ICC Profile Choice - ProPhoto RGB

[David Grover / Capture One]

Bold is for emphasis. As the movie line goes: you've never seen me angry. 

Ah - now I get it.  I edited my post - http://forum.luminous-landscape.com/index.php?topic=106031.msg872377#msg872377 to help.

[digitaldog]

Ah - now I get it.  I edited my post - http://forum.luminous-landscape.com/index.php?topic=106031.msg872377#msg872377 to help.

 

Good job. I'll try not to spit more blood in your direction.

[AlterEgo]

3)  I will endeavour to find the answer out for you, before any more blood is spilt.

thank you so much ! there are no bad feelings towards you or P1, but it is really a good thing if people intimate with internals can visit us once in a while and share things - no, we do not ask P1 to post a code that does demosaicking or NR for example - but some things really can be answered w/o harm - like for example the format of P1 "curves" .fcrv files so that people can create their own with their tools, w/o resorting to hex editors to dissect them first...

[Bart_van_der_Wolf]

Wrong again Bart  . It would aid the OP, answer his question. You do realize (crap, I keep assuming you do), OK hopefully you understand that IF the colorimetric processing color space of a raw converter who's gamut can be now defined is smaller than say ProPhoto RGB, it's rather pointless to select that as the encoding color working space?

It's not entirely pointless if further processing (e.g. in Photoshop) is going to be used that can increase saturation even further. It prevents yet another profile conversion with introduction of rounding errors if not performed in floating-point math.  Besides, a large portion of Prophoto RGB consists of coordinates for imaginary colors, how useful it that? It only helps a bit if one shrinks saturation again, because that might bring some imaginary colors back into useful gamut.

Problem is, in terms of C1, WE DON'T KNOW and some here are unable to provide this simple information, instead reference nonsense: Capture One works in a very large color space, similar to that captured by camera sensors.

Friendly advice, try not getting your knickers in a twist, and please dismount the high horse. You do not look good there. Asking things in a more reasonable voice might even help others who are reading this thread, this is not a pissing contest, unless you want it to be. Who knows, you might even get some answers, even if you have no intention whatsoever to switch to C1 or use anything else than LR. I'm cool with that.

Cheers,
Bart

[digitaldog]

thank you so much ! there are no bad feelings towards you or P1, but it is really a good thing if people intimate with internals can visit us once in a while and share things - no, we do not ask P1 to post a code that does demosaicking or NR for example - but some things really can be answered w/o harm - like for example the format of P1 "curves" .fcrv files so that people can create their own with their tools, w/o resorting to hex editors to dissect them first...

Agreed again.
David, no hard feelings, I know you're trying to help answer pertinent questions. When you first started with: I would suggest to Andrew (and yourself) if it is something you need to know, then simply ask Tech support but then moved to: I will endeavour to find the answer out for you, before any more blood is spilt, we made significant progress.
I do have to say, it might be either useful or hilarious to ask Tech Support the question, if time permits, maybe I will.

[digitaldog]

It's not entirely pointless if further processing (e.g. in Photoshop) is going to be used that can increase saturation even further.

No Bart, it really is pointless. Like the URL you posted which I guess doesn't count because it speaks of an older, discontinued product?

Besides, a large portion of Prophoto RGB consists of coordinates for imaginary colors, how useful it that? It only helps a bit if one shrinks saturation again, because that might bring some imaginary colors back into useful gamut.

Imaginary colors (device values) is the priced to be paid for a simple RGB working space where one can, if the converter supports it, provide real, actual colors captured that can be output. We've been over this far too often to have to describe this choice again. Truth be told, we don't know the gamut of the processing color space from C1 yet, the URL you provided was just an embarrassment for all concerned and some users do want and need to know such technicalities to select an appropriate RGB working space. 

Friendly advice, try not getting your knickers in a twist, and please dismount the high horse.

Friendly advise, when someone uses sound color science to dismiss nonsense you posted, it doesn't mean that person has their knickers in a twist. It appears the people providing the nonsense are the ones in a twist. It appears that reality continues to ruin your life, and to believe those providing reality are the people with knickers in a twist or spitting blood. The only dog I have in this fight is pointing out Colorimetric BS of which you've provided thanks to the P1 URL.

