I’m trying to work my way through it right now…it doesn’t look too difficult but i have no clue where to get a color chart file to upload in the darktable charts gui
ArgyllCMS as described in the Tutorial. If there is none for your color chart, you can create one with scanin …
I am a novice…way out of my lane. If I purchase a color card from color rite, will i get the file that I need? I tried to find a .cht file as mentioned in one of the darktable charts videos but I couldn’t locate it. I realize that this is probably an easy task for someone with the right skills, but I have none…
There are chart files for most of the X-Rite color checkers …
find /usr/share/color/argyll/ref/ -iname "color*"
/usr/share/color/argyll/ref/ColorChecker.cht
/usr/share/color/argyll/ref/ColorChecker.cie
/usr/share/color/argyll/ref/ColorChecker.ti2
/usr/share/color/argyll/ref/ColorCheckerDC.cht
/usr/share/color/argyll/ref/ColorCheckerDC.ti2
/usr/share/color/argyll/ref/ColorCheckerPassport.cht
/usr/share/color/argyll/ref/ColorCheckerPassport.cie
/usr/share/color/argyll/ref/ColorCheckerSG.cht```@asn please correct me if I’m wrong and forgive me if I ask for something already explained.
For profile creation (sensor vs color checker):
a. I check only RAW for correct L value.
For color emulation:
a. I check both RAW and JPG. JPG has the priority here. If the JPG L value is around 96 for ColorChecker Passport then it’s fine even if RAW says 89 or whatever.
b. Shall I set the input color profile for the JPG to matrix??? This confuses me. If I set it like that the style is nowhere near where it shall be. If I set JPG input color profile to sRGB it’s close (I set my camera to produce sRGB JPGs).
I tried this in darktable 2.7 (master) and input color profile has additional option working profile which I left at linear Rec709 RGB.
I hope that for 2.7 the process is not going to change drastically.
I did only an exercise. If I have time I would like to create full extraction of profile and color emulations for my camera since I like the Portrait mode rendering in my Lumix GX9.
YES. The JPEG has a tone curve applied it is probably brighter. Feel free to provide suggestions to improve the text if this is unclear!
You check the RAW for the correct L value. Then you check that the JPEG is L is between 96 and 98. L=100 means it is overexposed and we wont get useful information.
Don’t change anything, it has no color matrix as it is an already developed image! The camera should produce AdobeRGB if possible, as it offers more correct colors!
Hmm maybe the sRGB generated JPG is the problem here. If I did not change anything in the input color profile everything is washed out. If I set it to sRGB it looks more or less correct. Need to check it.
If the JPEG is sRGB then the input profile should be SRGB. It is a developed image.
I have tried adapting this process to a workflow in which the end goal is color-accurate photos taken under controlled conditions (camera positioning, focus, lens, lighting, etc are all the same for every photo). I’m running into a little trouble, though, so I’ll outline the process I’m using and maybe someone can tell me where I’m going wrong.
The camera is mounted above the subject, and the subject is evenly lit from the sides so that there is no glare. At the beginning of the session, I take a photo of the color side of the colorchecker passport at the same distance from the cameras as our subject (give or take a cm) so that the lighting will be the same. I turn off the base curve, sharpen, and highlights modules, and crop the photo down to just the colorchecker. I leave white balance on the “camera” setting. I then export it as PFM as detailed in the posting, and use the argyll reference cht and cie files to generate a dtstyle for LUT, tone curve, color correction and whatnot. I then take a photo of the white balance card in the colorchecker passport, and use the spot function of the white balance module on the grey card to create a white balance profile. The LUT style and white balance profile are applied to all the photos I take during the session, along with a style that turns off the base curve, sharpen, and highlights modules. Lens correction is also turned on for each photo, letting darktable automatically set the values from the exif metadata in the raws and the camera.
I think I might be doing the white balance wrong, though. The photos always turn out looking too reddish. This is apparent to the naked eye, and if I use the color picker on a photo of the grey card that’s gone through the process and look at the mean RGB, B is always lower, and R is often higher than G by about half as much. (To my understanding, they should all be equal.) I’ve tried setting the white balance on the PFM shot using the grey card before generating the LUT style, but that doesn’t look quite right either. I’ve found I can get closer if I turn off white balance completely for both the PFM shot and the shots of the subject, but I still think it’s not quite accurate.
Additionally, placing the grey card in different parts of the frame yields different results when using it for white balance. I suspect this might be due to the white balance spot picker using values from the raw image, prior to lens correction eliminating vignetting and such, but I also fear that it might be an indication that the lighting isn’t exactly even across the subject.
Any guidance on this would be greatly appreciated.
Hi, there! Although I’m aware that it’s theoretically impossible to render 16b per channel out from 8b per channel and also about the lower headroom while processing within lower bit depths, I still can’t stop thinking if these tools can facilitate profiling a jpg to another jpg or to “real” profile? Or could give to oneself the opportunity to at least die trying? 
Thanks!
This is what I was referring to. Thanks.
1 Like
In addition, we might also discuss about 16b tiffs. Some of us might sill hit film, develop and scan it. Although film has its own sweet inner color profile, we may need some profile sometimes. 
If you have film, you’ll want to profile your scanner INA similar way to this. Specific films have LUTs available. Darktable 2.8 will have a LUT module that’ll be able to use all the film emulations available to g’mic, for example.
Is there a way to use darktable-chart to create an icc-profile to be used in other software, e.g. capture one?
well, it looks like i can see only the colorchecker grid but not with the images to apply the grid on, both source and ref image. am i missing something? windows platform. thanks.
You have to use argyllcms for that.
Hm, from what I’ve read about Argyllcms (brief google search), it would help with profiling the camera, meaning matching the color target, but it would not be possible to create an icc profile to match the in-camera .jpg. Is this correct?
I’ve used darktable-chart to create styles to match the in-camera jpegs of my Olympus, which worked great. My goal is to create icc-profiles which do the same, but in capture one.
Thank you, this sounds exactly like what I was looking for!