In a way, most things are science-based: trial and error, (null) hypotheses and experiments. 
Instead of endless discussion, we should roll up our sleeves more often and fire up dt.
2 Likes
Any chance you could post the .xmpās? Thank you.
You can get them from the jpg just load them as sidecars on an image.
2 Likes
Well color balance is giving you global but also tone based editing of color and luminance ā¦much as the previous model did but in a more sophisticated way. The color calibration module is more rgb channel basedā¦the beauty is that you have many ways to fine tune color nowā¦
DSC_6236_01.NEF.xmp (13,6 KB)
DSC_4707_06.NEF.xmp (12,4 KB)
3 Likes
Indeed. Iām not really lost with the modules, and I think Iām coming up to speed with them, but at the same time Iād like to learn a little more about their functionality
Thank you, Boris,!
Thanks Todd!
Color calibration : fix light to match screen illuminant. Then, use the ratios between RGB channels to brighten or darken selectively. Then massage those ratios to make the light spectrum more even (less colorful, whiter, flat spectrum) or more spiky (more colorful, more like a laser sharp light spectrum).
Color calibration is very much grounded into light and emulating optical filters (like gels). Damn robust, great for a first basis, but not linked to color in a perceptual/psychological way, so not well suited for creative stuff.
Color balance : shift colors in perceptual spaces. We are much closer to paint mixing (which started the whole saturation thing : mixing white and red paints yields pink, yet your usual chroma slider gets you grey-ish red). Itās a much more arbitrary way to push colors around to reach new parts of the color spectrum to create an atmosphere and bring life to images, while ensuring colors stay in gamut and providing control to not saturate everything 200% at the same time.
Color balance lets you rethink your color palette in post like a painter, to do whatever you want including emulating film look. I believe https://www.youtube.com/user/marcobucci/videos will be of some help to get the color job.
7 Likes
Would you explain a little more about what ratios you are referring to?
Sorry Todd, Iām too dense to understandā¦can you explain the steps to do this. Thanks.
Roger
In most cases the xmp data or meta data is also in the jpg file so you can treat an exported jpg like an xmp and load it as one. So pull up the raw you want to test with and in lighttable go to the right hand panelsā¦where it says load sidecarā¦refer to the jpg file this should load the history on to your test imageā¦you may have to alter the filter in the dialogue box to show all files as it defaults to xmpā¦HTH
I think AP is referring to something like this
Say the output channel is Red. In the default setting, the three sliders are:
Red 100%
Green 0%
Blue 0%
The percentages denote how much of each channel we put in the mix. (ratios)
Obviously, Red +100 and 0 on all other sliders is exactly the same as Red. Source and Target are the same, so nothing changes in this case.
Now letās add some of the Green to the formula. Change the percentage on the second slider (Green) to 20%. At first thought, you might expect the image to get greener, but no. The target of this change is the Red channel. We make the Red lighter by adding some of the Green to it. Hence, the image gets redder. To be precise, it gets more redder where the Green component is light, and less redder (but still redder) where the Green is dark. Only where the Green is black, the new Red will be the same as the original one.
1 Like
You got R, G, and B values expressed in a 3D space. You can break them into a pixel norm, such as N = \sqrt{R^2 + G^2 + B^2} (euclidean norm), or any other norm (but color calib only uses this one because it has the nice property to be the same no matter the color space you are in).
Then, your ratios are r = \left\{ \frac{R}{N}, \frac{G}{N}, \frac{B}{N} \right\}. You can do whatever to your norm N (filmic in chroma preservation mode only deals with the norm), as long as your output ratios are equal to the input ones, the chromaticity will be preserved (so mess up with N as you want, then multiply it by r at the last stage). Itās easy to show that exposure compensation leaves the ratios untouched. But of course, preserving chromaticity does not mean preserving the color appearance (letās say it preserves the light spectrum and only brings it to an higher or lower energy), for this we would actually need to change the chroma and the saturation using some perceptual model.
The colorfulness in color calib increases or decreases the ratios r to make the color more or less colorful. Itās called colorfulness because it is neither a true chroma nor a true saturation, in fact it doesnāt match any property of color (from a psychophysics point of view) but rather simulates spectral sharpening.
2 Likes
Right, these ratios too.
You do it hereā¦
1 Like
@anon41087856, @priort
Thanks for the explanation. Basically, all the manipulation we do to the R, G and B, colorfullness, and brightness tabs is massaging the ratios.
Thanks very much, and the link is excellent btw. If you donāt mind, could you explain the differences between the Color Calibration brightness and Color Balance RGB brilliance (previously purity)?
P1, P2, P3 have similar saturation but increasing brilliance. Same for R1, R2, R3. When increasing or decreasing brilliance, we do so at constant saturation, following the equi-saturation lines. (Note : disregard the RGB equations, everything happens in JzCzHz).
The brightness in color calibration is a simple exposure compensation, it wonāt change chromaticity.
4 Likes
Perfect. Thank you very much