LR can be made to compensate for the channel imbalance, if you know how to do it. Unfortunately this has to be made on DNG files (I'm still not sentient enough within the LR program structure to be able to patch LR directly), so you get an extra step in the processing.
It actually has provision for BOTH an up-to eight column black-point calibration and up-to eight column amplification response calibration. And it also has provision for a way of telling the de-mosaic engine that the greens MAY have some colour response imbalance.
Now if Adobe could just apply those settings correctly, a lot of the problems would disappear. This does not take the heat off Canon, though, this IS a Canon hardware problem, not a converter problem. Software can be made to compensate for the built-in Canon hardware errors, errors that honestly just SHOULD NOT BE THERE! Unfortunately something really weird seems to be done with the topmost blacked out strip in the 7D, so the area that is usually used for calibrating blacklevels for each column is compromised by Canon. Also a hardware problem, made by Canon - not the converters.
Some 30 pages back or so I showed the difference between a "correctly applied & compensated" LR conversion and the beta conversion.
theSuede wrote:
LR can be made to compensate for the channel imbalance, if you know how to do it. Unfortunately this has to be made on DNG files (I'm still not sentient enough within the LR program structure to be able to patch LR directly), so you get an extra step in the processing.
It actually has provision for BOTH an up-to eight column black-point calibration and up-to eight column amplification response calibration. And it also has provision for a way of telling the de-mosaic engine that the greens MAY have some colour response imbalance.
Now if Adobe could just apply those settings correctly, a lot of the problems would disappear. This does not take the heat off Canon, though, this IS a Canon hardware problem, not a converter problem. Software can be made to compensate for the built-in Canon hardware errors, errors that honestly just SHOULD NOT BE THERE! Unfortunately something really weird seems to be done with the topmost blacked out strip in the 7D, so the area that is usually used for calibrating blacklevels for each column is compromised by Canon. Also a hardware problem, made by Canon - not the converters.
Some 30 pages back or so I showed the difference between a "correctly applied & compensated" LR conversion and the beta conversion....Show more →
You've partially answered a long-running question that I've had, which was "can this problem be fixed via a firmware update, or is it a hardware QC issue which requires swapping my 7D for a good copy."
The next question is whether it's a hardware QC issue, or a fundamental hardware design issue? There is a big difference between the two.
Many of us don't want a camera "marketed" at 18mp (which isn't truly 18mp visually, but still requires 18mp of storage space), we don't want extra workflow, and we don't want noisier LOW ISO images than what we're used to (re: 20D)
I understand that updated converters might be able to "mask" the underlying issue (at the cost of resolution), but that's not why we're paying Canon for this body.
As one reviewer aptly put it, "this is a body that we'd like to love." Sigh.
Hi,
I have it since 2 weeks this Eos 7D and I can easily say it is HORRIBLE!! after thousands of picture.
if you concern about image quality awfullt bad .no matter raw or jpeg format.jpeg picture is of course softer than raw which is usual but raw picture is also as bad as jpegs!!.
Please don't tell about high iso or etc.if you can't make sharp and clear picture at iso 100,so where you will use iso thousands?
I can tell about my lens,Canon EF 24-70mm f/2.8 L USM ,Canon EF 24-105mm f/4 L IS USM ,Canon EF 100-400mm f/4.5-5.6 L IS USM ....so which one is bad ones? if you use it at eos7D body all of them is giving very soft picturesYou have to edit them hours and hours photoshop or etc...I have also 5D MK2 and Eos 10D... do you know two of them much better than eos 7D picture. at least sharper.of course 5Dmk 2 sharper.But what about 10D?,This is a maybe 6 years old Dslr it is like a joke..
But if your piority is how handsome the camera, of course 7D will easily take a lead.feeling it at your hand you will feel more powerfull like a brick .All the button at the place where it has to be.Easy to use.Just connect with your full hd lcd tv via mini hdmi jack and look your soft!! picture at your sharp HD television screen this is what I am doing now.
I have to tell you this is my second copy of eos 7d,with latest firmware.Also you have to read some major camera review site.They cannot say very soft images but they were always telling about beta production model is problem maybe,but I didn't have beta production or early adopt one.
Before purchase please try it or rent it or borrow from friends 7D and put your best of the best L glass on it ,use tripod with the shutter release and mirror lock .Try to make sharpest picture with that brick!and don't be sorry about soft result ;this is not your fault it is because 7D!!
