Processing image in Darktable
16 April 2018
Table of contents:
- Let's begin
- Lens correction and cropping
- Exposure and white balance
- Graduated density module
- Local contrast, velvia, and highpass
- Local levels adjustment
- End result
This is an example of raw image processing in Darktable. The aim is to demonstrate some of the functionality and how to apply it to a landscape image. In particular it shows use of parametric masks and drawn masks.
If you would like to try this yourself, you are welcome to download my original raw file. It was shot with Nikon D7100 on a tramping trip in New Zealand mountains. I have also included XMP sidecar file with all edits you will see below. Once you import this into Darktable it will create a second copy so that you can compare your edits with mine. It was originally created using duplicate option.
This is a jpeg file produced by my camera with very conservative settings (click on the preview for full resolution image):
When we load the raw file into Darktable, the default module settings will be applied. Unless you changed them, they would look like in the picture below. You will see them in top left corner of darkroom screen (the rest of this article works in darkroom mode):
The image was shot exposing for highlights, also known as exposing to the right. But we want to check overblown highlights anyway. We do that with raw over-exposure indicator (located in bottom right part of darkroom screen):
In this case there are none, which is good. So let's start the processing:
Lens correction and cropping
I like to apply lens correction to all my photos. Not only it helps with geometry of the image, but more importantly it can reduce fringing due to chromatic aberration. If your lens is in the LensFun database, switching on lens correction with default values should be enough in most cases. If your lens is not recognized, you can help to get it calibrated and included. In this case I just switched the module on with default values. You will find this module in correction group.
Next I cropped the image using module crop and rotate from basic group. I like format 16x9 for various reasons, so this is what I did here and you can see my result so far:
Exposure and white balance
Now that we have our composition set, let's look at exposure. I will turn on over and under exposure indicator. You will find it in bottom right part of darkroom screen:
You can see some over exposed pixels highlighted in red and underexposed in blue:
As you can see, the image is pretty dull and uninspiring. The foreground is rather dark so I will adjust the image exposure to bring it into more natural mid-tones. In this particular case I also increased black level just slightly:
Pushing up overall exposure will cause a new problem – large over exposed areas. To keep this under control, I am switching on another basic group module, shadows and highlights. This time with default parameters.
All these exposure adjustments brought the image foreground roughly where I want to have it. Now I can have a good look at the colour. In general I prefer to pull back green tint a little. In this case I also pulled back colour temperature, going for slightly cooler look. Both of these adjustment can be found in module white balance in basic group. It's just adjusting to taste. But you could get more sophisticated with shooting a grey card in the same light conditions. That would help you to get the "correct" white balance – which you can adjust to taste, based on your artistic vision.
Here is our progress so far:
Graduated density module
We have increased overall exposure in the image. Even though we then darkened highlights with highlights and shadows module, we still have an over-exposure in the sky and on the snow:
This is not always a problem, it really depends on your artistic intention. However in this tutorial I want to show how to bring the tonal range of the sky and snow back using graduated density module (located in effects group). I will start by placing the "horizon" line of the graduated density filter on a slight angle to darken top right part of the image more.
I will also demonstrate another trick: graduated density would normally darken all pixels in the effected area. If we apply stronger effect (higher density value), it can make it sometimes too obvious and produces unnatural look. To mitigate this, I turned on a parametric mask and set it so that pixels with small grey values (channel "g", in other words pixels already dark) are effected only partially. Here are the properties I used – notice the small filled triangle on input slider moved a bit to the right:
To get a better understanding how the parametric mask impacts the image, you can use the two toggles in the bottom right corner of the properties – to the right of the mask blur slider. Here is what the display mask toggle produces – solid yellow colour means fully effected area, and less yellow means the effect will be only partial:
The resulting image with graduated density filter applied:
Note: One potential issue to look for if you use this trick with parametric mask, is banding of tones or colours. The more you push the effects, the more likely it could appear. When it happens I just play with the sliders until I achieve smooth results.
Local contrast, velvia, and highpass
The image is still a bit dull. One way to make it pop a bit more is to increase local contrast. This can be done with a module of the same name in tone group. I turned it on with default parameters to begin with, but it impacted over-exposed and under-exposed areas more than I liked. To reduce the impact on very dark and very bright pixels I used again parametric mask, this time on "L" channel (see the little triangles on input slider):
There are three ways of increasing colour saturation. Plain saturation which increases saturation of all pixels (module contrast brightness saturation), vibrance, and velvia which increases saturation of less saturated pixels more than those that are already saturated. I went for the last named – you will find it in color group. I have increased its strength to 35% in this case. You can read a comparison of these three methods here.
Last effect to make the image pop that I used here is another popular image sharpening technique. Select highpass module from effects group. Default values for sharpness and contrast boost are fine here, but I selected blend uniformly and blend mode softlight to make the effect work the way I want (using overlay would produce similar but slightly stronger results). I also adjusted opacity down to 40% to make it softer again:
You can check the result of this step below:
Local levels adjustment
In my mind the focal point of the image is the large clump of tussock. So I would like to help guide viewer's eye to it. One way to achieve it is to brighten the area with a local adjustment. First I selected module levels from tone group. I selected blend drawn mask and created a path to roughly cover the tussock and set a long falloff transition (represented by the dashed line; further away from the solid line – the softer the transition is). This is important to make the adjustment imperceptible. Finally I pulled the white point to the left by about 10% (white triangle on the histogram):
Here is a preview of the image showing the path:
Along the way we activated a number of modules. Here is the full history:
This obviously doesn't cover all possible topics. You can see for example a bit about noise reduction in another article.
And here is our final result as exported from Darktable (click on the preview to see full resolution version):
Please let me know if you notice any issues in this article.
Processing image in Darktable by Tomas Sobek is licensed under a Creative Commons Attribution 4.0 International License.