DPP: Digital image qualities - Exercise on sensor linear capture

This exercise was intersting to me not least as a way to start to understand how the digital camera works and how the image is processed by the camera before I get to see it.

However, the description of how to conduct the exercise in the study material is confusing and I am unsure if I have done the exercise correctly or if the exercise description is wrong.

I used an image of a duck that I took yesterday and using Photoshop CS 5.1 (and not Elements as the material says), converted it to 16 bits per channel and made a curve that was identical to that shown in the material. As suggested, the image became much darker and I was able to see what the image would have looked like when it was first captured. The histogram for this image showed significant bunching to the left of the graph.

I then put this 'new' image up alongside the original image and studied the histograms. If the definition of the 'linear' image is the one that is in the camera, unprocessed, then the histogram (as below) shows the tones squashed tightly to the left (and not the right as the exercise description says). The histogram for the camera-processed image shows the tones more even but with an emphasis more to the right of the graph.

To get the linear image as close as possible to the camera-processed version - and it was still a way off as I discovered - I adjusted the curve the opposite way to that when I created the linear image from the camera-processed version to the linear version. This did require some quite significant adjustments as the image below shows (see the curve diagram to the right).