Between Shock and Surprise.
08/27/2011: Wide enough for ya? I'm one of those people who thinks that if wide is good, wider is probably better. I've never been satisfied with how wide an angle I could get in the photographic frame. This wide-angle fetish goes way, way back. When I was 13 (1967), I asked for a 28mm lens for my first SLR, a Praktica FX3. My dad selected a Lentar 28mm F3.5 and my parents wrapped it up for Christmas. I found it in the back of their closet in mid-November and used it discretely until the middle of December when I slipped it back into the box so it would be ready to be wrapped for the tree. When I opened the present on Christmas morning, I managed a pretty fair approximation of something between shock and surprise. I still have it somewhere. "Back in the day" (as photographers of a certain age who have spent a lot of time with film are wont to say) I put a 21mm Vivitar, an 18mm Spiratone, and a 20mm F3.5 Nikkor -- the shortest, sharpest, widest-angle lenses I could find and afford -- on a Nikon F. There was always more sky or more sea or more foreground or more something that didn't quite fit. Came digital, and stitched panoramas offered some relief, but they are a lot of work to get just right. Casual they are not; if you act as if they are, you are sure to miss part of the world that you really wanted in the frame, and there you are with a hole in your whole. My 10-22mm Canon EFs lens (which might as well be a 10mm lens, because that's how it's used on my 50D 99% of the time) behaves as a 16mm would on a 35mm body. It's reasonably sharp, and reasonably wide. But "reasonably" translates as Sometimes Not Wide Enough. If the 10-22mm was close to the widest optic available for my cropped-sensor cameras, a wide-angle lens on a full-frame body seemed like the best way to go wider. It's a nice plan, but a Canon 5D Mark-II and a 14mm Canon lens (or a Nikon D700 and a 14mm Nikkor lens) comes in somewhere on the high side of 4.5k. Ah, but what about the original 5D? My what big, quiet pixels it has! The better for astronomical photography. And my, what a big, big sensor it has! The better for wide fields and shallow depths of field. And being far from the cutting edge, what big, big savings it offers. Used, a good 5D costs a third as much as a new 5DMk2. As for the lens, there are some surprising alternatives to Canon's big glass. There's a 12mm to something Sigma zoom that works with FX-sized sensors but gets mediocre reviews for everything except its extreme field of view. It's middling expensive and optically slow. My fetish for wide does have its limits: extremely wide, sharp and fast trumps unbelievably wide and not-so-sharp and slow. The Rokinon 14mm F2.8 gets rave reviews, and compared to Canon's and Nikon's similarly wide and fast lenses it might as well be free. Still, "Might as well be," isn't quite "free." What's free is what's already in the equipment cases. With the aid of a cheap adapter, the 16mm Nikkor fisheye I've had since Days o'Film mounts on my 5D very nicely. It's been sitting right there in the seldom-used drawer, next to extension tubes and I barely remember what all (steel developing tanks with spiral film reels will be showing up on "Antiques Road Show" pretty soon). There's this software package, PTLens (shareware, about $25), which rectifies various aberrations in the lusted-for 14mm lens (as well as in many, many more) and handles fisheye distortion nicely enough, too. Look at these three frames:
Now, here's the thing. A certain amount of barrel distortion is OK (except when it isn't) so the trick is to find the right settings in PTLens to give me the widest possible field on a 5D using the old Nikon glass while not looking too "fishy." One caveat right out of the box: the 16mm yields beautifully sharp stars that get compromised when defished. For razor sharp, really widefield astrophotos, live with the distortion. It's a suite of glass and software that needs to be worked with some to see what works and how well. We all know there'll be a Rokinon in my kit one day soon (it just can't be helped), but in the meantime, this will keep me entertained. How sharp is the rectified 16mm fisheye? Somewhere between pretty and very. At right is a full-res crop from near the center of the image (yeah, with some JPEG artifacting; this is a casual look). Edges and corners are less impressive, naturally, but there are switches to flip and dials to turn in the software and on the hardware before knowing just how good or bad the edges are. They're tolerable as is, not as good as my best glass, nowhere near as bad as the worst.
