Shadow/Highlight Detail in Photoshop -- Part III

Article and Photography by Ron Bigelow

www.ronbigelow.com

Photoshop CS or Photoshop CS2 Used in this Tutorial

Dodge and Burn Tools

Figure 1: Dodge and Burn Adjustments

The Dodge and Burn tools were designed specifically to lighten and darken tones in an image. However, the Dodge and Burn tools are a bit different from the tools that have already been covered. With the previous tools, the effects of the tools were not restricted to specific locations within an image unless a mask was used. The previous tools either applied the effects to the entire image or restricted the effects to specific tonal regions (e.g., the Shadows command in the Shadow/Highlight tool). With Dodge and Burn, the lightening or darkening is literally painted onto the image. For example, the highlight on Uncle Fred's forehead can be painted out, or the shadow under Aunt Edna's eyes can be touched up. These tonal adjustments affect only the areas were the effect is painted in.

Figure 2: Dodge and Burn Adjustments

The Dodge and Burn tools are found on the Tools palette. There are a number of ways to adjust the effects of the tools. Figure 1 shows some of the Dodge and Burn options on the Options Bar. This figure shows the adjustments for the size and hardness of the brush. Figure 2 shows another section of the Options Bar. This section of the Options Bar sets the magnitude of the effects (Exposure) and allows the effects to be limited to either the shadows, midtones, or highlights.

While Dodge and Burn can be applied directly to a layer, it is preferable to create a separate layer for the dodging and burning. The most common approach is to add a layer and merge the effects of the lower layers merged into it. The dodging and burning can then be done on this new layer.

Dodge and Burn does not do any better than the previous tools when it comes to the limitation of the tonal levels in the shadows (particularly for JPEG). The small number of tones available in the shadows of an image limit the amount of enhancement that Dodge and Burn can perform. The larger number of tones in the highlights allow for better results. As always, the situation improves significantly when the image was shot in raw (due to the larger number of tones).

Dodge and Burn appear to cause little stretching of the tones in the enhanced tonal regions. This is shown in Figures 3 and 4.  Figure 3 shows the histogram of a JPEG image that has not been edited. The histogram is smooth with no tonal gaps or compression. Figure 4 shows the histogram after Dodge was used to lighten the shadows. While the shadows have been shifted to the right, as expected, there appears to be little or no stretching of the shadow tones. Furthermore, the highlights are not compressed. This lack of compression of the non-enhanced areas (highlights in this case) is as expected. First, Dodge was applied with the Shadows option selected, so the highlights were protected. Second, the dodging was painted into the shadow areas only.

Figure 3: Histogram before Editing (JPEG Image)
Figure 4: Histogram after Dodge (JPEG Image)
Similar results occur if Burn is applied to the highlights (as shown in Figures 5 and 6).
Figure 5: Histogram before Editing (JPEG Image)
Figure 6: Histogram after Burn (JPEG Image)

One thing that is not obvious from the histograms, but can easily be seen by studying images where Dodge and Burn were used, is that Dodge and Burn do not substantially increase the contrast of the image like Curves does. Thus, the effects of Dodge and Burn tend to make the shadow detail stand out less than when Curves is used.

Clipping is not a big concern with Dodge and Burn. This is demonstrated in Figure 7 and 8. Figure 7 shows the histogram of a JPEG image before any editing. Figure 8 shows the histogram after Dodge was used to dodge the midtones. It can be seen that the right side of the histogram has grown taller (indicating more pixels with lighter tones after editing). However, the histogram also shows that the highlights were not clipped. So, while it is possible to clip the shadows and highlights with Dodge and Burn (I was able to deliberately clip some highlights by dramatically over dodging some lighter tones), such clipping would likely be caused only by deliberate intention or extremely careless editing. Normal editing is unlikely to cause clipping with Dodge and Burn.