[Bart_van_der_Wolf]

No Bart, it really is pointless. Like the URL you posted which I guess doesn't count because it speaks of an older, discontinued product?  Imaginary colors (device values) is the priced to be paid for a simple RGB working space where one can, if the converter supports it, provide real, actual colors captured that can be output. We've been over this far too often to have to describe this choice again. Truth be told, we don't know the gamut of the processing color space from C1 yet, the URL you provided was just an embarrassment for all concerned and some users do want and need to know such technicalities to select an appropriate RGB working space.  Friendly advise, when someone uses sound color science to dismiss nonsense you posted, it doesn't mean that person has their knickers in a twist. It appears the people providing the nonsense are the ones in a twist. It appears that reality continues to ruin your life, and to believe those providing reality are the people with knickers in a twist or spitting blood. The only dog I have in this fight is pointing out Colorimetric BS of which you've provided thanks to the P1 URL.

I tend to not engage in a battle of wits with an unarmed person, so I'll leave the BS to you.

Cheers,
Bart

[digitaldog]

I tend to not engage in a battle of wits with an unarmed person, so I'll leave the BS to you.

Unarmed perhaps, technically correct, absolutely. You've so failed to understand this topic, you should exit ASAP to avoid further embarrassment .


Oh and Bart, I've got nothing at all against PhaseOne and there's always a possibility I'll go back to their products from Adobe. FWIW, I'll bet you dollars to donuts I was using their products long before you! I see one of my first reviews of their products appeared over 16 years ago in PEI magazine:
http://digitaldog.net/files/Filmvsdigital.pdf

[tho_mas]

What a fruitless debate …

My take on the way C1's color does work… just assumptions based on my work with C1 for some years…

First - as in any other decent RAW converter - there are "hidden" profiles working "under the hood" correcting the nonlinearity of the different camera sensors and model the cameras colors to match a certain target.

In C1 this target seems to be what is represented in the so called "No Color Correction" profile (it is the very same profile that is also used for and called "DNG File Neutral.ICC", "JPEG.ICC", "TIF.ICC" and "Color View.ICC).
This is, if you want so, the smallest nominator (after said individual camera profiling "under the hood") in C1.
This profile has a pretty small gamut and a Gamma of 1.8.

What is called "camera profile" in C1 are actually "camera specific working spaces" - these are table based profiles with a neutral grey axis (Gamma 1.8 ). These camera specific profiles (in conjunction with the so called "film curves") "model" color (based on the starting point of what the "No Color Correction" profile represents … so that you don't see pale, boring colors but "vivid" colors that make sense (…) - therefore they are much larger than the "no Color Correction" profile...

Since these camera specific working spaces are designed as table based input profiles they get assigned to the image file - the file is not converted into these profiles. This is why - on this stage of the processing chain - there's no way that color clipping can occur.

More on clipping: the "advanced Color Editor" PHYSICALLY alters the input profile. So if you increase saturation in the Advanced Color Editor, you expand the input profile! This is very important … as this avoids clipping BY DESIGN. You could produce clipping… if you choose an output color space that is not large enough to match the input profile. But one of the strengths of C1 is that you can choose to process with the input ("camera") profile embedded … so that no further color conversion takes place (which is also my preferred workflow - I also edit my photos in Photoshop in these "camera working spaces").

My assumption about C1's "internal color space" …: the Advanced Color Editor can add 200% global saturation. So if you take the "No Color Correction" Profile and increase saturation by 200% you will end up with a profile that is indeed much larger than ProPhoto-RGB. Since this is a Color Space C1 can produce with max. Saturation in the Color Editor I think it's not completely far-fetched to assume that this represents C1's internal color space ... or something close to it (see attachment - coloured=C1, white=ProPhoto-RGB).

In reply to the OP: set ProPhoto-RGB as "output" color space in the "process recepies" tab and don't worry about clipping …

[bjanes]

The point is that if you think of camera primaries you can come to many incorrect conclusions because cameras capture spectrally. On the other hand, displays create colors using primaries. Primaries are defined colorimetrically so any color space defined using primaries is colorimetric. Native (raw) camera color spaces are almost never colorimetric, and therefore cannot be defined using primaries. Therefore, the measured pixel values don't even produce a gamut until they're mapped into a particular RGB space. Before then, *all* colors are (by definition) possible.

