Summilux-M 1:1.4/50mm ASPH
Introduction
The evolution of the high-speed standard lens came to a virtual halt in the late seventies. The reason is quite simple: the classical design for most 35mm lenses was a derivative of the double Gauss lens, as exemplified by the Biotar lens. In my report of the new Summilux-R 1: 1.4/50mm (II) in 1998 I noted that the era of the Double Gauss lens was not yet over. We were encountering a certain plateau however. In 1980 Dr. Mandler had written the definitive book about the stare of the art and the potential of the DG class of lenses. His conclusion was that for most lenses for general photography with focal lengths from 35mm to 90mm, the DG type was eminently suited and could not be improved upon within normal economical and manufacturing constraints. But the design has its inherent limitations. Very high-speed lenses could not be corrected fully for the oblique spherical aberration in the sagittal direction, astigmatism and curvature of field. These aberrations produced a low contrast reproduction of fine details and a soft edge contrast for subject outlines in the outer zones of the image area. The wider the aperture, the more pronounced these effects could become. In addition the improvement of performance at smaller apertures was quite modest because the balance of the aberration correction was naturally focussed on the image quality at the wider apertures. In addition, the focus shift could lower the quality of the image in the centre of the image.
These limitations made the design of a truly outstanding high-speed lens, based on the DG design an enigma. And in fact development of wide aperture lenses stopped around 1980. With eight elements the development potential was exhausted. The choice of glasses and especially the modern glass types could be beneficial, but not enough to give a radical improvement. The Canon FD 1: 1.2/55mm Aspherical design did demonstrate this, but with eight elements, one grounded aspherical surface and a floating element, it was very bulky and outrageously expensive. Wide open the overall definition showed the faint softness of the image that is the main characteristic of all very high-speed lenses at full aperture. If you look critically at the image, you have the impression that you just missed the critical focus plane. However good the mage quality wide open, there is always a visible improvement when stopping down to f/2 and f/2.8. It is as if the image snaps into focus with high contrast.
To be fair, it should be noted that most users of high-speed lenses do not use the wide apertures quite often and when then in conditions where the finest optical quality is not the overriding concern of the photographer.
In recent years we have seen only two new designs that show a marked improvement on the predecessors: the Leica Summilux-R 1:1.4/50mm (II) from 1998 and the Voigtländer Nokton 1:1.5/50mm Aspherical for Bessa and Leica cameras.
The new Summilux-R 1:1.4/50mm showed a significant improvement in the stopped down performance, based on a higher level of correction of the secondary spectrum and the reduction of astigmatism and curvature of field. The thick centre element (behind the stop) functions as a field flattener. The sharpness impression at the wider apertures was also improved, especially in the middle zones of the image. The overall performance wide open did follow the pattern described above and high contrast imagery can be expected from f/2. From 2.8 the quality is better than what can be expected from the Summicron-R. 2/50mm. Wide open the Summilux-R is remarkably free from veiling glare, flare around bright light sources and secondary reflected images
The Nokton has the classical double Gauss design with six elements and two aspherical surfaces on the last lens element. This lens shows improved performance in the outer zones when stopping down and a fair to modest contrast wide open. Excellent performance can be expected from f/3.5. At wider apertures we see a tendency to veiling glare and overall contrast is fair to medium. As a corollary the definition of fine detail is on the soft side, but quite evenly distributed over most of the image area. Astigmatism and curvature of field are partly responsible for this behaviour.
It has however the edge on the Leica Summilux-M 1:1.4/50mm, a design from 1962, modeled on the classical lines with seven elements and an air spaced lens between the second and third element. This lens has good contrast wide open, but the performance in the outer zones there is a fair amount of astigmatism and there is also some curvature of field. Definition in the centre of the image is very good, even wide open, but on stopping down the outer zones cover the finer details with low contrast. The intended use for this lens is the reportage and documentary photography with high-speed film, as the Kodak Tri-X (in this year 50 years old incidentally), where the coarser grain does boost the edge contrast and suppresses the reproduction of really fine detail.
Contrast versus resolution
There is a persistent story that one can optimize a lens for high contrast or high resolution. There is a certain truth in this assertion, but one has to understand the facts. The choice is only relevant for systems with a high amount of spherical aberration. In this case the rays from a point in object space do not converge to a point in image space, but the bundle of rays forms a pipe of light that is converging to a certain location on the optical axis where the waist of the pipe is smaller and then starts to widen. It is obvious that the location with the narrowest waist has the best contrast, but this is not the location of the film plane. Around the film plane we find a wider bundle of rays, but more dispersed and here we find the location with best resolution. In the past high-speed lenses al had a fair amount of spherical aberration and then the designer could select the back focal length such that the film plane was near the location of best resolution or best contrast. But the difference between both choices was quite minor in practice. Currently all Leica lenses have a very low amount of spherical error and then the choice is not relevant: all lenses have automatically a high contrast and a high resolution.
New approaches and a new vision
The first fundamental departure from the DG design was introduced with the Summilux-M 1:1.4/35mm Aspherical in 1989. In the patent literature (#5.161.060) the designers stress the fact that the new design gives a markedly improved performance in the outer zones of the lens. They also note that the employment of aspherical surfaces must be fully integrated in the whole design and must be part of the basic specifications. Here we detect the design principles of Mr Kölsch: you need to understand the basic problem of a lens first in order to find a sensible and fruitful solution. If a lens exhibits astigmatism as example, it makes no sense to let the computer find solutions, but you must first understand on a theoretical base why this astigmatism occurs, what lens surfaces are responsible for its magnitude and what is the best and most simple solution. This approach: to find the best and most simple solution for a design problem is the basis of all current Leica lenses. It is better to search for one special glass type with the required characteristics in one of the many lens catalogues and to understand it’s potential for the solution for the problem at hand, than to use two separate elements whose combined power might provide a solution too.
When I discussed future trends in optical design with the Leica designers in 2000, the natural question to ask was whether a future 1.4/50 design for the M series could be derived from the Summilux-M 1.4/35 ASPH. The answer was that it might be possible, but not at that moment. The Summilux design transposed to a 1.4/50mm lens would become too large and require a level of tolerances that was impossible to hold in series production, even with the high level of careful manual assembly and frequent inspections that is the hallmark of the Leica Manufaktur. Two years later the optical cell with the eight elements and some exotic glass types was ready and delivered breathtaking performances on paper. At that moment in time it was not yet feasible for production as the mechanical problems were not solved and the size of the lens was an additional hurdle to take. It was a clear design goal that the new lens should be as compact as the current Summilux-M 1.4/50mm. But the designer had an ambitious vision: this lens was to become the best high-speed standard lens in the world, a true demonstration of Leica’s optical ability in this field. For this goal the lens needed to perform above normal requirements in the near focus range and a floating element would be necessary. We know that a lens can be optimized for one distance only. This distance is normally the infinity position or in more practical terms one thousand times the focal length. But even at a distance of one hundred times the focal length (in this case 5 meter), the performance has not dropped to a level that is detectible. But at a close range and particularly at wider apertures one might begin to see a drop in contrast and a very slight haze that seems to reduce the clarity of the image. High-speed lenses in particular are prone to this effect, due to the presence of spherical aberration, astigmatism and curvature of field. If you are not a professional optical designer, there are many aspects you take for granted, but are a nightmare for the designer. The performance of a lens can be fully described by looking at the aberration distribution in the entrance and exit pupils. Here distortion and spherical aberration are two sides of one coin. Improve one and automatically degrade the other. But near distance performance needs to be distortion free and also deliver a crisp image. For a designer this is quite demanding a task.
The size problem
But the most demanding of all tasks is the combination of a small size with excellent image quality. I am now seriously designing lenses, as this is the only way to start to understand the problems and challenges of the optical designer. Modern optical design programs are extremely capable and you can ask the program to find the best lens design for a certain level of performance and selection of glass types. But invariably the optimization program will create a lens that is large and as soon as you restrict the physical dimensions of the lens, the performance drops significantly and all your work is for nothing. Then there is only one option: to make manual adjustments and very slowly proceed to a practical solution. As photographers we are justified to have simple demands: small size lenses and superb performance. But we do not appreciate fully the achievements of the designer when we do not have a clue how difficult it is to combine size and performance. The image quality of the new Summilux-M 1:1.4/50mm ASPH would be outstandingly good when fitted in a big mount, but is spectacularly good when one considers the size.