Ps.Forget about Af Micro adjustment .Because it is not at all for general settings for specific lens.you have to make all focal length!!! if you make at 70mm adjust ment,it will not working at 24 or 50mm.Trust me I tried hours and hours.
Fred Tedsen wrote:
For this image DPP is perfect. The same patches in Lightroom 2.5 and 3 beta fall apart with mazing at even modest amounts of sharpening (3 beta worse than 2.5). Artifacts show up in Capture One, but only with extreme sharpening.
I repeated this with my camera, but not so out of focus. The results were similar, with DPP completely free of problems, followed by C1, LR 2.5 and LR 3b. C1 still didn't show mazing but had a lot more artifacts.
Fred ... all LR/ARC support is currently defective (they are not production, fully supported 7D releases). Every single time we have looked at the issue it is the same ... heavy artifacting. On the Adobe forum, Adobe has hinted that they plan on releasing new software with 7D support in December. So in another three to six weeks we should have an Adobe update. The current problems are not a bad reflection on the camera/Canon or Adobe ... just a reality of Adobe's engineering software release approach.
rego wrote:
Hi,
I have it since 2 weeks this Eos 7D and I can easily say it is HORRIBLE!! after thousands of picture.
if you concern about image quality awfullt bad .no matter raw or jpeg format.jpg picture is of course softer than raw which is usual but raw picture is also as bad as jpegs!!.
Please don't tell about high iso or etc.if you can't make sharp and clear picture at iso 100,so where you will use iso thousands?
I can tell about my lens,Canon EF 24-70mm f/2.8 L USM ,Canon EF 24-105mm f/4 L IS USM ,Canon EF 100-400mm f/4.5-5.6 L IS USM ....so which one is bad ones? if you use it at eos7D body all of them is giving very soft picturesYou have to edit them hours and hours photoshop or etc...I have also 5D MK2 and Eos 10D... do you know two of them much better than eos 7D picture. at least sharper.of course 5Dmk 2 sharper.But what about 10D?,This is a maybe 6 years old Dslr it is like a joke..
But if your piority is how handsome the camera, of course 7D will easily take a lead.feeling it at your hand you will feel more powerfull like a brick .All the button at the place where it has to be.Easy to use.Just connect with your full hd lcd tv via mini hdmi jack and look your soft!! picture at your sharp HD television screen this is what I am doing now.
I have to tell you this is my second copy of eos 7d,with latest firmware.Also you have to read some major camera review site.They cannot say very soft images but they were always telling about beta production model is problem maybe,but I didn't have beta production or early adopt one.
Before purchase please try it or rent it or borrow from friends 7D and put your best of the best L glass on it ,use tripod with the shutter release and mirror lock .Try to make sharpest picture with that brick!and don't be sorry about soft result ;this is not your fault it is because 7D!!
Ps.Forget about Af Micro adjustment .Because it is not at all for general settings for specific lens.you have to make all focal length!!! if you make at 70mm adjust ment,it will not working at 24 or 50mm.Trust me I tried hours and hours. ...Show more →
Resolving power is not a problem I have with this camera. Here's an image taken with the 100-400 @ 180mm, wide open at f/5 in very bright unflattering light. Default Lightroom sharpening:
And one more, taken with the 800 f/5.6 IS, wide open, 1/250s, ISO 800. This was processed in DPP and then finished in PS, don't remember if noise reduction was applied.
I'm not saying these images are great, but resolving power even at slightly higher ISOs (800 in my case) on real world subjects with a real world workflow is quite good. I've printed each of these images and for all of them the prints are spectacular.
I love your style of complaining. . Glad your rant is only limited to this thread. You've clearly missed some of the pics I've been posting here for a month now. Good luck, bro !! And, yes, welcome to FM ! This thread has become a joke...adios !
rego wrote:
Hi,
I have it since 2 weeks this Eos 7D and I can easily say it is HORRIBLE!! after thousands of picture.
if you concern about image quality awfullt bad .no matter raw or jpeg format.jpg picture is of course softer than raw which is usual but raw picture is also as bad as jpegs!!.
Please don't tell about high iso or etc.if you can't make sharp and clear picture at iso 100,so where you will use iso thousands?
I can tell about my lens,Canon EF 24-70mm f/2.8 L USM ,Canon EF 24-105mm f/4 L IS USM ,Canon EF 100-400mm f/4.5-5.6 L IS USM ....so which one is bad ones? if you use it at eos7D body all of them is giving very soft picturesYou have to edit them hours and hours photoshop or etc...I have also 5D MK2 and Eos 10D... do you know two of them much better than eos 7D picture. at least sharper.of course 5Dmk 2 sharper.But what about 10D?,This is a maybe 6 years old Dslr it is like a joke..
