darktable color calibration - Brainblasting…

Happy new year,
well documentated, my IQ is not sufficient for that module. I thought, I’ am experienced user )-; Spent a lot of time with the new module - for me it is a kind of mystic.
Belongs to this: color calibration - Brainblasting...
I would appreceate a scene-workflow with filmic and the new color calibration modules. Issues are: magenta sunrise and banding of the overexposed sun.
The png-file(here now jpeg) is an output of Rawtherapee without banding or magenta for reference. Originally it was intensive orange, very “golden”.

Thanks in advance, Frank

These files are not licensed. Do with it whatever you think.
_IMG0037.DNG.pp3 (11.4 KB) _IMG0037.DNG.xmp (6.7 KB) _IMG0037.DNG (29.1 MB) ![IMG0037|690x460]

basic darktable scene referred workflow (ok - used the tone equalizer to boost the foreground)


xmp is included in jpeg

  • highlight reconstruction - reconstruct in color
  • exposure for general apperance
  • Whitebalance with colorcalibration (increased color temperature)
  • Filmic: automatic white relative exposure, reconstruction 7x

_IMG0037.DNG.xmp (9.7 KB)

Thank you. The error is located behind the camera. Blown reds result in magenta shifting.

brainblasting_IMG0037

brainblasting_IMG0037.DNG.xmp (12.7 KB)
dt 3.4

As you were asking about color calibration I’ll explain my process. Default setting had illuminant ‘as shot in camera’. CCT told me this was (daylight), so I adjusted illuminant to the same so I could access the temperature slider. Then I lowered temperature slider slightly to give more cyan/orange contrast. I handled the magenta highlights in filmic.

This is a fantastic shot. I couldn’t believe how little noise there was in those shadows!

Thank you. For my taste a bit greenish, but, it looks very naturally on the other hand.
Yes, I love my Pentax K3 APS-C. The values ISO400, 1/200 and +1.3 are a little bit weird - I didn’t pay attention enough. You look from German “Nordstrand” to “Pellworm”, if any interest.

Thanks for posting
DT 3.4.0

_IMG0037_01.DNG.xmp (8.6 KB)

1 Like

Tried to edit the image this way. The main problem are the blown out highlights. Filimic was not doing well for me, so I didn’t use it.


_IMG0037.DNG.xmp (7.8 KB)

Thanks for putting this one up.

_IMG0037.DNG.xmp (14.1 KB)

1 Like


brainblasting.pp3 (19.3 KB) RawTherapee 5.8-2757-ga6341bc5e

This being a RT edit I decided to wait a while before uploading mine. Then again, this is PlayRaw :smile:

After reading @age’s remark (see next reply) I checked my RT edit; Using RCD+VNG4 also shows weirdness:

I’m able to fix this by using AMaZE+Bilinear, many (all?) other options also show this (re-uploaded the corrected version)

I’m seeing some strange demosaicing artifacts with darktable 3.4.0
Could someone try this .xmp zooming at 200%?
_IMG0037.DNG.xmp (5.1 KB)

For comparison Photoflow

There’s extreme green noise in shadow too with other raw files

This RAW file is problematic but darktable has different and more artifacts using amaze compared with both rawtherapee and photoflow
(obviously with the same processing and color space)

I often have a slight bias towards green. Habits are hard to break. However, here is a way to enhance the cyan/orange contrast without the green. Use colour picker on clouds in colour calibration. Then in new instance of color calibration, set CAT to pass through, go to B channel, increase blue in green, and decrease blue in red.

brainblasting_IMG0037.DNG.xmp (13.1 KB)
dt 3.4

1 Like

The shot is good as it is. I just added a bit of bloom around the Sun and cropped it to 16:9.

1 Like

Hi there. I tried to produce a warm orange/red/golden feel. I used DT with filmic. I don’t think there are magenta artifacts.


_IMG0037.DNG.xmp (7.3 KB)

I opted for less contrast in the sky, and a bit of bloom in the highlights. Also a cooler white balance.
With darktable 3.4.


_IMG0037.DNG.xmp (44.5 KB)

1 Like

Nice interpretation. Looks almost a bit like winter. Interesting to see how many different variations we’re able to get out of the single raw file.

1 Like