HOWTO: color toning instead of misusing white balance?

Software RawTherapee
1

Hi, this is my first post, I hope my question makes sense… but first of all a BIG THANKS SO MUCH for this wonderful piece of software!!!

On the rawpedia white balance page it suggests to not use the WB controls for color toning but use the “color toning” tool for that purpose. I think I understand the reasons. Now, using a gray card and the “pick” tool it is easy to get perfect white balance.

But how do I (re-)introduce a color tone that is similar (though maybe less extreme) to the original lighting conditions?

How can I make these two photos look different:

  1. picture taken well after sunset in very blue light
  2. picture taken at a campfire in very red light

The WB tool seems ideal for this purpose. In the “color toning” tool I do not see a “color temp” slider or anything like it. If I understand the recommendation on the wiki correctly, how do I achieve this goal without misusing the WB tool for this purpose?

Thanks a lot for your help!

2 Likes
2

@drrossum Welcome to the forum!

Color toning is one of many tools that you can use. It is just an example. The purpose of the section is to let users know the downsides of using WB liberally; i.e., some tools depend on it being correct. However, if your “correction”, although not radiometrically true, makes the image look balanced who is to say that it isn’t a good starting point? Simply be aware of the consequences and your goals.

I know you were asking for a howto but I hope that my post gets you thinking about your workflow.

PS I forgot to mention that the excellent CIECAM02 module provides you the ability to change the temperature of your colours. It is an advanced module, so you may end up having more questions than answers but @jdc is a kind and patient dev. :wink:

2 Likes
3

If the tone you’re looking for resides in the color temperature line, I’d say go for it. Indeed, I think “candlelight warm” is a look; Stanley Kubrick demonstrated such in Barry Lyndon.

For white balance, temp/tint doesn’t work well for me. I prefer to modify the RGB multipliers directly. More intuitive to me, R=G=B for white balance makes more sense than 6500K…

If it does the specific thing you need, use the tool with confidence.

2 Likes
4

Hi @drrossum and welcome!

It sounds as if you could gain much by looking up
https://rawpedia.rawtherapee.com/Film_Simulation

Have fun!
Claes in Lund, Sweden

1 Like
5

I am not at home, but in vacation with my family.
No computer, just smartphone

Further i will give a longuer answer, but now short.

The good way in your case seems

  • good wb with 2-choices : sliders or wbauto (uses itcwb in autowb branch, good in 90%)
    -then simple solution cat02 auto branch cat02wb. Or complete with Ciecam02
    -or warm cool (use cat02 algo) in newlocallab branch
    Jacques
1 Like
6

Thanks Claes, ggbutcher, afre, Jaques, for the quick replies!

I will experiment with CIACAM02 and read up on cat02 and film simulation over the next couple of evenings and report back how that goes.

7

@jdc after having studied the CIECAM02 page and having experimented with it a bit I don’t think I fully understand your suggestions above.

I understand the suggestions by @afre and @ggbutcher that whatever tool yields the desired result is OK. But as a beginner, I would love to learn the procedure for doing this correctly.

I think I understand (but please correct me if wrong) that the CIECAM02 module takes the RT output (after demosaicing and some corrections), then 1. brings it into some neutral color balance in a well-suited color space (by setting the proper “scene conditions”), 2. applies some standard transformations in the well-suited color space, and 3. transforms it now to non-standard “viewing conditions”. So, if I calibrate the RT WB setting with a gray card then I can set “scene conditions”->“WP model” => “WB[RT] + [output]” (what does “[output]” even mean here? in the UI we are still in the input section, no?). It will be neutral and ready for step 2 now.

Now, the moment to apply the intended color correction (that brings back part of the original lighting conditions) seems to be after step 2 but indenpendent of, and before step 3. Also, while the “WP model” sliders in the input and output sections can give the desired results, I don’t think they are intended to be used for this purpose because:

  • input WP: brings the data into non-neutral balance (since the original scene was not neutral, e.g. candlelight) so subsequent corrections behave differently than intended (and brings us back to the OP question).
    -output WP: this should be a constant for one output device. If I take two pictures with different lighting conditions (one candlelight, one after-sunset) these would need two different output WP settings.

What I take away from this is that the CIECAM02 module is not suitable for this purpose.

8

Thanks, @Claes, for the Film Simulation suggestion. I think this could do exactly the right thing at exactly the right moment in the RT conversion sequence, but the control is not suited for it. I mean that there can be many different lighting conditions that one would like to recreate in a set of pictures. It would be too much work to create a CLUT for each of them…

9

I repeat, i am not at home, but in vacation with only my smartphone.
I can answer in 3 or 4 days :slight_smile:
Jacques

10

Hello

First of all excuse my bad english :slight_smile:

Difficult to provide a precise answer when the question leaves uncertainties hanging.

In summary your analysis of the operating principle of Ciecam is fairly correct. We take into account “scene conditions” to calculate a position in a some form of work profile (with a small gamut)

In this “space” we can make modifications which will take into account the physiological principles of men (their brain, their eye) and not only on mathematical principles.

After, we can take into account the “viewing conditions”.

It is (I suppose to simplify), in your case the variation of “illuminant” which will lead to a change in the appearance of the image.

In fact you can “warm” or “cool” the image, regardless of other possible arguments that are important (surround scene and viewing, mean luminance Yb scene and viewing, absolute luminance scene and wiewing);

But for this to work, I am not “god” and I do not work miracles, it is necessary that certain prerequisites of ciecam are respected in particular that the illuminants are with a CRI (color renderind index) close to 100, that there are no multiple illuminants, neither reflections, nor too restrictive luminance conditions. For more information see rawpedia
https://rawpedia.rawtherapee.com/CIECAM02/fr

First of all, the white balance “scene” must be correct (CRI, etc.), and in limits of daylight illumiant (about 4100K to 10000K) under we can use blackbody, and above be carefull.

As a reminder a color (for each pixel) will be perceived by the result of a mathematical formula

Matrix spectral Color_percept pixel_xy = Matrix spectral illuminant (T - g) * Matrix spectral color pixel_xy / Matrix standard observer 2°
After that, it is necessary - in all cases to make a chromatic adaptation - to take into account our brain and eyes.

Remark: grey white balance, or pick on a neutral patch, does not necessarily give the right answer, contrary to popular belief.
In particular “pick” which can only work with small temperature differences, otherwise we infer a linear tansformation, which will probably give erroneous results for colors with strong saturation.

I recommand to use “Itcwb” if possible(Iterative temperature white balance),with an algorithm that I developed (actually in branch “autowb”). This algorithm compare a number of image colors (20 to 40), with reference colors known from 200.
It calculate first green (green magenta) which is at least as important as “red blue”, by an iterative approach which optimizes the correlation (Student)

If I suppose, WB is correct, you can change the appearance with 3 ways (using ciecam principles)

  • use principal module Ciecam with “free temp + green + cat02” and the value you want in viewing conditions (warm or cool)
  • use branch cat0WB, which realize the same thing, but suppose viewng conditions are D50
  • une branch newlocallab, and use “warm - cool” which will take into account the initial image

But nothing prevents you from using tools like tone-mapping or color-toning afterwards to create special effects

jacques

2 Likes