What can you get out of this raw in darktable? xmps welcome.

Hi, I have this very dark image and processed it in darktable. I am sure I did not play well on this and wonder what is possible with this data.

(The license for this file is Creative Commons license.)

https://drive.google.com/open?id=14iyO24g8fpk_kJ8QRGc7lH9P9FJPdeyt

DSC_1510.NEF.xmp (1.4 KB)

1 Like

Here’s my take with DT 2.7.0-git896.bd1024330 (dev version)


DSC_1510.tif.xmp (9.4 KB)

EDIT: … and welcome to the forum!

Hi, welcome to this forum :+1:

Here’s my attempt on this photo:


DSC_1510.NEF.xmp (13.3 KB)

Some hints when shooting photos at night:

  • higher ISO decreases the dynamic range of the photo, with a lower ISO setting that photo could be shot with maybe 1/100 seconds exposure time. That longer exposure time makes the background a bit brighter, too.
  • If there are blown out highlights at night you can use a lower aperture to reduce that effect.

@gadolf: fantastic result. Interesting gradient on profile denoise. Is there a reason to exclude the right half? The white balance in the xmp is wong, i get a greenish result using your xmp directly. After resetting I get exactly what you have posted here.
Why did you rename the NEF to a TIF? Any advantages?
And thanks for the warm welcome :slight_smile:

Thanks. That was only for aesthetic reasons: I think the noise at the bottom part of the image adds to the ashes falling down the balloon.

What is your DT version?

Thanks @pphoto,

thank you for the tips. When I came to a temple the monks started the balloon in the very moment and I had to run to be able to still see the baloon on my lens. No time for settings. I now see that I have 1/1000. Sure that is not well set up for that situation!

The edit is the kind of color spectrum I wanted to create. The flames, smoke, and particles are visible, the dropping burning wax is not lost and you have made a very nice composition, reducing the visible area to the interesting part. Well done!

You made a lot of edits to get to the result and I have checked them all (top to bottom) to understand how you were able to create such a good result.

  • Never saw the bilateral filter in shadows and highlights. Makes a big difference!
  • Nice subtle but important effect of using shadow color adjustment
  • What is the effect of the negative contrast in the lowpass as a overlay? Increasing the low light regions without attenuating the bright areas?
  • in profiled denoise in your modules I see “migrate to the fixed algorithm” but if I add nother instance of the same module I dont have that checkbox. How come? Maybe it is because I use the latest dev version in which the one old erroneous anscombe transform algorithms can be fixed.
  • looks the color zone saturation settings are the ones that make the trick for why I like the final colors so much. I dont see the hue effect at all.
  • looks like you use the tonecurve to reduce the extremely bright shine. Awesome.
  • another suble but great change by reducing the temperature slightly.

I often drag and drop the histogram s.t. it is not clipped. This is not done here but still the edit is very nice. I need to understand how to utilize the histogram and exposure module better.

There is definitively much to learn from this edit. Thanks!

This is 2.7.0+910, the latest development version.

Not sure if I have the latest latest because it changes on a daily base. Last time I updated, if I’m not wrong, was two days ago.

Is it Windows? I’m on Linux, not sure if it is in sync with the latest Windows version…

i guess the white balance was just turned off unintentionally before uploading your xmp. Im running linux as well. As I said: when I reset the white balance I get exactly the same result like the jpg you have uploaded.


DSC_1510.NEF.xmp
(7.9 KB)

2 Likes

Forgot to mention, I used Rawtherapee to recover the blown highlights and saved a 32-bit tif. In that regard, I get poor results in DT (probably I’m doing it wrongly)

interestingly artistic. How did you do the linear tone curve? When I drag I always get spline like behavior.

You only have to grab the outermost points (lower left or upper right).

1 Like

thank you. Nice!

Never saw the bilateral filter in shadows and highlights. Makes a big difference!

When you use the gaussian blur you get halo effects with stronger values, with bilateral you got a higher range before halo occurs. In smaller ranges imho it is a personal taste if you use gaussian or bilateral

What is the effect of the negative contrast in the lowpass as a overlay? Increasing the low light regions without attenuating the bright areas?

It’s a part of the ‘subtle but important effect of using shadow color adjustment’. It’s explained in-depth in this video:

in profiled denoise in your modules I see “migrate to the fixed algorithm” but if I add nother instance of the same module I dont have that checkbox.

I don’t see this text, maybe it’s your development version. I use Darktable 2.6.1

looks the color zone saturation settings are the ones that make the trick for why I like the final colors so much. I dont see the hue effect at all.

The red tones in the balloon’s upper part had too much purple in my taste. Therefore I decreased purple saturation and moved the purple to red/orange in the ‘hue’ panel.
Then I realized that the area around the fire has become too greenish so I moved the green a bit towards orange.

looks like you use the tonecurve to reduce the extremely bright shine.

Actually I was trying to get rid of the halo effect above the fire. When you zoom in to the fire and deactivate the tone curve the dark area gets bigger.
Think it came in when I first used the lowpass filter. Didn’t put much time to eliminate that effect completely.

And thanks for your feedback :+1:

Thank you for the link. Impressive what he is achieving. I must say I have already heard and forgotten how to use that trick before. I will again. So many ways and there is no straight forward way. I tried to reproduce your result on my own and can hardly get such a good result like yours. Luckily the xmp gives a clear road and I can go step by step and understand what I did differently and worse.
Thanks again!

1 Like

Hey y’all,

here’s another one with dt-2.7, hope you like it…


DSC_1510.NEF.xmp (11,4 KB)

1 Like

DSC_1510.NEF.xmp (7.8 KB)

1 Like