Leica S2: the lenses: a technical analysis
A very famous comparison by Geoffrey Crawley of BJP fame confronted a modest and cheap medium format twin lens reflex of Chinese origin with a high grade 35mm SLR of Japanese origin. His conclusion was that the larger size of the negative fully compensated the much higher lens performance of the 35mm camera. All medium format and large format camera makers relied on this rule: the larger the negative the lower the lens quality can be. The logic is evident: a large negative requires less magnification and lens aberrations are therefore less magnified too. If a 35mm negative requires good quality at the 40 lp/mm level, a large format negative needs at most good quality at 10 lp/mm for a bigger sized print.
This rule holds even today in most medium format digital backs and cameras.
Leica is the only company at this moment to couple high quality lenses to medium format sensor sizes. I could examine two lenses: Summarit-S 1:2.5/70mm ASPH and Apo-Elmar-S 1:3.5/180mm.
These lenses represent the current benchmark in medium format lenses. This is not a feel-good statement, but a claim that can be proven with hard facts. For high fidelity photography the reference point is the contrast at 20 lp/mm. Below you can find a number of classical designs with the contrast value for the on-axis position at full aperture. The numbers are generated with the well-known Zeiss K8 equipment.
Planar 2.8/80 55%
Summarit 2.5/70 82%
Makro-Planar 4/120 72%
Sonnar 4/180 78%
Superachromat 5.6/250 80%
Apo-Elmar-S 3.5/180 90%
The famous Zeiss Superachromat has set the standard for performance in the medium format domain for a very long time and compared to other lenses it indeed is an outstanding performer that has had no challengers. The new Apo-Elmar-S outdoes and the Summarit 70 equals the Superachromat. Note that the Leica Summarit reaches the same value as the Superachromat wide open that is more than two stops wider. Now it is true that one merit figure cannot represent the full quality of a lens, but it is an indication when you compare the numbers generated by the same equipment under identical lab conditions.
The focus difference is for the Superachromat a stunning 0 (zero!) micron at f/11.
The Summarit has a focus difference of 22 micron over the aperture range from 5.6 to 22. The Apo-Elmar-S has a difference of 27 micron at f/5.6. These values are really excellent and at least twice as good as what you see from other comparable lenses.
The MTF graphs on the Leica website may give a comprehensive overview of the quality of the S lenses in general. Below you find a range of graphs that show the MTF numbers as measured on the K8 equipment. The correspondence between the two sets is very high. The Summarit has an interesting shape for the 40 lp/mm. It has the familiar Leica fingerprint of a drop in contrast at the image height of 15mm. That the curve rises upward after 18mm is caused by the slight amount of vignetting at the corners. It is well-known that vignetting will improve the performance in the edges as the most vicious rays are cut off from the image forming process. The drop in contrast is hardly visible in general photographic work, but can just be visible at big enlargements when you know where to look. The shape as it is, is not accidental but intentional because it will balance aberrations such that the critical 5lp/mm and 10 lp/mm deliver high contrast values.
The curves for the Apo-Elmar-S are exemplary and one would be hard pressed to find a better lens. The design is derived from the highly acclaimed Apo-Elmarit-R 2.8/180mm, a lens that is one of the benchmark lenses for the R system.
Below MTF for Summarit at 2.5; 4.0;5.6;8;11;22
Below Apo-Elmar-S at 3.5; 5.6; 11
One of the claims of Leica for the S-system lenses is its good close-up behavior at wider apertures. Tests did show that this claim is fully substantiated. (real-life examples will be shown in the next part). While not of truly macro-quality, the Summarit can be used with full confidence at closest distance wide open and you can expect best image quality with excellent definition and contrast.
When discussing lens performance of 35mm systems, it makes often sense to study the zonal behavior, that is the difference in image quality when you go from center to edge. For the S-lenses this is a redundant exercise. The lenses I used were sharp from corner to corner and even wide open you cannot fault the quality. It is most frustrating to sit behind the K8 equipment and note that the measured numbers being displayed on the monitor are so consistently high at all apertures and image points. You cannot help thinking that such performance cannot exist and that there must be weak points. But I am afraid that frustration should be exchanged for admiration.
The amazing lens performance has one drawback: the size and weight of the lenses makes handholding them a tough job. The classical rule that you can safely handhold a lens at a speed equal to the reciprocal of the focal length does not work here (and I might add it never was a valid rule). The S2 and its lenses require at least 1/1000 for good results.
Without the lens shades the lenses are a bit sensitive to flare and it is indeed possible when shooting with the rays of the sun obliquely entering the front glass surface to create light patches.
The analysis above is restricted to the optical/physical characteristics of the lenses. The S2 has a digital sensor and the lens behavior should be related to the sensor characteristics.
The OECF measurements at ISO160 show a dynamic range of eight to nine stops and a moderate noise, but the noise is well controlled as can be seen from the behavior of the RGB channels. Leica has given much attention to the clean differentiation of shadow detail and tonal values and has controlled noise effects. The behavior at ISO1250 is significantly lower, but one should realize that the S2 is not a camera that will be used at these ISO speeds quite often. At a reasonable mid position of ISO640 image quality is very good. And let us be real: ISO400 has served most photographers for decades with good results. I am not inclined to follow the demands of many commentators that one should strive for extreme values to feel happy. A bit of moderation is not a bad thing in these our modern times.
In AgX days there was an easy and simple method to measure dynamic range. It was based on the zone system. You photograph a grey card and over/underexpose for seven stops and you can then easily measure the density values of the negatives. The same is possible with digital capture systems. This I did with the S2 at ISO160 and ISO1250.
It is known that the highest value where you can see differences is 245 and the lowest value is 4. Using these reference numbers I got a range of +4 to -4 for ISO160 (eight stops) and -4.5 to +4 for ISO1250 (8.5 stops). The noise reduction algorithm of Lightroom 3 Beta could reduce the color artefacts but not without softening the details. The high ISO values of the S2 are interesting as emergency options. The S2 will be deployed under conditions where the best quality is required and then you may assume there will be amble illumination.
For numerologists the following measurements may be of interest: the maximum Nyquist resolution is 2500 lp/image height or 82 lp/mm in classical words. The Summarit-S at f/2.5 resolves 2000 lp/ih or 66 lp/mm which is an outstanding result. The lens optimum is at f/5.6 where the resolution is close to 2400 lp/ih, close to the Nyquist maximum. The Apo-Elmar-S has slightly lower values, but the higher magnification will compensate this to a large degree.
From the technical analysis of the S-lenses the conclusion is evident: the lenses are at the top of the medium format league and a quantum leap in quality compared to the competition and also to the M-series of lenses. The quality that you can get with the S-lenses will never be available in M-lenses. The lens quality is quite tightly related to the possible size of the lens. And M-lenses have to be small! It is on the other amazing how good the M-lenses are in comparison to the S-designs. You read about this in the next part.