|
|
MC Zenitar 16mm f/2.8 Test Images
Sarah Fox
It is said fisheye lenses were cool back when bell bottoms were groovy. There might be some truth to that, at least with respect to the circular fisheyes (a.k.a. "tiny world" fisheyes) that portrayed a 180 degree hemisphere in the middle of the frame. However, "diagnoal" fisheyes, which map a 180 degree field of view from corner to corner of the frame, still provide a very useful geometry in certain situations. For instance, I recently considered how to shoot a large group of people in a small space, putting great emphasis on one central figure in the center of the frame. An ultrawide angle lens would convey the somewhat surreal perspective the particular shot demanded. However, I could hardly use my Sigma 12-24mm lens for the job. Indeed, that lens has a rectilinear geometry and would stretch the heads of everyone appearing in the margin. It would also diminish the emphasis on the central figure, since the stretched margins draw the eye. On the other hand, a fisheye lens would keep shapes relatively correct and convey a better proportionality, at the sacrifice of straight lines that would become curved. I could of course do a rough fish-transformation of a rectilinear image by spherizing it, but that would degrade the image, limit the field of view, and ultimately not achieve what I would really like to achieve. A fisheye lens is the best solution.
Of course a fisheye is best used sparingly. Because it is such a specialty lens, it perhaps does not deserve the heavy investment that woudl better be applied to more commonly used optics. If there were ever justification for a "cheap" lens, this would be it. Fortunately the Russian made MC Zenitar 16mm f/2.8 fills this role beautifully, at least on a full-frame 35mm camera. Build quality, frankly, is quite bad. The EOS mounting flange is too thin, so the lens fits loosely. The miniscule sunshade protrusion is often misoriented. (It was on mine.) There are sometimes even focus problems that must be corrected by re-aligning the focusing threads. However, ignoring those shortcomings, the lens costs only $200 new, and it is surprisingly sharp for a fisheye.
Ken Rockwell has reviewed this lens here. (Also see Bob Atkins' lens review here and NK Guy's review here. It is useful to compare my lens with Rockwell's (a lens he actually borrowed), as he shows numerous 100% crop images at the lens' various aperture settings. Upon the basis of these images, he concludes the lens has a superior image quality to Nikon's 10.5mm diagonal fisheye at any aperture if it is stopped down to f/11 or more. He says the lens has minimal or no ghosting and no chromatic aberration. It's only natural that different lens copies exhibit slightly different characteristics, so it is not surprising that my images differ slightly from his. My lens is in some respects better than his and in some respects worse. It is for this reason that I am posting my own test results for others to view.
First let's look at the test images, and then I'll offer my interpretations. It might be helpful to compare side by side with Ken Rockwell's test images, which are shot with the same camera/lens and displayed at the same 100% magnification. Click here to open up another window with Rockwell's images. I shot my test images in the back yard with my Canon EOS 5D -- nothing elaborate or remotely artistic! Please note that the dull colors are more a matter of drought than anything else -- no fault of the lens. I shot what I consider one of the least forgiving and most informative subjects -- long needled pines. Here is the scene at f/16, reduced to 20% magnification, followed by both center and corner details at 100% magnification at each available aperture. I have shown the 100% crops both unsharpened (either in camera or in postprocessing) and sharpened (in postprocessing only):
The first thing I notice is that the lens is indeed rather soft wide-open. That's not particularly unusual for a lens. However, I do not find the same "comically soft" properties that Rockwell describes. My center 2.8 100% crop sample is very similar to his. He says that his test image doesn't even begin to show how soft the lens is at that aperture. However, I would say that my test image shows the wide-open softness fairly accurately. It's not horrible and unusable to the extent that the lens would be useless for shooting wide open to create a smaller image. The second full image, above, shows the lens at its worst, wide open, with sharpening in postprocessing. Although the image is not razor sharp, it is nevertheless usable for some applications.
On my lens copy, f/4 is vastly superior to f/2.8 in center detail, and f/5.6 is slightly better still. Above f/5.6 there appears to be little difference across apertures in the center, up to f/16 where diffraction effects are barely visible and f/22 where they are a bit greater still. The sharpest apertures are f/5.6 - f/11. It is noteworthy that my lens copy does not seem quite as sharp as Rockwell's at f/16, unless I apply sharpening in postprocessing. I do not know whether Rockwell's test images were sharpened.
Unlike Rockwell's lens, my lens performs relatively well in corner detail throughout the aperture range. Oddly, corner detail seems a bit better than center detail wide open, although it is still somewhat fuzzy. Sharpness gets slightly better at f/5.6 and f/8 and is relatively constant until the diffraction falloff, which just edges in at f/16. This makes the Zenitar 16 very similar to many other lenses that reach their peak sharpness 2-3 stops from their wide-open aperture.
