Exercise 2: Highlight clipping
Shown below are the 5 images produced for this exercise, separated by 1 f stop, as described. I have cropped the original framing to show the area of clipping i.e. the white painted bridge parapet to the left of the frame. It is also useful to have the grey painted concrete area in the crop as an inbuilt grey card. My camera shoots in RAW (NEF) and produces a jpeg file as well. I have used this file for the first part of the exercise.
1: Original image with highlights just clipped 1/800s f10
1A: Cropped image showing highlights: The white bridge has very little detail and is mostly pure white.
2: Cropped image – exposure increased by 1 stop: 1/800s f7.1 There is a lot less detail and the surrounding image is beginning to lose saturation
3. Cropped image – exposure decreased by 1 stop from the original: 1/800 f14 There is a lot more detail here, the algae on the paint, the shadows and the detailing on the pillars is now apparent. The sun’s reflection from the black bin bag in the bottom left is still clipped. Colour saturation appears just a little muddy.
4. Cropped image – exposure decreased 2 stops from the original: 1/800s f20 The image shows a lot more detail on the parapet now but the shadows on the image are now blocked up and the contrast needs adjusting.
5. Cropped image – exposure decreased 3 stops from the original: 1/800s f29 Here the white on the wall is very muddy and the image is underexposed. Curiously the white sunshades in the background are still white showing that they were very strongly clipped.
I took my images in RAW (NEF) but as I am using Capture NX2 there is no separate process required for conversion. I always have my camera produce a .jpeg for each NEF file. I tend to process in the other direction i.e NEF files to high quality .jpeg or .tiff files. As Capture NX2 is non destructive, I can store and recover all edits to a NEF file and save a number of versions. I do have a new version of PS Elements which came bundled with a recent purchase. I’ve looked at the Recovery slider and seen how it works. As my workflow has long been established in NX2, I will use the Auto Levels menu and the double threshold feature to reprocess the images and recover the highlight information.
For the first Image (1A) I will document the process using screen shots:
1. Here is the cropped image unprocessed showing the original histogram.
2. Clicking the Double Threshold box on the RGB histogram shows the lost highlights (white) and blocked shadows (black) against a grey background.
3. Adding Auto Levels to the edit list removes the highlights from the display.
4 Un-ticking the double threshold box reveals the edit and clicking Advanced in auto levels gives the option to adjust the contrast and colour cast.
5. I think this image is a little blue so I adjusted the colour (-22%) and increased the contrast by 6%.
7. The final image show no appreciable difference to the Quick Fix histogram but there is a marked shift in the distribution of the RG & B channels.
| Before | After |
 |  |
8. Checking the Double Threshold box again reveals no highlight values but does show that the blocked shadows are starting to increase.
Here are the remaining four images processed as above.
2. This image is more saturated than the original and a lot of the highlights have been recovered but is still not ideal. Again the threshold shows that the shadows are starting to block.
3. Although the lost highlights on the bridge have been recovered, it is at the expense of the overall contrast. I’ve adjusted it and the result is a compromise.
4. Auto Levels is now starting to work in the other direction for this underexposed image. I can see from the Double Threshold that the highlight areas have increased slightly and the blocked shadows have gone.
5, With this final underexposed image although the highlights have been restored, the surrounding areas are still very under exposed.
Conclusions: This exercise has served to demonstrate the limitations of digital processing with regard to the way in which a linear processer records an image compared to film or the human eye. I have also come to appreciate how these effects can be mitigated by software and the limits that apply to what software can achieve. However, with the rapid pace of change in digital sensors and processing software many of these limitations may already be in the process of being overcome.