But if your piority is how handsome the camera, of course 7D will easily take a lead.feeling it at your hand you will feel more powerfull like a brick .All the button at the place where it has to be.Easy to use.Just connect with your full hd lcd tv via mini hdmi jack and look your soft!! picture at your sharp HD television screen this is what I am doing now.
I have to tell you this is my second copy of eos 7d,with latest firmware.Also you have to read some major camera review site.They cannot say very soft images but they were always telling about beta production model is problem maybe,but I didn't have beta production or early adopt one.
Before purchase please try it or rent it or borrow from friends 7D and put your best of the best L glass on it ,use tripod with the shutter release and mirror lock .Try to make sharpest picture with that brick!and don't be sorry about soft result ;this is not your fault it is because 7D!!
Ps.Forget about Af Micro adjustment .Because it is not at all for general settings for specific lens.you have to make all focal length!!! if you make at 70mm adjust ment,it will not working at 24 or 50mm.Trust me I tried hours and hours. ...Show more →
theSuede wrote:
LR can be made to compensate for the channel imbalance, if you know how to do it. Unfortunately this has to be made on DNG files (I'm still not sentient enough within the LR program structure to be able to patch LR directly), so you get an extra step in the processing.
It actually has provision for BOTH an up-to eight column black-point calibration and up-to eight column amplification response calibration. And it also has provision for a way of telling the de-mosaic engine that the greens MAY have some colour response imbalance.
Some 30 pages back or so I showed the difference between a "correctly applied & compensated" LR conversion and the beta conversion....Show more →
Can you say a bit more about the calibration metadata in DNG? What do you mean by eight-column blackpoint calibration -- is it meant to apply the same compensation with a periodicity of eight columns? What do they do for the D3/D700, which has 12 channel readout? Also, it seems that some of the column banding from amplification does not have such a periodic structure in the 7D.
Furthermore, though I haven't had time to check it precisely, a bit of preliminary fiddling showed that there was no 4-channel matrix transform of the RGGB data that would align the green on all squares of the color chart, so it really starts to look like different spectral response rather than misaligned CFA. One can see that in the many-color chart examples I recently posted, where the grayscale patches seem to have different levels of mismatch, and the white patches are way off.
I looked at the review, also. It seemed like his big issue was IQ, sharpness specifically. So being stuck here with the flu, I decide to take a quick look at sharpness myself. I don't know what his problem was ... if I had to guess, he didn't AF micro adjust his lens or possibly the explanations I give at the end of this post ... whatever, he got pretty bad results.
Top image in each set is 7D, bottom image is 5DII. Both shot with 135L (f/5.6 on 7D, f/8 on 5DII). Direct flash, high speed sync. HANDHELD. Framed in a similar manner, but not perfect (7D seems to have gotten the short end of the deal). EXIF in full images. Top image set is the full frame, bottom set are 100% crops.
Here is how the images were processed. Processed in C1 PRO v5, standard setting for NR (and virtually everything else). Sharpening standard setting is .8, 180. I used .8, 230 on both images. Not a big deal (scale is 0 to 1000). Full frame images were downsized in Photoshop, same sharpening applied to both ... something like smartsharpen 50 (very mild).
I am not trying to do a review here ... just a reasonable, quick quality check. One thing that was going through my mind was the 135L is a very high quality lens (good for this kind of test), maybe I should also shoot a comparison with a lesser lens.
Of course if you were processing these images, you would apply output sharpening in Photoshop. Here is a smartsharpen of 100 on the 7D file. Nothing special about the value of 100 just wanted to illustrate how the file might take a little sharpening in Photoshop.
For those wondering what you would get if you followed the Darwin Wiggett (author of 7D review) approach to preparing your files. Here it is DPP processed no NR, no sharpening, everything else standard. 7D top, 5DII bottom.
What does it mean? There has been a trend with camera manufacturers where they have begin to rely on the extensive processing power of today's computers to really complete the processing of their images. I first saw this when there was a huge delay in ARC/LR support for my Panasonic LX3. The delay was caused because Adobe had to implement optics corrections, CA corrections and who know what on the RAW file in the converter. Very nice results on LX3. Adobe has added a whole technical capability for processing RAW files and making adjustments. The 7D may be very dependent on the performance of the converter ... where people have optimized settings to pull the images together. Defeat all of those judgments and you may end up with junk. For myself, I would be careful defeating the initial settings in C1 when I start processing an image (when a RAW converter costs $400, you hope you are getting some expertise in processing the images optimally).
One other possible explanation for Darwin's problems is that possibly DPP was not entirely ready when the 7D was released. There may be hope for DPP tuning and ACR/LR will probably be very good when we see it.
Here is what C1 produces with NO NR and NO sharpening (or anything else ... like a little contrast ... the output could be improved from here even with NO NR and sharpening). Obviously took a different approach than DPP.
droopy1592 wrote:
I really appreciate you taking the time to do this and bring light on the subject.
Likewise. It's pretty clear that most of the complaints about the camera here are due to one of the following errors:
- comparing 100% crops with much lower resolution cameras, i.e. unequal magnification
- misfocus
- camera shake
- using an incompatible raw convertor
- paranoia about noise levels which will not survive the printing process
- unrealistic expectations
People just don't understand why comparing 100% crops favours the lower resolution camera, so they go on doing it and making misleading pronouncements on the forums.
I think these crops should show how tricky the 7D design (or perhaps QC, but it would be like almost all copies bad so far which seems hard to believe, although it would be preferable I think).
7D (getting dotted hints along the maze-pattern if not solid maze lines, DPP tends to zipper fine details and that might be what is breaking the lines into dotted lines and/or they are fighting mazing using certain techniques but, at the least, they still leave behind more noise that the otherwise amazing sensor should need have, i'd honestly rather have 15% worse high iso than this myself): http://skibum4.smugmug.com/photos/711347424_seY3F-O.jpg
2nd set is Luma NR 0, Chroma NR 2, Faithful, Contrast 1, Sharpness 10 (pretty high), Otherwise 0 (high sharpness to bring out what is going on):
Now maybe ACR 5.6 will do better than either of these, but it certainly shows how tricky the 7D makes it. Maybe in the end you can coax near 18MP detail and reasonably suppress this stuff but it's not easy for a converter, at all. And it really seems hard to believe that what is gained is worth this. One thing, on DxO, PRE-demosaic testing it might actually give it a nice boost for SNR even if the effective SNR is much reduced after de-mosaic without trying all sorts of tricky stuff. Hopefully it wasn't a decision to look good on DxO realworld be damned (although I doubt that since they probably figure more people look at places like DPR anyway). Maybe I am wrong, but it seems a bit frustrating to general do such a great job and then make this seemingly odd design decision with the 4 spectral CFA. G1 does almost always have 10% to even more often 20% less standard deviation so maybe they let more through G1 to help things while keeping G2 stronger to try to maintain some better color discrimination and chroma noise performance but it still seems to me this makes things very tough for converters and that it is very easy to lose more than you gain. Again, I could be wrong.
But maybe ACR 5.6 will do a reasonable job regardless.
brainiac wrote:
Likewise. It's pretty clear that most of the complaints about the camera here are due to one of the following errors:
- comparing 100% crops with much lower resolution cameras, i.e. unequal magnification
- misfocus
- camera shake
- using an incompatible raw convertor
- paranoia about noise levels which will not survive the printing process
- unrealistic expectations
People just don't understand why comparing 100% crops favours the lower resolution camera, so they go on doing it and making misleading pronouncements on the forums.
I agree with most and much of what you've mentioned in this thread, and your input has been invaluable. Keep it coming!
I think you bring up a great point about noise levels surviving the printing process. Although many of us like to share images via the Internet, it's rare that we do so at full resolution. That's usually reserved for printing. And I'd bet that very few of us have actually printed ANYTHING of size with the 7D yet. So, paranoia in terms of the "finished product" (e.g. the print) may very well be erroneous and moot. Time and our experiences will tell. However, the current pixel peeping (for many samples) isn't exactly creating a lot of confidence when comparing against previous experiences with other bodies where viewing images at 100% showed "great/perfect" expected and perceived sharpness. The keeper rate should be higher with the 7D.
RE: using an incompatible raw converter. Is DPP the only converter we should judge images by? What about the JPEG output straight from the camera? Should we dismiss the on-board converter as irrelevant, or as a legitimate part of the overall analysis of the camera? I will state that even without full-resolution pixel peeping that my old 20D shows far less noise up to ISO 800 than my new 7D when looking at JPEG images straight from the camera. And for some people, that might be all that matters. There is also a subjective "harshness" to many images I see posted here. What happened to the "silky smooth" Canon images that many of us had previously grown accustomed to? (See skibum5's examples in the post directly above.) That might be part of the "unrealistic expectations?"
