ruthenium
Offline
• • • •
Upload & Sell: On
|
 p.2 #10 · Skin tones updates on A7V or A7rV? | |
old-gregg wrote:
This will not work, because a profile must be made for each camera model.
A sensor+CFA combination generate data, not color. Just numbers. There is no rule that says that 23234 means "red". Therefore, if you're building a RAW converter, the first step is to use a test target with known color values, to build a lookup table which maps digital values coming from your sensor to color frequencies. The result is called sensor's native color gamut.
Sensor's gamut is much wider than monitors or printers. So... the next step is to decide what to "chop off", i.e. you have to perceptually map values from sensor's native gamut to the target color space, like sRGB or AdobeRGB.
Next step is to apply gamma and a tone curve because sensor data is linear, but human eye perceive luminance logarithmically.
Last step is to boost some colors and suppress others to create a pleasing image. Usually two complimentary colors are slightly boosted, while others are slightly muted, saturation is boosted or muted, depending on the purpose of a profile, like landscape, portrait, etc.
And finally, all of this is repeated twice: for warm-vs-cool white balance. And the result is baked as lookup tables and curves into a camera profile, which is a file (DCP for Adobe).
I hope it's easy to see why the sensor/camera manufacturer is irrelevant. The differences between sensors get erased at the 2nd step in the process. The differences in output you see are the differences in profiles, which means that you can take any two cameras and build your own profiles that will produce identical color.
The color differences between cameras that people argue about online are just profile differences.
To learn more:
https://helpx.adobe.com/content/dam/help/en/camera-raw/digital-negative/jcr_content/root/content/flex/items/position/position-par/download_section_733958301/download-1/DNG_Spec_1_7_1_0.pdf...Show more →
I agree with you on that observed differences between colors are mostly due to different interpretations of the raw files in post. There's another factor that complicates comparisons - different lenses. In real life, for example, a Sony body with a Sony lens isn't going to produce a raw file truly equivalent to one from a medium format camera system, even when we do our best to satisfy all the technical conditions of photographic equivalence. In real life, those who compare colors may not even have or use photographically equivalent lenses. And, to add to this is Auto-WB that most people use, that I don't believe can be the same in different cameras from different manufacturers.
I am willing to believe that one can, in principle, somehow obtain equivalent raw files from different cameras (e.g. by using the same lens with adapters), then develop and apply suitable custom profiles. All of this, to prove the point that the images should be equivalent, regardless of the camera make.
At the end of the above technical exercise, we shall return to the reality of 99% photographers (1) using different lenses, and (2) using the pre-defined color profiles in the applications of their choice for post-processing.
Thus, for an average photographer, in practice(!), one camera system may well produce "better colors" than another, and there will be little that this photographer would be able to do in practice.
The typical common advice is to use custom profiles. But, what is the problem with the profiles available in Lightroom, Capture One, DxO Photolab, etc.? One can make a custom profile for use with a stable illuminant, e.g. in a studio. I doubt one can make a universal prifile for different lighting conditions that would work better than the commercial profiles.
Perhaps the best practical alternative to custom profiles is to use a relatively flat profile available (when available) from Lightroom, Capture One, etc., as the starting point for post-processing.
The last practical point is about WB. This is of personal interest, and I would be interested in opinions. I have come across this suggestion from professional experienced photographers to avoid Auto-WB, and use Daylight in the settings. Obviously, this is a subject on its own. But, I think there might be something wise about this approach. We cannot really trust Auto-WB, so what's the point of using it, when too often Auto-WB needs to be corrected in post anyways?
|
...
