I am trying to ‘live the dream’ and use Darktable the way M. Pierre is advocating with mostly Linear RGB tools. I have appreciated his Quick Approach article too.
Color Balance tool, though, is causing me grief. For a tool with a name that says it is all about colour, it sure has a big effect on tones. But, putting that aside, how do I use it to make colour adjustments, especially, how do I use it for the things I would have done with Color Zones tool?
A simple example: changing the blue sky in a photo. Perhaps small changes to its hue, its saturation, and its lightness. I know how to do this with Color Zones. But with Color Balance, I don’t even know where to start. I am trying to use it in Slope Offset Power, with HSL sliders – after all, I want to adjust H S and L of the blue. Where do I start? I thought if I picked up a colour picker and sampled the sky area… but that sure doesn’t work! I just neutralizes the sky!
Does anyone know of a written or spoken tutorial (not Boris H’s wordless videos: I need a walk-through commentary tutorial style, to know what is being done and why) for doing the above, using the current edition Color Balance in SOP mode? Or any other help would be much appreciated.
Thanks Boris. I saw that 2 weeks ago on Youtube (and commented to Bruce), and I thought it was interesting, but Bruce doesn’t use it for replacing Color Zones. In fact, 6 episodes after this one, he teaches us Color Zones. He is not trying to use Color Balance to adjust HSL of a particular colour.
Let’s take the example I mentioned above, of wanting to change the sky. Here is a sample (I am actually happy with this sky, but let’s pretend):-
It says the hue is 51.5 degrees. I assume that adding 180 degrees will give me the sky tone, so I manually set midtone hue to 231.5 degrees. Let’s see what happens when I then add some saturation:-
Different algorithms used by the modules cause different results. Why don’t you use color zones if it meets your requirements ? Nobody is forcing you to only use specific modules. The result of your edit has to comply with your requirements, not the way you reach this goal.
The hue sliders are there to create a split toning effect or to correct color casts (like in a picture with a mixed white balance). If you just want to increase saturation in the sky only, use a drawn+parametric mask and increase the saturation slider.
With color balance you can do white balance to a certain degree. It is mainly used to change the color mood of the image, as @Bruce_Williams explained.
To do white balance, you need grey as a reference in the highlights, mid-range and shadows. Here is an example:
Highlights are now corrected, but in the middle gray area yellow is still there. The wall of the castle is in this area, so I use the power function with the wall as reference:
I tried similar to you and also couldn’t get the same results. However you can change the color control sliders from ‘HSL’ to another option I can’t check right now, but I think is ‘RGBL’. I find this far more workable. If you don’t want it to effect luminosity, change blend mode to hue or chroma might help.
From M Pierre’s articles, e.g. the linked article says, “Modules to be used with Care: Color Zones… Prefer Color Balance.” I take it that means you can do with Color Balance what you would normally want to do on Color Zones… otherwise it’s better to say “only Color Zones can do what it is usually used for”.
By the way:
What is the difference in use between lift, gamma, gain and slope, offset, power ?
Both work in linear RGB, the identical colour space and have exactly the same setting sliders (but without the auto optimizers in lgg).
IIRC, he does say that color zones is OK if you’re making small changes. The interface to color zones is great, IMO. If you just want to make subtle HSL changes to blues (sky), go ahead and use color zones, IMO, it’s much more user friendly.
Aurelien does recommend color balance if you want to make heavy-handed changes. An example would be changing the sky to red. To get a similar color-isolation effect as color zones, you need to add a parametric mask using chrominance (and possibly luminance) values to isolate the sky.
I think in the end people are a little too strict on their interpretation of the linear vs traditional workflow…even AP himself has said some things to that affect in the course of his web casts etc…basically if you can stay as linear as possible then you will stay as accurate to the original sensor data as possible how ever if you like the results you are getting with whatever tools and a small hue shift here or there does not bother you then just use the tools that give you the look you want…something that seems to be getting lost to me is this quest for the perfect edit based on the tools or the workflow…painters don’t all use brushes…some use their hands some throw cans of paint in the jet wash…the point its they manipulate in the way they desire to get the result they want…so I would stress not to be too obsessed with this…it is certainly good to know the tools and know the limitations when you apply them but that is where this rigid notion of what to use when should be relaxed…for example at some point in the past I used a 3 image base curve fusion to enhance an old dark JPG from a point and shoot camera from 15 years ago…it looked great…I was just messing around but what no base curve on a jpg…I broke the rules…For me CB module gets used to do WB , saturation/contrast and color casts (since it automatically applies a 180 degree shift) and I use the tone sliders much like levels sometimes but I always use color zones to tweak the colors it works much better for me so why would I struggle to conform to some “rule”…others might use the channel mixer with without masks etc etc…I would say as a last point if you are struggling to get a good result then there must be a better tool…keep experimenting …prioritize results over rules…
please dont mix the sRGB with the pro photo rgb version of LiftGammaGain.
both prophoto versions works linear: the “old” lgg and the new american one. and i cannot find a difference in the results.
@anon31514447 There is a difference because the implementation is different. What I want to know is how you are comparing the two. Show us by dragging and dropping sample images and give us the steps or sidecars that you used to process them. BTW, welcome to the forum!
I think this may better explain it…if DT works as Blender does on the math then there is a hard coded max and min in the lift gamma gain model but not the slope offset power…I’m not an engineer but the explanation sounds reasonable… cycles render engine - What is the the ASC-CDL node? - Blender Stack Exchange
@ afre Ya its a rant isn’t it…sorry…going Covid crazy but so many posts asking how to get my raw to look like the jpg , must I use this tool or that tool etc etc…what order …I just felt the need to make a suggestion to be less rigid and explore…so apology offered to all for the rant ha ha hope you are well…thanks for harpooning me…
