Shooting Color Targets

[digitaldog]

This is going to be a most challenging task. I doubt the testing will end up fair enough for everyone.

I'm sure you're right. Still be an interesting exercise for a group who could comment on what they see. It might aid in the DNG vs. ICC profile debate a bit, at least if neither (or one) really sucks! That the DNG profile could be raw processor agnostic is interesting.

Another issue is what software is going to build the ICC profiles? Argyll? Different profile building softwares are going to make different decisions about how to move colors around. Are we also testing for how well a software builds a camera profile along with DNG vs ICC? We cannot do both at the same time.

Take your pick. That the various solutions act differently (even DNG Profile editor versus Passport software) is useful to look at.

As Brian said, "You can modify the camera tone curve, but you'll lose accuracy just like you do if you modified an RGB tone curve to a carefully calibrated image from a scanner." I don't see another way of getting both ACR and ID to a normalized (matching) state without tweaking some curves in ID for ICC profiles. We already know the neutral setting points for both software - all sliders zeroed and all curves linear. They match very closely, at least for my image of the CC, and differ slightly because of differing profiles.

That we may be forced to use this kind of 'workflow' with ICC profiles could again say something about the differing profile processes.

[samueljohnchia]

It might aid in the DNG vs. ICC profile debate a bit, at least if neither (or one) really sucks! That the DNG profile could be raw processor agnostic is interesting.

As it might happen, there is alternative view, that both are bad... I wrote to David Dunthorn of C F Systems and he said pretty much the same thing is true even for DNG profiles. But at the moment, I think that custom DNG profiles are excellent for normal usage in my experience, I have no experience with ICC profiles and I am going to try the perfectRAW software approach later this week.

I don't actually know the algorithms and formulas for the color transforms to create these profiles, and since DNG profiles are by nature perceptual, each profiling software is going to have its own unique recipe. This is getting confusing quickly, with increasingly greater number of issues to test, with multiple possibilities for things to screw up. We want to be able to point fingers and the right things causing these screw ups, and avoid them if possible. Can we reduce the number of variables and unknowns enough to make that happen?

[digitaldog]

As it might happen, there is alternative view, that both are bad...

I've had bad ICC camera profiles (and good ones) but never an issue building and using a DNG camera profile. But I've only done this with Adobe raw products.

I don't actually know the algorithms and formulas for the color transforms to create these profiles, and since DNG profiles are by nature perceptual, each profiling software is going to have its own unique recipe.

To be somewhat expected. I see slight differences between Adobe generated versus X-rite generated DNG profiles from the same raw. It's not a large difference however.

This is getting confusing quickly, with increasingly greater number of issues to test, with multiple possibilities for things to screw up. We want to be able to point fingers and the right things causing these screw ups, and avoid them if possible. Can we reduce the number of variables and unknowns enough to make that happen?

It's necessary to define a proper, agreed upon testing scenario, no question. The results will not be definitive but could still be somewhat telling. In the end, much of our preferences will be subjective unless one process produces poor results. I've seen that in the past with ICC camera profiles so I'd like to know what reason DNG profile creation and use seems to be far more reliable or if that's even the case among multiple users.

[samueljohnchia]

In the end, much of our preferences will be subjective unless one process produces poor results. I've seen that in the past with ICC camera profiles so I'd like to know what reason DNG profile creation and use seems to be far more reliable or if that's even the case among multiple users.

Andrew, as you mentioned elsewhere before, often folks want ICC camera profiles to adapt to similar but not identical shooting conditions. Especially for color critical work, that is bound to fail. In a strictly controlled setup for copy work and reproductions, with carefully selected lighting (probably handpicked light sources to match two or more lights), it is possible to make excellent ICC profiles, especially with a color target like the SG, I am told. The shot must be more or less "right" in camera by getting the lighting ratios and angles right, so that no additional curves are applied in post processing, just a neutral rendition from the raw converter, coupled with an ICC profile.

If we can avoid specifying a white point in an ICC camera profile it would be even better. I quote from ArgyllCMS's documentation:
"If profiling a camera in RAW mode, then there may be some advantage in creating a pure matrix only profile, in which it is assumed that the camera response is completely linear. This may reduce extrapolation artefacts. If setting the white point will be done in some application, then it may also be an advantage to use the -u flag and avoid setting the white point to that of the profile chart:

colprof -v -D"Camera" -qm -am -u scanner"
That is pretty much I can gather from ICC profiling so far. Haven't done anything since I don't use applications that support ICC profiles, and I don't have access to a mac all the time to try and run ID.

DNG profiles are by design more adaptable, even more so for dual illuminant profiles. That's why we find them so useful in varying situations. But then again most folks using DNG profiles are not doing critical reproduction work, so extreme accuracy isn't on the radar, especially since they are going to be applying all sorts of adjustments in ACR/Lightroom, which will throw off accuracy anyway, as neutral representation of raw photos are just not aesthetically pleasing most of the time.

