Brainiac wrote: I think it's quite easy to demonstrate the burlap noise reliably. Shoot a scene which has a fairly wide range of exposure values, where there is information in the shadows, and flat shadow areas. Set the contrast to minimum in DPP, export to photoshop, and run auto-levels. The burlap will become clearer if you have raised the brightness by a stop or two in DPP.
I tried exactly what you said, on a high dynamic range picture with levels from slightly blown to deep inky shadows, and every level in between. Contrast was at minimum at -4. I tried normal brighness, +1 and +2 brightness. I can't get it to show the slightest amount of banding at 100% pixels.
That is not to say another 5D2 will not show it, but I tried twisting the file every which way to ridiculous distortion, and I can't find a condition or process that shows banding.
With my experience in electronic manufacturing, there are variables in electronic components, such as amplifiers and A-D converters that might explain the problem in some 5D2 cameras.
tchan748 wrote:
i have a 5D2 since last Dec, and the bending issue does not bother me at all. But having said that, this thread is very interesting and i hope it doesn't turn out to be a debate of why we are discussing this. For those people that dont see it as a problem, you can either a. not read this thread anymore, or b. like myself, read it with a curious mind and try to learn something from it.
there are always new and interesting things we can learn about our equipment whether you actually use it to the full potential or not that is another topic....Show more →
+1, some posts here to search for 5D2 solution. some posts here to learn and share the topic. some posts here to help others. some posts here to establish his/her play ground. some posts here to screw up other's minds for some reason. For the later, has anyone understood what is this thread all about
I don't know much about sensor and image amplifier or A/D converter design, but one thing crossed my mind: What if it is not noise, but unevenness in optical transparency of the RGB filters or amplitude variation of the cells?
Don Clary wrote:
With my experience in electronic manufacturing, there are variables in electronic components, such as amplifiers and A-D converters that might explain the problem in some 5D2 cameras.
That sounds disturbing. Maybe there are indeed 5D2's that are worse than others...
Daan B wrote:
That sounds disturbing. Maybe there are indeed 5D2's that are worse than others...
Of course there are. At 21MP and with such high pixel density, even a tight tolerance of + or - 5% could very well be noticeable. I doubt, though, that 5% (which is close to military spec) is the tolerance of the 5D II. It's probably more like 10% or even as high as 15%. Then again, who knows? ...
Mirek Elsner wrote:
I don't know much about sensor and image amplifier or A/D converter design, but one thing crossed my mind: What if it is not noise, but unevenness in optical transparency of the RGB filters or amplitude variation of the cells?
Then the pattern would always be the same, and you could remove it using a frame as a negative mask.
Daan B wrote:
That sounds disturbing. Maybe there are indeed 5D2's that are worse than others...
Where is the evidence? On the dark frame thread we had samples from many 5D2's from all over the world, and they were all the same. Once again, pure speculation. Some people upload pictures with banding claiming they are fine. I think it is more likely that there is more variation in how people process and look at their pictures than there is variation in components. I'm not saying there's none, just that the evidence suggests the cameras are the same, and people react differently to the problem.
semorg wrote:
I got horrible banding problem with my iso 100 shots. I usually expose to the right. Processed the image with a bit contrast and red or orange like filters to convert to black and white.
FWIW my 5D1 often exhibits the same behavior when converting to B&W, especially if I'm really playing with/imbalancing the color channels. I've lately been using Nik Color Efex Pro (and keep meaning to try Silver Efex) for most of my conversions and find that I can usually avoid what I call B&W banding thru the right combination of conversion settings (but that combination varies by image). If I can't get rid of the banding, I apply a slight bit of surface blur to the skies/clouds on a separate layer then add a touch of grain to counter act the smoothing - a work around that can produce very nice results. I'm still new to my 5D2 and haven't given any B&W conversions a try yet so can't comment if one camera is more prone to the behavior than another.
brainiac wrote:
Where is the evidence? On the dark frame thread we had samples from many 5D2's from all over the world, and they were all the same. Once again, pure speculation. Some people upload pictures with banding claiming they are fine. I think it is more likely that there is more variation in how people process and look at their pictures than there is variation in components. I'm not saying there's none, just that the evidence suggests the cameras are the same, and people react differently to the problem.
