First good sunset of 2022

First good sunset I’ve photographed in 2022.

Went for a tungsten film look by setting WB to 3200K in my experimental program. Also because it falls apart at 5000k because it’s garbage.

The sun area was also more yellow in real life (golden yellow, not rat piss)

DSC09950.ARW (46.8 MB)

Make it look as good as possible…

This file is licensed Creative Commons, By-Attribution, Share-Alike.

2 Likes

My version…

Darktable 3.9.0~git24.477bc5c41a-1
DSC09950.ARW.xmp (16.4 KB)

2 Likes

Nice sky. Thanks for posting.


DSC09950.ARW.xmp (9.9 KB)

1 Like


DSC09950_y.jpg.pfi (48.9 KB)

White balance is by eye - I aimed for a golden yellow around the sun.

Display rendering is via the OpenDRT tone and desaturation curves applied in OKLab colourspace, followed by gamut mapping into sRGB using Björn Ottosson’s OKLab gamut mapper.

3 Likes


first.2022.sunset.pp3 (21.7 KB) RawTherapee 5.8 Development

4 Likes

ART

3 Likes

It looks like rat piss is what the camera had captured

raw

rgb (here we could and should rotate yellow to orange with the lch equalizer tool)

oklab (highly saturated yellow hues are now white so not hue preserving at all)

DSC09950.pfi (70.7 KB)

2 Likes

darktable:


DSC09950.ARW.xmp (13.7 KB)

1 Like


RT dev. DSC09950.ARW.pp3 (21.8 KB)

I like the oklab version very much, it’s my favourite so far. What is the white balance in kelvin? I can’t figure it out from the .pfi, and couldn’t figure out how to load it in photoflow.

It looks like rat piss is what the camera had captured

It doesn’t do that. Ratpiss, demonstrated by lightroom and rawtherapee:

Lightroom is yellow all over… but notice: before it skews to rawpiss, the rawtherapee version is clearly orange!!! So I’m pretty certain the camera records this orange hue.

I have three suspects:

  • Camera sensors render sunset colours too orange-ly (if you preserve hue at rendering), want to see if spectral profiling might fix this.
  • Bezold Brucke shift - identical hues appear yellower at high brightness (SUNSET), and oranger at low brightness (display).
  • My code’s math is wrong.

I don’t know, i haven’t midified the white balance, It should be the camera shot wb

I would say this one

1 Like

With reddish-orange colors, as luminance increases, when you clip the red channel but allow the green to keep rising, it hue shifts towards yellow.

This is a very common way that sunset photos end up very far from what the eye sees.

Here’s two renditions with Filmulator; one with the ~3300K WB from the camera and one with daylight WB.

  • Highlight Recovery 2
  • Exposure Comp -5/6
  • Temperature: default OR 5011
  • Drama 100
  • Overdrive on
  • White Clipping Point: 0.389 OR 0.554
  • Shadow Brightness: 502 OR 604

2 Likes


DSC09950.jpg.out.arp (11.1 KB)

ART

1 Like

Hmm, my previous attempt was clearly a bit over the top, the hour is too late. Here’s one made in rawstudio instead, which renders the sun more yellow, using camera white balance, and adobe standard profile.

rawstudio settings|345x1000

1 Like

Great image @ilia3101!

The colour balance and temperature here seems pretty delicate. I didn’t quite nail it here, but still the result seems quite pleasant. Using experimental gamut mapping algorithm for darktable’s Filmic.

DSC09950.ARW.xmp (20.3 KB)

1 Like

I wonder what rat piss is in RGB… :wink:

I’d say rat-piss is when orange skews to yellow through per-channel clipping or RGB curves. Red is usually at 255. Often green is too.

See how the orange cloud just becomes yellow in lightroom:

The backlit clouds go yellow in rawtherapee as the brightness increases:

I don’t like it.

image
lol

DSC09950.ARW.xmp (10.2 KB)

DT 3.8

2 Likes