I'm not sure what you mean by the statement that camera sensors capture spectrally. The input is a spectrum and the sensor has a spectral sensitivity. Doug Kerr has an excellent post on sensor colorimetry to which the interested reader can refer. The sensor does have spectral sensitivities and a typical response in shown in Fig 5 of Doug's post and is shown below. The sensor integrates the response at various wavelengths over the range of each of the RGB ranges and reports a single output value for each sensel. It does not record a spectrum.

If the camera filters followed the Luther-Ives criteria, one could produce a perfect metameric match to the spectral colors, but since no existing sensor conforms to these criteria, there will be metameric error which can be minimized when outputting to a colorimetric space. When outputting to the colorimetric space, any number of primaries can be chosen by making use of Grassman's Law. These primaries are those of the colorimetric space, not the sensor.

Bill

[digitaldog]

I'm not sure what you mean by the statement that camera sensors capture spectrally.

I was referring to cameras which don’t have primaries, they have spectral sensitivities. The image was recorded through a set of camera spectral sensitivities which defines the intrinsic colorimetric characteristics of the image. An simplistic way to think of this (while not purely accurate) is that the image was recorded through a set of "primaries" and these primaries define the color space of the image. Unless the camera spectral sensitivities are colorimetric, they do not define the intrinsic colorimetric characteristics of an image.

The input is a spectrum and the sensor has a spectral sensitivity. Doug Kerr has an excellent post on sensor colorimetry to which the interested reader can refer.

Agreed and I've read Doug's excellent piece.

If the camera filters followed the Luther-Ives criteria, one could produce a perfect metameric match to the spectral colors, but since no existing sensor conforms to these criteria, there will be metameric error which can be minimized when outputting to a colorimetric space.

Agreed, did bring up this issue below.

When outputting to the colorimetric space, any number of primaries can be chosen by making use of Grassman's Law. These primaries are those of the colorimetric space, not the sensor.

Absolutely agree, and that's where the URL that Bart provided is so flawed. Or his idea that some input profile tells us anything about those primaries the product assumes or 'knows' about the capture (spectral sensitivity) to get to this colorimetric space I've asked about in terms of it's color gamut.

[ErikKaffehr]

Hi,

The ColorChecker RGB values differ between different RGBs.

In LR I can choose to display the Colour in Lab values, which I prefer to RGB values. I don't know if this is possible Capture One. That said, Lab is dependent on illuminant AFAIK.

Another option is to download a synthetic colour checker in a known RGB and open it in Capture One. From that image the RGB values for each reference field can be read out.

I see a problem with Capture One's colour handling being quite arbitrary and not at all documented. Capture One uses ICC but it's deployment is not well specified. Adobe's DNG profiles on the other hand are a quite comprehensive recipe for raw conversion and well documented.

Using good DNG or ICC profiles, colour conversion should be quite good for continuous spectra, even I don't know how C1 handles different illuminants. For complex illuminants a specific conversion profile is needed. For DNG, there are a couple of free tools, like Adobe's DNG Profile editor or the Color Checker Passport software.
Anders Torger, a frequent poster here has developed another tool called DCamProf that is much more flexible than the commercial tools but needs to be used at the command line. Anders has found out and resolved some intricacies in both C1's handling of ICC and Adobe's colour processing engine.

Anders Torger's approach is to develop a colour profile that leads to a correct colour rendition on which a "look" can be applied. A look would apply very subtle colour changes, to tweak colour rendition for it's intended purpose.

Best regards
Erik

Another problem I have (related though not strictly relevant here) is that if I Pick a Colour Sample using the color picker in either the Basic or Advanced Color Editors, the RGB values in the Color Editor dialog are completely different from those above the image at the top of the screen.

This happens no matter how many selections I make from, say, a patch on a Gretag Color Checker.

I'm Windows 10. It would be interesting to know if others have the same problem.

So, which values are correct?

Thanks.

D.

[Doug Peterson]

The ColorChecker RGB values differ between different RGBs.

In LR I can choose to display the Colour in Lab values, which I prefer to RGB values. I don't know if this is possible Capture One. That said, Lab is dependent on illuminant AFAIK.

If looking to match patch values for Cultural Heritage or art repro applications you can use Capture One Cultural Heritage Edition which has a broad range of LAB Support and ICC profiles designed solely for color accuracy (no adaptions or embellishments for the purpose of pleasing color). These provide exceptionally high color accuracy out of box with supported light sources, as measured by FADGI and Metamorfoze Standards evaluations.