  Rockwell's lens and mine differ in a couple of other respects, besides resolving power. He says (and shows) that there is very little ghosting with his lens copy. On my own lens copy, there is a considerable flair/ghosting issue. The ghosts take the form of large, indistinct blobs, and they are apparent both in sunlight and indoor lighting. They reduce the contrast of the lens overall, but not as much as I've seen in some other lenses. Considering the wide angle of this lens, a bit of ghosting and flare are not surprising at all. Second, he shows that his lens has no detectable chromatic aberration. My copy, on the other hand, shows a bit of chromatic aberration -- much less than Rockwell's Nikkor, but certainly more than his copy of the Zenitar. This defect can be seen, for instance, in the branches of the pine tree at the top of the test frame, as shown in this 100% crop. Note the blue fringes on the undersides of the branches.
Unlike some, I am not particularly impressed with the build of the Zenitar. I note the following issues: (1) As mentioned above, the sunshade on my copy was misaligned, leaving small "nicks" at the upper right and lower left corners of the images (see test image). I was able to correct this by loosening three set screws with a precision screwdriver, turning the shade about 5mm to the right, and then retightening. (2) The EOS mounting flange is very thin and loose, and the flange falls sufficiently short of its ideal diameter as to give the lens a tiny bit of lateral wiggle in the mount. The mount wiggles when either the focus or aperture rings are turned. (3) There is no alignment spot for mounting the lens. (4) When unmounting the lens, it is possible to continue turning it counterclockwise well past where there should be a detent. My lens came with a missing screw. Perhaps that was the detent? (5) I find the aperture ring difficult to turn. This may be the result of parts that fit too snugly, but it also relates to bad ergonomics, with a ring that's frankly difficult to grasp. (6) Finally, the distance scale isn't quite accurate. Actual focal distance is approx 75% of what the scale indicates. I think all these defects are all indicative of poor machining tolerances. I would speculate that the same loose tolerances are evident in the spacing of the lens elements, which could contribute to the marked differences between the image quality of my lens and that of the one Rockwell tested..
Finally, many, including Rockwell, claim that it is extraordinarily difficult to shoot with a manual lens on a modern camera. I would love to perpetuate this myth, so as to keep demand down for the manual optics I often like to buy. However, in all honesty, it's quite easy to use a manual lens, particularly one with such enormous depth of field as a fisheye. First the auto-exposure: I simply switch my 5D to its Av mode, set the aperture I want (using the aperture ring on the lens), focus, and shoot. The camera meters the shutter speed automatically. There! That's not complicated, is it? When the camera "sees" nothing electronic on the other side of its lens interface, it defaults to stop-down metering, which recognizes that the lens cannot be controlled by the camera and that the camera must adjust shutter speed to deal with whatever light it "sees." For more on stop-down metering, including a discussion of older stop-down metering modes, please see this excellent discussion.
Focus is almost as easy. You have four choices. Use whichever one sings to you: (1) Open up the lens, set your best focus with a critical eye, stop down the lens, and shoot. (2) Install a split-circle focus screen in your camera to set your focus optically, just like in the good old days. You'll still need to focus with the lens wide open and shoot with it stopped down, but your focus will be more accurate. (3) Go by the numbers. If you're planning on shooting at f/11, for instance, you're going to have a huge depth of field anyway. Just guesstimate your distance, and set it with the focus ring. You seriously won't do much better, and this would be my method of choice with this lens. (4) Use an M42 version of the lens, combined with an M42 to EOS (or whatever your camera uses) adapter with the focus confirm chip. Then focus as you would when using a Canon EF lens in manual focus mode. However, be aware that the chip will defeat the stop-down metering. Also be aware that the chip is attached with epoxy and that it can fall off into your mirror box, causing damage. I wouldn't recommend doing it. (Note: be careful of mirror clearance on a full frame EOS camera, as there is one report of an M42 adapted lens binding the mirror.) You can also epoxy a focus confirm chip directly to the back of the EOS version of the Zenitar lens, but alignment might be difficult. The best solution (5) would be for Canon to stop disabling focus confirm when manual lenses are attached. Their continuing to punish us for using non-Canon EF lenses is beyond annoying.
All in all, the Zenitar is a pretty good little lens. Its best is maybe not quite as good as Rockwell's best, but its worst is far and away better than Rockwell's worst. I would estimate its optimal range at f/5.6 through f/11 and its usable range at f/4 through f/22 (and f/2.8 if the final image is to be fairly small), not at Rockwell's more narrowly constrained f/11 and above. Its mere $200 price tag well befits its specialty use. If it were $600, like a Canon or Nikon fisheye, I simply wouldn't buy it, no matter how good the image quality. Such a seldom-used lens is simply not worth that sort of money. I also wouldn't bother with it if the image were junk. However, this is one of those nice little bargains that is simply too cheap to pass up.
|
