Sharpening in Photoshop -- Part IV

Article and Photography by Ron Bigelow

Photoshop CS or Photoshop CS2 Used in this Tutorial

In the previous articles, we learned some basic sharpening tools and procedures; saw how sharpening can be adjusted separately for the mid-tones, highlights, and shadows; and discovered how the use of a filter on a separate layer can add more flexibility to the sharpening process. Yet, as much as we have covered, we are still left with a number of sharpening problems that have not been resolved. This article addresses two of these issues, color fringing and the destructive nature of sharpening, by introducing three new sharpening methodologies. The first two tools, LAB sharpening and the Fade Tool, address the problem of color fringing. The last tool presented in this section, layer based sharpening, addresses both color fringing and the destructive nature of sharpening (while this has already been addressed by the High Pass sharpening technique, layer based sharpening does it in a much more general and versatile manner).

LAB Based Sharpening

While sharpening has the desirable affect of increasing the contrast along edges, it can also have the undesirable affect of causing a color shift along the very edges that are being sharpened. Figure 1 (shown at 200% view) shows an example of color fringing. The sharpening in this example has been exaggerated to make it easier to see the color fringing. The power line in Figure 1 clearly shows a magenta colored fringe along the line. Figure 2 shows the same line after the color fringing has been removed. A halo around the power line still exists due to the high amount of sharpening used, but the magenta color is gone.

Figure 1: Color Fringing
Figure 2: Color Fringing Removed

Most photographers tend to work in an RGB color space (e.g., sRGB or Adobe RGB). RGB color spaces have three channels, one for each of the three colors (i.e., red, green, and blue). Sharpening in an RGB color space can result in color fringing like that shown in Figure 1. One way of dealing with color fringing is by using LAB sharpening. LAB is simply another color space. The LAB color space also has three channels. However, the channels are different than those of an RGB color space. Rather than three color channels, the LAB color space has a lightness channel, that stores the tonal information, and two color channels, channel "a" that stores the blue/yellow information and channel "b" that stores the green/red information. In LAB sharpening, the image is converted to the LAB space and the sharpening is conducted on the lightness channel only. Since the lightness channel contains only tonal information, color fringing can not occur. After sharpening, the image is converted back to the RGB color space for any final editing and printing.

Figure 3: LAB Channels Palette

LAB sharpening is fairly easy to do. The image is first converted to the LAB color space by selecting Image/Mode/LAB Color. The Channels palette is selected. The Lightness channel is then selected (see Figure 3). The sharpening is performed on the Lightness channel using the preferred sharpening method (e.g., USM or Smart Sharpening). After the sharpening, the image is converted back to the RGB color space by selecting Image/Mode/RGB Color. Figures 4 and 5 shows a section of an image that was sharpened with this method.

Figure 4: No Sharpening
Figure 5: After LAB Sharpening
For those of you that have come this far, it will probably come as no surprise that, while LAB sharpening solves the color fringing problem, it has other problems. Most importantly, LAB sharpening requires two conversions. Each conversion causes a small degradation of the image. As such, LAB sharpening results in some loss of image quality. Furthermore, LAB sharpening still suffers from the other problems associated with whatever sharpening approach was used on the Lightness channel. LAB has an additional problem. An image can not be converted to LAB space with the layers intact. To convert to LAB space, the image must be flattened or the layers discarded. This is enough of a reason for me not to use this sharpening method. Therefore, while I introduced this sharpening approach so that you will be familiar with it, I do not recommend it. Furthermore, I do not use this method on my own images. There are far better approaches to dealing with the problem of color fringing.

Fade Tool

Figure 6: Fade Menu

The Fade Tool is another method of dealing with color fringing. It is incredibly simple. To utilize this sharpening approach, sharpening is performed by whatever method is preferred (e.g., USM or Smart Sharpening). After the sharpening has been completed, the Fade tool is immediately launched by selecting Edit/Fade. The opacity should be set to 100% and the mode to Luminosity (see Figure 6). That's it. Once the OK button is clicked, the sharpening will be applied to the tonal values only. This produces a sharpening that is almost identical to LAB sharpening, but it does not require any conversions (for those interested in the difference between lightness and luminosity, see the Lightness and Luminosity box). Thus, the use of the Fade tool does not cause the type of image quality loss that is concomitant with the use of LAB sharpening.

