Sunset - inversion artefacts on the Sun

Dear All,

No matter what I do (in darktable), I cannot get rid of inversion artefacts in the following shot:

I’ve tried a couple of highlight reconstruction settings, filmic RGB recovery, even low pass, soften and bloom filters, I cannot get rid of the darker ring surrounding the centre of the Sun’s disc:

I’ve just realised the uploaded JPG + XMP have weird filmic settings, but my best attempts were not much better, either. I can, of course, blow out the Sun completely, but it’d be nice to preserve a hint of colour – so close to the horizon one can actually look at it.

2020-07-31_20-05-18_P1040218.RW2 (10.5 MB)

Suggestions are welcome (if possible, with explanation of why a specific method was chosen). Thanks in advance,

This file is licensed Creative Commons, By-Attribution, Share-Alike.
2020-07-31_20-05-18_P1040218_03.RW2.xmp (7.2 KB)

1 Like

Looks like a good image to play with. Not yet for me; I have stuff to do; I really should log off. :blush: A quick suggestion is to look at PlayRaws (and a few other threads) with sun and people’s sidecars or remarks there. This issue is actually pretty common and fairly easy to handle.

Is this okay?

First, turn off highlight reconstruction module.

Filmic should do the reconstruction.

In the photo you can see the filmic settings for reconstruction. What individual sliders do you can read in the respective context menu.

I also used the soften module with parametric mask to influence the highlights the most:


2020-07-31_20-05-18_P1040218.RW2.xmp (11.6 KB)

I’ve turned off highlights reconstruction and turned on color reconstruction, then a slightly adjustment with the hue equalizer


Sorry , hacking the thread

1 Like

oops, whats going on here?

1 Like

I think it could be due to the high noise levels in the image. @gaaned92 , what do you think?

Bad artifact. I have to identify the offending modules configuration and report to @agriggio

texture boost on negative values.
You can either use a mask to exclude them:
Screenshot from 2021-06-10 15.20.33

or a slight amount of negative black point compensation:
Screenshot from 2021-06-10 15.20.47


1 Like

this helps, thanks :+1:

my contribution

2020-07-31_20-05-18_P1040218.RW2.xmp (11.5 KB)



Thanks @agriggio .
Yes there are negative values in the raw data, I suppose due to noise.
I propose we continue in