8/28/2011. That romance lasted about a day and a half. This morning, after more experiments and a fresh critical look, I'm thinking that to control edge defects adequately I'll end up cropping the sides severely. I can have fisheye frames that are nearly 180° wide, but when remapping the image all the way to a rectilinear projection, pixels in the outer 30° or so get stretched far enough that aberrations are greatly magnified. Chromatic faults that are otherwise unobjectionable become gruesome. CA is usually easy to fix in Photoshop or in other tools (including PTLens), but so far these particular CA's have resisted repair. Throw away 30° on either side and you're left with a 120° frame -- comparable to a 14mm lens, but comprised of one third fewer pixels. Less extreme remapping, remapping that leaves a substantial amount of the fisheye's characteristic barrel distortion, leaves the edges crisper and works well for some subjects. When true rectilinear images are called for, this is not the ideal way to go. Put this lens and workflow in the toolbox as something to use when appropriate, but I'm thinking it won't become a regular way of working.
9/1/2011. In the recent history of photography (say the last 60 years or so), certain lenses have attained cult status. They're sharper, sweeter, faster, better, more something than the rest. The Leica 50mm F0.95 Noctilux comes to mind; the 6mm F2.8 Nikkor Fisheye; the 105mm F2.5 Nikkor "portrait" lens; the 180mm F2.8 Nikkor (everybody's astrograph); the 55mm F3.5 Micro-Nikkor... Hard-boiled photojournalists would have a different, maybe longer list. People who came of age with something other than a Nikon grafted to their eyebrow would name more Canon, Zeiss, or Leica glass. Cult lenses are usually (but not always) expensive and hard to obtain. The Leica nightglass and the 6mm Nikon fisheye are about as scarce as scarce can be, but there is nothing the least bit rare about some of those Nikkors, nor were they particularly expensive. I'm thinking the modern 14mm F2.8 Samyang (or Rokinon, or Bower, but not Bell & Howell) ultrawide is destined to join the hallowed few. The lens was first released in flawed form, and it was widely reviewed as a good "budget" lens -- capable if you couldn't or woudln't spring for the vastly more expensive Nikon or Canon OEM ultrawide lenses. You gave up all electronics and auto-anything, but for ten cents on the dollar, you got the same field of view and much of the central sharpness. The lens made little sense on crop sensor cameras (the 10-22mm Canon, for example, works just fine there). On a full-frame camera, edges and corners suffered some and a lot, respectively. The lens tended to pick up flare from any bright light or even from a bright sky or a sunny window. The lens had a lot of distortion (so-called mustache distortion: the central portion of the image showed noticeable, not just measurable, barrel distortion while the outer zones showed much less; because of the changing projection, a straight line passing above or below the center of the image acquired a disturbing resemblance to the hair gracing the upper lip of Snidely Whiplash, see below). If the original Samyang was merely OK optically, its construction was robust. Only a worrisome plastic lens mount gave away its budget-conscious nature. For the money, who would fault it? Few did (here's Ken Rockwell's underwhelmed review; note that though his photos show the UMC version, I'd bet much that he's reviewing the older lens; and here's something odd: the focus ring in the photos of the 14mm F2.8 Samyang ED AS IF UMC he shows rotates in the opposite direction from mine -- does Samyang reverse the threads to match the focusing custom of the manufacturer of the lens mount? Do Tamron and Tokina et al do that? After 40 years of doing this stuff, have I just never noticed? Should I admit that?).
