The Noctilux-M for the Leica-M rangefinder cameras has been introduced in 1976 and is the ‘brainchild’ of Dr Mandler of the then Leitz Midlands factory. It has a weight of 630 grams and the highest price tag of any Leica-M lens . Evidently some contemplation is needed before the decision to buy and use this lens. Optically it is a landmark lens. Its imaging characteristics are unique and can be favourably compared with the capabilities of the Summicron-M 50mm, itself a lens that arguably deserves the title of best standard lens in the world. The special properties of the Noctilux however are not easily detected. Just as is the case with a very fine wine, you need to develop a certain taste and experience before the joy can be yours. Conventional wisdom is not a good companion when embarking on the route to exploit fully the capabilities of the Noctilux. This article is based upon numerous discussions with people from the Leica optical design department, many computer prints and figures and of course a full year practical experience with this most demanding lens in many picture taking situations. Do I know everything there is to know about the Noctilux? I would definitely deny this. As with any Leica lens, the more you use it the more you can appreciate its finer points.
Optical performances: some general remarks.
The Noctilux has a frontlens with a diameter of 51,5mm. The Summicron has a diameter of 26mm. The negative area is the same. It follows then that the oblique rays entering the lens must be bent with a much sharper angle in the case of the Noctilux. This lens has some lenselements with a very high refractive index (> 1.9), where the Summicron only needs indices of max. ± 1,7. These glasses with high RI are quite heavy and expensive. The weight and price of the Noctilux can be partly explained by this fact.
The first version of the Noctilux (the 1.2) had been designed with six lenses and two aspherical surfaces ( front of front lens and back of back lens) to cope with spherical aberration. But the grinding process was slow, laborious and error prone. Costs became very high and to be honest the optical performance, while very good at f/1,2, became less convincing when stopping down.
The new lens, now fully ‘spherical’, needed seven lens elements and is optically a member of the Summilux family, whose ancestry started with the Summarex (1,5/85).
It is nowadays perfectly possible to design a standard lens with an aperture of 2,0 to have minimal residual optical faults. Such a lens (and the Summicron-M is a suberb example) will exhibit high contrast, very good definition and clarity of very fine details till the very limits of the recording power of the film emulsion.
By the way: the advances in film chemistry now have killed the old adagio that a modern lens is capable of better resolution than the film. One of the reasons the optical department of Leica is redesigning many lenses is just to take advantage of the new capabilities of modern films.
If we open up the maximum aperture of a lens one stop to get a 1,4 design, we must be prepared to accept some less rigorous corrections. Flare and distortion go up and colour corrections suffer a little. If we dare to venture again one stop more, we stare in the eyes of some nasty optical evils: transverse chromatic aberration, coma (also called: oblique spherical aberration) and spherical aberration.
So it comes as no surprise that the Noctilux when subjected to an optical bench test shows coma, spherical aberration and chromatic aberrations from an image height of 9mm outwards at full aperture. There is also some inward bending curvilinear distortion and quite pronounced light fall off at full aperture, which is almost gone when stopping down one stop.
Identifying optical defects in the lab is only a small part of the whole story. Quite often these characteristics are not visible or not important in practical situations.
Contrast and Flare.
We often do not realise the degrading effects of flare on the image quality of very fine texture details. There is an intimate relationship between flare and contrast, both macro contrast and micro contrast. Contrast in general refers to the relative luminance of two areas.
Macro contrast is the difference between the deepest black and the lightest highlight of a scene, that the lens is able to record faithfully. An ideal lens would only transmit all the light from the subject and bring it to sharp focus in the focal plane. In real life a quantity of non-imaging forming light ( for instance reflections between surfaces of two lens elements) will spread uniformly over the whole negative area. When we are taking pictures of a scene that has deep shadows, the light energy hitting the emulsion can be so low that no latent image will be formed. After development this negative area will be transparent. A lens with a high flare factor however will ‘illuminate’ the shadow areas and after development the negative will now have a certain density, but of course no details of the subject will be recorded. When comparing two lenses with the same nominal aperture (say f/1,4) we might think that the one that gives a slightly higher shadow density is actually a bit ‘faster’ as more light seems to be transmitted. In fact this lens may be just more flare prone and the other one could be more highly corrected for optical aberrations.
