Introducing a new color balance mode in darktable

According to this website (posted 2013) saturation should be after “S.O.P.”
Familiar looking diagrams :thinking:

It also has some stuff about the xml implementation on other pages (if that was of interest).

Which is an odd choice given that the further away from linear ratios, the less accurate a desaturation becomes.

Shoot%20Minh-Ly%20Toronto-0044-_DSC0252-ASCCDL_01

first results with all the ACD stack.

Edit : I don’t know what’s going on with the color here, what I see in dt and Gnome viewer is that :

Capture%20du%202018-09-24%2004-05-05

2 Likes

Hi folks !

I have added color-pickers to neutralize casts. Basically, once you pick an area, it computes the average color and inverts its hue, so you can revert color casts more efficiently. It doesn’t always get the saturation right though.

Capture%20du%202018-09-30%2018-53-44

1 Like

Hi again,

I have improved the accuracy of the color neutralization and added an optimizer. When you select 3 samples (black, grey, white), the optimizer tries to fit the CDL curves in order to neutralize the color of the sample patches.

Here is the result with no manual tuning (except for the sample patches selection) :

4 Likes

This feature has been merged today in darktable master branch and will be released (as far as I understand) in dt 2.6, next Christmas.

9 Likes

Hi Aurelien,
having tried your workflow exposure->unbreak input profile->color balance (as explained in your tutorial on darktable.fr) I can say I’m very impressed by its performance and reliability.
Even if automatic mode fails sometimes, the manual mode does the job very well.

However I would like to integrate the camera calibration into that process.
I’ve created a camera profile following the method explained here.
It provides better results than the camera base curve.

But, with your process, adding the color look table preset tends to jeopardize the work already done…
What are your thoughts about this. Do you have any advice ?

Hi Philipp,

Thanks for your feedback ! Today, we have added generic presets to the unbreak profile to fallback when the auto mode fails. They are provided for 8, 10, 12, 14, 16 EV dynamic range cameras.

I have seen the same issue with color lookup. I don’t know yet if the inaccuracies of the LUT are just amplified (it’s really hard to get an evenly-lit chart), or if we need to do the LUT on a log encoded chart picture (which defeats the purpose of the color balance after, since the LUT will then linearize colors). For now, you could try another LUT based on a log image of your chart (tune the settings of the unbreak profile with the mid-grey, white and black patches).

Also, I have discovered just yet that, after log + colorbalance, the global tonemapping with draco method gives much better results, betten than shadows-highlights.

Anyway, still WIP.

One in English would be great or an update to the darktable manual :slight_smile:

1 Like

I know, but as the GUI is beeing currently modified, I will do the tuto once all the labels/settings/modules are definitive

2 Likes

Here is my try.

image

image

The middle grey color picker value is far from 50 and there is no way to get it.
What is wrong there ?..

You need to use the gamma factor in color balance (with slope offset power) and set it at roughly -0.62. Could you send me the picture ?

The LUT obtained with this method seems to work well (may be by chance, to be double checked)

Not a manual, but here is a straight forward way to use Aurélien’s tools I’m happy with :

  • color balance (when possible)
  • exposure. Center the histogram (set as linear) visually keeping room on both side to avoid clipping (the automatic clipping fails some times).
  • unbreak. Idem. Keeping grey at 18%, play with black point and dynamic range against histogram still keeping room on both side to avoid clipping. Considering only color picker may be misleading. It happens to have max Lab at 96 and still saturation zones on the image. The histogram is more reliable (and faster).
  • lookup table with camera profile preset (optional).
  • color balance (slope offset power). Center the histogram with slope, enlarge it with contrast until before clipping, and balance the image with power. Then tweaks with contrast and SOP if necessary.

No unexpected or weird effect as far I can see. A beautiful (well, … depends ! :wink:) and natural looking image.
Bravo !!!

1 Like

Unbreak works on the log curve for data sanity and not for visual accuracy.
The grey is a real one and internal data corresponds to a real grey, so the color picker should have a mode (Lab Log ?) showing up actual data (50) instead of displayed data (75).
Does that make sense ?

No. Displayed data is the actual data, if darktable does not apply a silly gamma somewhere (which I suspect, but still no answer on the mailing list).

Log is a way to salvage the dynamic range and squeeze into in what the display has. What we need is a later way to remap the grey and manage the contrast at the extreme values. I’m currently working on that.

:thinking: … On the colorchecker view, the middle grey patch is really displayed lighter than a real middle grey (just compare with the dt background middle grey). The color picker is not too bad saying 75 as displayed color.

It’s not always clear what the float space should be. As a former 8-bitter, it took me a bit of digging to find out that the imaging convention is 0.0-1.0, because I think the practitioners just assumed we all would know it. G’MIC confused things, as it does floats, but in the 0.0-255.0 range.

8bit / 256 = normalizedfloat, for this discussion, so 75 / 256 = 0.29, or thereabouts…

1 Like

I think that’s only for the benefit of inputs/outputs that require it - G’MIC enforces no particular assumptions about data in general, although some specific commands/filters might do.

1 Like

Right now, rawproc internal data is all 0.0-1.0, but my curve tool is still 0-255, but I convert a curve value to 0.0-1.0 before applying it.

To the main discussion, it really isn’t always clear what data is being worked vs. what is being displayed. In that regard, I’m probably going to make 0.0-1.0 the configurable default for tools using the curve core, in a subsequent version of rawproc.

Sorry Troy, on the UI I don’t see any normalized value.
With only white balance and exposure, tweaking exposure, color picker Lab values match the colorchecker lab.


RGB values:
image

Adding the unbreak module (log) I can tweak it to get the same black and white values, but the middle grey is on 75 instead of 50. The displayed patch is really lighter the previous one.


RGB values;
image

Then Aurélien suggested to apply a gamma factor at -0.62. Having done that I can tweak again unbreak module to get this:


RGB values:
image