Using Adobe DNG profiles when developing raw files - advice and comment please

[dje]

Or jpg.
That's what I meant with "output related".

George

George

There are several types of ICC profiles, as classified in the ICC spec. The most common of these are

Input Profiles (for cameras and scanners)
Display Profiles (for monitors and screens)
Output Profiles (for printers)
Color Space Profiles (for transformations between color spaces) - these are device independant.

They have different content and different roles to play in the Color Management process. Adobe users may not be familiar with ICC Input profiles as Adobe software doesn't use them but rather uses their own profile structure (DCP).

Dave

[revi]

I don't know what you mean with "ICC profile that matches the camera". An ICC profile is as far as I know output device related.
(...)
George

As I understand it, an ICC profile corrects for differences between an expected value and a real value.

The raw data have to be converted at one point from "camera space" to the editor's working space. For that conversion, you need a profile that describes the correspondence between the raw data and the 'real' colours. Due to differences in the sensor filters (between brands/models/batches), that correspondence isn't universal, so you need something camera (or at least brand) specific.

If it helps, the raw file is the output from the camera. So a camera-specific ICC profile adapts that output to what the next stage expects. Whether that next stage is a program, as in raw conversion, or your eyes, as in screen display or prints, doesn't really matter as far as the ICC profiles are concerned.

[george013]

I still got the feeling it's only half of the story.

http://www.color.org/faqs.xalter#p4

George

[Manfred M]

I still got the feeling it's only half of the story.

http://www.color.org/faqs.xalter#p4

George

That data sheet covers a much broader view of the application of the ICC profile than the simpler one we are using in the purely photographic workflow we are dealing with in this thread. This makes sense because ICC is used in other areas, especially in the printing industry where there will be several more steps after the image has been created by other means as well as photography.

Dave and Revi have given you a good explanation from a generic photographic profile where a profile is needed every time data is passed to a new device. Capture One is much more compliant with the ICC approach than Adobe simply because their products (high end digital backs) have traditionally been used to generate images for the publishing industry which uses this approach in their end-to-end process. The software was developed to make the workflow consistent with end-user requirements.

[george013]

That data sheet covers a much broader view of the application of the ICC profile than the simpler one we are using in the purely photographic workflow we are dealing with in this thread. This makes sense because ICC is used in other areas, especially in the printing industry where there will be several more steps after the image has been created by other means as well as photography.

Dave and Revi have given you a good explanation from a generic photographic profile where a profile is needed every time data is passed to a new device. Capture One is much more compliant with the ICC approach than Adobe simply because their products (high end digital backs) have traditionally been used to generate images for the publishing industry which uses this approach in their end-to-end process. The software was developed to make the workflow consistent with end-user requirements.

Please don't start to make it more difficult as it is.
A colour is defined by its wavelength(s). Device independent.
A colour gamut is a range of colours. Device independent.
A ICC profile contains a relation between a colour and the digital r,g,b values. Device dependent.
If you want to move from one ICC profile to another you must recalculate those digital values. Rendering deals with that.
The raw file contains digital values for the three colours. That's what's given to the converter. I don't know how I could change that. I can transform it to another colour space but the original interpretation of the values must be known and is set by the manufacturer or the camera setting.

It's all in the link and it's from the International Color Consortium.

George

[Manfred M]

Rendering deals with that.

And the rendering engine uses the ICC profile to ensure that the colours are rendered accurately and bias that result from, for instance, the manufacturing process variability of the devices are compensated for.

That's what Dave. Revi and I are trying to tell you.

The camera specific ICC profile corrects for variability on the sensor / Bayer array.

The screen ICC profile corrects for variability in the screen hardware and manufacturing processes.

The printer ICC profile compensates for variability in the print head, ink and paper being used.

This is 100% independent of the data that the camera has captured and the editing software has changed.

[revi]

Please don't start to make it more difficult as it is.
(...)
I can transform it to another colour space but the original interpretation of the values must be known and is set by the manufacturer or the camera setting.
(..)

George

And the bolded part is exactly what the camera ICC profile is: the interpretation of the rgb values as provided by the camera. And as the raw file is supposed to give us the unmodified sensor readings (modulo amplification), that camera ICC profile must be applied on input into the raw converter.

