Sunset on Brusvikken - Darktable filmic vs sigmoid

A sunset scenery captured this evening right in front of my place. I moved in my new place this month and I am still stunned by the beauty of my new neighborhood.

Straight out of camera jpg, Provia Fuji film simulation with mostly default settings (but DR400):
DSCF9668

DSCF9668.RAF (17.4 MB)
This file is licensed Creative Commons, By-Attribution, Share-Alike.

Filmic, preserve chrominance RGB power norm:
DSCF9668_02

Sigmoid, preserve hue, 4% crosstalk:
DSCF9668_03

Both filmic and sigmoid give a closer rendition to what I perceived in real life.
I somewhat prefer the smoother highlights of sigmoid, but I think filmic is preserving more details that can be used for further processing. Similarly I prefer the default contrast of the shadows of sigmoid, anyway they are quite close.
Enjoy the challenge!

8 Likes

My attempt with dt 3.4.1, so I do not have sigmoid.

DSCF9668.RAF.xmp (9.5 KB)

4 Likes

with filmic, 3.4.0

DSCF9668.RAF.xmp (17.7 KB)

1 Like

Or… RGB Curves!

sunset-on-brusvikken-DSCF9668.RAF.xmp (14.7 KB)
dt 3.4.1

3 Likes

DSCF9668_dt2 Filmic…

DSCF9668sig
Sigmoid

8 Likes

Thanks for sharing
Filmic only (+colorcalibration , exposure auto, and color balance) , I didn’t install Sigmoid yet
dt 3.4.1

DSCF9668_03.RAF.xmp (20.0 KB)

10 Likes

Very nice location and a perfect evening mood. Thanks for sharing.

dt 3.5.0 sigmoid, tone equalizer to restore highlights, colorbalancergb for some extra-colour-pop, local contrast and contrast equalizer for crispness.

DSCF9668_01.RAF.xmp (11.9 KB)

3 Likes

DSCF9668.RAF.xmp (14.0 KB)

4 Likes

nice. very subtle. similar to my render with vkdt:

mostly:

param:colour:01:exposure:0.72327
param:filmcurv:01:x0:0
param:filmcurv:01:x1:0.1
param:filmcurv:01:x2:0.9
param:filmcurv:01:x3:1
param:filmcurv:01:y0:0
param:filmcurv:01:y1:0.03145
param:filmcurv:01:y2:0.92767
param:filmcurv:01:y3:1
param:filmcurv:01:black:-7
param:llap:01:sigma:0.12
param:llap:01:shadows:0.50314
param:llap:01:hilights:0.37736
param:llap:01:clarity:0.45912
5 Likes

Beautiful image. Tried to kepp it soft and “natural”. No filmic this time, just legacy tone curves :wink: .


DSCF9668.RAF.xmp (25.4 KB)

3 Likes


sunset.brusvikken.pp3 (15.7 KB) RawTherapee 5.8 (Development)

5 Likes

Thanks for sharing, nice place!

Photoflow, exposure adjustment, noise reduction, dynamic range compression, local contrast, tone mapping and colorspace conversion.


DSCF9668.pfi (35.7 KB)

4 Likes

It’s interesting that that are greater differences between different people’s filmic vs filmic (or sigmoid vs sigmoid) renderings than between filmic vs sigmoid renderings by the same person. I think the difference is more between the tastes than between the tools.

8 Likes

And between the different monitors/settings used as well, I think that’s the main reason why there are often important differences between entries in the PlayRaw section…

4 Likes

Yep

Absolutely. I often wonder what the heck I did when watching my own edits on different screens :roll_eyes:.

1 Like

Thank you for the play.

imported with RT (DSCF9668.tif.out.pp3 (12.1 KB) ) prepared for heavy color grading (and masking and retouching) in GIMP and G’MIC. So the RT edit is totally different :rofl::

I agree that the differences are more between tastes than tools.
Here is a collage to ease the comparison. Indeed there is a big range of variability regardless of the tool.

7 Likes

We need difficult images to process to really compare the two tools. I don’t know if I’ve seen any difficult images in the sigmoid threads.

Even base curve does just fine with the easy images, that’s not the point.

1 Like

It wont be easy to compare this way from random submissions. Filmic has a default exp bump of .5. That’s not enough for Fuji. Likely other cameras may have an optimal bump that is not .5. This bump might not translate to the sigmoidal curve. It might do better with more or less EV than filmic. This setting will might impact how the shadows, highlights, and overall tones are evaluated. If the goal is comparing two curves maybe some common conditions need to be specified?? Perhaps no changes be made and only apply the curves but even then unless one is significantly better every time it is going to be a tough call I think. The collage above is interesting but really there are more modules in play between edits. I guess the best you can do is perform an edit the way you would do it as your best effort and then export a version using either tone mapper and go from there as to results or to see how hard it was to get it “right” between the two…

So I guess what is a fair testing schema…