ICC Profile Choice - ProPhoto RGB

[David Grover / Capture One]

The answers some of us are looking or is on the ceiling? 


King Father: Some day, all this will be yours.

Son: What, the Curtains?

You can try the ceiling if you like, but you might have more luck if you scroll up to Post #63

[AlterEgo]

Please see below from Esben at Phase One.

Thank you... is it possible to ask Esben to post a small simple article for posterity and as a point of reference @ http://blog.phaseone.com/category/tech_talk/ ?

That will be a nice thing !

[digitaldog]

You can try the ceiling if you like, but you might have more luck if you scroll up to Post #63

Down actually... What I see is:

Anyways, the thing is that we actually do most of the work in the native camera space. In fact, we can pretty much work in the native camera space all the way to the output file.

What IS this so called native camera color space you're assuming?

The gamut you can see in the camera profile does not limit the gamut used for internal processing in Capture One.

Confused. The camera doesn't have a color gamut yet we're told you do most of the work in the native camera space who's 'gamut' isn't limited in the sentence above.

[digitaldog]

However, the conversion from the camera space (the “colorimetric interpretation” if you will) is based on ICC profiles.

I'm asking about this colorimetric interpretation which has a gamut (the camera doesn't). What is the gamut of your colorimetric interpretation? We're told it's based on an ICC Profile so it should rather easy to plot this (ideally in ColorThink), maybe provide a gamut volume. Then we have an apples to apples method of considering the processing gamut and our options for encoding this into a working space.

[AlterEgo]

What I see is: What IS this so called native camera color space you're assuming?Confused. The camera doesn't have a color gamut yet we're told you do most of the work in the native camera space.

what David is referring to is similar to what RPP does = demosaick, WB, curves, exposure correction, levels, you name it (including "Luma" curve - this is not with L from Lab as far as I can guess)  etc, etc are done with camera RGB numbers (what some people call "camera color space" albeit it is not a proper colorimetric color space like for example Prophoto coordinates with gamma = 1 or something else)... color transform (based on whatever data in icc/icm containter serving as C1 camera profile) is used at the very last stage in the C1 data pipeline - when you display the preview, calculate RGB numbers for color readout, output to .TIFF/.JPG (unless with special provision for camera profiling) or /as we were told today/ possibly start using C1 Color Editor to create C1 layers for "local" color adjustments (this is some kind of deviation).

my $0.02 based on interpretation of Esben's words passed by David (and Esben did 1-2 short postings before directly)

[tho_mas]

What IS this so called native camera color space you're assuming?

it is different (individual) for every single camera...

The gamut you can see in the camera profile does not limit the gamut used for internal processing in Capture One.

Confused. The camera doesn't have a color gamut yet we're told you do most of the work in the native camera space who's 'gamut' isn't limited in the sentence above.

should be easy to understand for a wise guy like you: the (individual) camera profile is not as large as C1's internal (working) space. It's basically the same as in Adobe software: there's a large working space (ProPhoto primaries) but so called camera profiles limit the actual colors to create a decent, "neutral" look.

I'm asking about this colorimetric interpretation which has a gamut (the camera doesn't). What is the gamut of your colorimetric interpretation?

again, different for every single camera...

We're told it's based on an ICC Profile so it should rather easy to plot this (ideally in ColorThink), maybe provide a gamut volume. Then we have an apples to apples method of considering the processing gamut and our options for encoding this into a working space.

this is exactly what you can do ... all "camera profiles" are available as icc profiles and you can load it in the gamut viewer of your choice.
Can you, vice versa, show me the "gamut" of any Adobe "camera profile" (so "input profile")?

[AlterEgo]

the (individual) camera profile is not as large as C1's internal (working) space.

based on Esben's words C1 does not work in any proper colorimetric color space till the last stage of the pipeline (output to...) - it simply works/manipulates with camera's RGB numbers, color transform (that's where camera profiles are used) is not applied yet at all when you for example do exposure correction or use RGB curve in C1 UI... as to the order of most basic operations : demosaick, WB, raw converter curves (transfer function and .fcrv curves like film, linear, linear scientific, etc) that is unknown vs exposure correction for example... one can assume they happen first, then UI operations like exposure correction, contrast, saturation (again on camera's RGB numbers)

[digitaldog]

it is different (individual) for every single camera...

