On this one I let the sensor heat up a bit (45 sec) in live mode prior to taking the shot. No difference that I could see from the other shots that I took.
After reviewing the posts I realized that I was shooting with my 16 35 lens capped and should have been using the body cap only. This one is taken with 1/100 shutter speed and the body cap.
ejmartin wrote:
I have not done any testing of the D90, but I have done on its sibling the D300, which shares the same Sony sensor. First of all, Nikon clips the raw data at the low end, sometimes quite severely, so that an average exposure of zero photons corresponds to a negative raw level, which of course the ADC cannot output. (This is in contrast to Canon, which sensibly applies a positive offset so that zero illumination corresponds to some positive RAW level that varies by model, recent 14-bit Canons have the offset at level 1024 out of 16383 levels).
Assuming that comment about Nikon to be correct, thank you. There are various ways to deal with noise. (I have a lot of experience with this from audio.) You can try to mask it, you can try to reduce it, you can try to increase the signal strength relative to noise, or you can set a threshold below which you simple chop off the entire signal. (Audio folks might know of a technique known as noise gating...)
It is, in a sense, a philosophical choice by the designer. The choice has implications for the photographer. If one camera simply reduces all values below some point (loosely, the point at which noise generally might be perceived to be an issue) to zero, this is not exactly the same as producing a signal with no noise. Instead the signal (and the noise) have been eliminated at low levels.
In some cases this can be a fine choice, but it others the photographer might prefer to have control over this.
For example, if you have shadow noise after pushing a bunch in post you have a number of choices about what to do. You could look at the resulting print and decide whether or note it is significant or even visible and do nothing about it. You could decide that a bit of noise reduction would cover the noise to such a degree that it becomes a non-factor. You could even do your own threshhold adjustment (noise gating, if you will) in post and reduce all values below some arbitrary low point to zero/black.
Which is best? It is a choice. My preference would be to have the largest dynamic range signal available to me to work with in post. If this creates a problem then I have the option of deciding how/whether to resolve it. Someone else might prefer a more automatic noise gating process in which the problems that result from very low levels of detectable light and normal noise of detectors lead to artifacts.
If there is a problem and Canon can do something to improve it, I'm all for that. On the other hand, some of the underlying issues here go so far as to state certainty that there is a "problem," that Canon misled consumers, and so forth. A bit of discretion wouldn't hurt in that regard - even while the "testing" continues.
Meanwhile, I'm making a lot of photographs with my 5DII and I'm just not having any real world problems related to the kerfuffle going on here.
p.11 #10 · Are you seeing 5D2 Criss-Cross Pattern?
While I don't have a 5D II (plan on buying one though) nor a 1Ds III, and thus by any means shouldn't have an opinion in this, I do have to say I agree with Dan 100%.
I simply haven't got a clue about my 5D and 1D MkIII concerning this "issue" as I have never gone to these lengths trying to push the limits this far. I usually limit myself to +/- 1 stop and sometimes will go as far as +/- 2. If I feel I need more, I bracket. Reality check: are we seriously investigating the problems of trying to squeeze 12 stops of DR into a single image? If so: what's next?
Now if it's clearly visible in prints of say 40 x 60 cm. at ISO 100 and pushed 1, maybe 2 stops: you've got a point. But I don't have the feeling this is the case here...
p.11 #11 · Are you seeing 5D2 Criss-Cross Pattern?
Buster H wrote:
After reviewing the posts I realized that I was shooting with my 16 35 lens capped and should have been using the body cap only. This one is taken with 1/100 shutter speed and the body cap.
Now it looks similiar to the samples from Brainiac and Jonathan. I have edited the comparison post accordingly.
p.11 #12 · Are you seeing 5D2 Criss-Cross Pattern?
Emile Gregoire wrote:
While I don't have a 5D II (plan on buying one though) nor a 1Ds III, and thus by any means shouldn't have an opinion in this, I do have to say I agree with Dan 100%.
I simply haven't got a clue about my 5D and 1D MkIII concerning this "issue" as I have never gone to these lengths trying to push the limits this far. I usually limit myself to +/- 1 stop and sometimes will go as far as +/- 2. If I feel I need more, I bracket. Reality check: are we seriously investigating the problems of trying to squeeze 12 stops of DR into a single image? If so: what's next?
