.
    I had my name on the 5D Mark II waiting list and just requested my deposit refunded. This is B.S. and typical Canon as of lately, sad to say. I might consider the 5D IF AND WHEN the issue is corrected in a well documented fashion.
    If anyone is awake at Canon, Bad, Bad, BAD.
     
  112. No, only two thing needs to happen:
    Wait a week (or less) to hear what Canon have to say.
    Have reasonable expectations and a sense of proportion.
    If their response doesn't make you happy, then you can start to complain. If you haven't bought a camera, then you have nothing to complain about of course, except being disappointed.
     
  113. Interesting - The Pop Photo review is now on the web
    "But this low-light performance isn't glitch-free. As have commenters on the web, we noticed tiny black blotches next to areas of blown-out highlights, such as the streetlights in our image of the skater. We found it hard to predict what images these spots would turn up in -- sometimes we'd find them, sometimes we wouldn't. Canon is aware of the problem, and our bet is that the issue soon will be resolved with a firmware upgrade. In any case, the blotches are easily removed in postproduction." - Pop Photo in tehir review
    http://www.popphoto.com/cameras/5677/canon-eos-5d-mark-ii-test.html
     
  114. If these black dots are visible to me at 100% in Photoshop, then they will be visible to the Photo Editors and Picture Libraries who routinely search for artefacts at 100%. If they find one of these the whole batch will fail. It is enough hassle removing dust spots without needing to remove these as well. Is it possible that photos from this this camera will become 'marked' by the Editors and Libraries for a more thorough screening than other cameras?
    The consumer has ultimate power in the buying market, despite the manufacturers desire to maintain a cashflow. Surely the consumer must be in a stronger position with the money still in his pocket. No, I don't own a 5D Mark 2, nor will I buy one just to see how real the problem is....then feel the donkey's ears growing faster than the number of black dots. As with Dana, above, I will reconsider my buying position as and when a comprehensive solution has been documented by a reputable source, comprehensively. If Nikon produce a high quality D700x at a reasonable price, that may be another option I consider.
     
  115. Just got mine in the mail today. I don't have a lens with me at work to try it out yet, so I'm stuck here pretending to take photos.
    This will be of interest to you all:
    Canon 5D Mark II Black Dots Fixed
    Since i haven't used the camera yet, I have no idea if this works. Check it out.
    jeff
     
  116. Could some moderator please fix the FTP call being made in one of these replies?
    The offending code is here:
    ftp://6887.com/5dII/blackspots-iso50-200.jpg
     
  117. Thanks for finding that Rob, it was bugging the hell out of me!
     
  118. Regarding the "black dots fixed" comment, the offending pixels appear to be present even when looking at uninterpolated intensity data in the RAW file, so things like noise reduction and auto lighting optimization aren't the cause (since they are both done during RAW file processing). HTP can affect RAW file data, but does not appear to be the cause of the dots.
     
  119. Yiannis,
    So then you're basically saying that as the pixels are being read, for a saturated pixel the amplifier applies less gain than for a darker pixel (for which it applies a higher gain)? In which case, if virtually no gain is applied for a completely saturated pixel, and the next pixel (to the right) happens to be very dark, then the amplifier can't respond quickly enough to increase its gain for that dark pixel to the right of the saturated pixel? Hence that dark pixel ends up appearing darker than it otherwise would?
    Also, for the rest of you with the camera who have reproduced the black dots -- what is their RGB value? Are they just low, or are they 0,0,0?

    Rishi
     
  120. Some of the pixels in the "black dots" are often 0,0,0 after RAW conversion to JPEG. At least one channel (R, G or B) aways seems to be zero. In the RAW file, they often show a level below that of the black point (i.e. the signal from a pixel that gets no light at all, which is always >0)
    Amplifier gain is (or should be) fixed. It's what sets the ISO. I think the suggestion made above is that when the output of the amplifier clips (at the high limit), it might then have some sort of "dead time" before it fully recovers, so low level inputs result in a zero output (or at least an output below the black level)
    Why some bright white overexposed pixel groups have a black dot next to them, yet others in the same image don't remains somewhat difficult to explain.
     
  121. Canon says:
    "Thank you for your inquiry. We value you as a Canon customer and appreciate the opportunity to assist you. We are sorry to hear that you are reconsidering the purchase of your 5D Mark II.
    It has been reported that some users are experiencing a phenomenon where small black dots appear when small, bright, point light sources are included in a photograph. Canon Inc. recognizes this issue, and is preparing a countermeasure. We will provide further details as they become available.
    We certainly apologize for any inconvenience this causes you.
    We hope this information is helpful to you.
    Thank you for choosing Canon."
     