The addition of a floating element in the same space of the previous Summilux-M can be regarded as a triumph of mechanical engineering. The task here is twofold: find a mechanical solution (in this case finding the correct specifications for the pitch and thread of the mating parts: you need to combine accuracy and smoothness) and find one that can be manufactured in a consistent way. In the past, the prototype stage is the platform to find out if the lens can be manufactured at all to the required specifications. But it is up to the manufacturing department to find a way to manufacture and assemble the parts in a cost effective way. This did not always function in the past: and many lenses had to be changed during series production to accommodate to the harsh realities of the limits of he workers at the assembly stage. The famous wooden hammer, often seen as the symbol of superior craftsmanship at the assembly stage is in fact the remedy for the shortcomings at the prototype stage. The current approach within the Leica optical design department is to order parts from the series production to be used during the prototype stage. So they accommodate to the realities of production tolerances and not the other way around. This is a most sensible approach and explains why the period between finalizing the prototype and the start of the series production is very short. Normally it takes Leica a long period to go from prototype to series production, a delay that was acceptable in the old times, but now is no longer an option, given the aggressive pace of new developments.
On test.
Let me be clear and honest: it is impossible for a 1.4 lens to deliver at that aperture the same image quality as what we can expect from the best 1:2 lenses at f/2. The increase in magnitude of aberrations and the additional types of aberrations (fifth and seventh order) forbids this level of quality. In particular the spherical aberration in the sagittal direction is a bad guy. This said, I have to note that the performance at 1.4 that we get form the new Summilux-M 1:1.4/50mm ASPH is very, very close to the performance of the best f/2 design (read Summicron).
At full aperture the overall contrast is high, with a faint hint of softness still hovering over the image. Astigmatism is very well controlled and the curvature of field is very flat, bringing a visibly enhanced image impact in the outer zones of the picture area. There is a small amount of coma detectible in small bright spots and specular highlights, but much less than in the previous design.
I made the test pictures with Fuji Provia 100F and 400F. Colour rendition of the new Summilux is outstandingly good. Especially in areas with dim light, the colours are bright and clean and quite saturated. With other high-speed lenses the colours in low light illumination look a bit underexposed and flat. They lack the sparkle of bright light illumination. Here the new Summilux sets a very high standard. And in combination with he excellent reproduction of high lights and reflections in bright surfaces, the images with this lens have a very fine and subtle three dimensional effect, that enhances the lifelike quality of the pictures.
Flare suppression is outstandingly good and even better than that of the Noctilux or the Summicron. It is almost impossible to create secondary reflections, even with bright sources in the picture or even worse, just outside the picture but shining into the front surface of the lens. In this last situation, you may see some smeared out bands of light in the outer zones, Of course, no lens with this aperture and front lens diameter is fully immune to flare and veiling glare but this lens is quite close to being for most intents and purposes flare free. You can see this when you make pictures where very bright windows figure prominently in the scene and the windows are decorated with small objects. Normally these objects are washed out and over saturated, but with the Summilux ASPH they retain shape and colour fidelity, a most remarkable performance.
Definition of very fine detail is quite good, but still has a somewhat soft edge. In night scenes the images retain their contrast and outlines of small subjects are clearly reproduced, but very fine detail is lost in the very low micro contrast of the subject textures.
Stopping down to f/2 adds crispness to the very fine detail and from this moment on the Summilux is ahead of the Summicron-M 2/50mm. The jump in quality when going from 1.4 to 2 is however a subtle one and not as visible as with all other high-speed designs. It is easy to assume that a picture made at 1.4 was made at smaller apertures. In this respect this is the first lens where you do not have to accept a compromise quality because of the high speed. The visible advantage of the Summilux is the excellent quality over the whole image area, where very fine detail is reproduced with clarity and crispness.
At f/2.8 the micro contrast of the extremely fine detail is enhanced, because now the effects of the higher order aberrations (residual errors are extremely small now) are neutralized. From here to f/8 image quality is extremely high and comparable with the best Leica lenses in the range (24, 90 and 135). The inevitable drop in contrast at smaller apertures is due, as always, to the effect of diffraction.
