Wow, that is a considerable difference. Thanks for posting. Can you post these at 100% instead of 400% so it is easier to compare them against the 7D samples?
I may have overlooked it in an earlier comment, but my primary concern about the image referenced in the first post would be the rather poorly executed raw capture. The shadows are clipped and there is at least a whole stop of unused highlight headroom. If that was my image I'd be focusing my attention on improving my own photographic skills rather than pixel peeping the camera's performance. What is it about ETTR that people shooting raw don't get?
There is also much that can be done by selecting a different camera profile and applying a mask to the default sharpening, in Lightroom at least.
Daan B wrote:
I agree... it doesn't look like 5D2 shadow banding. Still, there are distinct vertical lines (like a pattern). Hence I call it "banding". But maybe it is just noise. I don't know. But it is difficult to kill in PP (NR) because you will loose (some) detail in other areas in the process.
Just work with 2 layers in PS. Kill the noise on the top layer, then erase out what you want to show with details with the bottom layer locked. Not such a big deal and I dont even see the issue here either. It's a great camera, just buy it, enjoy it, and stop trying to find inherent "problems" with it.
rogie wrote:
Just work with 2 layers in PS. Kill the noise on the top layer, then erase out what you want to show with details with the bottom layer locked. Not such a big deal and I dont even see the issue here either. It's a great camera, just buy it, enjoy it, and stop trying to find inherent "problems" with it.
Skies are especially problematic. Even slow film suffered from grainy looking skies. I use a similar technique to yours all the time with my old 10D, 40D and 50D landscape files as skies are very noisy at ISO 800 and higher. The 7D is a lot cleaner and doesn't need the treatment until ISO 1600.
It is a known property from the early days of the 7D that is has column fluctuations that are independent of exposure (it is a multiplicative effect, not additive). Usually it is not visible with normal contrast and gamma curves. Enhancing contrast makes it more visible, lifting exposure in post-processing does not.
Here is Pixel Perfect's sky patch resampled horizontally and vertically and with additional sharpening to emphasize the effect. When it is so weak that you cannot see it, it manifests as noise grain being oriented vertically more often that horizontally. After resampling where the rows are effectively averaged, it becomes apparent that there are column fluctuations (below). The same does not happen when you average the columns, as there are no row fluctuations to be seen (to the left).
Daan B wrote:
Stupid question maybe... but what are column fluctuations? And how do you average the columns so that the row fluctuations can't be seen? By resampling?
Do you think this is a design matter and all 7D's "suffer" from these column fluctuations?
Column fluctuations means that the effective sensitivity (ISO amplification) is not the same from column to column.
The averaging is done simply by resizing the image without proportion constraints, and keeping the original size on one axis.
I think all 7D's do it, but there may be some copy variation in how much. I would not bother trying to find a good copy, just get the camera and see if the total package works for you
alundeb wrote:
Column fluctuations means that the effective sensitivity (ISO amplification) is not the same from column to column.
The averaging is done simply by resizing the image without proportion constraints, and keeping the original size on one axis.
I think all 7D's do it, but there may be some copy variation in how much. I would not bother trying to find a good copy, just get the camera and see if the total package works for you
Thanks for explaining
So, a constrained resize wouldn't average out the column fluctuations? How about using sRAW? Or using in-between ISO values (like 125, 160, etc)?
A constrained resize will also average out the column fluctuations and reduce the visibility. The little exercise with separate row and column averaging was just to show that the fluctuations exist vertically but not horizontally. (Err, or was is the other way around? )
I dont think any of the other things you mention make a difference.
alundeb wrote:
A constrained resize will also average out the column fluctuations and reduce the visibility. The little exercise with separate row and column averaging was just to show that the fluctuations exist vertically but not horizontally.
I dont think any of the other things you mention make a difference.
I understand.
On the sRAW: this is basically a re-sampling/re-sizing too, right? So, why would it not make the effect less visible (as a constrained re-size in PS would do too)?
On the ISO values: it is just that some have commented that in-between ISO values give better results than full ISO values on the 7D. Don't know if this is true though.
Daan B wrote:
On the sRAW: this is basically a re-sampling/re-sizing too, right? So, why would it not make the effect less visible (as a constrained re-size in PS would do too)?
This is a very interesting thread. I was just about to start one on similar issues with my 7D. I can assure Davenfl and others that it is not "total nonsense"!
I debated long and hard about whether to upgrade my 40D with a 7D or wait for a 7D MKII or whatever it would be called. Ultimately, I decided to proceed with the 7D just before Christmas because a) everything I could find indicated it would be a year to a year and half before the 7D replacement came out and b) I had some important low light shooting situations coming up in the next year and the 7D was supposed to be better with high ISO.