  122. I don't like that 'Black Sun' explanation b/c if it were true then at ISO 50 you'd see MORE black dots because ISO 50 essentially overexposes the image by 1 stop, then 'pull-processes' it back to 0 EV in order to bump up detail in the shadows.
    Think about it, if it really were the Black Sun effect, then pixels that were near saturated would be MORE LIKELY to be saturated by being overexposed by 1 stop; that is, the black dots effect would be MORE prominent at ISO 50.
    Yiannis' explanation seems to make sense, but he hasn't answered my question yet about amp gain, etc. So, I'm waiting to hear back from him.
    Rishi
     
  123. Rishi, I'll try to explain... The gain is fixed by the ISO setting, what Yiannis is really referring to is overshoot (look at the Electronics section).
    What happens is that when the signal goes from saturation to a low level (at the row amplifier), it has a negative overshoot period. If the sampling is done too early, the signal has not been stabilized and is very low.
    [​IMG]
    If importing the image is not done via DMA (Direct Memory Access), there might be a very good firmware solution that rereads the pixel if it is detected as negative overshoot (below the standard dark value). Otherwise, they might need to slow down the reading if that's possible or do some interpolation to replace the faulty pixels...

    Xavier (with some time in VLSI Design engineering more than ten years ago ;-)
     
  124. Thanks Xavier. I love this community :)
    Your explanation makes things make more sense. It's a fixed gain across a row of photocells, due to the fixed gain of the row amp. Yes? I think I was making it more complicated.
    So this means that the row amplifiers cannot apply non-linear gain, that is, higher gain for photocells exhibiting lower charge and lower gain for photocells exhibiting higher charge?
    Because that would require reading the charge of the photocell, and then applying an appropriate gain. That's not possible, is it?
    Also, would variations in the exact gain applied by these row amps give rise to banding (http://www.photo.net/canon-eos-digital-camera-forum/00Rjyp )?
    TIA,
    Rishi
     
  125. Plus, a picture is worth a thousand words. Thanks for that graph!
     
  126. Wait a minute, Xavier, is the Y-axis in your graph the voltage the amplifier is getting from a given photocell?
     
  127. Nevermind, that's just a generic graph explaining overshoot.
    So now I'm confused again, but most likely b/c I don't understand the fundamentals of how the charge is read & amplified. Isn't the same amplifier gain being applied to all pixels (determined by the ISO)? So then a bright saturated pixel will receive the same amount of gain as a dark pixel, correct?
    It's just that if you're reading the dark pixel during a 'negative overshoot' period of the amplifier, then you may darken the pixel and get a blacker pixel. If you're reading the sensor from left to right, then this would show up immediately to the right of a brighter pixel because a negative overshoot would be more obvious next to a bright signal (pixel), correct?

    But, then, theoretically this negative overshoot could occur anywhere in the entire image, not just next to saturated pixels, right? Or is it because the amp has to 'recover' from a high voltage output when a saturated signal comes in from a saturated pixel, causing a negative overshoot, which then affects its reading/gain application to the dark pixel immediately next?
    OK, I think I just rationalized it to myself in that last paragraph. Let me know if my thinking is wrong.
    Cheers,
    Rishi
     
  128. Oh boy, take a look at 1:01 onwards in this video... pay close attention to the center tip of the candle flame. A few times, huge black/white dots that look like huge pixels (just 1 at a time) will pop up... my guess is these were black dots that the video encoder turned into a big black macroblock during compression.
    http://uk.youtube.com/watch?v=CdaybWRRDWI
    Pretty hideous. Really hoping Canon can do something soon about this. The video linked above is inspiring, even though I'm a photographer, not a videographer.
     
  129. Rishi,
    Essentially, your last paragraph is correct: there is a negative overshoot because the voltage is coming from far off. It seems to me from the pictures I have seen that it happens only on small saturated dots. Which lends me to think that the conditions in which it happens is a small area of the dark/saturated/dark type. This would translate at the row amplifier to a positive overshoot (that we don't see) followed with the acquired "momentum" by an important negative overshoot.
    That's one possible explanation which I like, but there are others possible, including a bug in software signal processing...
     
  130. To the administrator(s):
    This page is still automatically linking to the 6887.com domain via ftp. This sort of behavior should be disabled by default in the forums, as it poses a security risk. Please correct this issue promptly as this opens your site to malware via redirection attacks.
     
  131. Peter - I think I've fixed that issue here. Let me know if you see it again (typo by poster).
    I have no control over what is and isn't allowed in the forums, but you are right that it's an issue we need to look at. Since I'm no expert on this, you might want to send a note with your concerns to Jin Choi (http://www.photo.net/photodb/user?user_id=19867), who is our systems admin contact. I'll send him a note too. Thanks.
     