Microcontrast refers to the relative luminance of two adjacent very small subject areas. It has special relevance to very fine textural details. If some object detail is just recorded, but the contrast is below the visual threshold, we are not able to see it. What we will see however is a kind of image noise, that reduces the clarity of the picture. It will be appreciated that good microcontrast is needed when the higher spatial frequencies must be clearly separated. Flare eradicates marginally resolved textural details.
The Noctilux has an impressive flare reduction capability. It is almost impossible to detect any non-imaging forming light in Noctilux pictures. When comparing Velvia transparencies taken with the Summicron-M 50mm and the Noctilux in an identical setup (some strong light rays leaking from the edge of a wall) the Summicron (itself famous for its flare suppression capabilities) showed reflections in the center of the picture where the Noctilux was absolutely free of it.
The Noctilux was also able to record very subtle gradations and colour hues in dark shadow areas and highlights with strong relections. Again a tribute to the excellent flare suppression. The Summarit, which we also used as a comparison showed, in this quite nasty test, a severe level of reflections.
Overall contrast at full aperture is lower than that of the Summilux and at f/1,4 the Noctilux is still somewhat behind this lens. When stopping down to f/2,0 we can compare the contrast of all three 50mm lenses, Noctilux-M, Summilux-M and Summicron-M. The Summicron-M is best, but the Summilux is very close as to be indistinguisable is most picture taking situations. The Noctilux has a slightly lower contrast. When comparing the Noctilux at its full (f/1,0) aperture, its contrast is much higher than that of the Summarit at f/1,5 and even the first version of the Summicron (the 7-lens computation from 1954) at f/2,0 is surpassed by the Noctilux at f1,0! .
The medium-to-high contrast at full aperture becomes a high contrast at f/2,0 and reaches a very high level at 2,8 (almost equal the Summicron-M at that aperture). The contrast drops slightly, but perceptibly so at f/5,6 and smaller. We made some comparison shots at this aperture of a high contrast scene (the typical narrow street in a small French village, with the sun high in the sky and taken against the sun) on a 100ISO B&W film. The Summicron produces a high contrast and very sharp image. The Noctilux has a bit lower contrast, but the same level of sharpness. Extremely fine detail however is rendered more sharply with the Summicron than the Noctilux.
This unusual balance between lower macrocontrast and very good microcontrast gives the Noctilux its ‘fingerprint characteristic’. The excellent flare suppression in combination with this high microcontrast produces the smooth rendition of fine textural details that gives the Noctilux images its ‘raison d’etre’. One would do this lens a grave injustice when classifying it as a poor-light-only lens. It is in fact an interesting alternative to the Summicron/Summilux pair.
A true f1,0 lens
Pictures taken with the Noctilux at f/1,0 and f/1,4 (and adjusted shutter speeds) were measured sensitometrically and the results were identical. This lens is truly a full stop ahead of the Summilux-M 1,4/50.
Close distance performance.
Most standard lenses are designed for optimum performance at infinity. The optical corrections however ensure that at a close distance range, say 1meter, the performance is hardly degraded. Very high aperture lenses, being less optimally corrected for astigmatism, coma, curvature of field, spherical aberration and second order chromatic aberrations, will perform less good at distances in a range below 2 meters. We compared the Noctilux and the Summicron at a distance of 1meter at all apertures. The Noctilux showed sharp object-outlines, but fine details were a little fuzzy at the edges and very fine detail was not resolved at all. From f/4,0 till f/11,0 the performance of the Noctilux markedly improved and became excellent at f/8,0.
Closeups at full aperture, however, are not the Noctilux’s forte. But then who will use this lens in such a situation. The depth of field at f/1,0 and 1 meter is a mere 1-2 cm! Portraits at this distance and aperture are bound to disappoint from a technical perspective that is.
Resolution and MTF.
The two main components of image quality- spatial frequency response and contrast transfer- (or in less technical terms: resolving power and gradation) can not both be optimized at the same time. Low resolving power and a fine gradation can produce a very pleasurable photographic image as long as the outlines of the object details are clearly delineated. The information content of a picture that has been trimmed for high resolving power is of course much higher. When the resolving power is very high, the microcontrast drops and the image details are no longer visible. The grain pattern of any film will produce a random noise and as soon as the microcontrast of fine image details is at or below the level of this noise, the image will only reproduce an amorphous density. The limit has reached. That’s why a fine grain high resolution/low contrast image often looks a little vague. In fact the first Summicron (7 element lens) was of the high resolving/low contrast type and the first Noctilux (1.2) of the low resolving/high contrast type.