Lightness and Luminosity

In any color image, there is both tonal and color information. We often see the color information represented by the values of the three primary colors (red, green, and blue). For instance, we may have a color of (25, 57, 123). The issue becomes how such a combination of colors is converted into a gray value. Lightness and Luminosity are both methods that are used to determine how the tonal values in an image are calculated from the color information. In other words, Lightness and Luminosity assign levels of gray for any combination of color. Lightness and Luminosity use different, but similar, formulas for assigning the gray values.

Still, not all is perfect. In a sense, the Fade tool is simply a bolt-on to the sharpening tool that is used (e.g., USM or Smart Sharpening). Therefore, this approach suffers from the same deficiencies as the selected sharpening tool except that the color fringing has been eliminated.

Layer Based Sharpening/Luminosity Mode

As seen in the last section, the Fade tool is pretty easy to use. Layer based sharpening isn't much more complicated. At first glance, it is so basic that it would be easy to underestimate the method. However, this would be a big mistake. Layer based sharpening is an important move in the right direction.

Layer based sharpening has three advantages. First, it eliminates color fringing. Second, the sharpening is applied to a separate layer. Like High Pass sharpening, this allows for a significant amount of flexibility. In addition, the sharpening can be undone simply by deleting the layer. While both High Pass sharpening and layer based sharpening are layer based and are sharpening methodologies, rather than tools, layer based sharpening offers much more flexibility than High Pass sharpening. With layer based sharpening, the photographer can choose which sharpening tool to integrate into the sharpening process (with High Pass sharpening, you have no choice but the use of the High Pass filter). However, by far, the greatest advantage of layer based sharpening is that it opens the door to the use of masks and filters that are the secret portal that leads to the advanced techniques and superior sharpening results.

There are actually two general ways that layer based sharpening can be carried out. In the first method, the sharpening is performed on a copy of the background layer (we will call this background-copy sharpening). This approach is good when you want to sharpen only the original image information and want the image editing to be performed on top of the sharpening. In the second method, the sharpening is performed on all of the layers (we will call this all-layer sharpening). This approach is good when you want to perform the sharpening after all of the image editing has been completed. The distinction between the background-copy sharpening and the all-layer sharpening will become important when we get to three pass sharpening (later in this series).

Figure 7: Layers Palette

Figure 8: Layers Palette with Sharpening Layer
Background-Copy Sharpening: The background-copy sharpening technique starts with the Layers palette (see figure 7). The Background layer is duplicated. This can be done by dragging the Background layer down to the Create a new layer icon at the bottom of the palette. The new Background copy layer is selected. The bend mode is set to Luminosity and the Opacity is set to 50%. Finally, the Background copy should be renamed as the Sharpening layer (see Figure 8).

This procedure has created a sharpening layer upon which sharpening can now be performed (more about the actual sharpening in a bit). How much easier could it get?

It is important to understand what has been done. Essentially, we now have an exact duplicate of our original image upon which the sharpening can be carried out. A very significant point is that the blend mode has been set to Luminosity. The Luminosity blend mode only lets changes to the tonality of the Sharpening layer show through in the image. In other words, any changes to the tones in the Sharpening layer will be seen in the image, but changes to the colors will not be seen. This is important because sharpening is all about changing the local contrast (i.e., tonality) between pixels. Thus, the Luminosity blend mode will allow the sharpening that is performed on the Sharpening layer to show in the final image. On the other hand, as previously covered, sharpening can cause color fringing. This color fringing causes strange color halos to appear around the sharpened edges. Luckily, the Luminosity blend mode does not allow these color changes to show through. Thus, the effects of the sharpening will be seen, but the effects of the color fringing will not. As a result, the use of the luminosity blend mode has completely eliminated color fringing.

Figure 9: Layers Palette

All-Layer Sharpening: Before the actual sharpening steps are covered, the all-layer sharpening technique needs to be introduced. The all-layer sharpening technique also starts with the Layers palette (see figure 9). In this case, there are several other layers in addition to the Background layer. For the all-layer sharpening procedure, it is necessary to create a new layer. Then, the effects of the other layers are merged into the new layer (notice that I said the layers' effects, not the actual layers themselves -- the other layers will still be there).

Figure 10: New Layer Menu
The top layer on the palette should be selected. The new layer is now created by selecting Layer/New/Layer. This will bring up the new layer menu (see Figure 10). The name of the layer should be changed to Sharpening. The Use Previous Layer to Create Clipping Mask option should be unchecked. The color should be set to None, the mode to Normal, and the Opacity to 100%. Clicking OK will create the new layer, which will now be found at the top of the Layers palette. If the Sharpening layer is not at the top of the layer stack (you forgot to select the top layer before creating the new layer), simply drag it to the top.
Figure 11: Layers Palette with Sharpening Layer

To merge the effects of the other layers into the new Sharpening layer, select the Sharpening layer. Then, hold down the Alt key (Option key on a Mac). While holding down the left mouse button, select Layer/Merge Visible. The Sharpening layer will now be an exact duplicate of all the layers below it. Finally, the blend mode is set to Luminosity and the Opacity to 50%. The Layers palette will now look as in Figure 11.