I remember the black frame thread... I also remember that not all 5D2's produced 100% the same results. I can't remember the reason for this anymore... I am sure we discussed this. Maybe people didn't follow the exact same procedure? Or maybe the variables in electronic components were exposed? It would be educational to gather a few 5D2's and test them side by side... to settle this once and for all.
At this point I am open-minded to all possible explanations.
akclimber wrote:
FWIW my 5D1 often exhibits the same behavior when converting to B&W, especially if I'm really playing with/imbalancing the color channels. I've lately been using Nik Color Efex Pro (and keep meaning to try Silver Efex) for most of my conversions and find that I can usually avoid what I call B&W banding thru the right combination of conversion settings (but that combination varies by image). If I can't get rid of the banding, I apply a slight bit of surface blur to the skies/clouds on a separate layer then add a touch of grain to counter act the smoothing - a work around that can produce very nice results. I'm still new to my 5D2 and haven't given any B&W conversions a try yet so can't comment if one camera is more prone to the behavior than another.
It's funny you mentioned that. Because that's what I tried doing. Adding grain seem to fix the problem, but it's so lame to have to do that. Another thing I found out with adding grain, is your intended print size. I found out, I need to use different size grains for different print sizes. But in general, I've never been 100% happy creating B&W images out of digital images, however the D3X seem to look like a great camera. Just waiting to see if they would put that electronics into the D700 replacement and if so, I'll probably move to Nikon.
Daan B wrote:
At this point I am open-minded to all possible explanations.
I have talked with Canon Professional Services and have described the problem to them. I'm told that Canon will adjust any 5D II camera for noise levels using proprietary software. The woman I talked with at CPS was not aware of noise or banding issues at low ISOs but suggested I send my camera in for adjustment, which I'm going to do ... even though I'm not anywhere near as troubled by the pattern noise and banding as so many forum members are. I do want to see if Canon's adjustment makes any difference and will report that difference here. I will be sending them sample images showing pattern noise and banding.
As I've posted here before, I can create these effects at any ISO but only at the risk of destroying the integrity of the image. I can see some evidence of banding and pattern noise in properly processed images, but only at 100% and 200% viewing ... not when an image is printed on my Epsons.
I might add that when an image is printed offset (magazines, brochures, etc), the issue is even more moot, since the smallest dot gain during screen printing will practically obliterate any noise or banding that can be seen at 100%.
Anyway, it will be interesting to see the difference Canon's adjustments make ... if any.
veroman wrote:
I have talked with Canon Professional Services and have described the problem to them. I'm told that Canon will adjust any 5D II camera for noise levels using proprietary software. The woman I talked with at CPS was not aware of noise or banding issues at low ISOs but suggested I send my camera in for adjustment, which I'm going to do ... even though I'm not anywhere near as troubled by the pattern noise and banding as so many forum members are. I do want to see if Canon's adjustment makes any difference and will report that difference here. I will be sending them sample images showing pattern noise and banding.
As I've posted here before, I can create these effects at any ISO but only at the risk of destroying the integrity of the image. I can see some evidence of banding and pattern noise in properly processed images, but only at 100% and 200% viewing ... not when an image is printed on my Epsons.
I might add that when an image is printed offset (magazines, brochures, etc), the issue is even more moot, since the smallest dot gain during screen printing will practically obliterate any noise or banding that can be seen at 100%.
Anyway, it will be interesting to see the difference Canon's adjustments make ... if any.
I'm told that Canon will adjust any 5D II camera for noise levels using proprietary software.
Well, now, that's hellaciously interesting.
I have recently experienced the problem with true dark tones printed at Zone 3 (when there is also true, detail-free black that should be printed lower), such as the satin trim of a tuxedo.
You want to simplify my position on this issue so that you can make it seem absurd, don't you?