Figure 7: Histogram before Editing (JPEG Image)
Figure 8: Histogram after Dodge (JPEG Image)
Figure 9: Noise in the Dark Areas after Application of Dodge

When it comes to noise, Dodge and Burn suffer from the same problem with the low SNR in the shadows that plagues the previous tools. Figure 9 shows the image of the dark sky that was used previously to show the shadow noise problem. This time, the shadows were lightened with Dodge. This figure shows that there is less noise than when Curves was used. This is due, at least partly, to the fact that Dodge does not increase the contrast like Curves.

The last consideration is with the data that is lost to the application of tonal curves. Dodge and Burn are no better than the previous tools. In other words, Dodge and Burn have no capability to access this data.

In summary, Dodge and Burn show neither significant stretching/compressing of tones nor clipping. However, these two tools suffer from the small number of shadow tones and an inability to access the data lost due to tonal curves.

Blend Modes

Up until now, each of the shadow and highlight enhancement methods involved a tool that was applied to the layers to produce an enhancement of the shadow and highlight detail. However, the Blend mode method uses the layers themselves to produce the enhancement of detail.

Figure 10: Layers Palette and Screen Blend Mode

Actually, there is not just one Blend mode method to bring out the shadow and highlight detail; rather, there is a multitude of Blend mode approaches. It is beyond the scope of this article to cover them all. Instead, two Blend modes will be introduced (one for the shadows and one for the highlights). The general approach of these two techniques is the same. Either the background layer is duplicated or a new layer is added and the effects of the lower layers are merged into it. Then, the Blend mode of the duplicated layer is set to the desired mode.

The Blend mode that is typically used to lighten an image is the Screen Blend mode. The Blend mode that is typically used to darken an image is the Multiply Blend mode. Figure 10 shows the Layers palette for an image where the Background layer was duplicated and set to the Screen Blend mode in order to lighten the shadows. Figure 11 shows the Layers palette for an image where the Multiply Blend mode was used to darken the highlights.

Figure 11: Layers Palette and Multiply Blend Mode

 

Unfortunately, the blend modes have to deal with the limited number of shadow levels just as the methods that have already been introduced. The Blend modes can only enhance the detail that is already there, and that detail is limited by the small number of shadow tonal levels. Just as before, the situation is far better in the highlights and when raw is used.

Stretching and compression of the tonal levels is an issue with the Screen and Multiply Blend modes. Figures 12 and 13 illustrate the tonal stretching and compression for an image where the Screen Blend mode was used to lighten the shadows. Figure 12 shows the histogram of a JPEG image before any editing. Figure 13 shows the image after the Background layer was duplicated and the Blend mode of the Background layer was set to the Screen Blend mode. Figure 13 clearly shows that the shadows have been stretched and the highlights have been compressed.

Figure 12: Histogram before Editing (JPEG Image)
Figure 13: Histogram after Screen Blend Mode (JPEG Image)
Figures 14 and 15 illustrate the tonal stretching and compression for an image where the Multiply Blend mode was used to darken the highlights. Figure 14 shows the histogram of a JPEG image before any editing. Figure 15 shows the histogram after the Background layer was duplicated and the Blend mode of the Background layer was set to the Multiply Blend mode. The rather significant stretching of the lighter tones can easily be seen in Figure 15. Additionally, the compression of the darker tones can also be seen.
Figure 14: Histogram before Editing (JPEG Image)
Figure 15: Histogram after Multiply Blend Mode (JPEG Image)
Figure 16: Histogram after Multiply Blend Mode (Raw Image)

When using the Screen and Multiply Blend modes, the compression of the non-enhanced tones can be eliminated by protecting those areas with masks. In addition, the use of the raw format will significantly improve the problems with both stretching and compression of the tones. This is demonstrated in Figure 16. This figure shows the histogram from the exact same image used for Figure 15. In Figure 16, the Background layer was duplicated and the Blend mode of the Background layer was set to the Multiply Blend mode (just as in Figure 15). The only difference is that, in Figure 16, `the image was converted from a raw file and saved as a TIFF. This allowed the file to keep all of its tones (i.e., 4,096 before application of a tonal curve) while the JPEG version of the file had a much lower number of tones (i.e., 256 before application of a tonal curve). The result is a nice smooth histogram where the tonal stretching and compression are no longer noticeable.

