I suspect this is a simple question and a simple answer, but I might be wrong. From my perspective I can adjust the brightness of the image using the exposure module or I could go into the color balance RGB module and adjust global brilliance slider. Is there any significant differences that I should be aware of between these two methods. I don’t want to to compound this question with issues of shadow and highlights unless that is a critical factor.
Is there a difference in adjusting exposure in the color balance RGB module and the exposure module?
I bet ‘brightness’ (greater ‘exposure’) and ‘brilliance’ aren’t the same thing. But I defer to anyone with expertise.
They are slightly different.
They are set in different modules, which differ in pixelpipe position, i.e. in which other modules they affect.
Mostly, they are the same sort of adjustment, but in different color spaces. Exposure is a multiplication in linear rec2020 RGB, while brilliance is a multiplyer on the Y channel of XYZ (if I’m reading the documentation correctly): darktable 4.0 user manual - darktable's color dimensions
The docs say that brilliance is orthogonal to saturation, which implies that its response is modulated by color somehow. IIRC, it’s mostly equivalent to exposure IF the image is properly white balanced. But it can differ somewhat if it isn’t or for highly saturated colors.
Iirc, that brilliance slider could lead to “explosion” of the highlights, when used at over 20%. That 20% includes a contribution from the global brilliance slider. (See e.g. this thread)
That makes me suspect that the brilliance adjustment is not a simple multiplication like the exposure slider…
Gone with color maths fixes in 4.8 / master
But does that mean that the “global brilliance” is now equivalent to “exposure”?
The code seems to be way more complex than that. There are multiple colour space conversions and gamut mapping. I know very little about colour, but that does not seem like simply scaling Y
to me.
A grey ramp; I applied +1EV via exposure, masked it to cover the top; then I used that mask as a raster mask for color balance rgb, inverted, and raised global brilliance until the two colour pickers matched (RGB = 142 for all components):
That was with the default saturation formula: darktable UCS.
With JzAzBz, global brilliance required a value of 60%, but that also did not match exposure:
(No filmic, sigmoid or any other curves were used, hence the blown highlights.)
As always , exposure is used to set your middle grey. A lot of other modules use that as a reference for their calculations.
Even if they might look visually similar, there is clearly a difference in concept .
That being said, brilliance will push brighter things brighter , but do less on darker areas (crudely explained ). So it matters what is dark, mid and highlight before enabling color balance rgb. I dont think they do the same, although at first glance you think “both brighter = the same”
Not paying attention to the moment for anything color related taking two grey ramps and flipping one of them so that they cross each other at the “middle” grey is a nice way to visualize what the exposure slider does and all the sliders in 4-ways and brilliance controls in rgb color balance…
Waveform… brilliance sliders…
I have an image of my colorchecker spyder24… I put the picker on the white patch and set samples on the primary color patches…
Then I first enabled rgb CB and applied a global brilliance adjustment and then turned that off and added exposure to give the same white patch L value…
There is no tonemapping or other modules just the wb and necessary ones…
A simple gradient is maybe better and easy to read in the histogram, in this case in the histogram is displayed the corrispoding rgb curve
Untouched
Exposure
Color balance RGB
Edit:
Exposure+sigmoid
… a lot about math and showcases. I think remember one of AP’s chats or videos where he mentioned that the color balance rgb brilliance sliders can be used in an artistic way kind of like mix white paint into the corresponding parts of the image … whatever that means in digital image editing. But I like that explanation
Ya I didn’t play with those sliders in my video only the brilliance sliders as the OP was interested in those…but I still think the cross over is great…it shows the tonal range changes and how middle gray moves and the slopes of the curves shows the contrast… when you play with mask in the module you can see changes to those as well so I like that but we all have ways that we see things better for sure…