[AlterEgo]

these are table based profiles with a neutral grey axis (Gamma 1.8 ).

except that you can perfectly use "not table" matrix+trc (gamma or actual 1D lut) profiles in C1 (color editor will not work though, but profile otherwise is totally usable)...

now with full LUT profiles you can simply clamp the demosaicked/with curve applied RGB (not yet in any proper colorimetric colorspace) data to an arbitrary smaller gamut within a proper colorimetric PCS - the question is what happens next, PCS can be cieXYZ or cieLAB - what C1 does next ? from PCS in icc/icm container to which color space ?

[AlterEgo]

My assumption about C1's "internal color space" …: the Advanced Color Editor can add 200% global saturation. So if you take the "No Color Correction" Profile and increase saturation by 200% you will end up with a profile that is indeed much larger than ProPhoto-RGB.

then save it (from color editor) to icc/icm profile and repeat the process again... now what ? can't C1 say for example - we are using LAB internally (with necessary round-trips for whatever calculations when needed - like ACR/LR does trips to LAB from PropPhoto coordinates with gamma 1 to calculate some maskes or whatever Eric Chan mentioned once or twice) for example and that's it.

[AlterEgo]

More on clipping: the "advanced Color Editor" PHYSICALLY alters the input profile. So if you increase saturation in the Advanced Color Editor, you expand the input profile!

PCS of the ''input" profile stays the same - cieLAB, you change (for the runtime only, unless you save it to icc/icm container to reuse) the lut with color editor and as a result the color transform clamps the data into a wider gamut (still subset of PCS).

[Dinarius]

David,

Regarding the discrepancies between the RGB readings at the top of the screen and those in the Color Tool dialog box, here is the reply I got from Support:

"Hello,

This is becuase the value you see at the top is the color after it has been converted to your output space, whereas the value in Colur editor is based only in the input profile you are editing."

1. There seems to be implicit in this reply an acknowledgement that the readings are different. Wouldn't you agree? (Though I know this is not your experience)

2. All I simply want to do is to be able to soft-proof/edit in ProPhoto RGB and then save my TIFFs in this same space. I'd like every RGB reading to be in PP RGB. So, what settings/preferences must I apply to the software to have this across the board? I've asked them the same question.

Thanks.

D.

Ps. Looks like I created something of a monster in starting this thread! 

 

[Dinarius]

If looking to match patch values for Cultural Heritage or art repro applications you can use Capture One Cultural Heritage Edition which has a broad range of LAB Support and ICC profiles designed solely for color accuracy (no adaptions or embellishments for the purpose of pleasing color). These provide exceptionally high color accuracy out of box with supported light sources, as measured by FADGI and Metamorfoze Standards evaluations.

Thanks for the reply.

C1 CH is priced for institutions, not individuals, but it's a nice thought! 

D.

[David Grover / Capture One]

David,

Regarding the discrepancies between the RGB readings at the top of the screen and those in the Color Tool dialog box, here is the reply I got from Support:

"Hello,

This is becuase the value you see at the top is the color after it has been converted to your output space, whereas the value in Colur editor is based only in the input profile you are editing."

1. There seems to be implicit in this reply an acknowledgement that the readings are different. Wouldn't you agree? (Though I know this is not your experience)

2. All I simply want to do is to be able to soft-proof/edit in ProPhoto RGB and then save my TIFFs in this same space. I'd like every RGB reading to be in PP RGB. So, what settings/preferences must I apply to the software to have this across the board? I've asked them the same question.

Thanks.

D.

Ps. Looks like I created something of a monster in starting this thread! 

Ok - Yeah that does make sense. 

1) What do you have set in View>Proof Profile
2) What is the output profile set to in the currently selected process recipe.
3)  Have you changed anything in Base Characteristics?

[Dinarius]

Ok - Yeah that does make sense. 

1) What do you have set in View>Proof Profile

It's set to ProPhoto

2) What is the output profile set to in the currently selected process recipe.

The ProPhoto box is ticked and highlighted in Process Recipes, and ProPhoto/16bit etc. is stated in the Process Summary tab below.

3)  Have you changed anything in Base Characteristics?

No. It's set to Canon 1Ds Mklll

Anytime I'm saving a file, I highlight it in the film strip at the bottom and choose Export/Variants. The export recipe in the dialog that appears is always the same - PP RGB, 16bit, TIFF, 300px, etc...

Thanks.

D.