The Screen and Multiply Blend modes do not cause clipping. This is demonstrated for the Screen mode in Figures 17 and 18. Figure 17 shows the histogram of a JPEG image before any editing. Figure 18 shows the histogram after the Background layer was duplicated and the Blend mode of the Background layer was set to the Screen Blend mode. As can be seen in Figure 17, this image is composed almost totally of lighter tones. The use of the Screen Blend mode forced the tones even lighter. Figure 18 shows that this rather extreme edit did not clip the highlights. It did, however, cause some rather severe compression of the highlight tones.

Figure 17: Histogram before Editing (JPEG Image)
Figure 18: Histogram after Screen Blend Mode (JPEG Image)
Figures 19 and 20 show that the Multiply Blend Mode did not clip the shadows even when applied to the already dark tones of an image.
Figure 19: Histogram before Editing (JPEG Image)
Figure 20: Histogram after Multiply Blend Mode (JPEG Image)

Keep in mind that Figures 17 -- 20 were used to illustrate a point. Under normal conditions, it is highly unlikely that a photographer would use the Screen and Multiply Blend modes in this manner (darken the shadows or lighten the highlights) unless he was producing special effects.

Figure 21: Noise in the Dark Areas after Application of Screen Blend Mode

When the Screen Blend mode is used to lighten shadows, it faces the same predicament with the poor, shadow SNR as the other tools that have been covered. Despite this, the noise produced when the Screen mode is used can be relatively mild as shown in Figure 21.

Unfortunately, the Screen and Multiply Blend modes are no help when it comes to the data lost to the application of tonal curves.

So, the Screen and Multiply Blend Modes cause no clipping. Also, the noise performance isn't bad when compared to some of the other tools that have been covered. On the other hand, both tools suffer from the small number of shadow tones, stretching and compression of tones, and an inability to access the data lost due to tonal curves.

Raw Converter Adjustment

This is where it starts to get interesting. All of the tools presented so far are hampered by the small number of tones in the shadows; the poor, shadow SNR; and the data loss due to the application of tonal curves. The tools are totally unable to overcome these limitations. Therefore, the tools are restricted in the results that they can produce. The rest of this series of articles will cover tools that allow these limitations to be reduced or even overcome.

Figure 22: Tonal Curve

The first of these more advanced shadow and highlight recovery techniques has to do with the application of the tonal curves and the use of the raw converter controls. As covered in Part I of this series, the data from the sensor is very dark. So much so that it is unusable, as is, for most photographic applications. To remedy this situation, each image has a tonal curve applied. The various tonal curves are quite steep (an example of a tonal curve is shown in Figure 22) and dramatically lighten the images to which they are applied. On the other hand, the tonal curves can also clip data in the shadows and highlights. Once a tonal curve has clipped data in an image, that data is lost in the image as well as the detail that resided in that data (although the raw file still has the data). The current technique is aimed at keeping at least some of that data that can be lost to tonal curves.

Now, one caveat with this technique is that it is used only with raw images. With raw images, the tonal curve is applied in the raw converter. Thus, the tonal curve, as well as the other raw converter controls that affect the results of the tonal curve, can be accessed by the photographer. With JPEG images, the tonal curve is set in the camera. Therefore, the tonal curve has been applied and the data clipped before a photographer gets the image from the camera. While many cameras do allow a photographer to set the tonal curve used by the camera when JPEG images are shot (see the sidebar labeled JPEG and Tonal Curves), once a tonal curve is selected and an image taken, the data is clipped in the camera. Unlike with raw, the photographer can not do anything to reclaim that lost data. Furthermore, even the cameras that allow the tonal curve to be modified in the camera lack the advanced controls that affect the results of the tonal curve that any good raw converter has. Finally, the raw converter allows a photographer to experiment with the tonal curves and other controls until the right results are achieved. This can not be done when the tonal curve is set in the camera.