This layer, as with the Sharpening layer that was created in the background-copy sharpening technique, only lets tonal changes to the layer show through in the image due to the Luminosity blend mode. Consequently, the color fringing problem has also been eliminated with the all-layer sharpening technique.

Sharpening and Fine Tuning: The actual sharpening is a rather straightforward procedure. The procedure is now the same for both the background-copy sharpening and the all-layer sharpening. The Sharpening layer is selected and the layer is sharpened using one of the sharpening tools (e.g., USM or Smart Sharpening).

Now that the sharpening has been performed, it can be fine tuned. The first adjustment is the Opacity. Moving the Opacity from its original setting of around 50% to a higher setting will increase the sharpening. Moving the Opacity to a lower setting will decrease the sharpening.

Figure 12: Layer Style Menu
The second fine tuning adjustment involves the Blend-if control. The Blend-if control can be found on the Layer Style menu by selecting Layer/Layer Style/Blending Options. The Blend-if control is shown on the Layer Styles menu in Figure 12. For sharpening purposes, the Blend If control should be set to Gray. The control labeled This Layer now provides the ability to determine what tonal ranges are sharpened. In the section on Smart Sharpening, it was mentioned that the shadows may need to be protected from too much sharpening because the shadows contain more noise than other areas and that it is usually undesirable to have sharp shadows that draw the attention away from the main objects of interest in the image. It was also mentioned that the highlights should be protected from oversharpening to avoid blowing out the highlights with the sharpening. The sliders on the This Layer control allow the photographer to protect both the shadows and the highlights.
Figure 13: Layer Style Menu

Moving the shadow slider to the right determines what shadows will be sharpened. Likewise, moving the highlight slider to the left determines what highlights will be sharpened. Figure 13 shows an example. In Figure 13, the shadow slider has been moved to a value of 15. What this means is that no pixel below a tonal value of 15 (the tonal values range from 0 to 255) will be sharpened. The highlight slider has been moved to a value of 240. Thus, no value above 240 will be sharpened. Thus, both the shadows and the highlights have been protected.

However, this creates a small problem. There is now a rather abrupt transition between the areas that are sharpened and those that are not. In the shadow areas, pixels with tonal values of 15 or greater will be sharpened while pixels with lower values will not. This could be visible in the final print. A similar situation exists in the highlights. Pixels with values of 240 and below will be sharpened while those with higher values will not. Again, an abrupt transition will be created.

Figure 14: Layer Style Menu

Fortuitously, a simple solution exists. Holding down the Alt key (Option Key for a Mac) and clicking on either the shadow or highlight slider will split the slider into two sliders (if it doesn't split at first, keep trying). The inner of the two sliders determines where the sharpening starts to be attenuated and the outer slider indicates where the sharpening stops completely. Figure 14 provides an example. The sharpening of the shadows starts to decrease at a tonal value of 20. Below a tonal value of 7, no sharpening is done at all. On the highlight side, the sharpening starts to decrease at a tonal value of 238; above a tonal value of 252, no sharpening is done.

The abrupt sharpening transition is eliminated. At the same time, a greater amount of control of the sharpening is achieved.

Figures 15 and 16 show an image sharpened with the layer based sharpening methodology. Figure 15 shows the unsharpened image. Figure 16 shows the sharpened image. Layer based sharpening was a good choice for this image as it allowed for a reduction in the sharpening to protect the shadow areas as well as the thin, vertical highlight.
Figure 15: No Sharpening
Figure 16: After Layer Based Sharpening

Layer Based Sharpening/Luminosity Mode -- Heading in the Right Direction

Layer based sharpening has some major advantages.

Nonetheless, we're still not out of the woods. The layer based sharpening that we have just covered does not address the problem that areas of different textures require different amount of sharpening. Neither does it address the fact that the sharpening requirements of the image may differ from those of the output device.

Despite these last two, as yet unresolved, problems, layer based sharpening is a significant step in the right direction. Furthermore, the full potential of layer based sharpening has yet to be introduced. The complete power of layer based sharpening comes to fruition when layer based sharpening is combined with sharpening masks and edge masks.


Sharpening -- Part III     Sharpening -- Part V