How to save reference image for profiling

Hello

I purchased Lumariver Profile Designer (the commercial version of DCamProf) for creating my own camera profiles and already successfully created DCPs for use in Raw Therapee. Now I am trying to create ICC camera profiles for use in darktable. However, I am unsure how go about exporting the required TIFF file from darktable. Unlike Raw Therapee, which has a dedicated option to “save reference image” for profiling, darktable does not seem to have this option and I could not find a guide to do so. One always has to choose an input profile as well as a working and output profile. What I did was select “linear Rec.2020 RGB” for all three. These are the available options:

EBW2kvHpnDeZrlpv

I disabled all other modules except white balance (spot WB on second left grey patch) and then exported as 16 bit TIFF. The resulting image looks similar but not really identical to the output from Raw Therapee. The files can be found here if you want to take a look:

The original CC24 Raw shot and ICC profiles created by LPD are also there. As expected, the colors from the two ICC files are different and I am not sure which is the “correct” one. What would you suggest? Should I simply use the linear TIFF from Raw Therapee (as this uses a correct way to export a file for profiling), even if it is from a different Raw converter? Or is it absolutely necessary to use darktable for exporting the TIFF if I want to use the created ICC file there? If the latter, what settings would you suggest for the input/working/output profiles?
Thank you any hints you can provide.

Kind regards
Stefan

With the novel color calibration in DT you might actually get just as nice a results by using it with the stock profiles… and skip messing with custom icc files

Lumariver has so many options… you don’t want to add tone curves and looks and use only the basic setting if you are trying to match color and exp if you are going to use filmic or sigmoid

Is the working profile in RT prophoto?? Might make a little difference…

EDIT

Ah I took a look …one thing is that the option in RT does not embed an ICC so the data gets a default profile with gamma and so its dark when displayed… in DT if you change the input profile to linear rec 709 or 2020 you will see it looks just like the one generated in DT which does have an embedded ICC and so displays correctly…

You want a linear TIFF with only white balance applied. Save the TIFF with no export profile; you want the camera space to make the profile. Either RT or dt should be able to do this.

Using the raw you shared and according to DT , CC module colorchart calibration generates a very nice color correction down to a pretty impressive delta E of 1.4…

image

image

DT does it by setting all profiles to be the same…it treats this as a no op and passes the data through…this is explained a few places

Once place is here … prep instructions for Pascal’s colormatch script using Argyll…

I tried to do that, but “no export profile” does not seem possible with darktable, one has to choose one of the options I listed above. Or am I missing something?

OK, thanks for investigating. So in conclusion, what would you suggest to use as input TIFF for Lumariver for creating the ICC profile? The RT or DT TIFF?

Hmmm… I just inspected the darktable docs (4.0) and there’s not enough information there for me to inform you. I don’t have an executable for darktable at present, so I’m going to have to rely on one of the users…

This is important; if the TIFF you save is in some other colorspace than the original camera response, you’re not actually making a camera profile.

pcode — Darktable Camera Color Profiling.pdf (134.8 KB)

Here are some old instructions…but basically it carries forward… At least for DT the recommendation that I am aware of to prepare the TIFF is to set all the colorspaces the same in all the profiles… This apparently passes the data through…

The images from each source looked the same to me but I guess you could analyse them. RT says it takes the image in in camera space and then passes it along with not color conversion…

DT instructions from here…

Processing The RAW

Open the respective RAW in your favorite RAW processing
software, and disable all possible processing. Set linear
gamma (1.0) and set input and output profile to the same
profile, and rendering intent to absolute colorimetric if
possible, effectively creating a color management
pass-through. Then apply cropping, and lens distortion
correction if needed, be careful not to apply lens
vignetting correction. Then save to a 16-bit TIFF
(again without compression).

Particularly for Darktable, make sure camera white balance
is unchanged, disable basecurve, disable sharpen, set a
linear rec2020 for input and output color profile. Make
sure not to override the output color profile while
exporting.

Particularly for UFRaw, make sure camera white balance is
unchanged, restore details for negative EV: clip, clip
highlights for positive EV: digital linear, load Argyll’s
ClayRGB1998.icm both as camera and output profile, with
an absolute colorimetric rendering intent. Set output
bit depth to 16-bit. Set gamma 1,00 and linearity 0,000.
Also make sure Embed exif data in output is enabled.

Particularly for DCRaw, you can use the following command:

dcraw -v -w -M -H 0 -o 0 -6 -W -g 1 1 -f -m 1 -T …/_MG_1234.CR2

Thanks, the PDF you linked to seems to contain all the info I need for now.

I would test your profile vs the default one and use then use the colorchecker mode of the CC module with each… You might find this novel tweak in DT which is much quicker than messing with profiles might be just a good… I use the exposure auto picker extensively to set exposure and this doesn’t work when you use custom icc profiles so this is my other reason for using the CC module if I want to correct color… I find setting it to 50% and doing a first run on the entire image is often a great starting point and if not I do it once more on a selected region or face or somethings that I want exposed nicely… one of the two is usually for me a perfect exposure and since I started doing this I am also far happier with the filmic results I get when I use it… I think often in the more distant past I never added enough exposure for this type of workflow…ie scene-referred using filmic…

DT chart is now working again… I took your image and created a style from it just to see what it would do… it did a nice job balancing the image which seemed to have a cooler wb as shot…

NEF.dtstyle (4.5 KB)

Thanks, interesting! Incredible what sophisticated tools DT offers, now if I just had the time to read the documentation for them all… :slightly_smiling_face: