Tone curve in Profile help.

[dwbell]

Hey all,

Hoping the more talented tech gurus can help me out a bit here?
I have custom camera profiles for each of my cameras, so that my colour is consistent across bodies.  They're all dual illuminant, some made in the old DNG profile editor, more recent ones in the Lightroom plug in from Xrite (Passport).
The colour is excellent, but I always find the tone to be too contrasty for my preference.  When I look at the values on the step chart (greyscale patches) part of the McBeth checker they are not uniform / linear but rather already hold a contrast adding tone curve.  What I do in Lightroom is to take the TAT tool in the curves panel and move these points per patch until they are almost "linear" again.  In effect I apply a custom inverse S or reverse S to the image.  I then save this curve per camera and invoke it on import setting along with the colour profile.  I LOVE, LOVE the files I now get with this calibration, skin tones especially.

BUT

I now have my curve area of Lightroom being used essentially for calibration purposes, which robs me of it's use for creative purposes.  Thus, I'd like to do this tonal adjustment at the profile stage, in the .dcp file to free me up to use the curve panel to suit my artistic vision.

SO

The question is how do I visually target the tone curve in DNG Editor (or passport) to match those which I have created in the curve panel now?  I can eyeball it so they look similar, but it's a bit vague.  Either that or a way to visually adjust the tone curve while viewing the grey swatches within the DNG editor?

The result should be the same as my custom camera profile plus my bespoke lightroom tone curve but all done in the .DCP file.

Am I missing something?  Willing to be schooled always!
Many thanks if you made it this far.

[rovanpera]

0 on the contrast slider adds quite a bit of s-type contrast to the image. The default I use is -33. 

[b2martin]

If you have the data points for the tone curve you can use Adobe's DNG Profile Editor to save them in the profile.  I don't know if you can edit passport profiles with the DNG profile editor, but you can do this with Adobe profiles.  Select your profile in the DNG profile editor, select the Tone Curve tab, click on curve (straight line to start with) and input the values in the in and out box for each point. 

[Redcrown]

In the Adobe DNG Profile editor, the process must start from a "base" profile. And that base profile has a base tone curve. If you make tone curve adjustments in the profile editor, the resulting profile will have a tone curve that is the net sum of the base plus your adjustments. If you don't make any tone curve adjustments, the resluting profile inherits the full base tone curve.

If you cycle through the Adobe profiles and look at their base tone curves you will see a difference. On my Canon profiles for example, the Adobe Faithful and Neutral profiles have a base tone curve that has lower contrast than the curve for Adobe Standard, Landscape, and Portrait. However, I learned once that the same is not true for Nikon. Base tone curves in the Adobe profiles seem to vary based on make and model.

If you look at the base tone curve graphs in the profile editor, the differences seem very small. However the difference in the image is significant. The higher contrast Adobe base tone curves look identical to a gamma 2.2 tone curve. I have no idea how or why the lower contrast base tone curves for some Adobe profiles were determined.

Note that you can start with any base tone curve and make adjustments in the profile editor to drive the gray squares on the image to whatever you deem is "linear". Then, in LR/ACR your tone curve adjustment is free to use for other changes.

In the Xrite ColorChecker Passport software the base profile is fixed (and hidden), and there is no way to adjust the tone curve for the resulting profile. On my Canon files, the Xrite software creates a base tone curve that is the same as the Adobe Standard base tone curve (higher contrast, gamma 2.2?).

I use a custom dual illuminant profile made with the Adobe DNG profile editor from the Canon Camera Faithful base profile. The main reason is the lower contrast tone curve. I find it a better starting point. Highlights are lower and less prone to blow outs. Shadows are lower too, but easy to adjust.

A few years ago I had a forum discussion with Adobe's Eric Chan about the concept of a base profile starting point in the Adobe profile editor. I had made a profile, then used it as the base profile to make another profile, then did it again. Each iteration generated a significanly different (and worse) result, which confused me. Eric told me not to do that. He strongly recommended using one of the Adobe profiles as the base for every custom profile. I'm still a little confused by the base profile concept, especially given that Xrite has no such thing. Xrite seems to always start from "square zero".

[deejjjaaaa]

you can dump .dcp to a human readable text format (xml) using dcptool, fix your curves as your like them and compile back to .dcp

[mouse]

If you cycle through the Adobe profiles and look at their base tone curves you will see a difference. On my Canon profiles for example, the Adobe Faithful and Neutral profiles have a base tone curve that has lower contrast than the curve for Adobe Standard, Landscape, and Portrait. However, I learned once that the same is not true for Nikon. Base tone curves in the Adobe profiles seem to vary based on make and model.

When I open the DNG profile editor and cycle through the Adobe profiles looking at the Tone Curve panel, they all appear perfectly linear to me.  (Nikon flavor).

If you look at the base tone curve graphs in the profile editor, the differences seem very small. However the difference in the image is significant. The higher contrast Adobe base tone curves look identical to a gamma 2.2 tone curve. I have no idea how or why the lower contrast base tone curves for some Adobe profiles were determined.

Note that you can start with any base tone curve and make adjustments in the profile editor to drive the gray squares on the image to whatever you deem is "linear". Then, in LR/ACR your tone curve adjustment is free to use for other changes.

In the Xrite ColorChecker Passport software the base profile is fixed (and hidden), and there is no way to adjust the tone curve for the resulting profile. On my Canon files, the Xrite software creates a base tone curve that is the same as the Adobe Standard base tone curve (higher contrast, gamma 2.2?).

I use a custom dual illuminant profile made with the Adobe DNG profile editor from the Canon Camera Faithful base profile. The main reason is the lower contrast tone curve. I find it a better starting point. Highlights are lower and less prone to blow outs. Shadows are lower too, but easy to adjust.

A few years ago I had a forum discussion with Adobe's Eric Chan about the concept of a base profile starting point in the Adobe profile editor. I had made a profile, then used it as the base profile to make another profile, then did it again. Each iteration generated a significanly different (and worse) result, which confused me. Eric told me not to do that. He strongly recommended using one of the Adobe profiles as the base for every custom profile. I'm still a little confused by the base profile concept, especially given that Xrite has no such thing. Xrite seems to always start from "square zero".

I too have trouble with the notion of a base profile starting point, and also have exchanged messages with Eric Chan.  I am no wiser.  

An interesting factoid:  If you open Adobe Standard profile (for Nikon D800) with dcpTool.exe you will find that it has no tone curve at all but does contain 2 Hue-Sat Delta tables.  Looking at Adobe Camera Standard and Camera Neutral profiles, you will see each contains a 96 point tone curve but no Hue-Sat Delta tables.  Don't know what conclusions to draw from that.

In any case, a suggestion to the OP:  He could select one of his profiles that he likes (except for the tone curve); open that profile with dcpTool and edit the xml file by replacing the tone curve (if one is present) with the values he established in LR.  (and of course recompile, as suggested while I was typing).

[deejjjaaaa]

I'm still a little confused by the base profile concept

this concept does not allow users to wreck the colors easily (no unless you put some work to inflict a damage)... because the most important part of the Adobe's color transform will be preserved intact...

[mouse]

this concept does not allow users to wreck the colors easily (no unless you put some work to inflict a damage)... because the most important part of the Adobe's color transform will be preserved intact...

Understood.  But the important question is what part of Adobe's color transform must be preserved, and what part may be modified to my taste.

[deejjjaaaa]

Understood.  But the important question is what part of Adobe's color transform must be preserved, and what part may be modified to my taste.

peopel are so lazy ...

http://exdeejjjaaaa.blogspot.com/2013/07/ec-others-forumsadobecom.html

[mouse]

peopel are so lazy ...

http://exdeejjjaaaa.blogspot.com/2013/07/ec-others-forumsadobecom.html

and some are polite, so thank you.

[Redcrown]

When I open the DNG profile editor and cycle through the Adobe profiles looking at the Tone Curve panel, they all appear perfectly linear to me.  (Nikon flavor).

You have to tick the "show base tone curve" box.

[mouse]

You have to tick the "show base tone curve" box.

Quite so.  I discovered that after I made my post.  Still puzzled where this comes from when using Adobe Standard.  As I remarked, if you explore that dcp with dcpTool you find no tone curve.  Perhaps the plot is showing the net tonal effect of all adjustments made by the profile.

