Above: Zeiss Jena 5cm f/1.5 Sonnar (prewar; 1937 example of the 1932 design) on a Leica M typ 240 with an Amedeo dedicated 50mm adapter. This particular lens is almost 80 years old.
1. The story
The derivation of the trade name “Sonnar”(which may have less to do with Sonne than being a portmanteau of Sontheim am Neckar) reminds one of the the way that the Mr. Sparkle is a joint venture of the Matsumura Fishworks and the Tamaribuchi Heavy Manufacturing Concern. Be this as it may, the Sonnar had but one goal in life: crush Leitz’s fast lenses in an era where ISO 12 film was the norm. And that it amply did. Even today, the performance of this uncoated lens is impressive.
When the Sonnar arrived in 1931 (f/2.0) and 1932 (f/1.5), the Tessar (or Elmar) was the gold standard in normal lenses: a well-corrected triplet that, in an era lacking anti-reflective coatings, sneaked in a little more correction by cementing two pieces of glass together. When it came time to exceed f/2.8, though, the real competition began:
- In 1889, Paul Rudolph, working for Carl Zeiss, determined that the best balance of contrast, correction, and cost was a three-element lens called an anastigmat (trade name: Protar).
- In 1895, Rudolph invented the Planar, which was a highly-corrected symmetrical lens. It was shelved soon thereafter, no doubt on account of the low contrast that occurs with many air-to-glass surfaces.
- In 1902, Zeiss released the Tessar, which provided more correction than an anastigmat (by adding a fourth element glued to the the third) without increasing the number of air-glass surfaces. The Tessar was technically inferior to the Planar, but it did not have the two extra air-glass surfaces (each robbing 10% of the light, compounded).
- In 1925, Max Berek modified the Leitz Elmax, which had a 1-1-3 (cemented) arrangement into the Elmar, which bore a heavy resemblance to the Tessar, allowing for a good 35mm-format lens with fewer elements and less assembly labor.
In parallel universe (but still orbiting around Zeiss)
- In 1916, an American (Charles Minor) started adding elements to the triplet, but just in the front. The result was the Gundlach Anastigmat, which had a blazingly fast f/1.9 aperture. The contemporary ads show that this was actually a cine lens.
- In 1922, Ludwig Berthele, working for Ernemann (of Ermanox fame) continued elaborating this into the Ernostar, which became on of the first plate lenses to hit f/1.8 (in 1924).
Scan used by permission of Peter Naylor.
- In 1926, Zeiss bought Ernemann and acquired Berthele in the deal.
- In 1931, Berthele made the first f/2 Sonnar, which was a new lens with an old name. It was for 35mm format and had a 1-3-2 arrangement, with the second and third group cemented together.
- In 1932, he made the f/1.5 version, which added an extra element to the rear group.
- In 1936, caught off-balance, Leica licensed the Xenon, a symmetrical Double-Gauss design from Schneider, licensed in turn from Taylor-Hobson in England (the Series 0), in turn had been cribbed from the 1896 Planar.
- In 1944-1945, the Zeiss plants were bombed back to the stone age.
- In 1949, the Xenon was updated with coagulation-style lens coatings and became the Summarit.
- In 1950, the Zeiss-Option Sonnars came out with a new computations.
The circle was now complete: the entire high-speed lens space was dominated by Zeiss designs and would continue to be – for pretty much all time. When you stop and think about it, until the advent of things like the 50/1.4G Nikkor, the history of high-speed lenses had been nearly nine decades decades of Sonnar and Planar clones.
Why did the Sonnar do so well? It’s not so complicated. It all boiled down to the number of air-to-glass interfaces. The classic triplet (the anastigmat) represented the best balance between correction, contrast, and cost. But adding more elements (to get more correction) meant more air-glass interfaces. And that meant less contrast and more flare. Zeiss increased the correction by cementing additional elements together to make a total of three groups. Leica could not do this because it did not have the intellectual property rights to do so. During WWII, Zeiss dabbled in coating its super-speed lenses, but it was not even really necessary given the Sonnar’s high transmission.
2. Using one today
These days, the Contax rangefinder is almost dead, 35mm film photography has gone all “Tony-Bennett-in-the-late 1990s,” and so the only place you’ll likely be using one of these is on a Leica body. Fortunately, it’s pretty easy to do. You just need the appropriate adapter. These are not particularly expensive for APS-C (although they do incorporate focusing helicoids); they are more expensive for Leica cameras because they need a mechanism to translate the movement of a 52.3mm lens to a camera whose rangefinder mechanism wants a 51.6mm normal lens (how two German companies known for their precision could get so sloppy about what constituted a “50mm” lens is baffling – but being a big-name German optical company means never having to say you’re sorry….).
By far, the best adapters for Leicas are made by Amedeo Muscelli, and of those, the best is that dedicated adapter for Contax 50mm rangefinder to Leica M. This is not the usual lens with the reproduction of a Contax helicoid and focusing scale; rather, combines with the lens to make a unit that looks a bit like an old Elmar (allowing, of course, for the streamlined – dare we say phallic) shape of a Sonnar. The dedicated adapter focuses in the same direction as a Leica, at almost the same rate of distance change per unit of turn, and it has a lever, which can be critical if you are using a collapsible Zeiss lens (since with a traditional adapter, you are grasping the lens barrel to focus – something you can’t do with a collapsible lens). When your lens is dialed in, this adapter focuses amazingly accurately right down to 0.6m – a lot closer than any Contax did.
