Best practice for "unexpected white balance coefficients" in darktable

[123sg (Steven)]

@Frisco

colors look absolutely terrible

Don’t worry, you’re not alone… dt can be a little shocking that way to start with
And I should say I don’t have a modern Sony so maybe there’s more going on than I’m aware of. I do find that sigmoid instead of filmic makes a difference (for me).
I’m not sure about this:

@Frisco

Note: The “R”, “G” and “B” labels on the channels of the color spaces in this module are merely conventions formed out of habit. These channels do not necessarily look “red”, “green” and “blue”, and users are advised against trying to make sense out of them based on their names. This is a general principle that applies to any RGB space used in any application.

It’s not that bad, but it’s also not quite as simple as:

@Frisco

if I want my foliage more green, I might turn down the R output for G input and compensate by increasing the R output for R input slightly,

either! The trouble is that hues can shift too - the colorfulness and brightness tabs are straightforward though.
Would you like to post a Raw file, one that you’re happy to share? I always like the challenge of matching this sort of thing. You just need to state a licence, like in this post for example: Out of Africa (Kruger National Park)