  132. Xavier,
    Your hypothesis is the ONLY one that makes sense so far. Again, because with the 'Black Sun' the effect would disappear at high ISOs. And, your theory predicts that the effect would be subdued at ISO 50, since a lower gain is applied by the amps.
    Perhaps the sensor is read too fast because of the implementation of HD movie mode in this camera?
    Rishi
     
  133. I received my 5DMII yesterday and immediately started testing. I too have the black dot issue... as expected. But a couple of things that I found:
    You can see it on the LCD screen at full zoom. I think this rules out RAW processing.
    I also tested my 1DsMIII and it did not have the problem. This confirms what others have said, it doesn't seem to be the sensor.
    So, I'm thinking that we need a solid test performed on a 50D to see if it is the DIGIC 4.
    I have listed out the exact test I performed here: http://douglasgoodinphotography.blogspot.com/
    Feel free to wail on me if I have been reduntant or completely wrong! :^)
    Thanks,
    Doug
     
  134. Xaivier,
    Excellent explanation of overshoot as it applies to the MarkIII problem. (I was an ASIC designer in a previous life :^)
    But, wouldn't the 1DsMIII have the same issue? I thought that they both used the same sensor... and therefore the same amplifiers etc. Am I mistaken?
    Thanks
     
  135. James,
    I believe that they are probably reading the sensor data much faster on the 5D² to catch up with the video 30 fps...
    So reading faster means sampling sooner, that is before signal stabilization!
    Oh, and it appears that they are reading 4 pixels at a time, which explains the black dots size...
    Search for "four channels" here.
     
  136. Xavier,
    Yep, that's it. You nailed it! Thanks for taking the time to point me to the improvements section of the 5DMII tech report. And with that we can ignore my request for a test on the 50D. :^)
     
  137. I believe that they are probably reading the sensor data much faster on the 5D² to catch up with the video 30 fps...​
    Woohoo! Looks like I may have been right, according to Xavier, when I said:
    Perhaps the sensor is read too fast because of the implementation of HD movie mode in this camera?​
    If so, can't they fix this in firmware? Like for photos readout more slowly, but for video keep the readout fast? Also, Xavier, Canon has probably figured out the cause already, right? Otherwise perhaps you should e-mail them? I'd rather they not just put some software fix to try to 'clone' out black dots...
    Rishi
     
  138. I'm sure Canon knows what's really going on...
     
  139. This might be interesting for some of you technically minded people. On TV last night I watched a short programme about someone who makes very small models and puts them inside the eye of a needle....a helicopter, a person, a flower. They fixed a TV camera to a microscope to see the models in detail (unbelievable!), The needle was backlit, and there along a high contrast edge was a line of black dots, exactly the same as I have seen exhibited by the 5D Mark 2
     
  140. Where's Vijay?! I'm sure he'd have some explanation invovling 'sentient black dots'!
    :)
    -Rishi
     
  141. And, your theory predicts that the effect would be subdued at ISO 50, since a lower gain is applied by the amps.​
    Rishi, I don't think that is correct. From what I can tell ISO 50 must be implemented in the raw software, not in hardware. When you compare an iso 50 to iso 100 with same exposure (same Tv, Av), the undemosaiced raw data is identical. ie. they are both iso 100 shots. The 'L' iso shot should receive the same amplification as the 100 iso shot (which is the native iso). By the way, I compared this on a 5DmkI. I am assuming 'L' iso is implemented the same way on the mkII.
     
  142. What I like about this problem is that Canon acknowledged it straight away. It is an indication that they have concluded that the way they treated the former problem was a mistake. That gives me hope.
    Happy shooting,
    Yakim.
     
  143. This is a very interesting discussion on the 5D2, a camera I am considering. As a long time circuit designer, I suspect this is not overshoot but rather the amplifier having very low gain as it recovers from saturation. If an amplifier is not designed carefully, its gain will go way down when you saturate it. Entire sections of amplification may become unbiased, for example. In this case, the next values would read some very low level. After a period of time, the amplifier bias levels return to normal and the gain goes back up. As this is not affected by ISO settings, the offending stages may be near the output, after the gain control.
    They may be able to slow things down and allow the amplifier time to stabilize. On the other hand, if they need this fast sample rate, the solution would be a new amplifier design and not firmware. There are things you can do in a circuit design to keep things biased under high input level conditions. Basically, you put circuits in to lower the gain of the amplifier gracefully so that nothing gets unbiased. In this case, the recovery can be very fast. The circuit design time would not be fast, however.
     