And how do you dial one in? If your lens is front-focusing, the simple answer is to remove the lens cell from the Contax barrel and unscrew the rear group slightly. It is never more than 1/4 turn, and you can maintain the setting by wrapping the threads of the rear group in Teflon tape and screwing it back in. Back it out about a 1mm (circumferentially) at a time, and check the focus on near and far objects. Do note that where a Sonnar has a lot of focus shift, you’re going to have to choose whether to
- Have the lens front focus at f/1.5, reach focus at f/2.8-4 ,and hit focus at f/5.6 and smaller
- Have the lens focus dead-on at f/1.5, miss at f/2-4, and become usable again at f/5.6 and on.
The first observation is that finding a prewar f/1.5 Sonnar that is not totally trashed is not particularly easy. Fortunately, at least cleaning marks are not an issue on uncoated lenses unless someone used Soft Scrub as an optical cleaner. Which does happen from time to time.
The second is that in the central part of the frame, this lens is very, very sharp. It has decent performance at f/1.5 if you optimize for that aperture, loses precise focus from f/2.8-f/4, and comes roaring back at f/6.3. If you keep with the original collimation (or an approximation of it, you get really sharp pictures around f/2.4, getting better through f/8.
The third is that the coatings on postwar Sonnars are not moving the ball much in terms of performance. Because this was the last fast Sonnar I obtained, it’s easier to compare this to the 1961 Carl Zeiss version. The 1937 model performs similarly in most ways. It is very slightly softer, with contrast that is almost at the level of a 1977 Jupiter-3.
Flare is only slightly improved by coatings, and they do not resolve the “rainbow circle” flare that afflicts every Sonnars (even multicoated Sonnetars) when a point light source is just out-of-frame. The one unique failure mode is strong side lighting (from the looks of it between 75 and 90 degrees to the lens axis), which can cause a veil across the entire surface. This also happens to a lesser degree with postwar Sonnars and copies, just not quite to the same degree.
Overall performance is strikingly close to the postwar, if you allow for slightly improved spherical aberration on the older lens. The postwar version is a tiny bit sharper, but seems clear that this comes at the expense of bokeh, which goes from smooth disks to ringed disks. If you care about that stuff.
4. Roy Batty
The f/1.5 Sonnar was the proverbial candle that burned twice as bright, and by 1962 it was essentially extinct. The “twice as bright” part is doubly applicable to the 1960-1962 Car That it was so widely copied in the postwar era is puzzling. Granted, German patents were handed over to the Japanese, but in terms of sheer performance with coatings, there were already better lenses to copy (like the Xenon). Canon, Nikon, and Zunow all made their own versions. The Soviets made one too. Perhaps there was a “prestige” element to the Contax that was desirable to copy (though you would not have the all-important brand name). Or perhaps there was something about the mechanical design of a 3-group lens such that the cost of machining extra parts for 6 groups cost more than triple-cementing rwo groups. The world may never know. The fetishization of the Sonnar did not really get started until the mid-2000s and by then, it was based more on imperfection and “look” than a perception that it was actually better.
The prewar f/1.5 Sonnar is a worthy lens, though its relative scarcity does not exactly make it a value leader compared to postwar variants. As with any 50mm Sonnar, as long as you take care to control the placement of light sources, it can be another creative tool, if not a broader-use lens.
This lens is perfectly usable on the M240. It doesn’t even take that much work.
The Leica M typ 240 presents some unpleasant choices in terms of 21mm lenses: you can spend $3,000 on a Super Elmar 21mm 3.4 and get the sharpest 21mm ever made for Leica – but suffer complex distortion and red edges. The 21-35mm M-Hexanon Dual (which is not a lot cheaper these days than a used Super Elmar) gives you two focal lengths, awesome sharpness and no color shifts – but it gives you a touch of geometric distortion. Everything else presents varying combinations of bulk, color vignetting, low resolution, and general misery. Here at the Machine Planet, we have a certain inbuilt arrogance to try things that conventional wisdom says should not work. The 21mm f/4.5 Biogon is a case in point. And yes, we made it work with a couple of off-the-shelf tools and less than a couple of hours of trial and error learning the ropes.
The good. If this were the film era, the 21mm f/4.5 would be the champ. It is small (barely bigger than a 40mm M-Rokkor), sharp (testing in some reviews to 3000 lines per picture height), well-made, and has about as close to zero distortion as any wideangle lens ever made (for example, it’s lower than the 35/1.4 Summilux ASPH). It also takes normal-depth 46mm filters common to the rest of your lens collection. Here is basically everything you need to know about its stunning performance:
The bad. In terms of conventional performance, the lens is relatively slow in terms of maximum aperture and has the usual light falloff from the center, often exaggerated by digital sensors. You can see from the chart above that it does not get much better as you stop down.