Vignetting is on paper on the high side with 2 stops at full aperture, but in practice the darkening of the corner-edges is not so pronounced that is should be a problem. Distortion is just visible, but not of relevance for most scenes, unless you do reproduction.
The floating element improves the quality in the near focussing range. Reduction in image quality will normally occur around the distance of 3 meters and less with high-speed lenses. The floating element does correct this reduction quite effectively, but to a certain limit. When you are in the close up distances from 0.7 to about 1 meter, the floating element can improve matters quite a bit, but then at the wider apertures we get soft images. For best quality in this focusing range, we need to stop down to /5.6 or smaller to get good imagery. We should not imagine that with the floating element, we get a high-speed lens with near macro capabilities. We should stay realistic in our demands. What the floating element accomplishes, is a visual improvement of the quality in the outer zones and a contrast improvement from 1 meter to 3 meters, where the wider apertures will now perform at the optimum of the lens.
Bo-ke aficionados will have mixed feelings about the performance of the lens. The unsharpness gradient is relatively smooth in the near focus range when the fore- and background planes are close to the sharpness plane. But when the distance between main subject and fore-/background increases the unsharpness subjects lose shape and form. Especially when the unsharpness areas are backlit, the shapes become very harsh and rough. This behaviour is due to the reduction of the astigmatism and curvature of field as is a small price am gladly willing to pay. With careful selection of the background and at full aperture, you can produce very intriguing natural portraits.
General conclusion.
The new Summilux-M 1:1.4/50mm ASPH is the best high-speed general-purpose lens in the Leica range. Its wide open performance is outstandingly good (in some respects like flare even better than the Sumicron at f/2). Stopped down it is better than the Summicron 2/50mm. It can be used as the universal standard lens and can be deployed without any restrictions in image quality at all apertures and over the whole image field. If you want only one lens for your M camera, this one should be the prime choice. I still have the opinion that the focal length of 50mm is the best single lens for the M (I am an old fashioned M3 user with 50, 90 and 135mm as the prime lenses) and offers a wide range of possibilities for picture taking.
Handling is superbly smooth and the size of the lens fits in well with the camera and the finger controls of the average user. The screwed finger grip allows for one finger fast focussing and the telescopic lens hood has a lock to prevent accidental moving. This is nice to have but not essential. The smoothness of the focussing mount is the most important aspect. Using the prototypes, you could notice some rough spots, but in the production versions, these are gone. The finish is of a very high standard and the aperture click stops match the quality of the rest of the lens.
This lens has optical qualities second to none and is a triumph of optical and mechanical engineering mastery. It even adds some new pictorial tools in the plasticity of the reproduction and the fine colour rendition in dim light and shadow areas.
Comparisons
The previous Summilux-M is clearly outclassed on all counts. As the Nokton is a small improvement on the older Summilux-M, this one too is not a serious competitor to the new Summilux-M 1:1.4/50mm ASPH, except on the matter of price. More interesting is the comparison to the current Summicron-M. The close up performance of the Summicron is definitely better than that of the new Summilux and if you do not need the high speed, it still has its virtues. The Noctilux-M is more difficult to profile. If you look at the objectified performance criteria, the Noctilux at the wider apertures is no match for the new Summilux-M. But then we have the more subjective considerations. Here the Summilux offers the real life dimension, where the Noctilux is more dreamlike and painterly in its reproduction of the scenes. The Nokton colour rendition is leaning to the pastel colours where the Summilux is more saturated. The drawing of the Noctilux is with a thick pencil where the Summilux uses a very thin tipped point. It is up to every photographer to make the choice. I can only try to describe the differences, objective as well as subjective,
What about the Elmar-M 1:2.8/50mm, This could be the perfect companion to the new Summilux-M: it offers excellent close up imagery even at full aperture, it is very compact and has excellent overall performance.
I often get questions like this: if you only had to use one lens, which one would you choose. The answer was not that easy. But now it is: the Summilux-M 1:1.4/50mm ASPH.
