Leitz Noctilux 1:1.2/50: the untold story
Introduction
The Leitz Noctilux 1:1,2/50mm is a very famous lens and around its performance several myths have been spun. The Noctilux was introduced in 1966 and stayed in production till 1975. This was the period of the great optical battle between the giants of the optical industry, mainly Leitz, Canon and Nikon. The search for a truly useful very high speed lens was on for some time. Most companies settled for the aperture of 1:1.2 as a sensible limit. The disaster of the several 1: 1.1 designs was well known. If there was a real need for such an aperture may be disputed, but as with the American car industry, that tried to build the most powerful V8 engines, the question was not: why, but how.
The classical 7 element designs, like the Summilux 1:1,4/50mm showed that spherical aberration at wide apertures was a villain as were many lens elements, as they introduced always some unwanted reflections. Leitz and others knew about the idea of an aspherical surface to counter these effects. But these surfaces were hideously difficult to manufacture. But Leitz decided to go that route and designed a six-element design with two aspherical surfaces. The manual manufacture of the lenses with the aspherical surfaces was very time consuming and demanded extreme expertise. Within the Leitz company only one person had the skills to do this job and even he had to discard many lenses as rubbish. The factual production of the Noctilux may ever be known, but a safe guess, based on documents and personal discussions, should be between 700 and 1500. This number makes the lens, any Leica lens, a collectors item and so the prices for the original Noctilux are sky high.
But this lens is also famous for its optical performance. It is claimed countless of times by even the most reputable Leica experts that the Noctilux is a special lens that delivers its optimum performance wide open and on stopping down will deliver unspectacular results. This behavior makes the lens a unique one and sets it apart from all other high speed designs.
This optical behavior is counter intuitive. Most lenses improve when stopping down and only very highly corrected lenses, close to the diffraction limit are at their best wide open. And whatever the qualities of the Noctilux, such a correction is not feasible for high speed lenses. So the question is: did Leitz perform a miracle or are we witnessing the creation of a myth.
The origins of the myth.
The British Journal of Photography published an article on December, 5 1969, titled "A look at very fast lenses", written by the highly reputed Geoffrey Crawley. Here he explained the principles of high speed lenses and their design options and compared the Noctilux with the Canon 1.2/55, the Nikon 1.2/55 and the Minolta 1.2/58mm. In the course of the review he notes that the Noctilux has the best wide open performance, the least flare and the highest clarity of all four contenders. He remarks that the Noctilux has been optimized for contrast as the subject outlines are very crisp and the detail definition quite fuzzy. The Japanese are generally of lower contrast, but here detail definition is much better resolved. He studies the behavior of the rendition of outlines, medium fine detail and very fine detail. In current words, the MTF at 5, 10 and 20 or 40 Lp/mm. He concludes that at full aperture the Noctilux is excellent at 5 Lp/mm but mediocre at the 20 or 40 Lp/mm. The others are good at the mid frequencies (10 to 20 Lp/mm), and worse at the 5 Lp/mm. This is a puzzling conclusion as I have never seen a lens that is very good in the mid frequencies and bad at the lower frequencies. Stopping down the Noctilux does not improve the contrast or resolution markedly and the others do improve such that they now appear to be better than the Noctilux. His conclusion is: "The Noctilux is therefore a lens, whose exceptional qualities are only exceptional at full aperture. The Noctlux continued to give good quality stopped down, but lacks the bite of very fine detail provided by the other lenses stopped down. It has obviously been designed primarily as a f/1.2 lens".
This claim has been copied by every writer about Leica matters and been adopted by users, some of whom even where afraid to stop down the lens. If that were the case, why should Leitz have added an aperture mechanism to the lens, stopping down to f/16. In the past, the Summarex 85mm was restricted to f/8, presumably while performance would drop dramatically (which is not the case by the way).
