Now that the raw conversion had been completed, the next step was to evaluate the image for further editing. This evaluation consisted of viewing both the histogram and the image. This first step was to evaluate the histogram. The histogram was evaluated primarily to determine if the black and white points needed to be adjusted (for further information on black and white points, see Levels). The histogram (see Figure 1) showed little detail on the right side. One might be tempted to use Levels or Curves to stretch out the histogram and set the white point farther to the left on the histogram. However, a deeper analysis of the image (Figure 2) shows that this would be a mistake. The image only has a small amount of light tones. The only real light tones are the snow patches. Since these patches are small, they do not show up on the histogram as having a lot of pixels. However, these snow patches are an important part of the image. They add drama to the mountain reflections. A lot of work had already been performed to maintain detail in these patches. Any attempt to set the white point farther inward on the histogram would destroy this detail and drive the snow patches to pure white -- destroying the detail. Consequently, no adjustment was made to the white point.
A look at the left side of the histogram makes it appear that the blacks were clipped. However, a look at the histograms of the individual color channels (Figures 3 -- 5) provides a deeper understanding of the situation.
Seeing the clipping in the blue channel, the immediate temptation would be to go back to the raw converter and try to reconvert the image in order to recapture the clipped, blue channel data. However, using Levels to look at this clipped data (Figure 6) shows that only a very small amount of data at the back of the grass was in fact clipped (this detail consists of very small areas of deep shadow nestled between the individual blades of grass). This data is pretty much insignificant. In fact, even if it was recaptured, it would not be noticeable even in a relatively large print. On the other hand, recapturing this data would probably result in a lighter image or one with different contrast. This would result in additional work. Since performing additional work in an attempt to recapture insignificant detail does not make sense, the image was not reconverted.
Another temptation would be to use Levels or Curves to adjust the white point by adjusting the individual channels (for an explanation of this technique see Shadow Values, Highlight Values, and Midtone Contrast: Individual Color Channels). Some experimentation with this approach resulted in severe color shifts (as expected) and an image with too much contrast. Thus, this method was abandoned.
After this analysis, it was determined that no adjustment to the black point should be performed.
At this point, a test print was made. Examination of the print indicated that there were three problems that needed to be corrected (see Figure 10).