"Downgrading" for a very specific reason

[Pi]

Show me pictures from two cameras with a difference of 10-15 points where the difference is not adjustable to the RAW file with a decent workflow.

Is that an admission that you were wrong, before you changed the topic?

Now, tell the OP that he wasted his time and money.

 

[privatebydesign]

Don't quite know how you implied that from my reply, but whatever.

The off the sensor colour response is independent of the RAW converter, but we can't see that, what we actually see is entirely dependent on the RAW convertor and that relies on any subsequent camera profile. A yellow interpreted by the sensor data as 205, 207, 37 loses no quality being told to display at 207, 207, 34. So we can argue the benefits of a metric you can't see in a visual medium, or we can try to educate people that the way to get more accurate colours is not to buy a different camera but to make camera profiles. I don't care, I use a camera with a high rating (though that played no part in me getting it) and I use custom camera profiles. I come here to help people make educated decisions for them to achieve their own goals, and point out some common fallacies and misunderstandings, not endlessly debate highly technical measurements that even the measure's don't use, I leave that to you and Aglet.

As to the OP, he and I were in contact, as I suggested on the threads first page, and the 1Ds MkIII RAW files I sent him played a part in his decision making process to get one. I see no contradiction.

 

[Pi]

The off the sensor colour response is independent of the RAW converter, but we can't see that, what we actually see is entirely dependent on the RAW convertor and that relies on any subsequent camera profile. A yellow interpreted by the sensor data as 205, 207, 37 loses no quality being told to display at 207, 207, 34. So we can argue the benefits of a metric you can't see in a visual medium, or we can try to educate people that the way to get more accurate colours is not to buy a different camera but to make camera profiles.

The whole point is that we can see what we can't see. We see as "yellow" a wide variety of colors (spectral distributions), and the camera sees another variety of colors as "yellow". This is loss of information, and there is nothing we can do about that. The colors in nature belong to an infinitely dimensional space; we see a 3D one, and the camera sees a 3D space but a different one.

The most obvious example of what I am talking about are the Sigma sensors. They are able to produce vivid "neon" colors but they are extremely inaccurate, meaning far from the human vision. There is no processing that can change that. In any specific shot, you can make the face, for example, look accurate but then everything else will be far off.

 

[privatebydesign]

You talk as though you can only adjust an entire colour space, not tune individual colours within that space; but this is not true, you can, you can even tune individual tones of a specific colour within a space.

If you want accurate colours you can accept that a sensor records 207, 207, 34 as 205, 207, 37, just so long as you then make sure the software displays that "false" colour correctly at 207, 207, 34, that is all profiling does.

I have no specific knowledge of Sigma colour output, but the fact that software designed to profile Bayer array sensor colours might not be as accurate on Foveon readouts is not, necessarily, a surprise.

I would, however, be surprised if most people could pick what sensor was used based on colour output if they each had a basic 24 patch dual illuminant profile applied.

 

[Pi]

I would, however, be surprised if most people could pick what sensor was used based on colour output if they each had a basic 24 patch dual illuminant profile applied.

I would be surprised if they could not see that any two cameras produce different colors.

 

[privatebydesign]

You clearly don't use camera profiles.

 

[Pi]

You clearly don't use camera profiles.

Camera profiles can do just that - basically replace certain colors on the target with the right colors, in specific light. It is like painting over an image. What it cannot do is to fix the other colors, and by that, I mean, colors in nature and how we see them. Also, you need to profile in any given light, which nobody would do. With a camera having color vision close to ours, you just need to worry about WB, no more that that.

...but whether the camera can or cannot exactly and accurately reproduce the scene colors is intrinsic to the sensor response and independent of the raw converter.

 

[privatebydesign]

Not only do you clearly not use custom camera profiles, you clearly don't understand the concept of dual illuminant profiles.

One of the areas I have had some small success in is flower photography, it is a discipline and clientele that holds colour accuracy as a key aspect of even basic quality work. I have never met a flower photographer who cared a jot about what sensor they used with regards colour accuracy, they all make it themselves.

My point is, almost every camera made gives us pleasing colours, if "accurate" colours are important to you there is much you can do to mitigate any inherent or software induced issues, further, there is very little actual real difference between most current cameras inherent recording capabilites and none that will restrict you in your desire to be as anal as you like regarding colour accuracy, if you choose to.

The same is not true of other sensor metrics, but those are included in the DxO sensor scores.

 

[Sporgon]

Here is a good read:
http://en.wikipedia.org/wiki/Color_vision#Mathematics_of_color_perception

Proof that you haven't the foggiest idea what PBD is talking about.