In two previous posts we looked at the basic building blocks of image processing techniques: the blurring and sharpening filters. In this final post of this small series I want to go through some of the more advanced filters that are commonly used in post-processing.
In some way or another, all filters are a combination of a low-pass (blurring) and a high-pass (sharpening) filter. The way these filters are combined is what gives specific final results and that is where the magic of image processing programs such as Photoshop or Lightroom lies. With the combination of a couple of commands, a great amount of manipulation and computation happens, making our lives as photographers much easier. And certainly Photoshop and Lightroom actions take this simplicity a step further.
when working with portrait photography, since it has the effect of giving an overall soft look at the image. Several proprietary plugin packages offer this under different names and it can as well be created in Photoshop by combining different effects.
The effect is basically achieved by combining an image with a blurred version of itself. The key step in the digital post-processing is the blending of the two images. The way to achieve the final effect is by using what is called the ‘Overlay’ blending mode.
A detailed explanation of the different blending modes is beyond the scope of this article, but in simple terms, the overlay mode is a selective mode that, when combining two images, will give a higher priority to the pixels with higher or lower brightness depending on how bright or dark they are in the original images. In other words, it increases the contrast of the bottom layer using pixel information from the top layer.
There are different approaches to implement the blending mode and computationally speaking it is just about what mathematical expression is used. Take, for instance, the following image of rooftops in Prague.
By simply calculating a blurred version of the image using a Gaussian filter and blending it with the original one with the condition that bright pixels will get brighter and dark pixels will get darker (increase in the contrast), the final image is obtained with the dreamy mood given by the Orton effect.
Notice how, apart from the dreamy mood, there is a clear enhancement in the contrast, with the darkest areas in front becoming much darker than in the original image. To solve this, a local enhancement in brightness would be necessary after applying the Orton effect, something that is indeed usual.
Another effect that is commonly used in order to give a moody feeling to images is vignetting. There are different types of vignette, some of them unwanted (like the ones that occur when using wide angle lenses with filters) but I want to talk about the artificial vignetting added during post-processing here.
If you think about a vignette from an image processing point of view, what you are doing is simply increase (or leave unaffected) the pixels that are on the middle of the vignette while gradually darkening those located on the edges of the vignette. Different parameters can be controlled here, like the contrast range (difference between the darkest and the brightest pixels in the vignette), size of the vignette and even its shape.
So the first thing that happens when you create a vignette is the creation of a grayscale layer that contains a shape (usually circular or oval) with decreasing brightness levels from the center towards the edges (it could be the opposite, with the brightest pixels on the edges and the darkest at the center as well, even though this is unusual). If you think of the brightness as a third dimension, the vignette mask will look similar to a Gaussian filter (or a Laplacian one if the vignette is inverted).
By applying the Gaussian-like mask to the Prague image after the Orton effect with a total radius of 350px (the original image is 1024 on the longest side), the following image results.
The effect of the vignetting can be controlled by changing the radius or the difference between the brightest and darkest pixels. In Photoshop, if a gradient mask is used to create the vignette, this would translate into changing the shape of the mask as well as the opacity of the layer after applying it.
I hope you enjoyed this entry and if you have any question, just write me an email.