  144. And, your theory predicts that the effect would be subdued at ISO 50, since a lower gain is applied by the amps.​
    Yeah I'm not so sure I believe my own statement here since ISO 50 is actually an ISO 100 shot overexposed by 1EV (an ISO 50 shot looks just like an ISO 100 shot except that the brights get brighter, and sometimes blow out... hence the limited dynamic range at ISO 50). So, even though the gain may be different (Bernie, it has to be lower to get the same RAW data from a +1EV overexposure, yes?), you'd have more saturated pixels... so I don't particularly see how this should decrease the black dot effect... yet it does.
    Rishi
     
  145. So, even though the gain may be different (Bernie, it has to be lower to get the same RAW data from a +1EV overexposure, yes?), you'd have more saturated pixels... so I don't particularly see how this should decrease the black dot effect... yet it does.​
    It was the same raw data with the same Tv and Av. This was to show that iso 50 is really iso 100. I set exposure manually to make sure it was the same in each image. If i had the camera on one of the auto modes, then it would have doubled either Tv or Av when switching to iso 50 from 100.
    So, the analogue gain should be the same between 50 and 100. The -1EV correction is performed in the raw converter after the fact.
     
  146. So then what the hell is the point of ISO 50? Why even put it in there? What a f'in gimmick if that's all it really is. It's like making a 60Hz television capable of accepting 24p HD video... just to take it back to 60Hz via 3:2 pulldown process.
    Some of the decisions these companies make just astound me. They just treat the public as if they're unassuming idiots...
    Wait a minute...
     
  147. I guess the only advantage is that you get less noise in an ISO 50 image vs an ISO 100. The trade-off is 1 stop of dynamic range. It's annoying, like you say, because this is something anyone could do in software themselves, but Canon (to the best of my knowledge) doesn't document this, and many unassuming users might think they are getting something special using ISO 50.
     
  148. ISO 50 is useful if you are shooting with a fast lens and/or slow shutterspeed in daylight. My Canon 30D doesn't have it and I've run into wishing for it more than once. Of course, an ND filter is the answer, but how handy it would have been.
     
  149. Canon announces firmware update in the works for:
    “Black dot” phenomenon (the right side of point light sources becomes black) When shooting night scenes, the right side of point light sources (such as lights from building windows) may become black. The phenomenon may become visible if the images are enlarged to 100% or above on a monitor or if large prints of the images are made. ​
    Also a banding issue discribed in the above link
     
  150. The above post was the Canon announcement from the Australian web site. It is now duplicated on the Canon USA web site. Here's the complete text:
    We have learned that some users of the Canon EOS 5D Mark II digital SLR camera have identified two types of image quality phenomena that appear under certain shooting conditions.
    1. “Black dot” phenomenon (the right side of point light sources becomes black)
    2. Vertical banding noise
    We are currently investigating and analysing the causes, and examining measures to reduce or eliminate these phenomena by providing correction firmware. An announcement will be made on the Canon Web site when such measures have been determined.
    Details of the phenomena and shooting conditions under which they are likely to occur are as follows.
    1. “Black dot” phenomenon (the right side of point light sources becomes black)
      When shooting night scenes, the right side of point light sources (such as lights from building windows) may become black. The phenomenon may become visible if the images are enlarged to 100% or above on a monitor or, if large prints of the images are made.
    2. Vertical banding noise
      If the recording format is set to sRAW1, vertical banding noise may become visible depending on the camera settings, subject, and background.
    • Vertical banding noise is not noticeable if the recording format is set to sRAW2.
    • Vertical banding noise does not occur if the recording format is set to RAW or JPEG.
    • Noise can be reduced if C.Fn II-3: Highlight tone priority is set to 0: Disable.
    Canon always strives to provide the highest quality products to our customers. We apologize for any inconvenience these phenomena may have caused. We appreciate your kind patronage and support.​
    Here is the link for this Canon USA "Service Notice" http://www.usa.canon.com/consumer/c...oryid=139&modelid=17662&keycode=2112&id=55060
     