JPEG and Tonal Curves

So, how does one select a tonal curve if he is shooting JPEG images? Many cameras have a setting for contrast. Typically there will be low, medium, and high contrast options. What this setting is actually doing is setting the tonal curve.

This technique starts with the raw converter. Many raw converters (but not all of them) allow the photographer to change the tonal curve (unfortunately, the raw converter may not label the control as the tonal curve). Figure 23 shows the control for the tonal curve on one raw converter. In addition, there are many other controls in the raw converter (e.g., exposure, contrast, and Curves) that affect the results of the tonal curve.

Figure 23: Tonal Curve Selection in Raw Converter

Once the tonal curve control is identified, the technique is incredibly simple. The photographer just chooses the tonal curve and sets the other raw converter controls to provide the detail that is desired. For example, if a photographer wants to enhance the shadow detail, she would merely select a tonal curve that is designed to reveal the shadow detail and adjust the other raw converter controls to maximize the shadow detail.

Figure 24: Image with Clipping Issues
Figure 24 shows a desert waterfall image that has problems due to clipping (I realize that this isn't the world's greatest photo, but it illustrates this concept well). The image has the very bright tones of the falling water and the very dark tones of the little cave in the rock on the left side. Figure 25 shows the raw converter histogram with one of the tonal curves applied. The tonal curve has caused the tones to be clipped in the highlights and the shadows.
Figure 25: Histogram with First Tonal Curve (Raw Image)
Figure 26 shows the histogram of the image with one of the other tonal curves applied and the exposure increased slightly. The highlight detail is no longer clipped (the colored spikes on the right of the histogram are specular highlights and should be clipped). In addition, it appears that there might be a little bit more shadow detail than with the first tonal curve. In short, the change to the second tonal curve allowed more highlight and shadow detail to be preserved that with the first tonal curve.
Figure 26: Histogram with Second Tonal Curve (Raw Image)
Figure 27: Histogram with Clipping Problems (Raw Image)
Another trick that is common with raw converters is to perform two or more conversions. One conversion is done to preserve the highlight detail and another is done to preserve the shadow detail. In addition, there could be another conversion aimed at the midtones. Figure 27 shows a histogram where the highlight detail has definitely been clipped. In addition, it appears that some shadow detail may have been clipped.
Figure 28: Histogram with Exposure Adjusted to Protect the Highlight Detail (Raw Image)
Figure 28 shows the histogram of the same image after the exposure control in the raw converter was adjusted to bring out the highlight detail. As can be seen, this histogram shows a lot of highlight detail that was clipped in the previous histogram.
Figure 29: Histogram with Exposure Adjusted to Protect the Shadow Detail (Raw Image)

Another conversion was then performed to see if additional shadow detail could be recovered. Figure 29 shows the histogram of the image after the exposure control in the raw converter was adjusted to bring out the shadow detail. It appears that a small amount of additional shadow detail may have been gained.

These two conversions, with perhaps a third for the midtones, could then be combined in Photoshop. The result would be an image with the greatest possible highlight and shadow detail possible from the original raw file.

The advantage of this technique is that the maximum amount of shadow and highlight detail can be preserved. Typically, once the conversion has been performed, the image will be edited in Photoshop using the tools previously covered (e.g., Curves and Blend modes). Therefore, the effects with respect to the other issues, such as noise, depend on which tools are used for further editing.

This technique is a step up. However, it still leaves an image with a relatively small number of shadow tones and a poor SNR in the shadows.

Articles

Shadow/Highlight Detail -- Part II     Shadow/Highlight Detail -- Part IV