Just today DpReview posted data for the m4/3 ultra-wide zoom from Olympus, the M.Zuiko 9-18mm f/4-5.6 ED. I have mixed feelings about the results and what they mean for serious photographers who would adopt the lens. I thought I might share this with you.
I’ve long been an enthusiast of what I will call ideal lenses. I call them ideal because they behave in an idealized manner, as if following a mathematical rule of thumb. The most widely used ideal lenses are the old spherical classics.
Wide open the ideal lens is not a terrific performer, with lower than optimal contrast and micro-contrast, but with generous dosage of softly-rendered defocused regions. As you step down the ideal lens its contrast and micro-contrast eventually renders beautiful detail in all its glory, as good or better as the latest imaging devices can record. While the quality of the image detail improves, the quality of the Bokeh becomes more neutral.
The relationship of the ideal lens to its aperture setting is one of direct proportional improvement of technical qualities towards smaller apertures. Up to a point of course. That point being the diffraction limit for combination of the lens and the imaging device – most commonly f/11 for current devices.
The old classic designs named Gauss, Biogon, Sonnar and Tessar are all ideal lenses in their native spherical implementations. Their characters change appreciably as you turn the aperture ring through every click. The photographer effectively gets multiple lenses in one of these, much like a zoom, with multiple choices of rendering style to apply to taste.
Lastly, the peak quality that a Gauss or Biogon design can deliver at f/8 is simply astonishing. This is one of the most solid virtues of the ideal lens: no matter the exotic new lenses that become available 30 years from now, a current-day Biogon will still stand among the best of them when it is stopped down. The ideal lenses can serve for a lifetime.
Enter into our tale the new-generation Japanese lenses in general, and the M.Zuiko 9-18mm in particular.
These are computer-generated designs with up to 3 or 4 molded aspheric elements that reduce the design by as many as 10 or 12 elements as a similar spherical design would employ. In my opinion this sounds like a cost-saving measure because with such small maximal apertures the spherical aberrations are not likely that difficult to control.
We can plainly observe from the test results that the 9-18mm zoom is not an ideal lens. It is at its best when wide-open and in this case it is not hyperbolic flattery. You start with what I’ll call average performance at the beginning of the aperture series, and then there is no more quality to be got from it no matter what you do with it. If the ideal primes are like zooms in character, this un-ideal zoom is like a prime in character.
My hypothesis is that the reduction in the number of surfaces made possible by the multiple aspheres (note: a pull away from the Zeiss design-relaxation philosophy) has the side-effect that the complete extents of all surfaces is needed for the lens to be properly corrected. As you step down the 9-18mm you may be robbing from it the surfaces it needs to produce corrected images. This is my poorly-educated hypothesis that attempts to explain the data.
How to judge the value of the un-ideal lens then? I see it as very limited to be honest with you. The 9-18mm fills a need that few other lenses do – exactly one other lens actually, which has similar issues but to a lesser extent. But in the long term both of them are likely to be forgotten, replaced by enhanced versions of themselves ad infinitum.
The truth is that the small sensors (by film standards) in the m4/3 cameras mean that ideal designs are not pre-existent on the market, you cannot borrow a 9mm ideal lens from a different system because none exist. It will be up to Olympus or Panasonic to come up with these designs, or their market to become important enough for Zeiss to step into that niche. That could not be a Biogon design, but it could be a Distagon.
Either scenarios seem extremely unlikely to me however. People who care about ideal lenses also care about larger sensors. For no better reason than that, and it is a poor reason, ideal lenses are tied to the old 24x36mm format.