We’re going to use this image for web-posting only, specifically, for this site. My pages can only accommodate images up to 450-pixels wide, so I’ll resample my image down to 72dpi and change the width of the image to 450px.
Now that I have an appropriately-sized image, I can merge all the changes we’ve made by flattening all the layers. Remember that I’ve already saved a full-resolution version of this image with the NR and base sharpening applied, so I don’t need to have all those layers slowing down my system for these final steps.
Once flattened, we’ll end up with a single ‘Background’ layer. To begin our OUTPUT sharpening phase, we duplicate the Background layer again.
Note that I’ve changed the new layer’s opacity to 70% again to give me the flexibility to increase sharpening if I need to. I’ve changed the blending mode of this new layer to Luminosity as well. Changing the blend mode to Luminosity allows us to apply sharpening without creating color halos around the edges.
For output sharpening, the critical part would be increasing the contrast and apparent sharpness of the edges of the image. For most portraits, we want to sharpen the eyes, edges of the hair, highlight edges that define the shape of the lips, nose, and other edges, but we usually don’t want to sharpen pores, shadows, solid areas of an image.
We need to find a way to isolate the image’s EDGES only and apply sharpening only to those edges.
Usually, brush masking is used to quickly mask off areas that we don’t want to affect, but using brush masks require manual, per image adjustments, which slows down our workflow and isn’t as accurate unless significant amount of time is wasted to manually pick out edges with a brush.
First we isolate the image by converting our image into a high-contrast, black and white image. Here we’re using the Calculations command (IMAGE >> CALCULATIONS).
Follow the options below, make sure you’ve chosen both Red channels and a blending mode of Hard Light. Click OK.
This created a new B&W channel called ALPHA2 if you check the Channels palette.
Next, we run the Find Edges filter (FILTERS >> STYLIZE >> FIND EDGES). This filter searches for all the hard edges in the image for us automatically.
Remember the basic mask rule – black hides, white reveals. If we apply sharpening to the image above, the black lines will NOT receive any sharpening, but all the white areas will. That’s not what we want, so we invert the image (CTRL-I) and get the image below.
We’re still seeing a lot of areas that we don’t want sharpening to be applied onto, such as the face and the skin of the arms. We can adjust the contrast of the image by knocking out those faint white lines using the Levels tool (CTRL-L).
Drag the black point slider (left-most triangle) towards the middle of the image. You’ll see that most faint white lines will disappear. Adjust the midpoint (grey) triangle as appropriate.
Drag the Alpha2 layer (still in the Channels palette) to the dotted circle icon at the bottom. This creates a selection based on that black mask we’ve just made.
Return to your Layers palette and you’ll notice that a selection (marching ants) has been made over the image. Hold the ALT key and click on the Mask icon to create a Hide-All layer mask (black mask)
All the black areas of the image will NOT receive whatever effect or sharpening we’re going to apply next. Only the white areas will reveal the sharpening.
Open the Unsharp Mask filter (FILTERS >> SHARPEN >> UNSHARP MASK) and use a fairly high sharpening amount with low radius setting. Don’t worry about the ugly result on the Preview window. Remember that the sharpening will only be applied on the unmasked section of the image (the edges).
If the edges are a bit too sharp or the transition between the sharp and unsharp areas are too abrupt, you can soften the mask using the Refine Mask adjustments. Adjust the Feather radius to soften the transition between the masked and unmasked areas. Conversely, you can apply a Gaussian Blur filter on the mask itself.
You can see that the mask has a much softer edge between the revealed sections and the hidden sections.
Here’s the final image. Mouseover the image to view the image we started off with.
The workflow may seem long, but unlike using manual masking where you need to reveal the areas to be sharpened manually, this workflow can be saved into an action and ran automatically in Bridge after import, you just need to add additional Save-As steps before the Output Sharpening phase.
For your convenience, you can download the free action set here (2kb RAR file).
Have fun and help me spread the article by sharing this on Twitter/Facebook/Digg using the buttons below.