Nikolas wrote in post #11159202
So can you answer my question or not? because the language was simple to understand.
You made a subjective comment I'm asking YOU personally to back it up
My comments were as clear as day.
Now put up or shut up.
Well, in a way, you answered your own question. You said:
I'd be interested as I can easily show banding at nearly any iso if i push the shadows hard enough.
The banding came from somewhere. It didn't come from the raw converter itself.
The difference with respect to the 5Dmk2 is that the pattern noise (not just banding) appears to occur earlier than most other forms of banding. That is, it occurs at "lighter" tones in the image.
I expect that some cameras are better than others in this regard, and some people have been lucky enough to get exceptional copies, such that the problem doesn't show itself any earlier than the more common forms of banding would.
But the fact that you can always produce banding by pushing the shadows is evidence that supports John's statement.
Just as with the 7D, the 5Dmk2 shows an imbalance in the green channels. This turns out to be relatively easy to show (thanks to Emil Martinec for the basis of the method I used. His method is here: http://forums.dpreview.com …m=1019&message=33384683):
First, pick up IRIS: http://www.astrosurf.com/buil/us/iris/iris.htm
Second, take any ISO 100 photo you want, preferably one that has a lot of out of focus areas, and load it into IRIS. Then open the command window by clicking on the button next to the camera button. Then type the following commands:
split_cfa r g1 g2 b
load g1
sub g2 0
mult 32
The image you see is the difference between the two green channels in the raw image, scaled up so you can see it. If you see vertical banding in areas without any detail then
your image will show pattern noise in the deep shadows. The only way it won't is if the demosaicing algorithm averages the two greens together in each quartet, in which case you
will lose detail. Since recent converters don't seem to show a maze pattern in the deep shadows on the 7D, this would cause the 7D images to show a little less detail than they would otherwise, and would thus require you to increase sharpening in order to compensate. This almost certainly explains why the 7D files require more sharpening than previous crop cameras in order to get the same level of sharpness as those previous cameras.
Want proof that Canon can make a camera that doesn't do that? It's straightfoward.
First, download this file:
http://www.artbyphil.com …aws/5DIIand1DsIIIRaws.zip, which was linked to by his review of the 5Dmk2 in comparison with the 1Dsmk3 here:
http://www.artbyphil.com …08_5DII_Review/index.htmlThen extract it. There are two raw files within, both taken at ISO 100. One was taken by the 5Dmk2, the other was taken by the 1Dsmk3. Perform the previously mentioned procedure in IRIS against each of those files. The 5Dmk2 file will show vertical banding in the difference between the green channels. The 1Dsmk3 file will show
random noise there, which is what you
should see if you didn't have a green channel imbalance.
Oh, and Canon hasn't historically reserved this level of quality to the 1Ds3, either. Download this file:
http://www.imaging-resource.com …40D/FULLRES/YIMG_7422.CR2, and perform the same test on it. You'll see no banding in it, either (look in the areas that lack any detail, in particular, and you'll see random noise there, not banding). That shot was taken with a 40D.
I'm sorry, but the fact that Canon's previous 1.6x crop cameras didn't have this issue proves that it
is not normal!!You can say all you want about how you shouldn't be pushing your shots and all that BS, but the fact of the matter is that Canon's current line of sensors
are flawed.
I, for one, won't be buying the 5Dmk3 if it shows issues like this. I'll jump to Nikon at that point. I won't stick with a company that insists on pushing garbage hardware at us. I've already verified that Nikon's sensors don't exhibit this issue, either.
EDIT: The 1Dmk4 doesn't show the issue either. See for yourself. Download a sample RAW image from it here:
http://movies.dpreview.com.s3.amazonaws.com/canon1dmk4/AK0C7227.CR2.zip (taken from the review here:
http://www.dpreview.com …noneos1dmarkiv/page16.asp), and perform the previously mentioned steps in IRIS against it.