It is now wll known that there are three regions of spatial frequency that define image quality. The low frequencies (5 to 10 linepairs/mm (cyles/mm)) define the clear outlines, the medium frequencies (20lp/mm) are responsible for fine object details and the high frequencies (40lp/mm) are needed when we want to record extremely fine subject details.
Forty lp/mm might seem unimpressive as a goal as we are accustomed to claims of 150lp/mm as the maximum resolution of a lens/film combination. This figure is out of proportion. A young person can see at a distance of 10cm and is then able to resolve up to 17c/mm. When mature a distance of 40cm and a resolution of ± 4c/mm are more realistic. This level of resolution assumes a high contrast of at least 50% to be clearly visible. Assuming the 40lp/mm as a limit, a 35mm negative need to be enlarged 10 to 15 times before the most critical inspection at close distance could notice smaller details than can be recorded with the 40lp/mm cum 50% contrast threshold. This contrast threshold is far more important than the resolution figures for the ultimate image quality. Here we stumble across the present day limits of optical progress. Look at the MTF tables below and remark that the contrast transfer of more than 50% for the 40lp/mm limit is feasable for the center only. And we are looking at topclass designs.
MTF measurements of the Noctilux and the Summicron can be compared to give this discussion a more quantifiable dimension.
The Noctilux however has some very special problems in this area. Consider the results in the tables below
|% Contrast in center Noctilux-M||lp/mm||1.0||2.0||5.6|
The 10lp/mm value at f/1,0 is very good, the 20lp/mm is still quite good and the 40lp/mm result is acceptable: the Summicron has the same value for the 40lp/mm in the extreme corners. (at f/2,0 mind you!). At f/2,0 we notice an overall drop in performance , especially at 40lp/mm, while at f/5,6 the performance is slightly better than at full aperture. Modern Photography tested the Noctilux a number of years ago and noticed the same behaviour. They just accepted the figures. We know that the Noctilux cannot be fully corrected for spherical aberration. One important result of this defect is a slight focus-shift. (‘Blendendifferenz’ the Germans call it). When refocused the Noctilux showed these results:
|% Contrast in center Noctilux-M (refocused||lp/mm||2.0||5.6|
The focus-shift of the Noctilux is 74 micron when stopping down from 1,0 to 2,0 and 120 micron when going to 5.6. The reason that the f/5,6 values do not drop as much as the f/2,0 values (in the first series of measurements) is the depth of field. At f/5,6 this depth of field is greater than the focusshift, so the results stay within tolerances. By the way: 120 microns is the total thickness of a typical fine grain 35mm film! These results show the pitfalls when testing an extreme aperture lens. As most objects in the real world are three dimensional and have depth, the effects of the focusshift are hardly noticable. What you could notice is a slight softening of the image at f/5,6, because the focus-shift will produce a somewhat larger diameter of the blur circle.
Sharpness and definition.
At full aperture the Noctilux records fine details with a high contrast. Its sharpness circle has a radius of ± 9mm from the centre. The total central image of very good quality is then 18mm. The area outside this circle and especially the corners have a very low contrast and details are not sharply rendered. At f/1,4 the contrast of the subject outlines improves, fine details sharpen up, but very fine details are still slightly soft. At f/2,0 the very fine details are now sharply rendered and the overall quality is comparable but not equal to the Summicron and much better than the Summarit at f/2.0. From then on the image quality stays on this very high level, but there is a slight contrastdrop after f/5.6.
The MTF values for the aperture of f/5.6 show the high level of optical quality when compared with the Summicron-M at the same aperture
|% contrast in center||Noctilux-M||Noctilux-M|
Note that the figures for the Noctilux at 5.6 in this table are slightly different from the table used to indicate the focus shift. The figures are taken from two different MTF-measurement apparatus. As a sideline we can comment here that differences of a few percentage points are relevant: comparison between lenses are only valid when using the same testing equipment and method! Our conclusion that from f/5.6 the practical performance of the Noctilux and Summicron are comparable, is substantiated by these figures. At least in the center of the image!! In the off-center regions the Summicron is of course the better lens.