Glass types
The glass selection for the new Summilux is very interesting. There is one glass element in the new lens, whose cost is higher than the cost of all seven glass elements in the previous model of the Summilux. One may ask whether the mechanical parts have been reduced in engineering quality to compensate for the high glass cost. There is indeed a persistent story among Leica aficionados that the new generation of lenses may be optically superior, but the mechanical quality is inferior to older generations. This is not the case. In the new Summilux, there has been advanced engineering to account for the additional movement of the last lens group, the floating group. The total movement is about 2 mm and the effect can be seen from 5 meters. The employment of automated machining is the reason that the lens can be in the same price league as the previous version. And as the lens elements, in combination with the mechanical components, can be made to a higher level of precision, the costly manual adjustments during assembly can be forgotten.
The new glasses used in the new Summilux are very sensitive and very difficult to machine. In fact, the glass is very soft, which makes it difficult to get a smooth surface on nanometer level. The glass is also very sensitive to atmospheric influences and after being polished may not be in the open air for more than a half hour. The manufacturing process has been adapted to this requirement and the glass is coated immediately after the final machining. This also implies very good and thin coating layers. In fact the new Leica process of 'cold' coating is needed to accomplish this.
One of the new glasses has an interesting story. The original glass lab of Leitz was very productive and creative and has produced numerous receipts for special glasstypes. But the glass lab was a research facility and not a production factory. When glasses were needed in substantial quantities, Leitz had to persuade glass manufacturers to adopt the glass in their catalogues. In the past Schott was often the partner in this business. Later that role was adopted by Corning, a company that almost became the main supplier of Leitz glass. But Corning has stopped making the Leitz glasses.
Leica however needed that special glass, formulated by the old Leitz lab in order to get the required optical characteristics. Leica approached Schott and they agreed to manufacture the that glass type. It can now be found in the regular Schott catalogue.
As a matter of interest, it may be noted that almost every glass now employed by Leica is eco-glass, or leadfree glass. The change from old glass types to newer eco-glasses implied in many cases a reformulation of the optical formula to hold the required image quality. In normal practice this change is not important as one does not notice any difference.
If you take a good look at the lens diagram, you will note that the rear diameter is not very large, less indeed than with the Summicron 2/50mm. Therefore the first lens element needs to be of high refractive power to bend the rays sufficiently to get through the aperture opening. Lens elements two and three correct the color aberrations and are also responsible for the flat field of the image. The color correction is excellent, but not of APO calibre. Some red and green color fringing can be detected at high magnifications. The aspherical (fourth) element does correct the fifth order spherical aberration (oblique spherical in the sagittal (horizontal) direction). Element five and six are needed to get the rays to enter the last group almost horizontally. This is a requirement to enable the last group to become 'floating'.
One question that may be asked is why the original layout of the Summilux 35mm asph has not been used. The negative front element will disperse the rays to such a wide angle that the diaphragm opening cannot capture them. Therefore a strong converging lens was needed as the first element.
Performance issues.
Some comments have been made why I did not compare the Summilux 50 with the 35mm version. The obvious answer is that the angle of field of the 35mm is so much wider than that of the 50mm, that a real comparison is not possible. The design choices are totally different. But many persons consider the 35mm wide angle as the true standard lens for the Leica. I declared the Summilux 50mm ASPH as the best Leica standard lens. This does beg the question. The verdict is difficult as the Summilux 35mm ASPH is an outstandingly good lens. But with a gun pointed at my head I would say that the new Summilux 50mm is slightly ahead at the widest apertures, especially in the zonal sections of the lens. The image area is normally divided in four sections (the centre, a circular segment from centre to the vertical edge of the negative format, an other segment from the vertical edge to the horizontal edge of the format and lastly the corners. Everything from the centre to the corner is designated as the 'image field'. Or as the 'FeldÓ in German.
To avoid confusion with the normal photographic parlance where the field is interpreted as the location where you take the photographs, I will use the concept of the zonal sections of the lens. The question has also been asked where the new Summilux-R stands in the performance league. If we set the previous Summilux at the beginning of a line and the new Summilux ASPH at the end of the line, the current Summilux-R would by on a point, located at about two thirds of the line segment from the start point.
What are the more subjective differences between the old and new versions of the Summilux-M? If you are familiar with Dutch painting, one could say that the older Summilux paints as Rembrandt, where the new one paints like Vermeer. This is an especially apt comparison as both painters lived in the same period. Rembrandt has been famous for his atmospheric and emotionally charged paintings. Vermeer on the other hand has been called the first optical painter as he painstakingly captured the finest possible detail with his fine brushes and used special techniques to bring the specular high lights to life.
