Accuracy of "Adobe Standard" Camera Calibration profiles in Lightroom

[gpagnon]

Hi all,
 
This may be an old and recurrent question, but I haven't been able to find a satisfying answer. So, please bear with me :-)

It is my understanding that the Adobe Standard camera calibration profiles are developed by Adobe using color charts for accurate color matching (for supported cameras, of course), as opposed to more specific profiles (eg, for the Fuji X100, "Camera Velvia / Vivid") that try to reverse-engineer the camera "secret-sauce" settings that are applied in rendering a JPEG with a given effect/film emulation.
 
However, I have seen authoritative recommendations on the web that if you are seeking to follow a workflow aiming at consistency and tonal accuracy, you *need* to create your own camera profile (eg, a dual-illuminant profile using an X-Rite passport colorchecker or similar), as the Adobe Standard profiles are *not* a suitable starting point.  Now, I can understand that there may be sufficient variations due to the use of different lenses on the same camera body that may warrant a re-calibration due to tonal shifts, color separation, etc., but what about cameras that have a non-interchangeable lens (eg, Fuji X100, Ricoh GR, etc.)?
 
I wouldn't think that production-related variations within a single camera model would impact color rendering significantly, and I would trust Adobe's engineers much more than myself in making an accurate profile (using more controlled lighting sources, a larger number of color samples than those on an X-rite passport, etc.). It would be really baffling if, with all their resources and knowledge, Adobe's effort at making the profile would produce inferior results to what a user can do simply by shooting an X-rite passport outdoor and under artificial light,  so maybe I am missing some crucial piece of information here.
 
I'd be really interested in hearing some expert opinion on this issue (I have watched the Ligthroom courses series here on LuLa, and I see you guys have close ties to Adobe's engineers..)
 
thanks in advance for any comments!

[RikkFlohr]

Individual Variability of Camera Sensors is one good reason. I have multiple Canon 5DMKIIs and they exhibit slightly different response in the red sensitivity. They require individual profiles if I want choose to be managing color on that tight of a level.

[Jim Kasson]

Hi all,
 
This may be an old and recurrent question, but I haven't been able to find a satisfying answer. So, please bear with me :-)

It is my understanding that the Adobe Standard camera calibration profiles are developed by Adobe using color charts for accurate color matching (for supported cameras, of course), as opposed to more specific profiles (eg, for the Fuji X100, "Camera Velvia / Vivid") that try to reverse-engineer the camera "secret-sauce" settings that are applied in rendering a JPEG with a given effect/film emulation.
 
However, I have seen authoritative recommendations on the web that if you are seeking to follow a workflow aiming at consistency and tonal accuracy, you *need* to create your own camera profile (eg, a dual-illuminant profile using an X-Rite passport colorchecker or similar), as the Adobe Standard profiles are *not* a suitable starting point.  Now, I can understand that there may be sufficient variations due to the use of different lenses on the same camera body that may warrant a re-calibration due to tonal shifts, color separation, etc., but what about cameras that have a non-interchangeable lens (eg, Fuji X100, Ricoh GR, etc.)?
 
I wouldn't think that production-related variations within a single camera model would impact color rendering significantly, and I would trust Adobe's engineers much more than myself in making an accurate profile (using more controlled lighting sources, a larger number of color samples than those on an X-rite passport, etc.). It would be really baffling if, with all their resources and knowledge, Adobe's effort at making the profile would produce inferior results to what a user can do simply by shooting an X-rite passport outdoor and under artificial light,  so maybe I am missing some crucial piece of information here.
 
I'd be really interested in hearing some expert opinion on this issue (I have watched the Ligthroom courses series here on LuLa, and I see you guys have close ties to Adobe's engineers..)

For a deep dive into profile accuracy (and whether and when that is even desirable), start here:

http://blog.kasson.com/?p=12456

If you scroll down to the bottom of the first post, in the comments section you'll find links to all the other posts in that series.