Notice too, that the performance of the Noctilux when not adjusted slightly for the focus shift, is comparable to the performance of the Summicron till the 20 lp/mm mark. Very fine detail is resolved better with the Summicron, but given the deployment potential of the Noctilux that is not a problem.
One cautious remark is mandatory now. If a photographer would like to exploit the image capabilities possible with the 40lp/mm limit, any handheld shooting is out of the question. A heavy tripod and a very carefully controlled exposure/development technique is necessary. 25ISO film is also required (Ektar 25, Kodachrome 25 and TechPan).
The hot question then has to be: can the Noctilux-M, given its optical potential be used as a universal ‘standard’ lens and double as a replacement for the Summicron-M. My answer would be: no. I use both lenses and choose according to demands and circumstances. While the deployment possibilities show substantial overlap, both have a specific optimum use. As an example: at distances from 70 cm to 1,5 meter at large apertures the Summicron-M excells. From aperture f/5,6 to f/16 the Summicron-M for all kinds of objects and distances the Summicron-M is hard to surpass. On the other hand: pictures taken at a range from 2 to 6 meters in low illuminance levels with adverse lighting when very fine subject details must be recorded at apertures from f/1.0 to 2.8 cry out for the Noctilux-M. Shooting reportage style, handheld, 400ISO or higher and with subject distances from 3 to 10, maybe 15 meters at apertures from f/2,8 to let us say f/8.0 there is obviously much overlap.
Coma: the testers trap.
Spherical aberration and astigmatism are detectable, but coma is quite visible when taking pictures where strong light sources are present. Again, we must caution for too hasty conclusions.We ‘see’ coma when an off-axis image point is not represented as a circular patch of light but as an asymmetrical (often comet-like) shape. The concentrated light energy falling on the film emulsion is distributed unevenly. The more intense this energy, the more pronounced the effect. If we take a picture of the well known string of strong light bulbs and focus with the Noctilux on these light sources, coma will be quite pronounced and visible as the butterfly distortions of light sources in the image. Light energy will be sharply focused and due to the high contrast of the Noctilux even more accentuated.
It seems more likely to take pictures at full aperture of an object focused at closer range when the background lights due to the small depth of field are out of focus. If we were to take a picture at 3metres, then any object at 10cm behind the sharpness plane would be unsharp. The light energy now is less intensely concentrated and the coma effect much more subdued.
The upshot is this: it is quite easy to demonstrate the coma effect with the Noctilux at full aperture. In practical picture taking the effect will in most cases be negligible. In any event it can be controlled by the experienced Noctilux user. Look at the illustration: a picture taken at night in a small village some kilometers from Solms. The lens is focused on the lights. The luminous energy of every point of light is sharply focused on the film and the coma is very visible. But is this a situation where you would use the Noctilux at full aperture. Hardly. You need some depth of field and the chosen exposure (about a 1/250 at f1.0 with a 400ISO film) is not necessary. A slower speed and a smaller aperture will give you a better picture.
Its general transmission characteristic is slightly warm (red).
Light fall off.
At full aperture the natural vignetting is clearly visible. Transparencies taken in clear daylight at f/1,0 show a circular darkening at the outside of the picture area. The magnitude in the extreme corners is 3 stops under exposure. It is impossible to neglect this fall off and certainly it is severe. Leica itself is honest enough to warn you for this phenomenon. From f/2,0 it is gone completely. In most situations where the f/1,0 aperture is really needed (creative sharpness or non-available light) the fall-off will occur in image areas that are pictorially not relevant.
The factory advises against the use of filters. We have a UV filter permanently in front of the lens to protect its large front glass and do normally not notice any adverse effects. In strong backlighting the reflections can become a disturbing factor. When the best image quality is needed, you are well advised to take the filter off.
The shallow depth of field (20cm at most at 3 meters) asks the utmost of the precision of the rangefinder. The Leica production tolerances are 1/00mm. The adjustment tools have their own tolerances of course and need to be added to the overall tolerance of the system. The rangefinder tolerance is then ±1/100mm. The human eye can detect differences of less than 0,06mm at a distance of 3meters. That is less than the thickness of a human hair. The virtual distance with which we look at the rangefinder spot in the camera is ±1,4meter (-0,6 diopter). We then are able to see differences of 0,05mm. As the rangefinder is built with a 1/00mm tolerance we may safely state that the accuracy of the rangefinder exceeds by a factor of 5 the one needed for the visual acuity of the eye.