The ugly. The worst thing is that the lens has color shift at the edges. It’s quite severe at first glance. These are the particulars:
- The red edge extends a couple of MM into the frame, from top to bottom, green and red on the left and red on the right. In the days of the Kodak DCS Pro 14n, this was called the “Italian Flag” effect.
- The intensity and intrusion of the edges is dependent on selected lens aperture and focused distance. Closer focus and wider apertures mean that the edges are far less obtrusive.
- There is an overlay of standard brightness vignetting that is characteristic of any symmetrical 21mm lens.
The variable nature of the color shading – why has no one else noticed this? – may well be the cause of claims that the problem “can’t” be corrected or that conventional tools result in under- or over-correction. Once you understand this, it’s easy to solve the problem. Never declare defeat prematurely!
Fixing things up. All solutions to this problem involve some kind of reference image, which is a test shot you make using a white field. You can shoot a white wall, shoot a ceiling with a flash, or use a diffuser. If you shoot through a diffuser, you need one that lacks texture (at small apertures, the ZM 21 can pick up the texture of the paper, even if you have it pressed right up against the lens. Your resulting references will look roughly like this:
One very good diffusion material is Yupo polypropylene watercolor “paper,” which, being plastic, has no grain. You can find this in most art stores.
- Layer Masks. Some, like Lloyd Chambers, advocate the use of Photoshop adjustment layers and masks to cancel out color and brightness shading. Although this demonstrably works, its shortcoming is that it needs a separate template and action for every permutation of shading (you can, most of the time, get away with four settings: f/4.5 and f/8, at 1m and ∞). It also presents a clumsy workflow that involves leaving Lightroom, going to Photoshop, and then back to Lightroom (and at that point, with a TIFF and not a DMG). For your most OCD applications, this is a workable solution; it’s just not the most batch-friendly or space-efficient solution.
- Cornerfix. Long the go-to solution for Leica M8 and M9 users, Cornerfix was originally designed to address the green shift that occurred when you put a UV/IR filter on an M8. This green shift was generally uniform and radial. Cornerfix takes the reference imageand then computes a mathematical mask. Cornerfix works with DNGs and exports DNGs (suffixed “_cf”)Unform, and it has a tremendous range of settings for addressing color shift, brightness vignetting, and the artifacts that result from correction. Cornerfix also shows you the effect of the selected mask on the current image. It also supports batch processing. The shortcoming of Cornerfix, though, is that because it does correction via equation, there are some kinds of color shading that it struggles with.
- Adobe Flat Field plug-in. The strangely named Flat Field plug-in is available on the Adobe Labs site. This plugin has virtually no controls and seems to be an automated variant of the layer mask technique. You select the image you want to correct, activate the plug-in, and then give it the reference image. The only controls are for “Color” or “Color and falloff,” which lets you leave in brightness vignetting if you want. The plugin is slow and kicks out another DNG, stacked with the first one, suffixed “_ff.” It does work very well – much better with the 21mm than Cornerfix – and it does not require you to exit Lightroom, but it’s a black-box solution that requires you to select your reference image carefully (because you can over- or under-correct by choosing the wrong one).
The winner: Flat Field. As the only solution that (a) works and (b) does not require shifting from program to program, Adobe’s free Flat Field plug-in for Lightroom is the best solution. Here is precisely how to use it:
- Shoot your profiles. Take your sheet of Yupo paper, hold it right in front of the lens (the easiest way is to sandwich the paper between your lens and the glass of a window). Pick your reference distances. We used 1m, 2m, 5m and ∞, but you could also pick your favorite hyperfocal distance. Shoot a test at one f/stop, all distances. Then switch to the next f/stop, all distances.
- When you are done, import the files into Lightroom. Immediately rename these with a designator that shows lens, aperture, distance. This will result in a name like “2145-80-inf” for 21/4.5, f/8, at infinity. Export all of these as original DNGs to a folder that is easy to find (think about “profiles” in your “Documents” folder.
- Install the Flat Field plugin.
- When you want to do a correction, select the picture(s) you want to fix. All of the ones you do together should have the same shooting aperture and distance (the M240 records a computed aperture value, and you should be able to tell by the composition where the lens was focused).
- Go to File–>Plug-in Extras–>DNG Flat Field–>Apply External Correction. This will pop up a Finder or Explorer window to select the profile from #2 (Lightroom does not let you choose from the catalog).
- Choose “color and falloff.” Although vignetting may seem cool in theory, symmetrical lenses need all the help they can get.
- Run it.
- You will then get a new file adjacent to the original with the “_ff” suffix. You can now manipulate this as if it were the original.
- If you get too much correction, try a reference photo shot at a closer distance. If you get under-correction, go for a farther distance.
Upshot. It is tragic that so many people started unloading these lenses based on a red-shift issue that is so simple to correct with modern tools. The ZM 21/4.5 is a fantastic optic that can now make the jump to modern digital Ms. And there is no reason why the same techniques could not be used to adapt other wideangle lenses to Ms or wideangle M lenses to things like the Sony A7 series.