CMOS vs CCD and colour profiles a quick and dirty comparison

[eronald]

With wit like that, I can't believe you are not writing for SNL.

See, this is what happens when you defend your beliefs.
In due course you too will get older and learn to run with the crowd.

Edmund

[Bart_van_der_Wolf]

Goody, I think I finally understand what is meant by green separation: the green separation on the white shed in the Leaf capture, right Synn?  Definitely the CFA, the Leaf's awesome at bringing out them greens

I agree Jack, I'm (almost) sure Sandeep is jesting. Your wit is (in this case) dwarfed by his.

Cheers,
Bart

[AlterEgo]

Yes,they say, the grass is greener in Sony country, but I cannot afford a Sony
This is a Canon 1Ds3, which has horribly low ISO, precisely because it has a cleaner CFA than most modern cams.

Edmund

I am sorry - but where is that "green separation" in that small jpg ?

[synn]

See, this is what happens when you defend your beliefs.
In due course you too will get older and learn to run with the crowd.

Edmund

This is what happens when one spends their own time and provides free samples to the internet.
Yet another reason why I don't bother doing such charity work anymore.

[vjbelle]

Interesting that the topic of greens comes up. Apart from the skintones, this is where I see the Nikon being absolutely brutal. There is very little separation in the greens, no matter how one processes the files or how they try to equalize them. This is probably due to the CFAs, I don't know and frankly, I don't care.

These images were shot a minute apart from each other.

Nikon:



Leaf:



The Leaf image is a stitch, but that's irrelevant.

I normally don't get into these discussions but if these images were shot one minute apart then I would have expected the clouds to be very similar...... they aren't.  Just an observation.....

Victor

[AlterEgo]

I normally don't get into these discussions but if these images were shot one minute apart then I would have expected the clouds to be very similar...... they aren't.  Just an observation.....

Victor

that was a "charitable illustration", don't bother.

[AreBee]

Posted previously in a different sub-forum, the initial post on this page may be of interest.

Far from being poor at discriminating green, apparently Nikon excels at it.

[synn]

I normally don't get into these discussions but if these images were shot one minute apart then I would have expected the clouds to be very similar...... they aren't.  Just an observation.....

Victor

Clearly, you are not familiar with the wind in northern Germany.

Posted previously in a different sub-forum, the initial post on this page may be of interest.

Far from being poor at discriminating green, apparently Nikon excels at it.

So, some words are apparently more accurate than actual pictures that one can see.
Ok, then. On par with LULA discussions.

[torger]

Posted previously in a different sub-forum, the initial post on this page may be of interest.

Far from being poor at discriminating green, apparently Nikon excels at it.

Very interesting post, it would be interesting to know who's saying that and what information/knowledge he has. My observations has been that the color separation "issues" aren't that a big thing especially now when sensor noise is low, and you actually can tweak with profile. How many that actually do it is another aspect though, and really there aren't much profile control a user can do anyway, you get what the manufacturer gives to you, or you make your own with a complex open source hack like mine.

I should say though that my focus in my profiling work haven't been to find answers to color issues we like to argue about in the forums but simply to solve my problem -- make a profile software that can make high end color profiles for all-around use so I (and others that want) don't need to rely on manufacturer provided profiles.

I did run across several challenges in making good looking color, and some was indeed about color separation, not sensor-related though but for example related to RGB modulation where reds are compressed and makes them harder to differ for the eye that it should be. I found it to be a lot more interesting to tweak the profile maker's color model and found it to have much more profound impacts on color rendering than I imagine the fine-tunings of CFA has.

[AreBee]

synn,

...some words are apparently more accurate than actual pictures that one can see.

Good attitude: humility is the first step to redemption.

[Bart_van_der_Wolf]

I did run across several challenges in making good looking color, and some was indeed about color separation, not sensor-related though but for example related to RGB modulation where reds are compressed and makes them harder to differ for the eye that it should be. I found it to be a lot more interesting to tweak the profile maker's color model and found it to have much more profound impacts on color rendering than I imagine the fine-tunings of CFA has.

Yes, and even the impact I saw (with some example profiles) on perceived noise because of your much less aggressive (but still effective) approach towards avoiding gamut clipping. Kudos!

