20
Replies
Thomas Geist wrote:
What I am talking about are profeiles that simulate the color response of in-camera JPEGs how the camera manufacturer set it. And what some users find pleasing and desirable.
Yes it's very subjective - a matter of taste. But it wasn't different with film, was it? You didn't always want a film with accurate color response. You might have wanted one with subjectively pleasing colors.
Is it really that people find exactly that particular colour response more pleasing and desirable, than all the other myriad, various and distinctive possibilities - or is it that they wish to match the camera's treatment with their Raw, because they miss the certainty and familiarity that JPG would have given them, and don't wish to step outside that?
I think it actually is different with film, or more broadly, that there is a "film" mindset or attitude, vs a "digital" one. People certainly did (and still do) like the effect produced by a particular physical process or stock. But if they wanted that, they used that - accepting all its consequences; and not just because they had to, or out of sentiment or elitism, but because that was part of the integrity of that medium.
For example, Polaroid or Daguerrotype or Kodachrome not only gave a distinct look, but also imposed distinct working tactics that were valued as part of the overall package. People didn't set out to manipulate method/product A into exhibiting the exact behaviour that was characteristic of method/product B. Method A's job was considered to be sufficiently fulifilled, if it produced the characteristics of method A reliably.
Now perhaps A included a broad range of possibilities, and B was very restricted and specific. But that was even better reason to use B, not A, when you wanted that particular effect. And that is how I have since come to regard camera JPG - if one likes it, that is IMHO what one should shoot - and if there are technique allowances to be made, surely that comes with the territory! Why accept an approximate simulation of the real thing, when there's no need to. And if shooting Raw because JPG won't deliver the goods - then Raw is now your medium, not JPG - so why should it matter if the results are not the same as with JPG?
This is absolutely NOT intended as a criticism - I sincerely do find a lot of these "simulation" ideas quite puzzling and strange. I won't go so far as to call all this quixotic, or a fool's errand... but it does seem to me like a lot of added trouble, rather than any kind of a simplification. And I - clearly - do not understand or appreciate what possible benefit it gives - but perhaps that's just me 
.
Not sure about ALL cameras, but most/many conform to Design Rule for Camera File Systems or DCF which is simple EXF data that species the color space (usually sRGB or 'other') for JPEGs. Not raw data!
Raw data doesn't have a color space per se. Raw converters have to interpret or assume some color space when processing.
To clarify, I was talking about the in-camera settings for JPG colour calibration, incorporated AFAICT as some kind of extended metadata on the Raw data. Of course this metadata also includes a lot of things that 3rd-party Adobe software can't / won't use. But whichever colour mode you had active at the time of shooting, gets in some way functionally recorded as a recipe into the file, and LR's support for the camera allows this to be presented as an "embedded" choice in the calibration menu - at least, with Pentax native DNG
So, the "Embedded" option is specific to the current image, applying whatever colour treatment the camera was internally set for "as shot" - but, via the DNG profile editor, can also be saved out as an independently-named profile - which can then be used on other images from the same camera, regardless of their "as shot" settings. The list shown includes profiles extracted this way, Adobe supplied ones which became available in subsequent LR upgrades, and just for chuckles, one made direct from a Colorchecker chart via the DNG profile utility.
With the accuracy thing I was referring to the "automatic" profile building tools like the one from X-Rite or the DNG profile editor's portion that reads in a Color Checker.
If I am mistaken, a colorimetrically accurate profile is all they can and want to build from that target.
Manupulating a profile or the sliders in LR's Calibration tab is the other story where I feel I can't do a thorough job. By that I mean, yes, I can tweak it to make one image or a certain group of images look good. But I personally already fail if I want to make a Color Checker shot look like the one that's showing Olympus' own genuine color rendition.
The Calibration tab sliders don't even work on a color's luminosity (for example I can only raise saturation for "reds" but not for dark reds etc.).
The secret or art (however you want to call it ) in building good profiles IMO is to make them universally applicable. At least as much as possible. The Adobe provided camera profiles do a good job there, as much as I feel the camera manufacturers with their own color rendition accomplish that fairly well.
I again want to refer to the thread where Eric Chan discusses with other users the evolution of the Nikon D3/D700 profiles. Making them similar, pleasing but also technically strong (shadow rendering and so much more that I barely understand) seemed to be quite a task. Four versions over the course of several years seem to prove it.
AdChoices