Highlights in the fog

This is one of my favourite images in 2024, I like the framing and the moment. However it has not been processed quite well. There are issues in the highlights. I had very interesting comments and analysis by @kofa who suggested to me that a play raw might be interesting for this community and I would like to see how you deal with it. There you are!
DSC_0359.NEF.xmp (9.9 KB)


DSC_0359.NEF (13.2 MB)

Oooops! As I upload it here I notice that I did take this picture in 2023-12-27 :sweat_smile:

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

11 Likes

I liked this photo when I first saw it in the best of 2024 thread. Glad you offered it up in Play Raw. DT 5.0

DSC_0359.NEF.xmp (14.5 KB)

4 Likes

Extremely minor edits to fix the lights follow below:

Sometimes, when you have bright lights that aren’t necessarily the same type of light as the surrounding area, they show up both blown out and saturated, and you get an effect like you see in your photo when using Filmic RGB. There are a few ways to fix this.

Option 1: If you don’t want to change anything else, a good and simple method is to go to Filmic’s “Reconstruct” tab and turn on “Enable highlight reconstruction”. That’ll usually do what you want right from the start, although you might need to adjust the threshold (depending on the image, although the mask mode (the square icon with a circle inside) will help you select mainly the light source, if just enabling it doesn’t immediately fix it).


DSC_0359.NEF.xmp (11.4 KB)

Other methods with Filmic RGB, although these will affect the image:

Option 2: Adjust the “white relative exposure” until the area looks correct. This will make your image lighter. But that might be what you want for some photos anyway.


DSC_0359_02.NEF.xmp (11.6 KB)

Option 3: Or adjust “Extreme luminance saturation” in the “Look” tab. But this will desaturate your photo. (Which might be what you want for some images anyway, or you might want to start with this and compensate with a boost of vibrance and/or chroma and/or saturation in “Color balance RGB”.


DSC_0359_03.NEF.xmp (11.6 KB)

Option 4: Or turn down the chroma or saturation on the highlights, using a module that desaturates. Here’s “Color balance RGB” turning down highlights chroma with a slightly feathered mask to constrain it to the highlights.


DSC_0359_04.NEF.xmp (11.5 KB)

Option 5: Or even turn off Filmic RGB altogether and use Sigmoid. Some people prefer Sigmoid, but I’ve personally gotten used to Filmic RGB and use it almost exclusively. But it’s still worth pointing out using Sigmoid will negate this issue by default. I’ve seen people getting great results from both modules in playraws.


DSC_0359_05.NEF.xmp (11.5 KB)

Option 6: (Back to using Filmic RGB — I thought of this mehod later. Whoops!) Switch the mode in the “highlight reconstruction” module (not to be confused with Filmic RGB’s “highlight reconstruction”). In this case, I switched it from “inpaint opposed” to “reconstruct in LCH”. There are options and other modes, and different methods and values may work better in other photos.


DSC_0359_06.NEF.xmp (11.5 KB)


darktable can do similar things in so many ways. (It’s both a strength and weakness.) Most people probably wouldn’t notice any difference between the above photos. Some of us here will see a difference when we look at them side by side. It’s quite subtle. I’m pretty sure someone in this thread will even recommend other options as well. :wink: There are probably dozens more ways to fix this, especially considering modules and masks.

Anyway, I’d recommend turning on the highlight “reconstruction” in filmic RGB “reconstruct tab” versus the other options, as it changes your image the least and requires the least amount of work… although the other options are all basically valid methods too.

10 Likes

DSC_0359.NEF.xmp (12.6 KB)

7 Likes

Thanks for the play!
I just took your XMP, switched filmic to v7, enabled reconstruction, picked white level from the reflectors, and moved highlights saturation to max to preserve the colour in the lights.


DSC_0359.NEF.xmp (11.4 KB)

Or even simpler: your XMP, filmic white picked from the lights, enabled reconstruction:


DSC_0359_01.NEF.xmp (11.5 KB)

8 Likes

My general approach to recovering highlights is to first set the exposure module to optimise the highlights and then using various modules to recover the shadows after that. However, the highlights were clipped here so recovery was not possible and being actual lights probably not necessary anyway.
DSC_0359.NEF.xmp (11.7 KB)

4 Likes

… Grayscale, USM, then a bump map.

7 Likes

That is a wonderful explanation and study, anyone should read it!! Many thanks

1 Like

Thank you!

1 Like


DSC_0359.NEF.xmp (34.7 KB)

6 Likes

DSC_0359.NEF.xmp (20,5 KB)

1 Like

With ART 1.24.5


DSC_0359.NEF.arp (14.9 KB)
Greetings. Roberto

3 Likes

uuuuhhhh, thats really foggy. This is how it would look on my old Ilford photo paper :wink:


(RT_dev)
DSC_0359.NEF.jpg.out.pp3 (15,6 KB)

A further try:


DSC_0359_01.NEF.xmp (18,2 KB)

2 Likes

This has the same problem as the original: there is a brighter ring around the reflector’s main area. L channel in Gimp with a threshold:

You are right. Are you able to avoid this with my edit? I tried it as well with filimic but had no success.

Edit Now I took guided laplacian mode on highlight reconstruction just like on the other edit from me. Better? (I think it’s different but still there)


DSC_0359_03.NEF.xmp (21,1 KB)

3 Likes


DSC_0359.NEF.xmp (13.5 KB)

5 Likes

My version…

DSC_0359.NEF.xmp (20,2 KB)

1 Like

I think the cause is that the blown area in the reflector’s light is very far from neutral (because of the white balance multipliers that take the maxed-out values from the raw to that colour):

That then messes with the rest of the process.

2 Likes

I played with colourcalibration and WB but I couldn’t get completely rid of the problem. Anyway that brought me to see some quite extreme colours, which resulted in this edit:


DSC_0359_03.NEF.xmp (20,6 KB)

4 Likes