I will quote from the first post a way to frame the discussion of accuracy:

The goal of a color reproduction system is to have the displayed and printed versions of the image reproduce visual sensations caused by the reference. Before discussing the theory and practice of color reproduction, we should consider exactly what “reproduce” means in this context. In The Reproduction of Colour, Hunt defines six possible objectives for color reproduction, which are paraphrased here:

Spectral color reproduction, in which the reproduction, on a pixel-by-pixel basis, contains the same spectral power distributions or reflectance spectra as the original.
Exact color reproduction, in which the reproduction has the same chromaticities and luminances as those of the original.

Colorimetric color reproduction, in which the reproduced image has the same chromaticities as the original, and luminances proportional to those of the original.

Equivalent color reproduction, in which the image values are corrected so that the image appears the same as the original, even though the reproduction is viewed in different conditions than was the original.

Corresponding color reproduction, in which the constraints of equivalent color reproduction are relaxed to allow differing absolute illumination levels between the original and the reproduction; the criterion becomes that the reproduction looks the same as the original would have had it been illuminated at the absolute level at which the reproduction is viewed.

Preferred color reproduction, in which reproduced colors differ from the original colors in order to give a more pleasing result.

There is specific information about ASP and the Sony a7RII here:

http://blog.kasson.com/?p=12524

Jim

[ErikKaffehr]

Jim,

Thanks for posting, I tried to write a "competent reply" but there are no simple answers, I guess.

Eric Chan's comment is very enlightening, have not seen it before.

Anders Torger has developed DCamProf and provided a couple of megatons of information doing that. He has a few findings:

• A colour checker is a good sample set for colour calibration
• Targets with more saturated colours no better for profiling
• Colour handling converters is sort of ancient
• Handling of highly saturated colours difficult, partly because of antiquated math in raw converters
• Phocus by Hasselblad and Capture One has very carefully made profiles, but they are still ancient
• Recent profiles from Adobe are very conservative on deeply saturated colours, giving somewhat "bland" results.
• Anders found that it takes quiet a few tricks to get stable profiles, in part caused by ancient math.
• I had some demo with a specific raw converter turning deep bluish purple into blue, indicates extreme sensivity to white balance

What I take home from that is that profiling methods normally used are a bit anachronous. It would make a lot of sense to have a profile delivering "accurate" base colour from raw pixels and than apply a "look" on it, but present technology doesn't really allow that. Both conversion and "look" are built in the profiles.


Best regards
Erik

For a deep dive into profile accuracy (and whether and when that is even desirable), start here:

http://blog.kasson.com/?p=12456

If you scroll down to the bottom of the first post, in the comments section you'll find links to all the other posts in that series.

I will quote from the first post a way to frame the discussion of accuracy:

There is specific information about ASP and the Sony a7RII here:

http://blog.kasson.com/?p=12524

Jim

[scyth]

Adobe's effort at making the profile would produce inferior results to what a user can do simply by shooting an X-rite passport outdoor and under artificial light,  so maybe I am missing some crucial piece of information here.

yes, you are missing something related to the word "simply", it is not "simple" at all = here is the topic http://forum.luminous-landscape.com/index.php?topic=100015.0 ( which shall be read along with dcamprof /mentioned above/ manual & tutorial )

[Jim Kasson]

yes, you are missing something related to the word "simply", it is not "simple" at all

I'll second that. Just getting images of a Macbeth target that are repeatable within 1 Lab or Luv Delta E from day to day takes a lot of painstaking effort. If two Macbeth targets are involved, it may (or may not) be impossible.

Jim

[dwswager]

Hi all,
 
This may be an old and recurrent question, but I haven't been able to find a satisfying answer. So, please bear with me :-)

It is my understanding that the Adobe Standard camera calibration profiles are developed by Adobe using color charts for accurate color matching (for supported cameras, of course), as opposed to more specific profiles (eg, for the Fuji X100, "Camera Velvia / Vivid") that try to reverse-engineer the camera "secret-sauce" settings that are applied in rendering a JPEG with a given effect/film emulation.
 
