maverick666 wrote:
I'm sure the shadow noise can be cleaned in RAW if Canon wants to, just like other company does but details will be gone uncontrollably.
So, if I understood you right..
Shadow noise this bad, and both horizontal and vertical banding are perfectly normal. At iso200. Yeah, right.
I'm sure you can, but all I'm saying that banding shouldnt be there. Neither vertical or horizontal. I'm really interested to see if brainiac got a dud, or if other 5D2's do this crap as well.
Shadow noise at low ISO (200) underexposed by one or two stops is going to be worse than the shadow noise in the same region at ISO 400 and ISO 800 respectively. It's going to look grungier than one might intuitively expect.
The issue is BANDING not noise. All our pictures have shadows, which are "underexposed" thats where the banding show goes on and that is where Richard is not happy with his camera.
Its just a test
brainiac wrote:
With respect you are all missing the point - both artists, and technicians.
If you don't understand this simple explanation, by all means ask questions, but please don't waste everbody's time with yet another great long exegesis of the 'expose to the right' technique which is something that everybody here has known for years, probably better than you, and which is wholly irrelevant to the general problem of shadow noise in 5D2 files at 200 iso.
I try to steer clear of the "you don't understand" responses. But I think I cannot avoid being explicit about that any longer.
While you and I have agreed on other issues in these threads, in this case I think that you do not understand. I'm not denying that you are seeing noise with banding in the uniform dark areas shown in this sample. (In fact, I earlier wrote something to the effect that this would be the factor that I'd want to look at more closely.)
I DO understand the issues with wide dynamic range scenes and how to deal with them in film or digital capture, and I understand the compromises that photographers always have had and always will have to make when the dynamic range of the scene exceeds that of the capture medium
It is the nature of digital capture that "failure" when exceeding the dynamic range of the capture system most significantly comes in two forms. At the bright end of the scale it manifests as "blown out highlights" where no luminosity values but the highest are recorded. At the dark end of the scale it manifests as a lower signal to noise ratio (e.g. more visible noise) and using fewer value levels to describe different levels (e.g. - banding).
I've made enough prints of photographs of high dynamic range scenes of the type you mention to understand very well these limits and how to deal with them in the camera and in post - and also to understand the effect that they will or will not have in a photograph.
It seems to me that you don't understand this context, which surprises me after seeing so many smart and insightful posts from you elsewhere. (e.g. - your clear reminder to people about the misleading effects of comparing 100% crops of test shots from different photosite density images, and some recent comments you made on a thread that degenerated into a Mac/PC argument.)
Anyway, good luck with this line of reasoning. I think I'll skip out on the chance to pursue it beyond this point - for now at least.
Andi Dietrich wrote:
The issue is BANDING not noise. All our pictures have shadows, which are "underexposed" thats where the banding show goes on and that is where Richard is not happy with his camera.
Its just a test
Banding is noise. Just a different form of it.
I understand Richard is unhappy with his camera. I interpreted you when you said "underexposed" by one or two stops to mean you took a picture, then push in post processing one and two stops. That's a lot different than saying a shadow that is say, Zone III, is underexposed because it's only a Zone III. I would call that proper exposure, which is what (I think) Richard is saying. "I exposed this image with the shadows being where I wanted them, with no pushing or pulling in post." I agree with him that his image should like fine like that (other than the fact that noise shows up in the shadows at low ISO).
If you have to push the shadows one or two stops to see the noise/banding, which I *don't* think Richard is doing (except for the one comparison he posted where he pushed it one stop which I think confused people), then to me it is a moot point. Though again, if I understand correctly, this is not what Richard is doing, so he has a valid point.
danmitchell wrote:
Good Lord, who said that?! I certainly don't "measure the quality of a photograph by the amount of the scene's information!" To do so would be pointless. Please don't misstate what I believe or what I said.
What I said is that if you want the best possible print that looks the way you want it to look (that's what you are talking about, right?) then the best starting point for exposure with digital capture is typically to aim for capturing as much scene data as possible.
This gives me the greatest potential to create the vision that I have for the print. I actually don't give a damn in an aesthetic sense what my histogram looks like. I do give a damn what my print looks like, and to get there I have to pay attention to some technical matters when I make the exposure.
I apologize Dan, I believe what you were saying (if I may attempt to paraphrase again), is that the quality of a capture is determined by the amount of the scene's information. And I would agree with that, although I don't believe that's the most efficient way to work. I have this crazy notion of trying to get the exposure closest to my vision in-capture. Call me old fashioned
Oh OH! Brianiac do have a way of stimulating controversy (just because you don't understand his point doesn't mean he's wrong or right)!! During the pre-release of the 5D Mk II, I believe, that Canon claimed it was to be their best high ISO product too. Why not wait for a new(er) product (PMA anyone?) so we have a bigger difference to rag about.
Just as 50D technology was to have "improved" the output of the (tooling depreciated but obsolete 1Ds Mk III) sensor used in the 5D Mk II the new 50D sensor technology is here for to Canon to "improve" a MK IV sensor.
The 50D has a pixel density that, spread over FF, would yield a 37/39 Mp. And, as Brainiac has shown us over copious previous posts, those pixels would not have to be enlarged as much as the 50D's 1.6x crop frame. This then should yield fine 3200/6400 ISO performance on a 30Mp ++ FF sensor (watch out Phase One, Leica S2 etc) and a different kinds of 50/100 ISO deficiencies?