Absolutely! Hence my question about the colorimetric processing color space. Are you suggesting it is different for each camera?

should be easy to understand for a wise guy like you: the (individual) camera profile is not as large as C1's internal (working) space.

We (you) know that how? Again, what IS the C1 internal working space's gamut?

It's basically the same as in Adobe software: there's a large working space (ProPhoto primaries) but so called camera profiles limit the actual colors to create a decent, "neutral" look.

So it is or isn't (basically?) the same as in Adobe software which is clearly defined: ProPhoto RGB primaries? We know this how?

Can you, vice versa, show me the "gamut" of any Adobe "camera profile" (so "input profile")?

I can show you the gamut of images processed though the Adobe engine. I can tell you want the designers of the product tell us what the colorimetric gamut of the processing engine is. I could shoot an image as a JPEG with the camera set to Adobe RGB (1998) and plot that too. But I don't know how any of that is pertinent in answering a simple question about the gamut of the processing color space from C1.

[AlterEgo]

Absolutely! Hence my question about the colorimetric processing color space. Are you suggesting it is different for each camera?

C1 does not use any colorimetric processing color space at the stages you mean (or I assume you mean)... again based on Esben's words interpretation

We (you) know that how? Again, what IS the C1 internal working space's gamut?

I guess it was established that it is wrong to use the word gamut for an input device like digital camera (till the color transform performed and it is not yet - see above), no ? lengthy topics were created about camera's "color spaces"

[ErikKaffehr]

Hi,

The descriptions given sound like a mess.

Common sense say that camera RGBs should be converted into a well specified and sufficiently large color space early on the process.

Best regards
Erik

it is different (individual) for every single camera...
Confused. The camera doesn't have a color gamut yet we're told you do most of the work in the native camera space who's 'gamut' isn't limited in the sentence above.should be easy to understand for a wise guy like you: the (individual) camera profile is not as large as C1's internal (working) space. It's basically the same as in Adobe software: there's a large working space (ProPhoto primaries) but so called camera profiles limit the actual colors to create a decent, "neutral" look.
again, different for every single camera...
this is exactly what you can do ... all "camera profiles" are available as icc profiles and you can load it in the gamut viewer of your choice.
Can you, vice versa, show me the "gamut" of any Adobe "camera profile" (so "input profile")?

[AlterEgo]

Common sense say that camera RGBs should be converted into a well specified and sufficiently large color space early on the process.

no, it does not... for example RPP does as much as possible before any color transform and moreover as much of what is done before color transform is done before demosaicking... however Adobe does differently (and may be one of the reasons was {historically} / is simply the need to show adjustments in close to real time on average Joe's computer) - post initial raw data read : demosaick first, apply WB+initial color transform (the part of dcp profiles that goes before the UI driven exposure correction), then what user does in UI is processed (exceptions naturally are in what you see in "Camera Calibration" tab in ACR and WB - as WB is mixed with color transform application in Adobe's realm /that btw is why ACR/LR can do identical conversion with original raws and linear DNGs obtained from them without any differences in output if linear /non lossy/ DNG was made first with the same release of ACR/LR/Adobe DNG converter/)... that allows for smoother UI reaction.

[digitaldog]

I guess it was established that it is wrong to use the word gamut for an input device like digital camera (till the color transform performed and it is not yet - see above), no ?

Agreed.

[tho_mas]

based on Esben's words C1 does not work in any proper colorimetric color space till the last stage of the pipeline (output to...) - it simply works/manipulates with camera's RGB numbers, color transform (that's where camera profiles are used)

that's not correct. There is always an input profile working in C1. You can choose to choose whatever input profile you prefer ... but you can not set no input profile at all. Therefore - as long as you embed the input profile on output - you are always working within the gamut of the input profile.

[tho_mas]

Absolutely! Hence my question about the colorimetric processing color space. Are you suggesting it is different for each camera?

yes, it is.

We (you) know that how? Again, what IS the C1 internal working space's gamut?

I don't know. Since C1 can produce colors that exceed ProPhoto-RGB it must be larger than ProPhoto. But personally I don't see any benefit of knowing the internal color space of C1 anyway...

I can show you the gamut of images processed though the Adobe engine.

:-)

I can tell you want the designers of the product tell us what the colorimetric gamut of the processing engine is.

what for?