On a more quantitative level, I made pictures with the older version of the Summilux at 5.6 and with the new version at 1.4. Even experienced users could not see a difference. I did check this again at the optical bench and indeed here the same conclusions holds. The new one is almost four stops ahead of the previous version. This is an incredible result, given the fact that the older Summilux got very high praise in the past for its performance. There is on the other hand a fair amount of discussion in the newsgroups, whether the performance of the new lens is not hyped-up or sexed-up (to use the language of the spin doctors in the political arena). Many people were and are very happy with the image quality of the seven-element Summilux-M. So what is the fuss about? Just marketing speak from Leica and some overcritical zealots?
Let me say this: the original Porsche 911 sports car was and is a great driving machine. Driving one is a fine experience and may recall the glorious days of the real men, who could handle an unpredictable beast and knew exactly where the limits were of cornering and road holding. The newest Porsche 911 is a vastly improved car that still captures the original atmosphere, but is improved in every screw and detail. Now road holding and cornering and braking are impeccable and the important performance parameters have been significantly improved. The older one is a very fine vehicle, but the newer one is simply better. You can be very happy with the older 911, but the new one out-engineers and out-performs the older one.
In photographic terms: the older Summilux can deliver fine images, and if you are happy with them, stay at all means with the older version. If you however wants to take pictures at f/1.4 with the uncompromising quality, previously associated with an excellent f/2 lens, there is no way but to buy the new Summilux. I did an additional, special check between both versions on the bench to try to find the important differences that are relevant for picture taking in real life. First of all there is a major difference in the edge contrast at the low frequencies: where the older one at 10 linepairs/mm has a quite soft representation of the black line pattern, the new one shows lines with excellent edge definition. The overall impact is an image that is really crisp and powerful, where the older lens delivers an image that is more modest, more reduced in its drawing and contrast. At the finer frequencies, let us say, 30 to 60 lp/mm, the older one has a quite fuzzy representation of the lines in the sagittal direction and even become blurred. The new one has very good micro contrast and shows the fine lines with good clarity and separation. With hand held shooting, one is hardly able to capture these fine details and so one does not see them in the picture. And what you do not see, does not exist; that is the adopted view!
If you were to use the new lens only and exclusively at shutter speeds of 1/30 and less, the chances are high that vibration will blur the edge contrast and reduce the impact of the image. When using a good high speed film, like the new Kodak BW400CN, that has excellent fine grain, and allows a higher shutter speed, the difference is easier to see, if you do your own printing. (the natural gradient is not steep to allow printing on color paper!). Best is of course a medium speed film, like TMax100 or Delta100. And if you are a fan of slow speed films like Technical Pan, the differences between the old and new version are quite easy to see. But even with Tri-X 400, the new version delivers images with much higher impact and tighter grain. You will not see this clearly when you concentrate on the small circle in the centre of the image. But when you look for significant details at a location half way the axis point and the edge of the vertical delineation of the negative format, the older one will disappoint you, where the new one will show quite fine texture in the subject details.
The current view of a high-speed lens is to use it only at low shutter speeds in scare ambient light. Under these circumstances any performance advantage may be lost, depending on the workmanship of the photographer. If you do care for the best images, even in adverse conditions, one would yearn for the highest quality lens. Indeed if you use the lens at widest apertures to be able to use a slower speed film or a higher shutter speed or to take advantage of the limited depth of field, one is not inclined to compromise.
The 1.4 lens has always been the lens to judge the standing of a lens line. If you read the older literature of a fine manufacturer like Canon, you will notice that they regard the 1.4 design as the pivotal lens for the photographer, but only if the quality is uncompromisingly good. As a matter of interest, I checked the current Canon 1.4/50mm AF design for comparison. Overall this is an excellent lens, very close in performance to the current Summilux-R. The drawback is a quite high level of distortion, even more than that of the Canon 1.4/35mm ! There is also a trace of coma and the oblique spherical aberration in sagittal direction (the killing aberration for all high speed Double Gauss designs). The Canon philosophy then allows a higher level of distortion to compensate the other lens errors. For the intended use of the 1.4 lens, this is good thinking. But Leica engineers would not agree, as they know things can be done better. The new Summilux-M ASPH is the proof.