But there is more. The weak spot, so to say, of the whole rangefinder mechanism is the mechanical linkage between the feeler arm and the curved cam at the rear of the lens. The total movement of the feeler arm is fixed and therefore the same for any lens. The movement is restricted to 2,5mm in the Leica system. The focusing movement of the lens from 1meter to infinity can be translated to a 2,5mm movement of the feeler arm. This distance of 2,5mm is not surprisingly exactly the distance the normal 50mm lens moves forward/backward from 1m to infinity.
We thus have a 1:1 correspondence between the movement of the standard lens and the feeler arm. When I move the lens forward for 0,1mm the feeler arm will also move 0,1mm and so will the movement of the rangefinder spot in the viewfinder. The 135mm lens moves 18mm for a distance range from 1,5m to infinity. But the feeler-cam is restricted to 2,5mm. So now the correspondence is about 1:9. Here is the problem. The viewfinder movement of 0,1mm is now translated in a lens movement of 9 x 0,1mm or about 1mm. Within the 1mm movement the rangefinder patch will not move. Therefore the distance measurement has a certain inaccuracy.
The distance the Noctilux lens moves from 1m to infinity is 2,5mm, the same as the Summicron. The Noctilux however needs a turn of 9mm on the distance ring where the Summicron has 4mm. To move the rangefinder patch over a distance of 0,1mm, we need to turn the distancering of the Noctilux over a distance 2,25 (9/4) times as long as that of the Summicron.
The art of fixing the shadows.
Sometimes the usefulness of any aperture larger than f/2,0 is questioned as the quality of high-speed films is so much higher now than in earlier days. So simple arithmetic tells you that f/1,0 at ISO400 or ISO100 is equal to f/2,0 at ISO1600 or ISO400. True? Not at all. A lowspeed film still has more contrast, better sharpness, more vivid colours and/or smoother gradation and finer grain than its high speed counterpart.Especially high speed transparancy film is still not good at any speed above ISO200. So if you need very good image quality in low-level light or twilight situations, an ISO 100 could make your day. The omnipresence nowadays of portable flashlight, automatically filling in the shadows, is slowly killing that fine example of the art of photography: capturing the shadows. Fine tonality in deep shadows is a hallmark of the photographers craft. The light gathering power of the Noctilux gives you shadow detail and subject texture where the combination high-speed film/lower aperture lens and/or flash could let you down or kill the feeling of the moment. It is quite easy to overexpose when using the Noctilux. As a habit I set the ISO value of transparency film 1/3 to a 1/2 stop higher (under exposure).
There is a persistent opinion, reproduced in almost all publications, that the Noctilux has only special qualities at full aperture and is not useable at smaller apertures. That is definitely not true. The original Noctilux 1,2/50mm (the one with the aspherical surfaces) recorded image outlines with medium to high contrast, but lost out in the recording of fine image details. When stopping down the image quality did not improve very much. It might be possible that these characteristics have been projected into the new Noctilux-M 1.0/50 because the name is the same. The latter lens however is a quantum leap better than its illustrious predecessor. To be honest however: it is easy to overstate your case. The Noctilux from f/5,6 shows a slight softening of the contrast of subject outlines due to the inevitable focus shift. While very good at this and smaller apertures the Noctilux is not on the same level as for instance the Summicron-M (latest version) which can be described as outstanding from f/5.6.
The Noctilux-M is a superb lens with a fascinating wide open performance and a competent performer at smaller apertures. Its shadow penetration, absence of flare, and rendition of fine textural details give low-level-light pictures a clarity and richness of colours that is quite unique. The excellent flare reduction makes it a preferred lens in conditions of brilliant lighting. You need a trained and disciplined eye to appreciate the image qualities. You need also some technical qualifications to exploit fully the potential of this lens. The optical/mechanical demands on the camera/rangefinder system are quite high.
The M series is built to a precision adequate for the optical/mechanical demands of a lens of the caliber of the Noctilux.