Thanks for the explanation. I’ve followed AP video and I have calibrated monitor, but the difference was barely noticeable.
Well if there is no problem with CC module when using custom WB I’ll give it a chance once more. But I have a question raleted to camera settings. If you create photo of the screen, in which WB mode to shoot? In my opinion it doesn’t matter but I’m not sure. Because I used Custom WB, which means the camera calibrates (adjust Kelvins) it’s WB according to the monitor screen. And then when I did white balancing in DT, channel coefficients didn’t changed much as I excepted.
You can ignore the error msg…. if you intentionally modify the starting values in wb… Its just telling you that it things you are doing a double wb. The “trick” is only going to help if it turns out to modify your these initial values from what DT uses as camera reference in a non trivial way and if the change has the desired result… I think often what you see when there has been an issue is the red scalar changes significantly from what DT uses… but as you say it might make little difference if your starting values are okay….
I should probably expand a bit on my previous reply.
When you shoot raw, the camera doesn’t alter the raw data based on the WB setting, but simply writes it in the metadata, so post-processing software (such as darktable) can use it if needed.
In darktable this information is normally only used for two things:
If you choose “as shot” in white balance, the output is altered accordingly. Any other option ignores it.
In color calibration it is used to calculate the initial CAT settings. As far as I know, this is always the case and doesn’t depend on the settings in white balance.
My version with ART. I used the white balance picker on the collar of the t-shirt, this was enough to fix the colors.
Then slightly lightened the image, and increased contrast a bit (but used a mask to keep the contrast increase from the face, since it is in shadow). P7010037.orf.arp (12.6 KB)
This is such a wonderful capture with a lot of emotion! It looks like your camera’s light meter got confused by all the magenta in the midtones (from the red brick and leather chair), and compensated with way too much green.
I used the color calibration module in DT to send those midtones back toward the skin tone latitude in the oranges, countered with cyan/blue in the mids, then added magenta to the highlights to bring facial skin tones back into normal ranges, since it looks like the light source from frame left might’ve been yellow/green bounce light off of a tree or bush through a large window, perhaps?
Also boosted exposure to even skin tones in shadow and take a little drama out of the scene, and a couple diffusion masks to soften the transitions into the hot highlights frame right to add a bit of a filmic roll-off.
I don’t think you have to play with the trick of taking a shot from a calibrated display – you’d have to be sure the display is very well calibrated, better than the reference values than the profile.
I just used color calibration to get a sample from the top right corner (seems to the the edge of a pool or tub):
Taking a sample from the grey shirt produced similar results.
The outputs with filmic and sigmoid can be tweaked to one’s liking. I also applied color balance rgb with the preset basic colorfulness: natural skin.
Todd, you do not have perceptual and relative render versions of the jpgs. In the first post, both jpgs and the xmp shown below the second jpg are sigmoid perceptual renders. In the second post, the jpgs and the xmp below the second one are filmic relative renders.
If you look at them they are clearly different… When I export I just change it in the export settings not the xmp so it will always show whatever I left the edit at which is usually my default. The change for the export is made just for that export. I use one of the color.org srgb icc as the one coming with DT only has a matrix and no lut and will not offer both renders no matter which you choose in the output profile or export settings for rendering intent…
So to use custom ICC profiles you put them in to the folders color/out or color/in as sub directories of your DT config folder. Then they are available. in this case it’s an ICC for output so color /out… Then just select it and choose your desired rendering intent… I do it from export but you could change the output profile intent if you wanted to
I did… though I tried to understand the true difference from using appearance. It did suggest appearance was superior to maintain color when going from a large to small colorspace so maybe it might be better. I found perceptual on both about the same as the DT profile and then relative had the strongest influence on the output with preference profile and using appearance seemed to offer something in between… It may also very picture to picture and in some cases the bolder look of relative is appealing and others its too harsh and perceptual is better … again all just personal observations…