Cheers,
Bart

[Jack Hogan]

Very interesting post, it would be interesting to know who's saying that and what information/knowledge he has.

Ah yes, THE Suede, a giant in the industry, considered one of the best color scientists in town, participated in some forums to bring what he termed 'sane and reasoned feedback' to discussions by us amateurs.  He was quite helpful and very forthcoming with his encyclopedic knowledge until a fateful exchange with another giant over a year ago - about the particle-wave nature of light of all things.  I haven't seen him around since, quite a loss to the photographic community at large if I may say.  I hope he'll come back from his self imposed exile one day.

If TheSuede said it, I believe it.

Jack

PS Clearly keeping in mind that he wrote that almost three years ago when the latest generation of Nikons was not even on the drawing board, and Nikon may have tweaked CFA recipes since then.

[Bart_van_der_Wolf]

Ah yes, THE Suede, a giant in the industry, considered one of the best color scientists in town, participated in some forums to bring what he termed 'sane and reasoned feedback' to discussions by us amateurs.

Yes, I've always liked the posts by 'the Suede', very informative. But he does explain more about the CFA design than the profiled results.

Cheers,
Bart

[synn]

synn,

Good attitude: humility is the first step to redemption.

redemption is for those who have erred, not for those who are charitable towards perennial complainers too poor to own anything they complain about.

[AreBee]

synn,

redemption is for those who have erred...

Humour is for everyone.

[eronald]

I am sorry - but where is that "green separation" in that small jpg ?

ok, yellow separation, then

Edmund

[ErikKaffehr]

Hi,

Yes, I always appreciated info from THE Suede…

Best regards
Erik

Ah yes, THE Suede, a giant in the industry, considered one of the best color scientists in town, participated in some forums to bring what he termed 'sane and reasoned feedback' to discussions by us amateurs.  He was quite helpful and very forthcoming with his encyclopedic knowledge until a fateful exchange with another giant over a year ago - about the particle-wave nature of light of all things.  I haven't seen him around since, quite a loss to the photographic community at large if I may say.  I hope he'll come back from his self imposed exile one day.

If TheSuede said it, I believe it.

Jack

PS Clearly keeping in mind that he wrote that almost three years ago when the latest generation of Nikons was not even on the drawing board, and Nikon may have tweaked CFA recipes since then.

[AreBee]

Jack,

...keeping in mind that he wrote that almost three years ago when the latest generation of Nikons was not even on the drawing board...

I'm not so sure: theSuede mentioned one day later in this post "...the absolute DR of the latest Sony sensors..." and "...D800...".

[ErikKaffehr]

Hi,

I would suggest that this thread is about CCD vs CMOS in MF digital. The question here is if the new CMOS based backs have different rendition of colour compared to their CCD brethren. The link you refer to discusses 4/3 compared to MFD, mostly.

The next question is if the possible differences in colour rendition depend on image processing and weather those differences can be corrected by correct profiles or not.

Personally, I have a P45+ and some Sonys. Both systems are used with Lightroom. With the Sonys I generally used Adobe Standard profile, while with the P45+ I mostly used a dual illuminant profile generated by Adobe DNG Profile editor. With that stuff, I can only tell the images apart by looking at the file names or the EXIF.

One point here is that I have used Lightroom since 2006 and I was pretty happy with it's colour since Process Variant 2010, so I am adopted to Sony/LR/Adobe Standard.

On the other hand I built a new dual illuminant profile with DCamProf for the A7rII, and will do it for my older cameras, too. The enclosed image shows Adobe Standard on the left and DCamProf on the right. Same processing otherwise.

Best regards
Erik

Jack,

I'm not so sure: theSuede mentioned one day later in this post "...the absolute DR of the latest Sony sensors..." and "...D800...".

[ErikKaffehr]

Hi,

These plots compare Nikon D2X, Sony NEX5n and Canon EOS 5DIII. Note that the 2006 vintage D2X is very similar to the quite new NEX5n.



These compare (unidentified) Phase One, Sony NEX5N and H2D. My guess is Phase One is DALSA and H2D Kodak. I would say that Phase One is optimised for studio work, while the others have higher sensitivity for blue, which would be beneficial with incandescent lighting.


The data were taken from Torgers references.

Best regards
Erik