Captured Moments » Eventually interesting stuff about Linux, Programming, Software, Photography.


  Photography     F-Spot     Usability     Panorama     Photoblog     Gimp     GPS     UFRaw     Bibble     Windows     Sony     Nikon     Web     Work     Software         Plugin     LMMag     Hugin     Live     Linux     Programming     Weather  

My Flickr Photos

Books on

aus der ev. Kirche Sulzbach (Ts.)
Julia und Paul
Wellner Bou

Admin area


Especially if you want to convert photographs to black and white and create a nice dark sky, it may happen that you will have some noise in the sky, even if you shoot with low ISO. This happens if you use a channel mixer or a similar tool for black and white conversion. As the sky is (often) blue, it will appear darker in the black and white version if you reduce the blue channel. But in general you will have more noise in the red channel than in the blue and the green channel.

Of course this depends on the sensor type and the camera hardware and software you have. This per channel noise phenomena is very deeply documented on, I recommend to read it, if you are interested in the technical stuff of (your) digital camera. There is another article there testing the results regarding channel noise with a magenta filter.

By the way, this same method can be used for selective sharpening. Rolf from treats this in his podcast and there are a lot of tutorials treating this, too.

Creating the layer mask

  1. Duplicate the layer.
  2. Detect edges, for example with [Filters] -> [Edge Detect…] -> [Edge…]. Default Settings are (mostly) ok.
  3. Apply a gaussian blur filter to this layer. The radius depends on the image size, but it shouldn’t be too small nor too big You may have to try several times if the first try is not giving you the result you want to have.
  4. Open the “Brightness/Contrast” dialog and move the contrast slider to the very right. Then, play with the brightness slider increasing it to get the areas you want: In the black area(s) we can apply a much stronger noise reduction or blurring later, in the white areas we can apply more sharpening. It may happen, depending on the amount of noise and the photograph you have, that you will have to paint parts of the sky with a black brush afterward.
01_DetectEdges.jpg02_BlurredEdges.jpg03_AppliedBrightnessAndContrast.jpgCreating the layer mask

Blurring and applying the layer mask

  1. Duplicate the original layer again.
  2. Apply the kind of noise reduction and/or blurring you want to use. This may be Wavelet denoise or just a selective gaussian blur[1] blur filter. Don’t care about the areas which are marked white in the area mask we created before.
  3. If you are done denoising and/or blurring, create a layer mask for this blurry layer (right click: “Add Layer Mask…”) and copy the layer mask layer into it.
  4. Invert the layer mask.
01_CropOriginalSizeBefore.jpg02_CropOriginalSizeAfter.jpgSky comparison, before and after, 100% Crop of 12 Megapixel JPG.

Maybe you will need some more little adjustments in the layer mask. You could smooth the edges again blurring them, increase or decrease them or paint on them with white or black brushes to include or exclude areas which should be blurred or not. Additionally you can now sharpen the original layer without sharpening the sky.

This photo on Flickr.

[1] The selective gaussian blur filter needs much more time to run, but it is in this case better than the gaussian blur because it will remain — at least partly — the edges of the original image. This will it make easier to overlap the blurred layer with the mask with the original layer.