However, I have seen authoritative recommendations on the web that if you are seeking to follow a workflow aiming at consistency and tonal accuracy, you *need* to create your own camera profile (eg, a dual-illuminant profile using an X-Rite passport colorchecker or similar), as the Adobe Standard profiles are *not* a suitable starting point.  Now, I can understand that there may be sufficient variations due to the use of different lenses on the same camera body that may warrant a re-calibration due to tonal shifts, color separation, etc., but what about cameras that have a non-interchangeable lens (eg, Fuji X100, Ricoh GR, etc.)?
 
I wouldn't think that production-related variations within a single camera model would impact color rendering significantly, and I would trust Adobe's engineers much more than myself in making an accurate profile (using more controlled lighting sources, a larger number of color samples than those on an X-rite passport, etc.). It would be really baffling if, with all their resources and knowledge, Adobe's effort at making the profile would produce inferior results to what a user can do simply by shooting an X-rite passport outdoor and under artificial light,  so maybe I am missing some crucial piece of information here.
 
I'd be really interested in hearing some expert opinion on this issue (I have watched the Ligthroom courses series here on LuLa, and I see you guys have close ties to Adobe's engineers..)
 
thanks in advance for any comments!

I have never had good results with Adobe Standard.  I make dual illuminant profiles for my cameras and sometimes shoot specific profiles.  Though I profiles made with ColorChecker Passport seem to need some contrast reduction, especially versus Adobe Standard.

[Tim Lookingbill]

Recent profiles from Adobe are very conservative on deeply saturated colours, giving somewhat "bland" results.

As long as all color hues are bland to the same degree as was in the original scene (no hot colors) that's when I employ the saturation slider in ACR/LR.

Balance in saturation is most important with any image rendering. Hue can be easily fixed in HSL but usually corrects itself with just a saturation boost overall. It's the relationship of colors that must be preserved. Whether it's accurate doesn't matter much unless you're doing repro work of fine art paintings but even then a balanced relationship is still necessary.

Apply any profile whether custom or Adobe canned that first renders all color relationships equally and edit from there. It really doesn't need to be more complex than that.

[gpagnon]

I kind of sensed I was in for a beating :-)

It will take me some time to digest such wealth of information, but in the meantime I wanted thank you all very much for taking time to reply and explain the issue in such detail (special thanks to J. Kasson for the links to his blog, and to scyth for the link to DCamProf).

best
g.

[dwswager]

I kind of sensed I was in for a beating :-)

It will take me some time to digest such wealth of information, but in the meantime I wanted thank you all very much for taking time to reply and explain the issue in such detail (special thanks to J. Kasson for the links to his blog, and to scyth for the link to DCamProf).

best
g.

The larger point is to pick the profile the either gets you closest to your desired output OR makes it easiest/possible to get to your desired output state.

[MarkJohnson]

Without plumbing the theoretical depths, if you do landscapes, it's easy to make a Color Checker Passport camera profile for daylight and then just see which you prefer when working on your photographs. I often prefer the CCP profile; colours can be a bit livelier.

[rdonson]

For me with my Canon DSLRs ASP was terrible.  The reds just were terrible and especially noticeable on flowers like roses.  I bought the x-rite Colorchecker Passport Photo to overcome this.  It worked great.

With my Fuji X-T1 and now my X-T2 my life got simpler.  I simply choose one of the Fuji film simulations (often Provia) for a profile and I'm much, much closer to the look I desire without the need to use the Colorchecker.

I'm not interested in absolute color accuracy when I usually take a photograph.  If I were doing art reproduction that would be different.  I'm interested in something that looks pleasing and what I thought things looked like. 

[dwswager]

For me with my Canon DSLRs ASP was terrible.  The reds just were terrible and especially noticeable on flowers like roses.  I bought the x-rite Colorchecker Passport Photo to overcome this.  It worked great.

What I found with Canons too.  While with Nikons, it's more the greens that seem screwed up.  The Colorchecker also fixes blues in both.  For landscape, shoot 2 images.  One in very cool light and one in very warm light and make a dual illuminant profile.