My first serious test was a low light indoor sporting event. I was shocked at the noise at 5000 ISO (definitely didn't expect clean but didn't expect how bad it was) and even very disappointed with how bad it was at 3200 ISO. so I set about running a series of tests. My first test:
• My 7D and my 40D camera bodies
• My 24-105L lens, set to F11
• Tripod used
• Pictures shot in RAW
• All images shot ‘to the right’ of the histogram, but allowing little or no blown highlights
• High ISO noise reduction turned off (according to a Canon white paper, this does not alter RAW images anyway)
• Pictures viewed in Lightroom 2.7, using the Library’s Compare View (no adjustments made)
My 40D outperformed my 7D at 3200 and 1600 and was pretty much dead even at 800. So I proceeded to my second test:
• My 7D and my friend’s 7D camera bodies
• My 24-105L lens, set to F11
• Tripod used
• Pictures shot in RAW
• All images shot ‘to the right’ of the histogram, but allowing little or no blown highlights
• High ISO noise reduction turned off
• Pictures viewed in Lightroom 2.7, using the Library’s Compare View (no adjustments made)
My friend's 7D noticeably outperformed my 7D. So I documented all this, dumped a bunch of the images onto a CD and had the store I bought it at send it back to Canon. When it was returned, they send they upgraded the firmware, adjusted the CMOS, cleaned the sensor and made some other electronic adjustments. So I ran the same set of tests. My 40D still outperformed it. My 7D now was slightly better than my friend's 7D at 6400 and 12800 but was worse below 6400 to (but not including) 800. But now there was a new problem; I had very noticeable vertical banding in highlights and even some lighter midtones, from 100 to 1000. And the firmware was now 1.2.3.
Again, I documented all this, included pictures and sent it back to Canon. This time I asked for a replacement because I can't sell fine art enlargements with that much noise, not to mention the banding and I had lost confidence in my copy. They sent the same camera back. They had fixed the banding. Apparently there are two little dials for the sensor and they had adjusted one and not the other, thus causing the banding.
They did not indicate on the work order that they adjusted anything re high ISO noise; they only said that it was within specifications. I ran the first series of tests again. As mentioned, the banding was gone. It was apparent that they had tweaked the camera for ISO noise again, as the 7D now matched my 40D at 3200 and 1600, but definitely was not better. Now I have a new problem; the temperature and tint are both off (hot and magenta).
I asked for a refund from the store I bought it at, but they refused. The guy I was originally dealing with actually sounded like he cared, but the girl that gave my camera back the last time was quite snotty.
I am sooooo put off with Canon and the retailer. I truly regret having bought the 7D. While it is impressive that they added 80% more pixels and maintained the same noise as the 40D (after going through hoops!), I bought it for the better high ISO and for this it is an utter failure.
So 1) there does seem to be noteworthy descrepencies between copies,2) banding can definitely be a problem and 3) if you're buying a 7D, you are buying it for other reasons/features (certainly not the noise management).
Now I have to run a series of tests using the in-camera WB adjustments to see if I can fix that. I don't want to send it back to Canon and have it come back with yet another problem!
akin_t, your logic is baffling. So you have been using the camera for 8 or 9 months and you have not experienced it, therefore it isn't? I'm guessing you have never been struck by lighting either, therefore that never happens and "nobody thinks it's an issue because it really isn't"?
Clearly, there isn't a high precentage of 7Ds experiencing these problems or there would be more mainstream commentary in photo mags and so on. But if you search the web, there is actually a reasonable number of posts on this issue.
Daan, I won't get a chance tonight to post a banding example, but I will try to do so tomorrow. Regarding the noise, yes it was compared at the same magnification.
EverLearning wrote:
Daan, I won't get a chance tonight to post a banding example, but I will try to do so tomorrow. Regarding the noise, yes it was compared at the same magnification.
EverLearning wrote:
akin_t, your logic is baffling. So you have been using the camera for 8 or 9 months and you have not experienced it, therefore it isn't? I'm guessing you have never been struck by lighting either, therefore that never happens and "nobody thinks it's an issue because it really isn't"?
Clearly, there isn't a high precentage of 7Ds experiencing these problems or there would be more mainstream commentary in photo mags and so on. But if you search the web, there is actually a reasonable number of posts on this issue.
Daan, I won't get a chance tonight to post a banding example, but I will try to do so tomorrow. Regarding the noise, yes it was compared at the same magnification....Show more →
Eh slow down bro ...
He said, and I quote "It surprises me that nobody thinks of this as an issue" ...
To which I anecdotally replied, it isn't an issue because it doesn't affect a lot of users.
Stop fishing for things that aren't there, I'm just telling him why people aren't up in arms about the phenomenon he's experiencing.