[AlterEgo]

that's not correct. There is always an input profile working in C1.

the question is totally different - when (at which stage of the C1 pipeline) the color transform guided by the data in that profile is applied...  not if, but when... so, for example, when you use a "saturation" slider in C1 "exposure" tool, you work with camera's RGB numbers ___before__ that color transform... RGB data altered by C1 code as a result of your operations with that slider are not yet color coordinates in any proper colorimetric color space... they will be later down the pipeline (for example when you see the image on your monitor or color readout sticker will display you RGB in some proof color space)

[AlterEgo]

Since C1 can produce colors that exceed ProPhoto-RGB it must be larger than ProPhoto.

camera native RGB is not larger or smaller than any proper color space - they are simply different animals, so depending on how you do a color transform you can end up within or outside at the end of the pipeline...  consider this - if C1 camera RGB numbers are {r1, g1, b1} before saturation slider operation and {r2, g2, b2} after how do you know at that moment in C1 code - before the color transform /a pair : first input profile - which is as Esben fro P1 says applied at the end of the pipeline and then, for example at the next stage, proof, output profiles/ where you are going to end up ?

[tho_mas]

the question is totally different - when (at which stage of the C1 pipeline) the color transform guided by the data in that profile is applied...  not if, but when... so, for example, when you use a "saturation" slider in C1 "exposure" tool, you work with camera's RGB numbers ___before__ that color transform... RGB data altered by C1 code as a result of your operations with that slider are not yet color coordinates in any proper colorimetric color space... they will be later down the pipeline (for example when you see the image on your monitor or color readout sticker will display you RGB in some proof color space)

ah, okay - I did get you wrong. Makes sense - thanks!

[ErikKaffehr]

Updated a bit…

Hi,

This is a bit of a nonsense. ProPhoto RGB encompasses all visible colours ProPhoto RGB encompasses nearly all visible colours and more importantly all colours in the Pointons gamut. So, a gamut may be larger but that just means it is less efficient.

Not saying that ProPhoto RGB is the most perfect or most efficient RGB, but it has been defined by a company called Eastman Kodak who used to know what they used to be doing, once upon the time.


Best regards
Erik

yes, it is.
I don't know. Since C1 can produce colors that exceed ProPhoto-RGB it must be larger than ProPhoto. But personally I don't see any benefit of knowing the internal color space of C1 anyway...
 :-)
what for?

[digitaldog]

yes, it is.
I don't know. Since C1 can produce colors that exceed ProPhoto-RGB it must be larger than ProPhoto. But personally I don't see any benefit of knowing the internal color space of C1 anyway...
 :-)
what for?

I explained to David the reason it's useful to know the color gamut of the processing below (or above) 
Here's what I think we know from PhaseOne thus far (since it came from them):

Capture One works in a very large color space, similar to that captured by camera sensors.
I believe many of us here agree cameras and their sensors do not have a color gamut.
The gamut you can see in the camera profile does not limit the gamut used for internal processing in Capture One.


   AR: That appears to dismiss?: The Raw camera RGB is profiled, and that can be expressed as a gamut by plotting the hull and calculating the size.


However, the conversion from the camera space (the “colorimetric interpretation” if you will) is based on ICC profiles.


   AR: again, if this is a colorimetric interpretation, shouldn’t it be possible to examine the gamut and it's size?


While it allows us to extract just about everything provided in the file, it also means that we need a significant effort to get the best possible quality for a camera. However, it does mean we do not limit ourselves to standardize on a specific internal color space.

   AR: so the colorimetric interpretation varies from camera to camera? You are saying it does, I have no reason to doubt that (I'm here to learn). That would appear to indicate some profile, supplied or under the hood for each camera could be examined for it's gamut size. I'm not referring to a profile built from shooting a target of course, I'm referring to what I believe is a profile or some description we can interrupt a gamut size from: the conversion from the camera space (the “colorimetric interpretation” if you will) is based on ICC profiles.
I think we agree that this gamut is indeed a “colorimetric interpretation”


Seem fair?

[tho_mas]

camera native RGB is not larger or smaller than any proper color space - they are simply different animals, so depending on how you do a color transform you can end up within or outside at the end of the pipeline...

while this is true I am talking about processed images ... and these of course contain the result of color transforms etc.