With regard to expectations, I'm not sure how to define unrealistic. I expected more detail and IQ than I'm getting with my 7D copy. I expected less noise at LOW ISO settings (<= 800) with my copy. Design compromises likely made with this sensor may be a contributing factor to this discussion, as well as potential QA/QC issues. I'm not sure, and at some point I may send my copy back to Canon for "calibration," or have it replaced at Best Buy. Individual expectations and results may vary, as we've seen with different copies and different workflows.
Other than these specific "IQ disappointments," I'm thoroughly satisfied with the 7D body (my copy). I'm going to hold on to it and shoot for a couple of months and see what happens in the converter marketplace, and stay tuned for any discoveries made here, including workflow. So far, the 7D seems to require much more time spent with workflow. That's not necessarily a good or bad thing, but I think we'd all agree that time is usually better spent shooting than fixing.
I'm going to put you on the spot now. Even if you may not own one (I'm not sure if you do), does the 7D meet your expectations, and why? If not, why not? Were you expecting "more" or "less?" Have you printed anything of size yet with the 7D?
cameron12x wrote: RE: using an incompatible raw converter. Is DPP the only converter we should judge images by? What about the JPEG output straight from the camera? Should we dismiss the on-board converter as irrelevant, or as a legitimate part of the overall analysis? I will state that even without full-resolution pixel peeping that my old 20D shows far less noise up to ISO 800 than my new 7D when looking at JPEG images straight from the camera.
There's a difference between stating it and demonstrating it. If you want to demonstrate it, I'm sure people here will be grateful, but please please please be sure to uprez your 20D file to the same pixel dimensions as the 7D file before making the comparison. You did remember to compare them that way didn't you?
>There is also a subjective "harshness" to many images I see posted here. What happened to the "silky smooth" Canon images that many of us had previously grown accustomed to? That might be part of the problem.
It might be all of the problem. Do you see that harshness in prints, or is it merely a result of doing false comparisons at 100% without first uprezzing your previous 'silky smooth' files?
>With regard to expectations, I'm not sure how to define unrealistic. I expected more detail and IQ than I'm getting with my copy. I expected less noise at LOW ISO settings (<= 800) with my copy.
Are you talking about more detail in your prints, or in your 100% views of files? Can you show us the shortfall of detail in your 7D files next to a similar file from your 20D, uprezzed of course, to 18 Mpixels?
Design compromises likely made with this sensor may be a contributing factor to this discussion, as well as potential QA/QC issues. Individual expectations and results may vary, as we've seen with different copies and different workflows.
What design compromises? Do you have actual information about those compromises, or is this speculation provoked by 100% comparisons with much lower resolution files?
>So far, the 7D seems to require much more time spent with workflow.
...that is again, I'm afraid, speculation. It seems as though people are spending more time on their files to make their pixels look as good, not to make their files look as good.
That's not necessarily a good or bad thing, but I think we'd all agree that time is usually better spent shooting than fixing.
Presumably you mean shooting beer.
>I'm going to put you on the spot now. Even if you may not own one (I'm not sure if you do), does the 7D meet your expectations, and why? If not, why not? Were you expecting "more" or "less?"
I do not yet own one but I have used one, and I have seen a hell of a lot of uploaded material which shows it to be an excellent camera, and a few uploaded examples which show only that badly processing files and viewing them at 200% can reveal a certain level of noise. I have seen the effect of moving from 13 to 21 Mpixels, and it takes getting used to. The noise gets finer in the image, but more noticeable in crops. That's expected.
The reason that I don't own a 7D is that I have two 5D2's and I prefer full-frame cameras. If the 7D were a full-frame camera and had a manual focus screen option, then I would own two or three of them, and no 5D2's.
I tried really hard to generate mazing in the the kindly uploaded colour chart file. I found that I could spot it at 200% in mustard yellows if I set all contrast sliders to maximum and pumped DPP sharpening up maximum, at which level you might expect it to wreak havoc. Considering the magnification, the very modest sharpening artefacts (maze?) are a triumph and a credit to the 7D: http://cyberphotographer.com/7d/mazy.jpg
The morals of this story are:
(1) don't believe everything you read on the internet - use a camera to make a print.
