by Mirek Pruchnik, December 2007
The history behind this mini-guide is quite simple: DaveK posted a photo of his friend Frank and asked for critique. Because I liked the photo I thought that I'll give it a go through PSE to improve colour of the skin and also sharpness of eyes. Because my result was not that bad ("slightly better" - as Cekari kindly commented ;)) I thougt that it could be a good material for a mini-guide, especially that it could show some interesting techniques/tricks which may be useful for those less familiar with PhotoShop. Therefore I asked Dave for permission to use his photo and because he agreed - here it goes ...
The first step is to open the photo in PSE (or PS if you decided to spend more on software than on lenses ;)), open the Layers palette and prepare duplicate of the background layer (right-click on the layer and select "Duplicate" from the pop-up menu). It is worth assigning the duplicate layer some meaningful name at the time when it is created. Please note that all graphic or filter operations are applied to the selected (active) layer.
As you can see I created two duplicates: one called "Work copy" and one called "Mono". As their names suggest, the upper one will be converted to a monochromatic version (B&W with some warm, dark sepia toning) while the lower one will be used as our background. It is advisable to keep the original photo intact at the bottom of the layers stack - this way it is always possible to go back to it and use it as a basis for some other purposes.
To convert the upper layer to B&W I used OptikVerve Labs plugin - the best freeware PS plugin around. Here I started with the "Character" preset but then tweaked it a lot as the output was way too dark and contrasty. Here I decided to keep dark sepia toning, but switching to pure B&W may work better. Even just an ordinary desaturation of the layer may give interesting results.
After applying the plugin the B&W photo will be visible as the "Mono" layer is on top. Now you can experiment a bit with various blending modes of this layer (but as I checked this did not work well in this particuar case) and/or its opacity (which is what worked well).
So, let's leave the blending mode as "Normal" and set opacity to 35%. This gives more or less the effect I wanted - better looking skin tone. At this stage the processing could have been finished but since I wanted to improve sharpness of the model's eyes I decided to use high-pass selective sharpening.
To apply this type of sharpening we need yet another duplicate of the original photo which should be placed on top of all layers. So, the procedure is as follows: select background layer (this is why it is useful to keep the original intact), right-click and select "Duplicate", give the new layer a meaningful name, e.g. "High pass", and finally drag it to the top of the layer stack. Once it's on top, we need to apply the high-pass filter to it (menu "Filter", "Others", "High Pass"):
I think that a radius of 2 should be just right (this would depend on the photo), although some suggest using much bigger values and then adjusting the effect with the opacity slider. The actual sharpening is done by changing blending mode of the "High Pass" layer to "Hard Light":
Easy way of checking how it compares to the unsharpened photo is by turning visibility of the "High Pass" layer on and off (click the eye icon on the left of the layer miniature). Is the effect too strong? No worries - use the opacity slider of this layer to set sharpening just right. Another possibility is to set the blending mode to "Soft Light", which gives more subtle effect.
As you can see, sharpening is applied to the whole photo. If this is what you are after then you are done - you can flatten it (right-click on any layer and "Flatten") and save it as a JPG. Saving the photo in PhotoShop own format first (before flattening) may also be a good idea - all layers will be saved and the photo can be reworked later on. However this is not what we want to achieve - we want only model's eyes to be sharpened, not the whole image.
Why would we want to do this? Well, there is at least one good reason: sharpening a photo, whatever method you choose, will amplify the noise. So the best approach is to sharpen only those parts of the photo that must be sharpened. In particular there is no sense to sharpen areas that are out of focus.
So, how would we apply the sharpening to the part of the photo? In full-blown PhotoShop we would simply add a mask to the "High Pass" layer and paint in balck on this mask over all parts that we do not want to be sharpened.
Layer mask digression
Generally, a layer mask is a piece of "template" which controls how a layer is applied to the one below it. Black parts of the mask are "translucent", i.e. corresponding parts of the layer below will be visible. White areas of the mask are opaque, i.e. corresponding areas of the upper layer will cover everything below it (well, meaning of the "cover" depends on what belnding mode is selected). Gray areas on the mask give partial opacity, i.e. upper layer will be mixed with the lower layer on these areas.
So, the layer masks are the answer for our needs but this is where we hit the problem: in PS Elements (surprize! surprize!) ordinary layers cannot have masks. This "priviledge" is reserved only for adjustment layers. Hopefully there is a trick which allows us to work-around this problem. But before we apply it let's revert blending mode of the "High Pass" layer back to "Normal" and its opacity back to 100%.
Layer mask trick in PSE
The above mentioned work-around is quite simple: it's enough to create a dummy adjustment layer right between two layers where we need the mask to be created and group the upper layer with the just created adjustment layer. The resulting effect will be that the mask "coupled" with the adjustment layer will serve our purpose - just like it was added to the upper layer.
Let's see this in action: first step is to select the "Mono" layer and then add a "Solid color" adjustment layer (new adjustment layers are always added above the selected layer):
You can use whatever color you want for this layer - it completely does not matter in here. Now, after our "dummy" layer was created, you should select the "High Pass" layer and group it with the "dummy":
Now, all grouped layers behave just like one. For example clicking the eye icon of the layer which is the lowest in the group will turn off visibility of the whole group. Similarly the mask which accompanies this lowest layer will serve as a mask for the whole group. Now, let's check how it works in detail:
- Select the "Color Fill" layer (oops, I forgot to rename it :$ ),
- Use the "Paint bucket" tool to fill the layer mask with the black colour (just use the tool on the photo). Black layer mask is translucent (bottom layer is visible) and so the photo should change from the "High Pass" view to the "Mono" view.
- Now, switch to the "Brush" tool, select some soft brush of a relatively small size, select white color, and draw over model's eyes (zoom-in for a precise work). Model's eyes should become grey (i.e. "High Pass" layer covers what's below it).
- If you accedintally draw with white over too big area - no worries, just switch to the black colour and repaint wrong fragments back to black.
Generally this is how it should look like at this stage:
The final step is to turn on sharpening: change the blending mode of the "Color Fill" layer (remember: bottom layer in the group controls whole group) to "Hard Light" and its opacity to the level that gives appropriate sharepening.
The final effect should look like this:
It's worth noting that having all these layers in place allows one to change all the settings and parameters at will, e.g. easily add more B&W by changing opacity of the "Mono" layer or sharpen more by changing opacity of the "Color Fill" layer. Because of this it is reasonable to save the file in PhotoShop format before flattening the image and exporting it to JPG to publish it on Dyxum ;)