Goodbye noisy skies with "distorted pixels"!
I guess I am not alone tackling the issue of denoising the skies in outdoor shots,
Here are a couple of examples, where my first attempts to "improve" the skies led to refusals initially. That time I knew little about the origin of "distorted pixels" and possible noise at nominal IS0. Simply by the method of a "reasonable guess", I decided to redo the editing with less contrast and saturation boost.
Later, however, I have discovered for myself a pretty simple method of noise reduction especially suitable for the skies.
So, I introduce another layer over the original picture, then in this new transparent layer create a gradient over the sky area with the two utmost colors picked up from where I like in the original photo (normally the darkest and lightest spots in the sky). Finally, I vary the transparency of the top layer to my liking, while checking for the noise and "distorted pixel" artifacts at 100%. The two edge colors for the gradient can be made slightly darker/brighter, so as to make the sky look more "dramatic". Of course, before merging the layers into the final picture, one has to erase from the created gradient layer those parts where it overlays with the non-sky part of the photo (for ex. by using some smart selection tool or simply brushing out in the mask mode and then cutting out; alternatively, one could of course pre-select the sky area beforehand and apply the gradient only to the selection).
An important tip: be careful about not making the two ends of your gradient too different from each other. Because any gradient is a set of stepwise levels, so if you are unlucky you may end-up with annoying clearly seen color banding in the skies, instead of the desired smoothness.
Well, it might be an obvious method for some, but also might be helpful for others.
PS: I did not mention above, but first of all, one should probably consider using neutral gradient or polarizing filters when shooting in daylight, just to minimize the editing needed...
Photo credits: Androniques.