The 5D2 has strong pattern noise at ISO 100

Petra: Have you read and understood this link?
http://www.clarkvision​.com/imagedetail/dynam​icrange2/

Does it look to you as if film has an easy time beating digital cameras? Note that the test was with a 1Dmk2 so there has been a couple of years of improvements in digtal sensors.

Have you read this link, with regards to noise, dynamic range and tonal range?
http://www.dxomark.com …urement-definitions/Noise

Have you taken a look at their measurements of a number of cameras? Here is a comparison of a 1Ds3, a 5Dmk2 and a 1Ds3 and a Nikon D3x:
http://www.dxomark.com …nd2)/Canon/(bra​nd3)/Nikon

As you can see, the 5Dmk2 really have good dynamic range - ability to see the difference between a light and a dark region in a scene.
And the tonality isn't bad either.
But the big issue that can't be seen in these graphs is that the noise isn't random, which means that the noise is inventing features in the image. With random noise evenly aligned around the expected value, the noise will not affect the tonality we see. It will actually increase the tonality our eyes will se in case the sensor is running out of AD bits. With the 5Dmk2, the patterns are instead killing bits/stops of usable dynamic range. But that isn't a general, hard, limitation with digital cameras, but a specific limitation with the 5Dmk2. Wanting the camera not not produce banding is not an impossible dream.

Uhm.... I do understand dynamic range, thank you. And no I haven't taken measurements of a number of cameras, I'm too buys USING my camera.

As per pattern, I honestly didn't see much of a pattern, not more than what I've seen from some other camera brands.... I guess my brain doesn't register that type of patterns well. I saw noise yes, when I tried to play with the OP's raw image, but not in some very strict pattern.

And I think that this "issue" is pretty extreme..... at least in real life, and how I shoot. So I rather use my camera, produce some good images that I get paid for. That's better time spent in my humble (ok, not so humble then) opinion.

Pixel peeping is not good for anyone.... the results on photo paper is.

CyberPet wrote in post #8783191
M'kay. Sorry if I'm not believing you.

Here is the proof: an LX3 raw file that demonstrates 10.6 stops of dynamic range with no pattern noise.

Yes I am pixel peeping, but in some cases I can see a pattern in the shadows with no brightening of the image with an editor even at iso100. Certain dark reddish browns seem the worse. No pattern in same conditions using my 5d (although less detail of course).

Can you see it in a normal size print? Arguable either way.

But it does does make me a little sad.

1) because is could have been avoided 2) because it will mean that I will need to upgrage to another camera when they keep the IQ and solve the pattern noise problem 3) because the used 5d2 price will be less at that point because it has the issue and the newer version doesn't.

CyberPet wrote in post #8784270
Uhm.... I do understand dynamic range, thank you. And no I haven't taken measurements of a number of cameras, I'm too buys USING my camera.

As per pattern, I honestly didn't see much of a pattern, not more than what I've seen from some other camera brands.... I guess my brain doesn't register that type of patterns well. I saw noise yes, when I tried to play with the OP's raw image, but not in some very strict pattern.

And I think that this "issue" is pretty extreme..... at least in real life, and how I shoot. So I rather use my camera, produce some good images that I get paid for. That's better time spent in my humble (ok, not so humble then) opinion.

Pixel peeping is not good for anyone.... the results on photo paper is.

That comment don't match well with your previous posts, where you very explicitly have used "impossible", which is a very hard word not leaving something up for discussion. So your "impossible" was based on you not having spent any real time testing different cameras or reading technical reviews of different cameras? I have never owned a car capable of higher speeds than 220km/h, so by implication it would be impossible to build faster cars?

What is your view of the the two edited pictures below? Do they clearly indicate to you that the scene was impossible to capture? Or do they just suggest that the 5Dmk2 is capable of great dynamic range but forcing the owner to work around the pattern noise?

By the way - this thread isn't about pixel peeping. Banding problems can be seen without zooming in on 100% crops. It's just a question of how much usable dynamic range that is reasonably to ask from a 5Dmk2, given the cost of the camera and given the marketing information from Canon.

kevindar wrote in post #8729644

HTTP response: NOT FOUND | MIME changed to 'image/gif' | Redirected to error image by FLICKR

full size

nicksan wrote in post #8730698

CyberPet wrote in post #8784270
Pixel peeping is not good for anyone.... the results on photo paper is.

This problem does not require any pixel peeing to see. It ruins prints of all sizes, even as small as 4x6. If you are going to resort to ridicule, may I suggest a more appropriate slur? How about "dynamic range peeping"?

Yes, I used the word impossible *without any noise*. BIG difference. If I had intended to take a shot of that scene, I would have bracketed the shots, as it's not any moving objects and it's a perfect candidate for a noise free subject with much more dynamic range than any digital camera can handle.

I would never expect to be able to push a photo like that just because I spent nearly 3K on a camera. It just seems absolutely unrealistic. I know this has been said 20 times on here already, but that photo really should of been an HDR. Even if I had a camera that wouldn't band like that when pushing the DR that much, I would still bracket and make an HDR because the end product would look better.

