6.5 A Case Study

This chapter has covered many elements of photo touchup and enhancement. To put all the elements into perspective, this section presents a case study to illustrate the work flow of a typical sequence of corrections. The photo in Figure 6.37

  

Figure 6.37: Original Image

is the subject of the study; it is a difficult case because it has many defects.

Before we begin, several observations can be made concerning this photo:

• It is dark and clearly underexposed.
• The colors are muddy and there is an overall lack of sharpness.
• The photo has a green tinge or color cast to it.
• There is an unsightly bright green spot on the lefthand side of the photo that will blemish the final product if not removed.
• The subject matter, the cowboy roper with the children, is poorly separated from the background.

In a nutshell, this photo needs a lot of help. It will be significantly enhanced by the steps described in the rest of this section.

The first step, as described in Section 6.1, is to maximize the tonal range  of the image. This is done by opening the Levels tool and choosing Auto Levels. Figure 6.38(b)

  

Figure 6.38: Tonal Range Maximized and Midtones Adjusted with the Levels Tool: (a) Original Image and (b) Result of Levels Tool

shows the result. Figure 6.38(a) shows the original image for comparison purposes. In addition to selecting Auto Levels, the midtone slider control in the Value channel has been adjusted to lighten the result a bit. Notice that the green color cast has been greatly diminished by the Levels tool adjustment, and the colors seem much clearer.

The second step is to correct for any remaining color casts.  As described in Section 6.2.2, this is accomplished using the Color Picker  to measure neutral colors in the image. The cowboy trick-roper and children are standing on a stone surface that appears to be granite. It is reasonable to guess that this stone should be a neutral gray.

With the Color Picker, you can see that there are actually a range of values from shadows to midtones in this stone. Measuring the image colors at several locations yields a representative shadow value of 59^R 49^G 42^B in the dark stone just to the left of the roper's lower legs. Similar measurements made in the lighter stone just beneath the little boy's feet yield a midtone value of 119^R 120^G 109^B. From these two measurements, it seems there is a brown/orange cast to the image. As for a representative highlight value, the right rear part of the lasso loop should be neutral but is measured at 177^R 183^G 167^B. This is a pale green.

To remove these color casts, the Curves tool is used as described in Section 6.2.2. For this case study, the shadow values of red at 59 and green at 49 are both moved to match the blue at 42. The midtone values of green at 120 and blue at 109 are both matched to the red at 119. Finally, the highlight values of green at 183 and blue at 167 are balanced to the red at 177. The result is shown in Figure 6.39(b).

  

Figure 6.39: Color Cast Correction with Curves: (a) Previous and (b) Color Corrected Image

In comparing this image to the previous one, shown again in Figure 6.39(a), the stone is now clearly a more neutral gray. Also, you can see that the skin tones had a slight orange tinge, which has been removed.

The next step is to remove the green glint at the middle left side of the image. This is done as described in Section 6.3 using the Clone tool. The image without this blemish is shown in Figure 6.40.

  

Figure 6.40: Blemish Removed with Clone Tool

At this point in the sequence of corrections, it should be clear that you cannot get additional subject detail using the methods from Section 6.2.6. The cowboy and children already fill the entire range of tonal values from the darkest shadows in the cowboy's shirt to the brightest highlights in the rope and the little girl's collar. The only way to get more out of this image is to try and separate the subject from the background, which can only be accomplished by making a selection. 

Figure 6.41

  

Figure 6.41: Selection of Background

shows the selection that was made using the Bezier Path tool. The selection was made in several stages. The main outline of the subject was made first. Three regions were then removed from this first selection by using the method for subtracting selections described in Section 3.2. These regions are the two enclosures the rope makes with the roper's body and the small hole between the roper's and little girl's leg.

Inverting the selection by typing C-i in the image window, the background can be lightened by using the perturbation  method described in Section 6.2.5 to adjust the Value channel of the Curves tool. As shown in Figure 6.42(b), the result produces a subject that

  

Figure 6.42: Background Lightened with Curves: (a) Before and (b) After

is better defined against the background. This is compared with the image from the previous step, which is shown again in Figure 6.42(a).

The final enhancement to the image is to sharpen the subject a bit. This is done using the Unsharp Mask as described in Section 6.4.1. The Unsharp Mask parameters were chosen to be 3 for Radius and 0.5 for Amount. The final result is shown in Figure 6.43(b),

  

Figure 6.43: Comparison of Original and Enhanced Photos

and Figure 6.43(a) shows the initial image from Figure 6.37. The comparison of the two images is dramatic. The original seemed unretrievable, and, although the final result is perhaps unworthy of National Geographic magazine, it is, nevertheless, greatly improved.

The new image has several qualities worth noting. First, the colors of the enhanced image are much sharper and better defined. In comparison, the original image's colors are muddy. This is due primarily to the enhancement to the tonal range. Second, the green tinge seen in the original image has been eliminated; the subject of the enhanced image is also sharper and better defined against the background. This is due to the Unsharp Mask and the reprocessing of the background with the help of selection tools and the Curves tool.

©2000 Gimp-Savvy.com