Reverse engineer an effect


(Alex Mozheiko) #1

Hello Everyone,
I’m being fascinated again and again by how well some phototographers handle esposure. However, even though, I sometimes can find some barely noticeable notions how they do it, at best I stumble upon either nonFOSS-oriented live session (with Photoshop and Lightroom) provided by these photographers or lack of any shared knowledge at all.
I don’t want to copy anyones style, I want to reverse engineer a specific effect from their workflow with Darktable/GIMP. I assume this approach is ethical enough, but feel free to share your considerations.

So here’s an example of a photo by Russian photographer Aleksander Ra
We can see the sun is down low and the buildings should hve gone in deep shade, but, they are . . . well exposed on both black and white end.
Next is the sky - if pointed at buildings, exposure meter would kill the bright sky. But it"s there, also very well exposed. If exposing to the sky, then the buildings must be dark, but they are not.
Then vehicles - they don’t overexpose the ground with traffic lights, the street is well exposed.
Overall lighting is well balanced on every end - no dark corners, no dark shadows, no flat or sharp midtones, no exhaggerated street lights, and all the details are distinguishable and extremely sharp.
Every detail of the scene has this perfect, unnaturally well balnced exposure effect.

And Aleksander is not the one using this “unnaturally good exposure” effect. Check out Daniel Kordan, Michalel Shainblum and others.

How do these guys do it with Photoshop?
I’m out of ideas. What would be an approach with Darktable/GIMP/other FOSS?

(Alex Mozheiko) #2

Another effect example

(Shreedhar Inamdar) #3

To my eyes the first photo looks like a composite of two (or more) shots and the second one is exposed for the sky with strong fill light for the three foreground trees, say the headlights of a parked car.

(pphoto) #4

It’s hard to determine which part is coming from the camera and which part has been made during post processing.

In the light trails of the first picture and the smooth water in the second I see that the photographer used a long time exposure. This increases color richness and details in darker areas. To protect highlights from overexposure you stop down the aperture.

Maybe one or more filters could have been used, polarizer or a neutral-density filter.

Moreover, the photographer could have taken multiple shots for a HDR image.


There’s a ton of lightroom tutorials showing the classic “pull the shadows up a lot and pull the highlights down a lot” trick. This works wonder as these two slider do their job quite well (That is why they have been requested quite often in FOSS circles).
An example of someone who does this trick all the time is Serge Ramelli (you don’t have to like his slightly over the top processings).
Here’s an example video where the trick is shown (around 2:50):

The next step, if the dynamic range of the camera is not enough is to composite multiple exposures.

(Alex Mozheiko) #6

But that will bring a lot of halo :wink: which is nowhere to be seen at the examples.
That is why I believe these effects are not result of a single module action but some sort of sequence of tone balancing actions.

(Glenn Butcher) #7

Yes, the first has to be a composite; I don’t know how one would keep the shutter open long enough to get the automobile light trails yet have a prayer of definition in the sky, with any camera…

Dynamic range in one exposure is very camera-dependent, in my perspective. That’s the reason I upgraded cameras last time, not for IQ but for DR, and I’m about to do the same in considering the Nikon Z6. With that, it’s more feasible to expose for the highlights and then in PP pull the shadows into view without too much noise to correct. My old trusty D50 just fell apart in that regard, and the D7000 made it somewhat feasible.

(Morgan Hardwood) #8

This looks like the same technique used by all other master landscape photographers - manual exposure blending using luminosity masks.


You are right, though LR does a very good job to hide most halos.
But, I agree with the others, that the first image is probably a combination of multiple exposures done manually.
I do this in Gimp usually using luminosity masks, with imho very good results.


Looking at the street image, it looks like it received a lot of photo-shopping:

  • image stacking (and tone mapping)
  • sky replacement (quite popular; expose to earth, add dramatic sky)
  • dodging and burning (less)
  • digital painting (more)

Personally, I tend to do none of the above.

There are many strategies for avoiding and / or preventing halos.

(Boris Hajdukovic) #11

What others have mentioned above, there is a lot of work with luminosity masks and dodging and burning.

It reminds me a little of my own work.

I intentionally took this picture very underexposed to avoid overexposing of very bright gap between darker clouds and a block of flats. The girl with the bicycle, trees and grass was accordingly very dark. With the use of luminosity masks and lot of dodging and burning I managed to direct the viewer’s attention to a diffusely illuminated girl with a bicycle:

Fahradmädchen by Boris Hajdukovic, auf Flickr

Similar to this picture; I used the same methods to brighten up the front of the houses and the street:

Franz Mehlhose Café by Boris Hajdukovic, auf Flickr

To learn how to use luminosity masks in GIMP I highly recommend the tutorial by Pat David:
Getting Around in GIMP - Luminosity Masks

[PlayRaw] Tiger and Turtle Bracketing
(Todd Prior) #12

Lots of great info about setting exposure…