charlyw wrote:
Well you would see the same if you were to lay of whatever you are smoking. The 1D Mk.III has nice ISO/noise performance for the time it was released. It has only 10Mp though - so to level the playing field you have to severely downsize the image from the 7D. Before doing so you can apply liberal amounts of noise reduction as the details wiped out by this step were never recorded by the relatively low resolution sensor of the 1D Mk.III anyway. The images at ISO 12800 that are now trickling in are showing just this advantage and this translates nicely to lower ISO settings as well......Show more →
Jorkata has stated that he doesn't care about downsizing, etc. He is looking at 100% views, which is easily shown to be nonsense for nearly all applications. Adding pixels to a camera will always be a negative in his "calculations". I have finally hit the "hide me" button for this reason.
python2000 wrote:
Jorkata has stated that he doesn't care about downsizing, etc. He is looking at 100% views, which is easily shown to be nonsense for nearly all applications. Adding pixels to a camera will always be a negative in his "calculations". I have finally hit the "hide me" button for this reason.
I think it would be the sane option for me to follow your lead here. At times I like to pixel peep - when comparing shots from series to determine which of those is technically the best - but when it comes to comparing cameras you have to look at the bigger picture, something which he clearly isn't capable of...
In engineering terms, we sometimes use analysis referred to as "BLACK BOX” theory. This means you ignore the internals of a device and concern yourself only with the input and output. In this situation, it means the question is: for a given lens, exposure and ISO combination, which black box produces the better print of the same specified size.
You can change any of the parameters and run the comparison again, but not look inside the box. Understanding the guts may be important for other types of evaluations. But it is generally irrelevant to parametric analysis of input and output.
What technique did you use where the real estate shown for the 7D is smaller than the 20D? Are these randomly sized areas or are they supposed to be proportional to something, say 100% pixels, for example?
I made them quickly. Just cropped a patch out quickly, whether I cropped the same size doesn't matter, so why bother taking time to crop same size patch out? 200% pixels with no interpolation to make it easier to see no matter what monitor you are using.
the only issue of relevance with those crops is whether you see smooth smooth surfaces or whether maze/basket-weaving artifacts appear when using ACR (and perhaps other third party converters).
it seems my 7D has them, my 5D2 doesn't not did my 20D or 40D or 50D.
I don't know how many 7D's have it, probably not many, although there are some hints that it may be more common than with previous models where it was really exceedingly rare I believe.
skibum5 wrote:
I made them quickly. Just cropped a patch out quickly, whether I cropped the same size doesn't matter, so why bother taking time to crop same size patch out? 200% pixels with no interpolation to make it easier to see no matter what monitor you are using.
the only issue of relevance with those crops is whether you see smooth smooth surfaces or whether maze/basket-weaving artifacts appear when using ACR (and perhaps other third party converters).
it seems my 7D has them, my 5D2 doesn't not did my 20D or 40D or 50D.
I don't know how many 7D's have it, probably not many, although there are some hints that it may be more common than with previous models where it was really exceedingly rare I believe.
I agree with your methodology, it just didn't seem right based on the sizes. Hopefully there is a solution - whether it be caused by the RAW converter or some hardware issue.
Others have noted that even when the noise seems higher than expected, the quality of noise is more grainy/random than other cameras, and therefore more appealing or film-like. I wonder if this is how they are doing it, by somehow randomizing the channels a bit.
mfurman wrote:
Exactly. It appears as if we are reading about two (or more) different cameras. I do not think that there is any consensus ever achievable on the pages of this forum
More than two different cameras for sure. The original crop crowd, the FF crowd and the 1D replacement crowd, the Nikons are better crowd and a mix of everything in between. That makes it about 20 diffeent cameras by my count
surfnron wrote:
Ken - I've been following this thread for the last 50, (?), pages or so. You just have to pick out the parts that you want...
For instance, the discussion about "no one views at 100%". Maybe not, But I do crop images, so how the camera handles cropping is important to me.
I skipped the 50D - just could not get myself to "pull the trigger", but I'm looking at the 7D very closely. The images I'm seeing on N&W are encouraging - you might want to pop over there if you haven't...
I'm going to wait until Isee something that helps me make up my mind - that has not happened yet.
Ron...Show more →
Hi Ron,
The problem with digging through 50 pages is, well, having to dig through 50 pages
I've thought about checking out the images on the N&W forum but I don't get over there much and those guys are so good they could make a Kodak instamatic look good
I'm thinking I might just hold out for the 1D III replacement and skip this one until I find a compelling reason to buy a 7D. Maybe just to have a play with the video feature for wildlife videos but not sure I have have the time to learn another workflow process.
UCSB wrote:
Ken ... it is a great camera ... if you are interested, just go buy one and give it a try. We are not going to get a consensus because many people that are commenting do not own one and are not going to be able to form final opinions. Others, myself included, are taking and processing images, but really waiting for full support from Adobe to produce final images and do detailed evaluations.
I don't have to worry about waiting for Adobe to catch up. My main interest would mostly be for the video feature. I already have the 1D Marik III which is an outstanding wildlife camera and the reason I am into photography at all. What I would really like to see are some good video examples of action photography of things like birds in flight with a long lens.