I suppose in a sense you might simplify what I wrote to "the problem doesn't exist." But before you continue down that path, please read what I wrote about this. My point was not that the 5D2 (or any other camera) is perfect or without room for improvement. In fact, my point was pretty much exactly the opposite of that, and it had more to do with what I characterize as an obsession with issues like this one in general.
I have yet to meet the perfect camera, the perfect lens, the perfect film/sensor, or the perfect much of anything. However, I have met less than perfect cameras, lenses, and film/sensors that do a phenomenal job producing photographic images.
If you - or I, right? - post an opinion in a forum we can all assume that there may be other different opinions. If your point is that the only people who should post in a forum on "noise banding on he 5D2" are those who agree that noise banding on the 5D2 is a major issue or who otherwise agree to your notions of what is proper to post, I think it would make more sense for you to get a blog and just post an article on your position and not accept comments or consider alternate points of view. In a forum it isn't up to you to say which posts are and are not appropriate - though I grant that you are also free to have an opinion about this...
In a later post you made a comment about (to paraphrase) "the pattern noise bothers everyone who has come across it" and "if you haven't come across it" just be happy and go away. Again, you seem to need to misrepresent that alternate point of view so that you can denigrate it. Although I'm clearly in the "stop obsessing about this" camp, I actually posted a sample image (in this thread, IIRC) that I said showed noise banding. So the "if you haven't come across it" point is merely a distraction. I can produce it... but only by a series of steps that typically include: significant underexposure, pushing the darkest areas of the underexposed frame by rather extreme amounts, typically also or alternatively altering curves/fill/black point. viewing the image at 100% magnification on the screen, and looking very closely.
My point is not that this particular artifact does not exist - I'm quite sure I've never written that, but please point out to me where I did if I'm mistaken. What I have written more than once - and here, yet again - is that I've made many thousands of exposures with this camera during the past year including quite a few with a dynamic range larger than that which the 5D2 sensor (or other DSLR sensors) can handle and that I've never encountered this problem in a way that would be remotely visible in a large print, and even when it is visible at 100% on the screen, better exposure technique would have resolved it.
I know this won't change your mind, but I don't want my point of view misrepresented.
Dan
(I later edited the quote to avoid the confusing attribution...)
Daan B wrote:
Oh, you guys are above it all...
Too bad this thread is AGAIN turning into a "the problem doesn't exist" vs "the problem does exist" debate... This is just sad
Can we just stick to finding ways to work around the "problem" - for those who are bothered by it anyway, PLEASE?
I'm jumping into this thread late, but just to add my few cents regarding the banding issue. I had one of the first 5d2's that came into the country back in December of 2008. I noticed the ISO 100/200 banding issue almost immediately.
The crosshatched banding pattern is mostly contained in the red channel and appears in darker areas of photos, especially foliage, or in blue skies or bluish water. It is a real issue that affects different bodies in varying degrees, with reports varying from very bad to none at all. This would indicate a wide variance in manufacturing and QC. It appears that the older the body, the more likely it will exhibit this behavior.
The higher the dynamic range of the photo, the more likely the banding is to appear in the darker areas. It will show up on prints in some circumstances even if the shadows are not pushed. Pushing shadows as little as 1 stop, which is very common in post processing, can render some photos useless. The banding cannot be removed with noise reduction software.
The most important aspect of the type of banding presented is that it is unique to the 5d2. No other camera to my knowledge exhibits this behavior. I have asked repeatedly in other online forums for an example from any other camera made that shows the same crosshatched banding pattern at ISO 100/200 and to date have received no examples from anyone. My 1d Mark III and 400d do not do this, nor does any other digital camera that I have used. Exposing to the right partially mitigates the issue at the risk of blowing highlights, but is not a real fix.
