I got stuck at a subtle point while thinking about complex barycenter methods for direct optimization. So I checked out news on the just announced Fujifilm X-Pro1. It is a fascinating piece of engineering.
A year ago I started looking at cameras because I wanted to take pictures again. Except for a toy camera, a gift to Hannah, and the iPhone, I had so far passed on digital photography. Actually I had barely photographed since she was born, my interest turned off by the gift of a Canon Rebel 2000 - a low point in the history of photo technology, the photographic equivalent of Windows Millenium Bug.
In any case, the photo industry has long been rather uninteresting. Engineering progress after the Asahi Pentax Spotmatic and the Olympus OM-1 was evolutionary, and uneven. Electronics was used to make film cameras more foolproof (bakachon in the unpc Japanese expression) with the predictable result that mostly fools wanted to use them. Exposure controls familiar to photographers were replaced by pre-programmed modes designed by marketers; classic rangefinders and focusing screens were killed by dubious autofocus schemes.
Then low-quality light sensors were matched to low-quality optics to start the digital photography revolution. The most successful brands, Canon and Nikon, were concerned with preserving their market share in the single lens reflex market rather than with advancing camera design - not that they didn't have very good engineers. Smaller makers fought the silly megapixel race. European brands abandoned any attempt at innovation for some notion of prestige - only Leica is left, catering to people stuck with pricey legacy lenses.
But things are changing. From unexpected corners: large Japanese corporations without a heavy commitment to the DSLR format, which is a lousy compromise that doesn't take good advantage of modern technology. Sony matched their electronics with optics and camera designs from traditional manufacturers. Fuji realized that the film business was disappearing, and made the best out of their technological expertise (unfortunately Kodak, which had a bigger lead, was occupied by HP executives who were passed over in the quest to destroy HP and proceeded to destroy Kodak instead). The new Fuji expands and explains the engineering design ideas that were present in the X100, the best camera ever made. Optics, electronics, materials science, mechanical design, signal processing, software and interfaces - it's all there. Read about it!
This all happened in the last year or so. Before that, both Sony and Fujifilm produced nondescript cameras. Olympus and Panasonic have the interesting 4/3 standard, but Sony and Fuji seem to be far ahead of everyone else. Nikon and Canon may be waking up - they have great engineering, but still fear that good designs will cannibalize their market share.