Anyway. The optical and mechanical engineers of Samyang evidently noted the reservations in early reviews. They added or improved the antireflection coatings on several surfaces to fix the lens's proclivity to flare; they put metal parts where they needed to be; and either by adjusting internal spacing or changing surface figures, they addressed the edge sharpness issues. Didn't they though! The second edition of the lens, flagged "UMC" (for "Ultra Multi-Coated"), costs 25-30% more but still comes in at a small fraction of the price of Nikon and Canon primes. The new version displays even more mustache distortion, but it is in every other way a dramatically improved product. Handed the newer lens, reviewer after reviewer, with or without quantitative optical assay equipment, expressed something between shock and surprise at its performance. Here's one good example that also includes a link to a similarly detailed review of the original version. The lens is sharper wide open than either the Nikon or Canon versions even when those pricier alternatives are stopped down. Furthermore, the Samyang's images are nearly as sharp at the edges and in the corners of the frame as they are in the center. Out toward the periphery of the frame, the "budget" lens was significantly sharper than the OEM primes. This was almost unheard of in any lens, especially in a scary-cheap, optically fast, exotically wide-angle one.
Flare, sharpness, and mechanical concerns are no more, but the severe and complex "mustache" distortion not only remains but has become worse in the newer lens. It is not particularly distinctive of the Samyang. Modern big name 14mm primes and ultra-wide, ultra-expensive zooms also show some of the same distortion (not as much, but still...).
14mm @ F2.8, Canon 5D
The most serious flaws with the Samyang / Rokinon / Bower UMC lenses seem trivial: the focus scale is off, and I quibble with the long throw of the focus ring. Open the lens to its widest aperture, spin the focus ring to its stop at the infinity mark, and objects at infinity simply are not in focus. Ten minutes of experiments under a starry sky confirmed what I'd read from several reviewers: the best infinity focus is found near the 7ft / 2m mark. What matters much to astrophotographers may matter less to photographers whose subjects are exclusively terrestial: stop this lens down to F5.6 or F8, turn the focus ring to anywhere on the far side of 5 feet, and the world farther away than your outstretched arms will be razor sharp. All of it. If you want to take full advantage of its faster F-stops -- and with this lens, you might as well considering that it's deadly sharp wide open and you've paid for the speed, at least a little -- you'll need to take a little care to find and keep the point of best infinity focus. Casually running the focus out to the infinity stop just won't do. And you have to wonder why there is so much focus travel. The focus ring turns a long, long way to get from one stop to the other. It's a far longer, slower, finer trip than it needs to be given the immense depth of field of super-wide lenses.
Practical concerns & caveats: having fallen in love with the idea of this all-manual superwide super-affordable super lens, I became concerned that it would be discontinued or changed. I thought, if it's really that good, the only way forward is down: someone will cut a corner or second-guess some aspect of its design or production and its performance will suffer and then how will I find one of the good ones? If you want one, I thought, better buy it now. Paranoid or experienced? You decide. I first ordered the Bell & Howell version from BuyDig at a suspiciously low price. I soon discovered that other photographers' had bought the lens at the same price under a few different names (Bell & Howell among them). Some concluded that the cheaper variant was really the recalled first edition. Their universal advice was to spend a little extra money for the revised "UMC" lens and never look back. I cancelled my order from BuyDig (who handled my request quickly and gracefully) and then tried to reorder the proper lens. Unfortunately, BuyDig's catalog descriptions of ultrawides in Canon mounts left room for doubt: was the lens they offered at the higher price really and for sure the second version? I ordered from tried-and-true B&H Photo who clearly offered the newer lens at the same price as BuyDig. Even with free shipping, it arrived in less than 48 hours. I snapped a few casual photos, loaded them on the computer, and enjoyed that now-familiar sensation between shock and surprise.
9/4/2011. The Harris Brothers. Reggie & Ryan at the Happy Valley Fiddlers Convention. Reggie needed a wide shot that could be turned into a banner for a promoter's website. We did something quick since the boys were due at Beech Mtn for another show in just a couple of hours. A little HDR-style work let me keep the clouds and the shadow detail (I unpacked the CR2 RAW file twice, once with plenty of fill and once with exposure knocked down to get the sky):
Canon 5D, 14mm F2.8 Rokinon
You might notice a curved top edge to the suitcase / amplifier. That's my doing, not the lens's. I'd moved in very close to make sure Reggie and Ryan were large enough in the frame, and things got a little out of hand. I applied a quick fix in Photoshop to remove some of the perspective from the subject-area of the shot. Could've been done better. |
:: top ::