Also, since FYI few Canon lenses can take ultimate advantage of a 50D sensor's resolution (my own experience), this will go double for 50DS like equal pixel density on FF-especially the edges. So we can soon enjoy may posts decrying the crippled IQ of so, so many Canon lenses. Meanwhile Canon will have a new marketing vehicle for their Mk II (III) lenses.
Andi Dietrich wrote:
The issue is BANDING not noise. All our pictures have shadows, which are "underexposed" thats where the banding show goes on and that is where Richard is not happy with his camera.
Its just a test
kidtexas wrote:
Banding is noise. Just a different form of it.
I interpreted you when you said "underexposed" by one or two stops to mean you took a picture, then push in post processing one and two stops.
Thanks for the free lesson but I never mentioned that I pushed the image, this is not interpretation, I call this invention.
Noise or banding? I guess it should be clear as banding is at least a very specific form of noise
Andi Dietrich wrote:
Thanks for the free lesson but I never mentioned that I pushed the image, this is not interpretation, I call this invention.
Geez dude. That's why I asked for a clarification on what you meant by L-1 and L-2. You said underexposure. I shoot a lot of film and I take that to mean, "My photo was a stop underexposed," and in order to get it to look right, I either push the film a stop, or try to rescue shadow detail in another manner. I don't normally interpret "underexposed" to mean that shadows are exposed for as shadows. That sounds like proper exposure to me.
And it wasn't a "lesson," just trying to clarify what you meant.
ChrisDM wrote:
I apologize Dan, I believe what you were saying (if I may attempt to paraphrase again), is that the quality of a capture is determined by the amount of the scene's information. And I would agree with that, although I don't believe that's the most efficient way to work. I have this crazy notion of trying to get the exposure closest to my vision in-capture. Call me old fashioned
Chris, thanks for the explanation.
To clarify my approach - as if it needs it at this point - I don't care one tiny bit what the sensor capture "looks like." (The RAW file actually doesn't "look like" anything at all - it is just scene data.)
I care deeply about what the photograph I'm trying to create will look like.
Whatever gives me the best opportunity to achieve that goal is the "right" exposure. With that in mind, you might go further and say that it is important to capture a scene in the form that provides the greatest technical and creative potential for your photograph.
That is surely not a radical or controversial idea by the standards of film or digital photography. It is the norm.
Brainiac - I don't know if you missed my question in the midst of all this bickering about what a proper exposure is.. but have you tried other raw converters? It is quite possible that the raw converter is responsible for the banding/blobbing you see. I haven't seen any problems with my 5D2 at high or low iso...
kidtexas, the "zone system" is not really a concept of the digital world. It is based on the fact that you can enlarge the DR by film processing. This is not the case with digital, the levels do the work of a spotmeter and a exposure table here (IMO) and DR is given by the sensor specs.
The problem is that all this talk of exposure does not provide any information, the question is if the 5Dii shows banding were it shouldnt and if Canon downgraded the sensor of the 1dsIII for some reason without letting us know
The problem is that all this talk of exposure does not provide any information, the question is if the 5Dii shows banding were it shouldnt and if Canon downgraded the sensor of the 1dsIII for some reason without letting us know
The 5D2 sensor is not the 1DsIII sensor. They told us that.
But everyone knows that automobiles have four wheels, so if it has four wheels, it must be an automobile, even if the manufacturer calls it a truck.
Andi Dietrich wrote:
kidtexas, the "zone system" is not really a concept of the digital world. It is based on the fact that you can enlarge the DR by film processing. This is not the case with digital, the levels do the work of a spotmeter and a exposure table here (IMO) and DR is given by the sensor specs.
The problem is that all this talk of exposure does not provide any information, the question is if the 5Dii shows banding were it shouldnt and if Canon downgraded the sensor of the 1dsIII for some reason without letting us know
I was only merely trying to figure out if your samples that you posted had their values adjusted by post processing.
thedigitalbean wrote:
I've yet to see anyone do a low ISO comparison that demonstrates the 5D2 or 1Ds3 being worse than their predecessors. If you have both cameras, this should be a trivial comparison to make so I don't understand why it would just be 'murmurs', concrete examples should be easy to show. I have however seen lots of compelling evidence to show that neither the 5D2 nor the 1Ds3 can compare to the D3x in low ISO.
I did some testing in that respect. There were some comments on this forum and elsewhere that the noise goes dramatically up if you "push" the image 1.5 or 2EV in post processing. I found it interesting and did some tests and compared the effect of the pushing to my 5D archive. I would say that 5D2 does not like the pushing as much as 5D, the 5D raws were always cleaner in comparison. On some images the noise is distracting - but again, I am talking about exposure that was corrected +1.5EV in post processing - film would do the same. Well exposed images may show some noise at ISO 100 and 100% in Photoshop as well, but it does not bother me, in fact it looks more natural in a way.
I don't have 5D anymore and could not do side by side comparison with the same scene. In addition, I am not convinced that the noise should be attributed to the camera sensor or processing. It could be because of immaturity of the 5D2 raw converters.
