Blurring the Lines
The next step is to apply a Gaussian Blur to the mask. This will accomplish two goals.

• It will soften the edges around the edge mask, preventing sharp, unnatural-looking transitions between sharpened and unsharpened areas.

• It will get rid of some of the noise that Find Edges picked up.

The radius setting depends primarily on the resolution of the image and secondarily on its content. High-resolution images typically require a higher-radius blur than low-resolution ones, and close subjects with soft detail, such as head shots, typically need a higher radius than high-frequency images with lots of fine detail. In this case, a Radius of around 2.8 pixels blurs out the noise while preserving the main edges, albeit in softened form, as shown in Figure 5.

Figure 5: Applying a blur prevents unnatural transitions and eliminates noise.

Next I'll invert the mask. This step is optional: I simply find it easier to see what's happening when I make the areas I want the mask to protect black and those I want it to affect white. Inverting the mask also means that you don't have to invert the selection after the mask is loaded. Choosing Invert from the Adjust submenu on the Image menu produces the result shown in Figure 6.

Figure 6: Applying a blur prevents unnatural transitions and eliminates noise.

The final step in creating the mask is to exaggerate the contrast of the mask, so that the edges we want to sharpen are really white and the noise we want to protect is really black. The Levels control offers an easy method for doing this. The trick here is to adjust the black and white input sliders in Levels so that the edges you want sharpened are pure white and the areas you want protected are solid black. You also want to leave a little gray in between, so that the transitions between sharpened and unsharpened areas are soft rather than sudden. The actual numbers will depend on the image, but typically you'll want to move the sliders quite a way in, so that a lot of pixels are forced to black and a good many are forced to white.

Figure 7 shows the Levels adjustment for our example, and its effect on the mask.

Figure 7: We used the Level adjustment to exaggerate contrast, to ensure the edges we want to sharpen are white and that noise is black.

Note that the mask still contains a fair number of gray pixels: Those areas will receive slight sharpening, and will prevent sharp transitions between the sharpened and unsharpened areas.