Are you on linux?? There is a script that does a blended icc profile from a raw jpg pair…I have found the results to be not bad…GitHub - pmjdebruijn/colormatch: ColorMatch
It could just be that the input profile is not a good fit. Here the color calibration with the color checker should help but only for that lighting and maybe in general if it is a daylight one. You could also try to render you images perpetually vs relative and see if one of the mappings give you a closer result in you output files??
I think I stumbled across this script some time ago when I tried to calibrate my monitor on Linux the first time - so back then I didn’t pay too much attention.
So instead of ‘standard color matrix’ one chooses the hereby created ICC then?
Correct…I did have to modify the script to work with a color checker 24 patch…supposedly works much better with more patches but I still felt good about the results…I wrote up what I did and posted it a while back I will try to find the post for you…
thanks a bunch! I also cheaped-out and only posses a ColorChecker24
(Got this Datacolor set with checker, cube and Spyder X)
This is my post but the thread itself might contain something of interest as well…
I have been following this with interest as I feel the same. ART or OOC jpg colours look spot on - but darktable often delivers colours that look decidely odd: including skies with a green tinge and grass with a yellowish orange tint. People don’t mind a slightly dull image and adding contrast etc but having to individually fix the colour in each image feels like something is wrong. I do think people have a valid point to make here.
In the end I worked out what was most consistently out of kilter and tweaked the RGB values in colour calibration by eye, and I now seem to consistently have images that are a more than passable starting point.
Although this works, it does feel like patching up with sellotape. I am happy to be shot down here for failing to grasp the last few years of dark table’s work on colour science but would it ever be possible to combine filmic with the algorithm in RawTherapee or ART that fits the tone curve to the embedded jpg as an option?
Please post problematic images.
Sure will do. I think it would be useful to gather the best examples with my some comments on my thought processes so that people may see if I am doing something wrong.
Won’t have much time on the laptop over the next few days so don’t think that I have forgotten…
We will be here…I have notice a funny thing and in many instances I find I like the color in my image better when I blend filmic in lightness. Likely this is going against some part of the code but to me it takes a bit of the desaturation of filmic and resaturates in a pleasant way vs the results from blending in normal
If you want to post a shot of it I can run it through colormatch and you can try the icc file…
Thanks for offering! I think I will have to do a better shot first.
OK sorry for the noise. I realise that on balance, using colour preservation (RGB power norm) in filmic RGB gives better results (less lurid grass, more saturated skies) than having it disabled, counter to my first impressions, and when used in conjunction with colourfulness preset in RGB colour balance mostly gives pretty realistic colours, so I think I need just to RTFM…(retreats back into hole)
No need to be sorry, and I’m sorry if that came off aggressive. I’ve found lately that talking about these things in the abstract isn’t very productive, and getting an images to work on proves or disproves the point quite quickly.
No worries, didn’t seem aggressive to me. You are right though, images are much more use than a descriptive recollection… which often turn out to be the ramblings of a slightly distorted memory!
Its a bit of a preference thing. I prefer none at least 50% of the time and Euclidean over power norm. I suspect this might vary depending on what camera you have and the starting point for the color before filmic. I have also taken to blending filmic in lightness mode…I like the look better and it removes a bit of the desaturation that I think can be a bit aggressive or too much for some images.
I wonder why people hate no preservation so much. It seem to be much closer to the out of camera colors when white relative exposure are not push above 4. i just wonder what are the disadvantages when compared to power norm. i find that power norm does preserve colors when is push way above 4 and the contrast seem to reduce very much. maybe the experts can shed some light
The manual has some explanation: darktable 3.6 user manual - filmic rgb
My guess would be that no tends to saturate the shadows and desaturate the highlights, which is not great for landscape type of shots.
I think they have…from the manual…
“There is no “right” choice for the norm, and the appropriate choice depends strongly on the image to which it is applied. You are advised to experiment and decide for yourself which setting gives the most pleasing result with the fewest artifacts.”
Well, to be honest, I choose “no preservation” everytime I use Filmic RGB. Firstly, I don’t like the look of any preservation mode, which is subjective of course; secondly, I see it gives nasty artifacts when it comes to overexposed areas such as lightbulbs, chandeliers etc.
And since I can’t escape from the bulbs indoors, I don’t preserve chrominance