  151. The above post was the Canon announcement from the Australian web site. It is now duplicated on the Canon USA web site. Here's the complete text:
    We have learned that some users of the Canon EOS 5D Mark II digital SLR camera have identified two types of image quality phenomena that appear under certain shooting conditions.
    1. “Black dot” phenomenon (the right side of point light sources becomes black)
    2. Vertical banding noise
    We are currently investigating and analysing the causes, and examining measures to reduce or eliminate these phenomena by providing correction firmware. An announcement will be made on the Canon Web site when such measures have been determined.
    Details of the phenomena and shooting conditions under which they are likely to occur are as follows.
    1. “Black dot” phenomenon (the right side of point light sources becomes black)
      When shooting night scenes, the right side of point light sources (such as lights from building windows) may become black. The phenomenon may become visible if the images are enlarged to 100% or above on a monitor or, if large prints of the images are made.
    2. Vertical banding noise
      If the recording format is set to sRAW1, vertical banding noise may become visible depending on the camera settings, subject, and background.
    • Vertical banding noise is not noticeable if the recording format is set to sRAW2.
    • Vertical banding noise does not occur if the recording format is set to RAW or JPEG.
    • Noise can be reduced if C.Fn II-3: Highlight tone priority is set to 0: Disable.
    Canon always strives to provide the highest quality products to our customers. We apologize for any inconvenience these phenomena may have caused. We appreciate your kind patronage and support.​
    Here is the link for this Canon USA "Service Notice" http://www.usa.canon.com/consumer/c...oryid=139&modelid=17662&keycode=2112&id=55060
     
  152. ISO 50 is useful if you are shooting with a fast lens and/or slow shutterspeed in daylight.​
    No, it's not, if all the camera is doing at ISO 50 is telling the meter to overexpose by 1EV at an actual ISO of 100, only to pull it back in software during the RAW conversion.
    You could do this yourself by prolonging the exposure at ISO 100 by 1EV, then doing a -1EV adjustment in the RAW converter.
    Unless the camera's doing something else, which, it seems, it is not, according to Bernie's results.
     
  153. Does this happen in video? I purposely over exposed some shots in this video can cant seem to notice the black dots.
    http://www.vimeo.com/2581629
     
  154. Does this happen in video? I purposely over exposed some shots in this video can cant seem to notice the black dots.
    http://www.vimeo.com/2581629
     
  155. Yes, happens in video, as has been shown in Vincent LaForet's (sp?) video in the scene with the couple underneath the cityscape/bridge; and also, on Canon's own marketing video there's some pretty bad pixelation resulting from the, I believe, the black dot problem, right around a candle flame.
    Links to both of these examples are posted above, somewhere.
     
  156. Intersting.
    I finally got my 5D and after a day or so, I got around to taking a pic of the Christmas tree lights at high ISO and then on Auto with a f/1.8 lens.
    And - nothing Not right, not left, up, down. Not even a black spec.
     
  157. Canon has responded to this, the response has been published on dpreview.com
    http://www.dpreview.com/news/0812/08121902canon5ddots.asp

    "Canon responds to black dot and banding concerns

    Thank you for using Canon products.
    We have learned that some users of the Canon EOS 5D Mark II digital SLR camera have identified two types of image quality phenomena that appear under certain shooting conditions.
    “Black dot” phenomenon (the right side of point light sources becomes black)
    Vertical banding noise
    We are currently investigating and analysing the causes, and examining measures to reduce or eliminate these phenomena by providing correction firmware. An announcement will be made on
    www.canon-europe.com as soon as measures have been determined.
    Details of the phenomena and shooting conditions under which they are likely to occur are as follows.
    1. “Black dot” phenomenon (the right side of point light sources becomes black)
    When shooting night scenes, the right side of point light sources (such as lights from building windows) may become black. The phenomenon may become visible if the images are enlarged to 100% or above on a monitor or if extremely large prints of the images are made.

    2. Vertical banding noise
    If the recording format is set to sRAW1, vertical banding noise may become visible depending on the camera settings, subject, and background.
    ▪ Vertical banding noise is not noticeable if the recording format is set to sRAW2.
    ▪ Vertical banding noise does not occur if the recording format is set to RAW or JPEG.
    ▪ Noise can be reduced if C.Fn II-3: Highlight tone priority is set to 0: Disable.

    Canon apologizes for any inconvenience caused and thanks photographers for their patience. "

    now after officially canon admit the problem I'm confused now whether to get the 5D mark II or just buy the old 5D ? any advice pls..
     
  158. Just buy and enjoy the pictures.
    If Canon responded like that, it is the firmware problem, not the sensor.
    Buy the camera and update the firmware when it becomes available.
    Black dots don't show up unless it is blown up to 100% or really large print.
    http://www.kenrockwell.com/canon/slrs/5d-mark-ii/black-pixels.htm
    Check his opinion about black dots. I agree with him and I really like the image quality of this product.
     
    • Rishi Sanyal [​IMG], Dec 19, 2008; 03:52 p.m.
      • ISO 50 is useful if you are shooting with a fast lens and/or slow shutterspeed in daylight.
    • No, it's not, if all the camera is doing at ISO 50 is telling the meter to overexpose by 1EV at an actual ISO of 100, only to pull it back in software during the RAW conversion.
    • You could do this yourself by prolonging the exposure at ISO 100 by 1EV, then doing a -1EV adjustment in the RAW converter.
    • Unless the camera's doing something else, which, it seems, it is not, according to Bernie's results.
    Are you kidding me ? Have you ever tried to use F1.2 or F1.4 in broad daylight extremely bright ?
    you are mensioning "- 1.0 EV" to make it happened.
    But what if the shutter speed already hit the maximum shutter speed with the aperture you desired @ ISO 100 ? And it is still over exposed ?
    All you can do is make the aperture smaller than you desired. (meaning that you have to use F2.0 instead of F1.4 to have "-1 EV" but the effect you you want to have is @ F1.4 ... it is going to make a huge difference.)
    But all you need is just 1/3 ~ 1 stop below to have the apreture you desired and you do not have your ND Filter with you ?
    ISO 50 becomes really handy with this kind of situation.
     
  159. Brandon-

    a) That's a pretty fast shutter speed. I mean, even at 1.4, 1/8000 should be enough... or at least if it isn't, maybe f2.0 isn't such a terrible aperture? But I digress.

    b) I believe that the point was that "ISO 50" is _actually_ just shooting ISO 100, then taking the raw file and pulling it down one stop. It isn't actually a different ISO. So you can use it, but it isn't really any different from shooting it overexposed at ISO 100 and pulling it back in RAW. No increase in quality, but actually the opposite.

    Or, that's what I interpret the previous message to mean.

    All that being said, I really enjoy my 5d2. I haven't run into the dots, and since it's a loss of quality instead of an increase, I simply don't use ISO 50.
    00RzY5-103115584.jpg
     
  160. Whoa, some people are going formatting-trigger happy.
    Thanks Brad for explaining to Brandon my point so I don't have to spend my time doing so :)
    Nice shot -- I agree, high ISO will allow for amazing shooting scenarios. Shooting at f/5.0 indoors at night! Ha! Never woulda thought of doing that with, say, film...
    That being said, my eye-brain/whatever is trained to pick out defects/flaws in photographs/video imagery, so I just don't think I'd be able to deal with the banding & the black dots. So I'm waiting for a firmware fix.
    Interestingly enough, I saw some slight, although higher frequency (so perhaps less noticeable to some) horizontal banding (across the long axis of the CCD) on some sample photos of that orange Ferrari (I think?) shot taken by Nikon's new D3x.
    So it'd seem there are still some technological limitations. If you read the white papers for some of these products, it really is amazing how much complexity goes into these systems... no wonder they break apart within a matter of years whereas some 1980s SLRs are still functional today...
     
  161. I am not saying ISO 50 is good.
    I said it comes in handy.
    I tested them and ISO 100's picture quality is better than ISO 50, but it looks pretty smooth.
    Well I know that you can adjust the bightness of the picture with the picture taken with RAW and adjust it further.
    But there is a difference between taking pictures as way they are only with the camera setting and adjusting the pictures later on ...
    This is a digital world, and it may be right to adjust the pictures with photoshops and other graphic utilities...
    But I hope some people know what I am saying.
    If not, well... it was just my opinion.
    Thanks.
    Is there any news about when Canon will release the new firmware yet ?
     
  162. But there is a difference between taking pictures as way they are only with the camera setting and adjusting the pictures later on ...​
    Says who? Do you have any proof/evidence for this? Do you know how the ISO 50 is implemented by Canon? i.e. the specific algorithm? If not, then you're not qualifed to comment on this, & are just spreading misinformation.
    The same scene, at the same EV when taking the shot, has blown out highlights at ISO 50 but no blown out highlights at ISO 100 (the test has already been done on the 5D Mark II). Therefore, this indicates to me, along with other evidence (such as that presented by Bernie above) that ISO 50 is just a 'hack'... and you'd get the same results just taking the shot at 1/gazillionth of a second, having some blown out highlights, then applying -2/3 EV in post-processing. B/c ISO 50 is, apparently, also not saving your highlights from blowing out.
    So either the camera does the -2/3 EV, or you do.
    I'd love to be proven wrong though, that ISO 50 ACTUALLY does something better (for example, apply less gain than ISO 100... but that's likely not the case, since highlights are blown out MORE at ISO 50 than ISO 100)... but I need more evidence than 'it comes in handy'.
    Y'know?
    Cheers,
    Rishi
     
