Nice waterfall, shame about the vegetation blown highlights

Taking a bit of time whilst on leave to go through some of the trickier captures in the last couple of months. For a shot without tripod, just leaning on a fence, I’m very happy with the waterfall. I’ve had a lot of trouble with the vegetation highlights, however. And I’m not sure what’s the best crop. Something to play with, if it piques your interest.

2023-05-01_10-30-46.75_DSC3410.nef (24.3 MB)
This file is licensed Creative Commons — Attribution-NonCommercial-ShareAlike 4.0 International — CC BY-NC-SA 4.0

2 Likes

Morning, Mate!

Here is a very swift first attempt — I ought to be able
to make it a trifle better but have to rush off…

Have fun!
Claes in Lund, Sweden

3 Likes

Not entirely happy with this but…


2023-05-01_10-30-46.75_DSC3410.nef.xmp (13.7 KB)

5 Likes

My version…

2023-05-01_10-30-46.75_DSC3410.nef.xmp (19.8 KB)

10 Likes

@dqpcoxeas You restored the leaves in lower right corner superbly! I brilliantly fail on them :sweat_smile:

4 Likes

Er-hm… Would this be considered cheating???


2023-05-01_10-30-46.75_DSC3410.nef.xmp (13.8 KB)

Have fun!
Claes in Lund, Sweden

1 Like


2023-05-01_10-30-46.75_DSC3410.nef.xmp (19.3 KB)

3 Likes

Another less brilliant version…
Thanks for your comment, I love challenges…

2023-05-01_10-30-46.75_DSC3410_01.nef.xmp (22.5 KB)

10 Likes

In a word, yes: like solving the gordian knot by slicing it with a sword

1 Like

RT user here. I can’t for the life of me get anywhere close to what @123sg and @dqpcoxeas managed to do with the blown-out leaves, even when focussing exclusively on that area of the image.

Are there any RT users who know how something similar could be done in RT?

Disclaimer - I don’t consider myself an “RT user” as I don’t use it regularly… but I thought I’d have a go!
Different to my DT version… but here it is in case it helps a little :laughing:

I set highlight reconstruction to colour propagation, and used both shadows and highlights, and dynamic range compression. Also desaturated the highlights a little in the Lab Adjustments.


2023-05-01_10-30-46.75_DSC3410.nef.pp3 (14.0 KB)
But I do feel I’m randomly rummaging around… hopefully others will have more helpful suggestions.

P.S. I’m running RT 5.9 release candidate 1 - I never installed the release!
Edit: sort of OT but I’m actually quite impressed with the tools in RT for dynamic range management. I think I should use it more…

Edit number 2… second try…


2023-05-01_10-30-46.75_DSC3410.nef.pp3 (14.4 KB)

3 Likes

I’m on my phone right now, will post my edit tonight. Thank you for the effort! I’m a little bit relieved though tbh since your results in RT, while better than what I got so far, are still quite similar-looking.

Maybe I have to finally give DT a serious try. OTOH, so far I also haven’t been able to reproduce any RT edits I’m happy with in DT.

2 Likes

I’ve always been of the opinion that RT and DT are equally ‘good’, just different approaches. I tried both a couple of years ago and found I got on well with DT so I mostly stuck with that, without any perfect reason really.
Try all the options, if you have the time! :wink:

2 Likes

In your RT edits, which are better than mine, I see the desaturated blotches that I hate in my edits and which do not appear in your and even less in @dqpcoxeas 's DT edits.
image

Maybe someone can fix this in RT?

2 Likes

Recovery in darktable (dropped the exposure, no filmic or other tone mapping):

The processed image:


2023-05-01_10-30-46.75_DSC3410.nef.xmp (10.7 KB)

I went for a minimal stack, so no detailed (and excellent!) adjustments like those from @dqpcoxeas.

6 Likes

@Sunhillow I get the same (and even worse-looking) blotches on those leaves.

@kofa Oh what! The leaves look close to perfectly recovered in your edit!

@123sg Here’s my RT attempt (full image and cropped), but the leaves look terrible imho:


2023-05-01_10-30-46.75_DSC3410.Ogven.jpg.out.pp3 (15.3 KB)


2023-05-01_10-30-46.75_DSC3410.Ogven.square.jpg.out.pp3 (15.3 KB)

2 Likes

Wow … I am grateful for so many responses! @kofa, thank you for your share especially. However I found myself unable to reproduce your results. I could get something close in the vegetation if I apply exposure unmasked, but then I lost all the rock detail around the waterfall. It was not until I picked through your stack that I picked that you achieved magic with the tone equalizier. I have never used such large adjustments there, so I am now emboldened. Thank you!!! :smile:

4 Likes

Thanks @kofa i think you got what segmentation mode can do :slight_smile:

For the rest of the crew here, this is not magic or just good luck. Let me try to explain:

  1. The visualize button besides the clipping threshold slider shows, we have almost all green channels with full data = blown out. So where to get those from?
  2. In opposed mode we use some constant correction data from the whole image. This doesn’t work really good here as the rest of the image is very dark (there are more points in this but this is the most significant one)
  3. In segmentation mode we will try get those correction data from the green data very close to those blown out. So we need segmentation. And because the greens in the dark are pretty dark we have to push the candidating slider pretty much.
  4. An additional step could be tried here as there are parts blown out in all color channels (the pale blotches reported). Try rebuild small parts …
3 Likes

I recently watched Boris’s latest video on the Tone Equalizer module. It had never dawned on me to use multiple instances of TE on the same image, with the mask adjusted to work on different parts of the image. I ran a few tests on old Play Raw images, and the improvement can be amazing. I rarely use this word, but Brilliant!

3 Likes

Actually, there was quite a bit of trial and error. Having learnt some image processing at university, segmentation and morphological operations are not alien to me, but I don’t claim to understand all that the module does.
For example, the combine mask does not change much between 4 and 6:


But the output changes considerably (as we move from 5 to 6). I assume the colour chosen as the representative one for the segment changes. At combine = 4 or 5, we have green; at 6, a sickly colour:


And even though the leaf on the left is marked with the same colour in the mask as most of the clipped greenery, it gets a purple tint. Or are the colours not important there (the only meaning being that a contiguous area is one segment, but another contiguous area with the same colour, in a different part of the image is another)?
image
image

Turning on any of the rebuild modes then makes the situation worse, I think, since much of the blown-out area gets desaturated, and some isolated green areas show as artefacts:

What is the meaning of the candidating mask? It’s supposed to indicate the segments that have good candidates.
According to this mask, if it shows the candidate colour, the leaf on top of the shoot (?) should have the same green candidate as the leaves to its right (yellow arrows), but then the ‘shoot’ should have a transition between the green and the cyan (red arrows). Even if the colouring is not according to the actual candidate colours, I don’t see a transition on the shoot:


Thanks for any explanation!

3 Likes