Shall we start a new thread on the Color Management forums about DNG and ICC profiling? I think Hening as already got what he needs for shooting targets.

[digitaldog]

In a strictly controlled setup for copy work and reproductions, with carefully selected lighting (probably handpicked light sources to match two or more lights), it is possible to make excellent ICC profiles, especially with a color target like the SG, I am told. The shot must be more or less "right" in camera by getting the lighting ratios and angles right, so that no additional curves are applied in post processing, just a neutral rendition from the raw converter, coupled with an ICC profile.

That's been my experience among other's I respect. But part of the problem, perhaps a very large part, can be the raw converter and how it has to be (or should be) set to produce the data to feed to the profiling software.

Shall we start a new thread on the Color Management forums about DNG and ICC profiling? I think Hening as already got what he needs for shooting targets.

Not a bad idea. I'd like to see if there are raw converters like ID that can accept both types of profiles. Then see if there are willing people here who could test the waters once the proper procedures for handling the data in this converter are well defined. One person could do copy work and examine the effect on "accuracy" while another could shoot in the field.

In the end, it may not matter. I don't think ICC camera profiles will find their way into Adobe products and I doubt those products that don't handle DNG profiles will change their mind about them. But it could be an interesting exercise.

[Vladimirovich]

I've seen that in the past with ICC camera profiles so I'd like to know what reason DNG profile creation and use seems to be far more reliable or if that's even the case among multiple users.

software was written to make it simple for an average Joe and that software does a lot of corrections for an average Joe behind the scenes - absolutely not the case w/ the software written for ICC/ICM profiles (more so raw converters manufacturers did not bother unlike Adobe to supply such applications, not even in perpetual Beta form), there no corrections for your errors... other than that you can see that neither ACR/LR w/ whatever profiles nor C1/RPP w/ ICC data containers do not have any critical advantage vs their opponents when it comes to the colors beyond tastes and personal preferences/experience/vested interests of their users (RPP might claim better work in deep shadows by design to apply everything before demosaicking/WB and doing that w/ better precision and using matrix profiles to further lessen the errors).

[Vladimirovich]

I'd like to see if there are raw converters like ID that can accept both types of profiles.

rawtherapee shall if I am not mistaken

[samueljohnchia]

Not a bad idea.

let's continue into a new thread then. 

[Hening Bettermann]

> Shall we start a new thread on the Color Management forums about DNG and ICC profiling? I think Hening as already got what he needs for shooting targets.

Yes he has :-) and more than that - thanks to all of you! See you again on the thread on profiling :-)
And, Samuel, I will be eager to follow your exploration of ColorPerfect - see you on THAT thread as well :-)

Good light - Hening

[Iliah]

As it might happen, there is alternative view

"the BAYER interpolation usually used on digital camera images does adjust the red value of a pixel differently according to the readings of nearby blue and green pixels. But this is done according to the geometric relationship of several surrounding pixels and the values each pixel is sensing. Geometry within the image is the key. A specific pair of blue and green pixels will influence the red pixel value in different ways depending upon the geometry of placement of the pixel values within the image."

Really?

[samueljohnchia]

"the BAYER interpolation usually used on digital camera images does adjust the red value of a pixel differently according to the readings of nearby blue and green pixels. But this is done according to the geometric relationship of several surrounding pixels and the values each pixel is sensing. Geometry within the image is the key. A specific pair of blue and green pixels will influence the red pixel value in different ways depending upon the geometry of placement of the pixel values within the image."

Really?

Only David knows what he means. Iliah, you are an expert in the field of color. It would be greatly appreciated if you could comment about camera profiling, and the shortcomings of it. Is ColorPerfect really protecting the color integrity of photos and camera profiles destroying that?

[Jim Kasson]

"the BAYER interpolation usually used on digital camera images does adjust the red value of a pixel differently according to the readings of nearby blue and green pixels. But this is done according to the geometric relationship of several surrounding pixels and the values each pixel is sensing. Geometry within the image is the key. A specific pair of blue and green pixels will influence the red pixel value in different ways depending upon the geometry of placement of the pixel values within the image."

Really?

Not sure what the quoted poster meant, but here are some papers that discuss demosaicing methods where one color plane, say red, is influenced by nearby pixels in the other two color planes:

L. Chen, K.H. Yap, and Y. He. "Color filter array demosaicking using wavelet-based subband synthesis"
Proc. ICIP, pages 1002–1005, 2005.  Here is a link to a more recent wavelet paper.

R. Lukac and K.N. Plataniotis. "Demosaicking using vector spectral model" IEEE International Conference
on Multimedia and Expo, pages 1185–1188, July 2006.

Ron Kimmel. "Demosaicing: image reconstruction from color ccd samples" IEEE Transactions on Image
Processing, 8(9):1221–1228, 1999.

Jim

[Vladimirovich]

and camera profiles destroying that?

camera profiles shall be considered only along with the code that applies them...