  163. Rishi
    I think what your saying is the extended ISOs are similar to PUSH and PULL processing of film.
    If I have time today, I''ll run some "Ansel Adams" curves of the 5D II sensor. I'm thininking of shooting a plain white paper at a ten stop EV range (Ansel's Zones I-X) at 4 different ISOs (50, 100, 800, 25,600) and maybe 2 or 3 different light intensities (because of the wide ISO ranges) - with two of the ISOs in Canon's "extended range" (50 and 25,600). With that ISO range, I'm not sure I can do this. Probably need a flash set up.
    Photoshop's histogram data gives the mean intensity value as well as the data for each color channel RGB and CMYK. I'll take both RAW and JPGs but I think I want RAW here.
    Then a plot of the data (X-axis: EV and Y-axis:photoshop mean intensity) should reveal if anything is different in the extended ISOs (I'm sure it is - otherwise it wouldn't be an option on the camera).
    The heel and toe of the plots will reveal if there is any loss of highlights or shadows at the various ISOs
     
  164. Has anyone explored the effect on the black dot issue of using LiveView? I would run these tests if I had my MKII yet, but will be picking it up on Tuesday 1/06. And, if my new MKii has this issue. Never thought I would be hoping it did :(
     
  165. Dennis -- yes exactly, but even worse than 'push/pull'.
    Digital ISOs themselves, if I understand correctly, are basically the film equivalents of 'push & pull'. There's a 'native ISO' for the sensor... outside of that, increasing the ISO just increases the gain applied by the amplifiers to the photocell voltages. In other words, say the 'native ISO' is ISO 100. When you expose at ISO 400, you decreased the actual exposure of the sensor by 2 stops, so there was 4 times as less light falling on the sensor. So, very simplistically, the amplifiers would apply 4 times as much gain.
    ISO 50 is even worse. Because if the native ISO of the sensor is, again, say 100, then you overexpose the sensor by 2/3 of a stop. So maybe now you've blown out more actual photosites than you would've had you exposed at ISO 100. On top of that, you're applying whatever gain you would've applied at ISO 100 (I'm basing this conclusion on Bernie's data that if you set the shutter speed/aperture the same for both an ISO 50 and ISO 100 shot, the RAW data are identical)... so now the voltages going to the ADC are all higher because you overexposed yet applied the same gain. IF any of these voltages exceeds the maximal input that the ADC maps to the highest bit, then you've lost more highlight data.
    But that last one is a big 'IF'. In the best design, the ADC wouldn't behave like that and the maximal output of the amplifiers would be lower than the maximal input of the ADC to leave some 'headroom'. That's my guess anyway. An electronics engineer should step in here... I'm actually underqualified (I'm in biology/chem) to comment on this as this what I've gleaned from what I've read and other discussions I've had.
    My point being: this all gets complicated when we're sitting here contemplating without knowing the actual electronic design.
    Anyway, Dennis, if you do that test, def let us know the results!
    Cheers,
    Rishi
     
  166. I equate film emulsion thickness, rather than development push/pull, to analog gain in the sensor path. The thicker emulsion allows film to form a viable latent image with less light, at the cost of a grainier image, and the image becomes unuseable at lower light levels (blown highlights) than with a thinner emulsion.
    I equate the 'expanded' digital ISOs ('digital' gain) with pushing or pulling film. They all result in a reduction in potential dynamic range given a constant analog gain or emulsion thickness, and both involve exposing at an ISO for which the sensor+gain or emulsion was not designed for. Optimal dynamic range was always a battle with film, just as it is with digital. With film we exposed for the shadows and developed for the highlights, the opposite of digital.
     
  167. Regarding ISO 50: on the regular 5D, ISO 50 has one stop less dynamic range which is why most landscape shooters keep it at 100, even with a tripod. Canon stated that they made ISO 50 something you had to turn on because of that fact, so people couldn't just easily select it. Knowing that fact of the original, there's good reason to believe that this is the same story with the 5D2.
    The noise improvement at ISO 50 (if any) doesn't seem to be worth losing dynamic range. In my landscape shots, I always use 100 and forget that 50 exists. The 5D (1 and 2) are so noiseless at 100, it's crazy. I suggest a good B+W ND filter if you need the longer shutter speed. That filter comes in handy even more now with video on the 5D2, if you want a narrower depth of field during daylight video shooting.
     
  168. FIRMWARE 1.0.7 HAS BEEN RELEASED
    http://web.canon.jp/imaging/eosd/firm-e/eos5dmk2/firmware.html
    it specifically addresses the black dot issue and vertical banding if shooting sRAW.
    sRAW shooters need to also upgrade Canon DPP software too.
    off to download it myself and give it a few tests.
     