If you expand the drop down list in that pane you can choose:

Camera Raw Default: This profile will use the default Camera Raw tone curve. This option is useful if
you wish to use the color adjustments (but not the tone curve adjustments) of an existing profile as a
starting point.

I'm not sure what this explanation means, but it suggests that if you have made a camera raw default tone curve that corrected the problem with the DNG profile, you could import it here and save out a new profile incorporating that curve.  That would free up your ACR tone curve to be used for other purposes.  (see original post; but he is using LR??)

[deejjjaaaa]

Still puzzled where this comes from when using Adobe Standard.  As I remarked, if you explore that dcp with dcpTool you find no tone curve.  

ACR/LR have things embedded in their code too - like hidden exposure correction is a combination (sum) of what is in the code and what is the profile used.

[mouse]

ACR/LR have things embedded in their code too - like hidden exposure correction is a combination (sum) of what is in the code and what is the profile used.

I'm sure they do.  However I was under the impression that the tone curve displayed in the DNG Profile Editor, Tone curve panel, was limited to the tone curve applied by the profile only. 

[dwbell]

Hey Guys,

Thanks for all the input and learning you've provided.

I've solved my concern with some help from Rob over at the Adobe forum;
https://forums.adobe.com/thread/1562639

Basically, he wrote a script that can do exactly what I want.  But I have also managed to do the same thing with TextEdit.
I create the curve I want visually using TAT in Lightroom and save it.
I open the camera recipe in DNG Editor and create the same number of points as my LR curve, their position is irrelevant but easy to "match by eye".
Open both files side by side in text edit and adjust the Profile curves values in the code to match the LR curve in the XMP file.
Export the recipe as a profile and you're done.

I now have what I want.  Colour corrected (Matched) via computer (which I'm visually happy with).  Tone corrected visually by me. This colour and tone correction applied automatically on import via computer (camera profile).  The ability to use the power of curves within LR to suit my artistic vision independent of camera used to shoot with "Linear" as my starting point always.

Verification - I had one file in LR with the old profile and the tone curve applied.  I had a Virtual Copy of that file with the new profile (incl tone curve embedded) and a linear LR tone curve.  I opened both as layers in PS and applied the blend mode Difference to the upper layer.  A totally black resulting image confirmed to me that the pixels were indeed identical between the two files.

(edit - I should add that I have achieved this with one camera , my back up 5DMk2 and on back up files obviously.  I have not yet applied the thinking to my 5DMk3 or Sony A7R and A7S cameras.  So problems with "Nikon" base profiles may appear at such a time as I do this on the Sony files - will update if so)

[sandymc]

When I open the DNG profile editor and cycle through the Adobe profiles looking at the Tone Curve panel, they all appear perfectly linear to me.  (Nikon flavor).

Anything you can do with a tone curve, you can do in the HueSat tables - the tone curve is only really there for backward compatibility; HueSat tables weren't in the original DNG specification, but were added later.

DNG Profile editor probably makes some kind of assumption that a profile without a tone curve, but with a HueSat table, has the standard Adobe tone curve embedded in the HueSat Table. But you'd need to ask someone like Eric Chan to be sure.

Sandy

[mouse]

Anything you can do with a tone curve, you can do in the HueSat tables - the tone curve is only really there for backward compatibility; HueSat tables weren't in the original DNG specification, but were added later.

DNG Profile editor probably makes some kind of assumption that a profile without a tone curve, but with a HueSat table, has the standard Adobe tone curve embedded in the HueSat Table. But you'd need to ask someone like Eric Chan to be sure.

Sandy

Sandy, thanks.  I suspected as much when I wrote earlier:

Perhaps the plot is showing the net tonal effect of all adjustments made by the profile.

[deejjjaaaa]

curiously D810 has ProfileToneCurve in Adobe Standard dcp profile

[dwbell]

Interestingly what i said earlier about the difference blend mode verification is actually false.
There are a few 1% point differences between the files in the C and K channels.  I'm lost as to why this should be so.
Also interesting is that I can see this difference when switching between the two files.
It's not enough to not do it though, as this is the starting point for post work the 1% disappears during the first global contrast adjustments.