Now if it's clearly visible in prints of say 40 x 60 cm. at ISO 100 and pushed 1, maybe 2 stops: you've got a point. But I don't have the feeling this is the case here......Show more →
This whole thing started when Brainiac posted images where when the shadows were lifted in a workflow similar to what he was doing with a 1Ds3 he was seeing patterned noise in the shadows. Noise that would be obvious in a print.
Just because people have identified a series of steps (which by any measure are subjecting the files to extreme abuse) which makes it easier to see the patterns and to determine if it is a common property among all 5D2s doesn't change the original reason for this, Brainiac seeing the patterned noise in his 'real world' images.
One shouldn't allow what they see in this thread to dissuade them buying a 5D2. However one shouldn't write it off entirely as people just trying to find fault with the camera.
p.11 #13 · Are you seeing 5D2 Criss-Cross Pattern?
I should also add that *I* haven't experienced any real world issues with ISO 100 shadow noise during the time I had a 5D2, including the many landscapes I did in Death Valley with the camera.
p.11 #14 · Are you seeing 5D2 Criss-Cross Pattern?
thedigitalbean wrote:
This whole thing started when Brainiac posted images where when the shadows were lifted in a workflow similar to what he was doing with a 1Ds3 he was seeing patterned noise in the shadows. Noise that would be obvious in a print.
The last sentence was my caveat. I stand corrected!
p.11 #15 · Are you seeing 5D2 Criss-Cross Pattern?
In all the years I've been shooting digital, I've yet to find a need to push a RAW capture more than +1 EV in post. (More than that for me is a "write off"--I just screwed the exposure.) Consequently, I've never seen anything like this in my own 5D2 files.
I do not wish to discount Brainiac's findings or minimize the concerns expressed here, just second the perspective of a number of others: the 5D2, in real world shooting, delivers amazing IQ (IMHO).
I do think this thread is interesting in illuminating what happens at the extreme limits of the camera's design envelope.
p.11 #16 · Are you seeing 5D2 Criss-Cross Pattern?
thedigitalbean wrote:
I should also add that *I* haven't experienced any real world issues with ISO 100 shadow noise during the time I had a 5D2, including the many landscapes I did in Death Valley with the camera.
+1
I can't make it happen when I push my images but others can so it is possible.
p.11 #17 · Are you seeing 5D2 Criss-Cross Pattern?
Perhaps I can add to this that no camera has allowed me to get away with pushing shadows the way my 1Ds3 did. But don't understimate the value of that technique. DR is often a problem, viz. the popularity of HDR. HDR composites are pretty boring to do, and a camera with large DR is a real boon. That's why I'm disappointed with this aspect of my 5D2. I don't believe anybody here hasn't taken pictures which could have benefitted from more DR, and if you haven't tried the technique of pushing the shadows, then you don't know what you're missing. Especially on a mk3.
p.11 #18 · Are you seeing 5D2 Criss-Cross Pattern?
Well, the "criss cross" pattern seems to be there on all 5D2's if you look for it (in this case by the "lenscap" procedure).
It doesn't seem to matter if the 5D2 has FW 1.0.6 or 1.0.7 installed. It doesn't seem to matter what RAW converter is used. It even doesn't seem to matter if the 5D2 has been serviced by Canon for this problem. There also seems to be some inconsistency between 5D2's. Brainiac's 5D2 shows less noise at ISO100, while other 5D2's show less noise at ISO 200 (when compared to ISO 100). ERichard's 5D2 even shows the criss cross pattern when NOT pushing anything. It also seems that the 1Ds3 is doing about a full stop better in this regard. And although some ISO pattern can be detected in the 1Ds3 files as well, it is much less agressive (maybe smoother) than what we see in the 5D2 files.
So, I guess Brainiac's concerns are valid.
If this "phenomenon" gets in the way of shooting depends highly on personal shooting and PP methods. But I think it is fair to say that when you need the largest DR possible, especially when you need to pull out some detail out of dark areas of the frame, the 5D2 is not the most suitable candidate. Of course depending on how far you want to open up the shadows, how a particular 5D2 copy generates noise and what ISO is used.
If you shoot a 5D2 it might be a wise thing to see what ISO setting will be the cleanest. It may well be that ISO 200 gives you cleaner images than ISO 100. And then again, it may not
I am still worried about the appearant IQ inconsistencies between 5D2's...