CyberPet wrote in post #8784529
Yes, I used the word impossible *without any noise*. BIG difference. If I had intended to take a shot of that scene, I would have bracketed the shots, as it's not any moving objects and it's a perfect candidate for a noise free subject with much more dynamic range than any digital camera can handle.

Fair enough. Realize, though, that doing that isn't always an option. The example shot was an example to show the type of noise in question. It was not intended to be an example of the situation in which there are no other or no better (e.g. HDR) ways of getting the shot!

The problem here isn't the fact that there's noise. It's that the noise has a pattern. Pattern noise is much harder to deal with than random noise, and is much more objectionable to look at as well. As has been already mentioned, it also tends to introduce false detail.

The end result is that the usable dynamic range of the 5Dmk2 is lower than it would be without the pattern noise even if it retained the same degree of noise at the low end. Because the amount of noise isn't what matters here, it's the kind of noise that matters.


Now, naturally there are a lot of people who don't care, generally because the shots they take don't need the maximum dynamic range the camera offers. And that's fine.

But that doesn't mean the problem doesn't exist. It only means that it's not relevant to those people, however common they may be.


The bottom line is this: someone who needs the kind of dynamic range the 5D2 theoretically has but in practice doesn't have must choose a different camera. It's as simple as that. And without a thread such as this one, such a person wouldn't have any idea this problem existed until he tried to actually use the camera's theoretical dynamic range, and at that point he's made a substantial financial commitment by purchasing the camera. It's far better to know about issues like this in advance than to have to discover them for yourself.

CyberPet wrote in post #8783191
M'kay. Sorry if I'm not believing you. But I guess we have to agree to disagree. If you push an image that far with any digital camera, you'll get pattern noise. The image is poorly exposed to begin with... period.

Petra, I'm not sure what you are disagreeing with here. Daniel shows pattern noise that rises above the "normal" random noise that is expected from boosting dark areas, and that the pattern noise interferes with getting any meaningful details out of those darker areas. He has compared this with other cameras that certainly show "normal" random noise but without pattern noise to this degree. Are you saying you disagree with this basic finding?

Saying that the example image is "poorly exposed" is off the point, actually, although if you look at the Raw image it pretty much does what it can to retain highlights while hopfully keeping some shadow detail, if little. You can certainly say that the image is a good example of stretching the capabilities of a camera insofar as DR goes, and you can say that, naturally, there will be noise when you try to boost the shadows, but the point of the post is that the pattern noise is an unexpected and unwelcome interference -- if we saw just random noise we could still collect some real detail.

This reminds me of the "black spots" situation with the 5D2 when it first got released. It was a real anomaly that affected some photos but not most. It showed up under close analysis, but for "most photographers" it was not an issue. OK, that's good that for most photographers and most photos it was not an issue, but it was still there. However, in the discussion of that anomaly, those that were examining and analyzing it were met with a huge amount of outright hostility from those representing "most photographers". That seems pretty strange, especially in light of the fact that Canon ended up acknowledging the problem and addressing it in firmware.

This is a similar situation, IMO, with a similar "big picture": most photos taken by most photographers won't have a problem, because we try to avoid these extremes in lighting or else we don't try to get the most out of the shadows that might be possible. But that doesn't mean it's bad to make that attempt, or that it's bad to note that some cameras have a problem at the extremes of their supposed capabilities.

Who knows -- maybe Canon will actually address this in a firmware fix, maybe it's something in the processing that just slipped through. Or, maybe it's something in the hardware that will have to be "fixed" in the next release. Daniel has noted that the problem hasn't been seen with the 7D. We don't know the root cause. But the good news is that for most of our photography this won't be an issue and the 5D2 will continue to be the great camera that it is.

pwm2 wrote in post #8783935
Not at all. He has a correct exposure for the dark parts too. Same photo with a different camera would not have tainted the dark parts with the banding.

And how can you say he is expecting the impossible, when the impossible is already possible to buy?

Bottom line is that its a crappy, uninteresting photo so who cares about the noise?

versedmb wrote in post #8784659
Bottom line is that its a crappy, uninteresting photo so who cares about the noise?

Heh. So clearly, if the shot were an interesting photo that couldn't have been obtained any other way but by taking a single shot, then everyone here would do an about face and agree that it's a problem.

Right....


:p

versedmb wrote in post #8784659
Bottom line is that its a crappy, uninteresting photo so who cares about the noise?

And you are being a goofball rabble-rouser!

tonylong wrote in post #8784841
And you are being a goofball rabble-rouser!

You said it in the nicest possible way Tony..

I was going to comment on that statement and would have been thrown out.. Sometimes it's better to keep my trap shut!!

yogestee wrote in post #8784981
You said it in the nicest possible way Tony..

I was going to comment on that statement and would have been thrown out.. Sometimes it's better to keep my trap shut!!

Heh! Under his thick and stubborn skin, I believe Michael is a nice and reasonable fellow!