Having said all that, there may be hope for a repair from Canon. Many early adopters who saw the same problem tried sending in their 5d2's to Canon for repair only to get them back in the exact same condition. I decided to wait it out and keep in contact with them to give them a chance to address the issue. About 3 weeks ago I finally sent mine in, at which time Canon replaced a circuit board and recalibrated the camera. Upon its return, I took some high contrast test photos to determine if the banding issue had been resolved. At this point it appears that it has been, because to date I have not been able to duplicate the banding that appeared previously.
brainiac wrote:
Then the pattern would always be the same, and you could remove it using a frame as a negative mask.
Exactly. But I can't tell if it is, because I can't reliably reproduce the problem in large enough clean area to compare. Is there a way to reproduce it in uniformly gray or black image?
By the way, I noticed that I usually see it on downsampled images and most of it disappears if I look at the pixels in 100% view. So perhaps part of the problem is the bicubic algorithm in PS?
rminton wrote:
..... About 3 weeks ago I finally sent mine in, at which time Canon replaced a circuit board and recalibrated the camera. Upon its return, I took some high contrast test photos to determine if the banding issue had been resolved. At this point it appears that it has been, because to date I have not been able to duplicate the banding that appeared previously.
This is very good to know. As I posted earlier today, I'm sending my 5D II in to Canon Professional Services in Jamesburg, NJ, and hope to have a similar result. I'm also of a mind, based on what I've seen and read here, that the early production runs suffered the worst from banding and pattern noise. Could very well be that Canon released cameras with defective sensors, circuit boards or both.
gdanmitchell wrote:
You want to simplify my position on this issue so that you can make it seem absurd, don't you?
I suppose in a sense you might simplify what I wrote to "the problem doesn't exist." But before you continue down that path, please read what I wrote about this. My point was not that the 5D2 (or any other camera) is perfect or without room for improvement. In fact, my point was pretty much exactly the opposite of that, and it had more to do with what I characterize as an obsession with issues like this one in general.
I have yet to meet the perfect camera, the perfect lens, the perfect film/sensor, or the perfect much of anything. However, I have met less than perfect cameras, lenses, and film/sensors that do a phenomenal job producing photographic images.
If you - or I, right? - post an opinion in a forum we can all assume that there may be other different opinions. If your point is that the only people who should post in a forum on "noise banding on he 5D2" are those who agree that noise banding on the 5D2 is a major issue or who otherwise agree to your notions of what is proper to post, I think it would make more sense for you to get a blog and just post an article on your position and not accept comments or consider alternate points of view. In a forum it isn't up to you to say which posts are and are not appropriate - though I grant that you are also free to have an opinion about this...
In a later post you made a comment about (to paraphrase) "the pattern noise bothers everyone who has come across it" and "if you haven't come across it" just be happy and go away. Again, you seem to need to misrepresent that alternate point of view so that you can denigrate it. Although I'm clearly in the "stop obsessing about this" camp, I actually posted a sample image (in this thread, IIRC) that I said showed noise banding. So the "if you haven't come across it" point is merely a distraction. I can produce it... but only by a series of steps that typically include: significant underexposure, pushing the darkest areas of the underexposed frame by rather extreme amounts, typically also or alternatively altering curves/fill/black point. viewing the image at 100% magnification on the screen, and looking very closely.
My point is not that this particular artifact does not exist - I'm quite sure I've never written that, but please point out to me where I did if I'm mistaken. What I have written more than once - and here, yet again - is that I've made many thousands of exposures with this camera during the past year including quite a few with a dynamic range larger than that which the 5D2 sensor (or other DSLR sensors) can handle and that I've never encountered this problem in a way that would be remotely visible in a large print, and even when it is visible at 100% on the screen, better exposure technique would have resolved it.
I know this won't change your mind, but I don't want my point of view misrepresented.
Hey Dan, I seem to have misunderstood your intentions... Sorry about that It is just that I don't want to let this thread end in another pointless yes/no debate.
veroman wrote:
This is very good to know. As I posted earlier today, I'm sending my 5D II in to Canon Professional Services in Jamesburg, NJ, and hope to have a similar result. I'm also of a mind, based on what I've seen and read here, that the early production runs suffered the worst from banding and pattern noise. Could very well be that Canon released cameras with defective sensors, circuit boards or both.