(If the camera produces jpegs, then the camera profile is applied in the camera, of course)

[george013]

And the bolded part is exactly what the camera ICC profile is: the interpretation of the rgb values as provided by the camera. And as the raw file is supposed to give us the unmodified sensor readings (modulo amplification), that camera ICC profile must be applied on input into the raw converter.

(If the camera produces jpegs, then the camera profile is applied in the camera, of course)

I've looked in the manual of Capture One and some tutorials https://imagealchemist.net/capture-o...aracteristics/

The ICC Profile field lists the profile that Capture One uses as input profile. It is a profile that reflects the color characteristics of your camera under normal light conditions. Each camera for which Capture One supports the raw file format, has its own unique color profile(s) made.

Looking at the examples in the above link,flash, outdoor,tungsten etc, it looks to me more a matter of white balance.

Well,I've something to study. I only use the old CaptureNx. There it's no issue.

George

[dje]

I've looked in the manual of Capture One and some tutorials https://imagealchemist.net/capture-o...aracteristics/

Looking at the examples in the above link,flash, outdoor,tungsten etc, it looks to me more a matter of white balance.

Well,I've something to study. I only use the old CaptureNx. There it's no issue.

George

Hi George

That's a useful article which I haven't seen before. Thanks for posting.

The reference to different light conditions (illuminants) is about the light condition used to create each ICC profile. For the Phase One camera in question, there are several profiles produced under different lighting conditions and you would select the one that most closely matches your actual lighting condition. This is quite distinct however from a White Balance adjustment which still needs to be made separately.

Dave

[george013]

Hi George

That's a useful article which I haven't seen before. Thanks for posting.

The reference to different light conditions (illuminants) is about the light condition used to create each ICC profile. For the Phase One camera in question, there are several profiles produced under different lighting conditions and you would select the one that most closely matches your actual lighting condition. This is quite distinct however from a White Balance adjustment which still needs to be made separately.

Dave

Don't thank me. I'm just searching for more info and this one pasted by.

I do think of ICC profiles as describing the input or output device under standardized conditions: white light.
White balance is dealing with non standard light conditions: not white light. The way I see it is that a part of the white balance is moved to the ICC profile.
I still have doubts about that "input profile". To me it must be in the camera and used for determining the digital values to a standard. So we can work on it further.

I still can't place the by Bill used profiles: D750 and DNG. DNG is just a uniform raw file. And what is a profile D750. Or does it mean to take in account the camera settings?

George

[Manfred M]

Hi George

That's a useful article which I haven't seen before. Thanks for posting.

The reference to different light conditions (illuminants) is about the light condition used to create each ICC profile. For the Phase One camera in question, there are several profiles produced under different lighting conditions and you would select the one that most closely matches your actual lighting condition. This is quite distinct however from a White Balance adjustment which still needs to be made separately.

Dave

What is interesting is that they have only done that level of work for the Phase One backs, not all the other camera models that are supported. Different backs have different settings, so this lack of consistency is a bit puzzling too; for the IQ 160 back they have four different flash profiles.

If the light source has a specific wavelength of light, it should be picked up during the profiling operation. If it does not, then it should make a difference in the profile. When I develop a custom profile for daylight and another for studio flash, I see no difference at all in the image, regardless of which of the profiles I use.

Time for a bit more research?

[dje]

What is interesting is that they have only done that level of work for the Phase One backs, not all the other camera models that are supported. Different backs have different settings, so this lack of consistency is a bit puzzling too; for the IQ 160 back they have four different flash profiles.

If the light source has a specific wavelength of light, it should be picked up during the profiling operation. If it does not, then it should make a difference in the profile. When I develop a custom profile for daylight and another for studio flash, I see no difference at all in the image, regardless of which of the profiles I use.

Time for a bit more research?

Manfred correct me if I'm wrong but I think the color temp of flash is not too different to daylight? So any difference in color between the two might be very subtle and perhaps only picked up on an image of a color checker passport. Even larger differences in illuminant used in making the proflies may only result in subtle changes. I'll have to do some tests with custom profiles myself. I've only got daylight versions at present and will have to make some new ones using significantly different illuminants.