My copy of the 7d can also show those maze artifacts in that orange color, but less pronounced.
skibum, do you remember from the early efficiency estimates, that the noise std.dev was different for the two green channels? My copy of the 7D measures the same, and the immediate interpretation is that the system gain is different for the two channels. Why would they do a thing like that? It could be that the CFA uses two different green filters now, and they compensate the amplification in some way, to make the channel amplitudes identical for white light. That would make it possible for generic RAW converters to produce a reasonable result. Only specialized RAW converters like DPP wil handle all colors and WB's without artifacts.
That would at least explain why I too only see this in C1 via DNG, not in DPP.
The problem with digging through 50 pages is, well, having to dig through 50 pages
I've thought about checking out the images on the N&W forum but I don't get over there much and those guys are so good they could make a Kodak instamatic look good
I'm thinking I might just hold out for the 1D III replacement and skip this one until I find a compelling reason to buy a 7D. Maybe just to have a play with the video feature for wildlife videos but not sure I have have the time to learn another workflow process.
The 1D series is not in my budget for the foreseeable future Ken. I'd really like to see what the "big boys", (you included), over in N&W can do with the camera. What some of the newer members are posting is encouraging though. I'm in no hurry now as my "season" is winding down, and I can wait until spring.
Ron
alundeb wrote:
My copy of the 7d can also show those maze artifacts in that orange color, but less pronounced.
skibum, do you remember from the early efficiency estimates, that the noise std.dev was different for the two green channels? My copy of the 7D measures the same, and the immediate interpretation is that the system gain is different for the two channels. Why would they do a thing like that? It could be that the CFA uses two different green filters now, and they compensate the amplification in some way, to make the channel amplitudes identical for white light. That would make it possible for generic RAW converters to produce a reasonable result. Only specialized RAW converters like DPP wil handle all colors and WB's without artifacts.
That would at least explain why I too only see this in C1 via DNG, not in DPP. ...Show more →
i think DPP checks the balance between the channels or particular readouts and applies compensation while other softwre does not
python2000 wrote:
I agree with your methodology, it just didn't seem right based on the sizes. Hopefully there is a solution - whether it be caused by the RAW converter or some hardware issue.
Others have noted that even when the noise seems higher than expected, the quality of noise is more grainy/random than other cameras, and therefore more appealing or film-like. I wonder if this is how they are doing it, by somehow randomizing the channels a bit.
it actually makes it less film-like, you don't a nice dotted, random noise but extended lines in pattern form
the high density and perhaps other stuff can make it get a more film-like grain (when it is not troubled by maze stuff)
I don't see how one can draw too many assumptions about a new camera or sensor based on the evidence provided by a beta version of a 3rd party RAW converter. It's simply too soon to tell, and in any case if the camera vendor's results don't show artifacts, then the onus is on the 3rd party software vendor to get it right.
Any camera maker's hardware may have design characteristics that require compensation in software for optimal results (such as a expanded green channel filter). In fact, as vendors push the envelope on pixel density and noise, this is likely to become the norm rather than the exception.
python2000 wrote:
Jorkata has stated that he doesn't care about downsizing, etc. He is looking at 100% views, which is easily shown to be nonsense for nearly all applications. Adding pixels to a camera will always be a negative in his "calculations". I have finally hit the "hide me" button for this reason.
I did it a long time ago! The amount of nonsense and BS that person is spewing is above everything I've seen at this forum.
I am far from an expert in pixel technology but there is certainly a lot of "noise" about noise. I will probably not use 6400 more than a couple of times a year but from my meager experience in post processing, it appears to be a non-problem. I took a shot of a object at f/2.8 at 6400 ISO, on a monopod, with only natural light from a window. I processed the RAW in DPP and used the DPP NR of 4 Luma and 11 croma (I think those are the terms...:-) The first image is a 100% crop of the RAW in ACR. The second image is a 100% crop after DPP processing and transfer to CS4 as a TIFF but converted to JPG for posting. The third is the full image after sharpening using CS4 resize and the Bicubic Sharpen. All reduced to 1024 for posting.
What is the problem? Totally non-scientific but real. I know I am not using my images commercially but damn, I am more than happy with the final IQ. YMMV and I am sure it does...
garyvot wrote:
I don't see how one can draw too many assumptions about a new camera or sensor based on the evidence provided by a beta version of a 3rd party RAW converter. It's simply too soon to tell, and in any case if the camera vendor's results don't show artifacts, then the onus is on the 3rd party software vendor to get it right.
Any camera maker's hardware may have design characteristics that require compensation in software for optimal results (such as a expanded green channel filter). In fact, as vendors push the envelope on pixel density and noise, this is likely to become the norm rather than the exception. ...Show more →
Well the thing is some few people had this going as far back as with their 10Ds and converters such as C1 and ACR do not compensate for it to this day.
A few people had it terribly with their 5D2 and eventually Canon replaced their units.
It is really something that Canon should fix, either fix it in the RAW as it writes it or have better QC.
I'm really not so sure it has anything to do with beta ACR support, since the final ACR for the 5D2 and 10D and so on didn't help any of those people any.
It's not an expanded green channel filter, it's that the readouts don't all read out with the exact same gain or offset.
I assume that you got the color accurate as well, since the brass/bronze looks a fair amount less "yellow".
To tell the truth I really wasn't concentraiting on the brass, but it is pretty close. It is not a deep, deep brass bezel and the light is reflecting off of it pretty well.