  169. Anyone have any 100% crops with the new firmware?
    Anyone want to take a stab at what exactly Canon did in the firmware update? Noise removal update? Change the amplifier behavior? Sensor array read speed?
    Would be cool to know.
    Rishi
     
  170. >>Anyone want to take a stab at what exactly Canon did in the firmware update?
    I did a quick glance - it looked to me like the black dots are now white dots [I shouldn't take that leap without a better look, however]
     
  171. i upgraded the last night - popped a 50 f/1.4 and went for a drive. ended up getting called by the office and had to scrub the planned test shoot (6 client projects on the go). did however roll off a bunch of shots from the car while driving home... this is a totally crap pic but i selected it from the dozen i shot as it has lots of little lights in the distance. not sure about the purple glow around some of the lights but I attribute it to shooting through my car windshield. was also -15 outside with a strong wind so stopping and rolling down the window wasn't a good option either.
    ISO3200 f/1.4 - full resolution JPG direct from DCIM folder
    http://farm4.static.flickr.com/3500/3180310711_e6b3d5e511_o.jpg
     
  172. How do people feel about the purple fringing around highlights and the 'purple blotchiness' so typical of 5D Mark II pictures (both problems there from the beginning)?
    Re: purple blotchiness, check out here:
    http://forums.dpreview.com/forums/readflat.asp?forum=1032&thread=30301707
    Christopher, the purple blothiness is pretty bad on your shot above.
    Why aren't more people raising this point as a problem? It's even present in ISO 100 images!
    Rishi
     
  173. Christopher, do you have a RAW version of that image?
    I have a feeling that Canon's own demosaicing algorithm magifies the purple blotchiness, whereas ACR or DCRAW does a better job at preventing the purple blotches.
    Would like to try it on your file since the purple blotchiness is so apparent in it.
    Thanks,
    Rishi
     
  174. Rishi.
    Apologies, i was tired - long day and long night - thought i had JPG+RAW set and it was only JPG that it was recording. AgreeD the blotchies are icky. I thought maybe it had to do with the high ISO, the cold and my car windshield.
    Will recreate and be sure RAW is enabled (normally is but I wanted both JPG and RAW for the test).
     
  175. Thanks Christopher. Would test myself if I could get my hands on a 5D Mark II at no cost :)
    Currently sticking to film & the nightmares of scanning... (don't get me started)
    Can't wait to go digital.
    -Rishi
     
  176. Never thought I'd end up as a 'pixel peeper.' I got curious about what looks like a good fix from Canon, version 1.0.7, and wondered whether they had merely turned the black dot into a white one, thuswise making a circular shape sort of oval. So, I thank Andrew Yip for his tests, I copied a version 1.0.6 onto a version 1.0.7 in Photoshop and after trying to align them perfectly I reduced the opacity to see if a circle remained a circle without a black dot attached. I'm not sure if a circle remains a circle or not, I'd appreciate some second opinions on this, just to satisfy my curiosity....
     
  177. I took Andrew Yip's ISO 800 images, rotated one a little bit to align correctly, then layered them. Find the layered TIFF file here:
    http://staff.washington.edu/rjsanyal/Photography/ISO800-BlackDots.tif
    Hope Andrew doesn't mind :)
    Open the file up in Photoshop and show/hide the top layer (the newer firmware) to distinguish between the two. It does kinda look like the circles turn to ovals :( Probably some sort of cloning procedure going on?
    Not a big deal for circles of light. Wonder though, if it is cloning, if it'll work in all situations. Such as strips of bright light. Or what have you...
     
  178. Good bit of skill Rishi, very informative, thanks. It looks to me as though Canon may have tackled this problem at source and not via cloning as there does appear to be evidence of the black dot changing to the background colour as I lowered the opacity. This ties in with an opinion I read in another forum somewhere. Whatever, it is so difficult to see now at high magnification it makes the black dot affair a non issue now.
    I agree with you about the strip of bright light, I would like to see an example of that to see if that too is cured (I suspect it is).
    I hear there have been issues in the past with firmware updates introducing new problems, I'll give my purchase 3 or 4 weeks and hope this is not the case here.
    I have looked forward to this camera for a long time....nearly there!
     
  179. It looks to me as though Canon may have tackled this problem at source and not via cloning as there does appear to be evidence of the black dot changing to the background colour as I lowered the opacity.​
    Perhaps. But the alternative explanation is that they bias the cloning to select pixels from the right (the background) as opposed to the left (the bright light source) since the black dots always fell to the right of bright light sources. See what I'm saying?
    Either way, as much of a pixel peeper as I am, I really don't care... to me it's as much as solved :)
     

Share This Page