(2) don't apply extreme sharpening in DPP before making enormous prints, but I thought everyone knew that.
brainiac wrote:
What design compromises? Do you have actual information about those compromises, or is this speculation provoked by 100% comparisons with much lower resolution files?
I do not yet own one but I have used one, and I have seen a hell of a lot of uploaded material which shows it to be an excellent camera, and a few uploaded examples which show only that badly processing files and viewing them at 200% can reveal a certain level of noise. I have seen the effect of moving from 13 to 21 Mpixels, and it takes getting used to. The noise gets finer in the image, but more noticeable in crops. That's expected. ...Show more →
I'm heading outside to shoot with my 7D, so only a couple of minutes to write. Will write more at a later time.
Nearly everything on Earth ever engineered or made by humans is subject to design parameters and, ultimately, compromises. I don't want to get into a philosophical discussion about that here, and I'll leave it at that. And of course I don't have any inside information about the design criteria which Canon designers and engineers have used with respect to this sensor, and I doubt if anyone on this forum has such information, either. If I did have such knowledge, a NDA would be in place I would not be able to comment. Probably not a realistic question ask.
I'm not positing that the 7D is a bad camera (and never have). If I considered it such, I would have returned mine long ago. My observations and "disappointment" only centers around certain aspects of IQ.
In your last post you mentioned yourself that "it takes getting used to moving from 13 to 21mp." Please elucidate in more specific detail? That could be helpful.
Well, I've just made an even larger jump, from 8 to 18mp. And I must admit that I'm not liking everything that I "see." And the body of evidence is mounting that other people (and reviewers) are "seeing" some of the same things (most notably, higher perceived noise than expected at LOW ISO). And for the masses, perception can often times be interpreted as reality.
Maybe we all should have expected that. If not, Canon has some educating to do? I don't intend to turn a blind eye. That being said, there still is MUCH to like about this body. I don't need to say that again.
Please note that I've been shooting Canon bodies since the late 90s, and have no intention of switching to another vendor at this point.
Have you made any prints or have you seen any prints made with the 7D? If so, what size? I believe that is the next step for me. Making large prints with the 7D.
brainiac wrote:
I tried really hard to generate mazing in the the kindly uploaded colour chart file. I found that I could spot it at 200% in mustard yellows if I set all contrast sliders to maximum and pumped DPP sharpening up maximum, at which level you might expect it to wreak havoc. Considering the magnification, the very modest sharpening artefacts (maze?) are a triumph and a credit to the 7D:
The morals of this story are:
(1) don't believe everything you read on the internet - use a camera to make a print.
(2) don't apply extreme sharpening in DPP before making enormous prints, but I thought everyone knew that....Show more →
1. I am not against high MP counts. I own a 5D2 and actually considered the 18MP of the 7D a big plus compared to say the 8MP of a 20D
2. One thing for sure is that it is absurd to call the 7D a triumph of resistance to artifacting! Come on, I can boost contrast and set sharpness to 10 in DPP with my 15MP (close to 18MP) 50D and see NO artifacts, silky smooth yellow and yet the 7D under those conditions is riddled with artifacts and no upsamling the 50D to 18MP or the 7D down to 15MP does not make them look even in this case.
Now maybe something other than DPP will do it better. ACR 5.6B looks potentially good (the DPR samples are soft compared to what I can get, but sharpening them even a ton doesn't appear to bring out maze, although maybe the jog compression got rid of it, but still hopeful). Maybe you don't have to decry as much as I do, but you surely can't call it a remarkable triumph at avoiding such artifacts when it avoids them less well, clearly, than any other Canon body.
3. his copy also has the most even column to column response which probably helps his a bit compared to most copies
Did a little more analysis of UCSB's colorchecker RAW file. Here is the fractional difference (G2-G1)/(G2+G1), after median filtering both channels to reduce the column gain fluctuations:
The deviation from pure gray is an indication of channel imbalance. The patches that seem to be different are the black patches, where we are dividing by a small number which amplifies fluctuations; but the rest is remarkably uniform, spot checking indicates the fractional difference of the two greens is a few tenths of a percent, which would not be enough to cause serious nastiness.
This is quite different from skibum's first copy, which showed a substantial difference in yellow in one direction, and a difference in cyan in the opposite direction. I'll have to go back and dig up my data on that one, but I seem to recollect the difference was several percent. Here, UCSB's copy doesn't show any dramatic difference between yellow and cyan patches.
Tentative conclusion would be a bad batch among the first run of cameras which afflicted skibum's copy but not UCSB's.