Dave

[george013]

Manfred correct me if I'm wrong but I think the color temp of flash is not too different to daylight? So any difference in color between the two might be very subtle and perhaps only picked up on an image of a color checker passport. Even larger differences in illuminant used in making the proflies may only result in subtle changes. I'll have to do some tests with custom profiles myself. I've only got daylight versions at present and will have to make some new ones using significantly different illuminants.

Dave

CaptureNx
Daylight direct sunny 5200
Daylight cloudy 6000
Daylight shadow 8000
Flash no filter 5400

George

[dje]

I'll have to do some tests with custom profiles myself. I've only got daylight versions at present and will have to make some new ones using significantly different illuminants.

Dave

First Image : Captured in Daylight about 5000 degK, ICC camera profile made from image captured under fluorescent light (about 3300 deg K)




Second Image : Captured in Daylight about 5000 degK, ICC camera profile made from image captured in daylight (about 5000 deg K)



The first one looks a bit over-cooked to me especially the reds and oranges.

Photos processed in AfterShot Pro.

Dave

[george013]

First Image : Captured in Daylight about 5000 degK, ICC camera profile made from image captured under fluorescent light (about 3300 deg K)




Second Image : Captured in Daylight about 5000 degK, ICC camera profile made from image captured in daylight (about 5000 deg K)



The first one looks a bit over-cooked to me especially the reds and oranges.

Photos processed in AfterShot Pro.

Dave

I wouldn't know how that color checker can be useful to me. I don't have a reference of good and bad. Maybe I must have a copy of it myself to compare with the image.
I still don't get the idea of that ICC profile. Looks more of a preset.

George

[dje]

I wouldn't know how that color checker can be useful to me. I don't have a reference of good and bad. Maybe I must have a copy of it myself to compare with the image.
I still don't get the idea of that ICC profile. Looks more of a preset.

George

You can use a color checker for visual inspection but it is most useful for creating profiles. In simple terms, the process goes like this:-

Take a raw photo of the CC with a known illuminant

Apply white balance factors to RGB raw values

then, using suitable software

measure the RGB values of the color patches

employ an iterative process to create a transformation matrix or LUT to produce XYZ values from the raw RGB values for each color patch, such that they are as close as possible to the known XYZ values of the color patches (this calibration info is available from the makers such as XRite)

this matrix or LUT forms the basis of the ICC profile

as part of the process, the XYZ discrepencies for each patch are usually tabulated as Delta E values (in Lab space)

Dave

[george013]

You can use a color checker for visual inspection but it is most useful for creating profiles. In simple terms, the process goes like this:-

Take a raw photo of the CC with a known illuminant

then, using suitable software

measure the RGB values of the color patches

employ an iterative process to create a transformation matrix or LUT to produce XYZ values from the raw RGB values for each color patch, such that they are as close as possible to the known XYZ values of the color patches (this calibration info is available from the makers such as XRite)

this matrix or LUT forms the basis of the ICC profile

as part of the process, the XYZ discrepencies for each patch are usually tabulated as Delta E values (in Lab space)

Dave

Then you've a color correction of your images for that specific illuminant, isn't it?

George

[dje]

Then you've a color correction of your images for that specific illuminant, isn't it?

George

Yes that's right George. As seen in my example above, color correction using the camera profile doesn't seem to be particularly sensitive to illuminant temp. Also the profile is calculated on white balanced RGB data. I've edited my previous post to add in the white balance step.

Dave

[george013]

Yes that's right George. As seen in my example above, color correction using the camera profile doesn't seem to be particularly sensitive to illuminant temp. Also the profile is calculated on white balanced RGB data. I've edited my previous post to add in the white balance step.

Dave

It's like in CaptureNx where I can find the right color temperature, save that editing and past it to the images that were made in equal circumstances. It's post processing, and cheating of Capture One. If I look at the images of Bill they've a different color temperature but both are 5998K.
Maybe Capture One can't copy the editing to other images. I don't know.

George

[xpatUSA]

The OP asked about "Using Adobe DNG profiles when developing raw files".

I assume that an Adobe DNG profile is a DCP file (.dcp) which makes me wonder what all the talk of ICC is about unless it's referring to JPEGs embedded in the raw file?