The test results
I compared the Noctliux 1:1.2/50mm with a Canon FD 1:1.2/55 SSC Aspherical. But first the performance of the Noctilux wide open and stopped down. At 1.2 overall contrast is medium and the central image area with a diameter of 10 to 12mm exhibits a resolution above 80 Lp/mm with good edge sharpness and contrast. The outer zones from 6 mm image height are very soft and while the horizontal line patterns are acceptable, but soft, the vertical lines are very fuzzy, very low in contrast and with strong color fringing. More important is the behavior of the patterns that represent the 10-20 Lp/mm. Here we see very good contrast and edge sharpness in the central portion of the image and a rapid grow of fuzziness at the edges when going to the corners. Still the overall performance does indicate good edge sharpness at the 10Lp/mm, soft and fuzzy edges at 20Lp/mm and a quite fuzzy reproduction of the finer details. This is what Crawley described. In a comparative sense, the Noct wide open is not a superior lens: the Slux at 1.4 is the better lens.
But now the rest of the story. At 1,4 performance hardly increases, but a faint contrast jump can be noted. At 2.0 the finer detail becomes visible, but is still blurred and the edges of the major outlines (10 Lp/mm) improve, but there is still a trace of softness. At 4.0 the image quality gets a boost and now quite detail is recorded with good definition over a large part of the image. Astigmatism is gone now and the line patterns in all directions are crisp and contrasty. Corners and edges are very weak and stay so at all apertures. At 5.6 overall performance improves, very fine detail (100 Lp/mm) are recorded with ample clarity and the 10 -20 Lp/mm are very contrasty and with the exception of the corners rivals the performance of the then current Summicron. At 8.0 the corners now show good definition and this aperture may be designated the optimum. At 11.0 the contrast drops due to diffraction etc.
One of the claims of the BJP, that the lens is optimized for maximum aperture is not borne out by these facts. It is true that the performance is flat from 1.2 to 2.8, but improves significantly after 4.0. The performance at 4 and 5.6 is so much better than at 1,2 that no one can fail to notice it. Now the BJP does not claim that there is no improvement, only that the increase in performance lags behind the others and also behind the quality increase of the Summilux and Summicron contemporaries. If we compare the MTF graphs at 5.6 for the three lenses (Slux, Scron and Nlux) they are very comparable if not identical!
My conclusion is that the Noctilux is quite good wide open, but not spectacular so and certainly not as good as the Slux at 1.4. And stopped down to medium apertures the Noct 1.2 is as capable as the Slux and Scron at the same apertures.
I did not test the contenders, but if the BJP claims that these were better stopped down than was the Noctilux, it in essence claims that the Japanese lenses equalled or surpassed the contemporary Slux and Scron performance. I did study the 1.4 designs of the Canon company and found that they lack the crispness and clarity that you expect from a current Leica lens. Compared to the Leitz offerings of the same period, I am not sure that Leitz is as superior as the Leitz collectors would want to believe.
If you study the article carefully, it seems that the flare suppression of the Noct is in fact the main base for the claim that the Leitz design is superior to the others. This might be true. The classical seven-element lenses are not famous for their flare reduction. And the six-element Noct has a distinct advantage here. The BJP acknowledges that the Leitz design 'cheats' because of the aspherical surfaces. So I compared the Noct to the Canon 1.2/55 with an aspherical surface.
Canon FD 1:1.2/55 SSC Aspherical
In 1971, when the Noctilux was still being produced and sold, Canon introduces their own version of an 1.2 design with aspherics. Presumably they could not stand the humiliation of being defeated by the Leitz construction, at least wide open.
With this lens Canon pulled all stops. See my report on the Canon design here. The Canon aspherical is wide open clearly better than the Noct 1.2. Especially in the outer zones the Canon is much improved and in the center of the image the Canon has an impressive performance, that surpasses the one you get from the Noct. Stopped down the Canon wins easily. Wide open at 1.2 you will note a difference in approach of correcting aberrations. The Canon has a very good image quality ove the whole picture frame, where the Noct concentrates on the center part of the image. Whatever the approach, the Canon design gives the better imagery, aperture for aperture.