Leica M
The new M is unquestionable a better camera than the predecessor, the M9. It would be most remarkable when this would not be the case. No manufacturer today will introduce a new product that does not claim to improve on the previous products and the comparable products of the competition. The important topic is not that there are improvements, but the question how relevant these improvements and additions are is the most pressing topic. My report will be split in two parts: the first part is a reflection on the status, philosophy and direction of the digital rangefinder, now in its latest incarnation, the M. The second part will look at the performance, but this part will be completed when the real production version is ready and relevant RAW developers are available.
I took pictures with an almost-production-ready M and could use the RAW processor in Lightroom 4.3. This version opened the RAW M files (I never use JPEG files), but I am not sure it does do all the fine-tuning that would be necessary. As a sideline I would remark that the focus on the qualities of the RAW or JPEG engines seems more an excuse to cover possible failures in assessment or judgment, because the M files generally are very good and the post-processing software does hardly add significant improvements or adjustments. It is here as in the chemical processing: when the film emulsion is good, the choice of developer is of minor importance and so it is with RAW developers. While I am able to make some comparisons, I final verdict is not yet possible.
My comparison cameras are the M9-P and the M Monochrom and I mostly used the latter one when studying the characteristics of the M. Again as a side line it is amusing to note that almost everyone refers to the M as the M240 which is precisely what Leica hoped to avoid when dropping the digit subscript from the model designation. What is the gain when the popular designation of the M is M240? The next model (with internal type 243 (just imaginary!!) will be called M243 and not just M.
The full-sized (non-compressed) RAW files of the M are 48 Mp (using a two-byte value for every pixel) and the M9/M/Mono files have 36 Mp (same two-byte value). The pixel size of the M is 6 micron and all other digital Leica M models have a 6.8 pixel size, resulting in a Nyquist limit of 83.3 lp/mm and 73.5 lp/mm. The differences in resolving power are negligible and the differences in colour reproduction are subtle. It is a change from Kodak Portra Vivid Colour to Portra Neutral Colour when moving from M9 to M.
The main difference is the amount of noise in the higher ISO speeds. When I wrote in my review of the M9 that the higher ISO values were hardly useable, I was almost keelhauled by many fellow reviewers, but now that the M is on the market, common knowledge has it that the M can be safely used at ISO 3200 and the M9 should be restricted to 1200 at most. It is remarkable that every reviewer seems to assume that changing the ISO setting does change the sensitivity of the sensor (“high ISO value”). The fact is that the M sensor has a basic sensitivity of ISO 200 (M9 has ISO 160) (based on one of several different ISO norms). A higher ISO setting does not increase the sensitivity, but increases only the under-exposure range. This is exactly the same technique as pushing is in the world of film emulsions. With pushing you do not increase the basic sensitivity, but by a longer development time and more active developer you can reach even the smallest grains, faintly touched by light. The consequence is a steeper curve and a grainier negative. The technique of setting a higher ISO value (exposure index in emulsion terminology) creates a kind of push processing in the digital signal processing chain. It simply does increase the level of voltage or lowers the number of electrons at the ADC stage (gain that is). The SNR (signal-to-noise-ratio) changes and the noise level will be increased. The M does an admirable job in balancing these processes and the useable push is indeed around 1600 and 3200, effectively an under-exposure of three to four stops. At these levels, the image quality hardly deteriorates. (This verdict applies to the M Mono too one might add). Some reviewers might hope that the M is a quantum leap forward compared to the M9 or M Mono, but that is hardly the case. If you are new to the M-system and does not want all the additional features of the M, the current M-E is a very fine proposal indeed.
The main attraction of the M then will for many users be not the enhanced image quality compared to M9 and M Mono, but the much increased versatility with its potential for future developments. The Live View (LV) feature is a great advantage for persons who want to study the performance of lenses around the limits of the optimum focal plane. It is best used on a tripod, because with the enlargement of 5 or 10 times with and without focus peaking the image on the monitor is a bit nervous. It does distract from the art of composing a picture. Careful tests have indicated that the rangefinder focus is not less accurate than the LV with focus peaking. LV with focus assist is however easier to use especially in dim light and with low-contrast motives. The fact that in LV the photographer has three additional exposure metering options (multi-field, center-weighted and semi-spot (just as with the R8/9) is of little importance and falls in the category nice-to-have.
The two main additional features give the M its praised versatility, but when using the M in the great tradition of the M3 (the rangefinder that optimized the art of the snapshot) are in the category that it can’t do any harm and it may do some good. The electronic EVF-2 which in combination with the R-adapter allows the use of R-lenses gives the M a kind of dSLR feeling and usability (but without mirror box).The EVF-2 by itself may be helpful to take low-angle pictures, but in my view a CRF camera has to be used as an extension of the eye.
The movie capabilities may be state-of-the-art, but not being acquainted with movie-makers wishes and demands I will not commend on these features which reside in a different domain: a movie maker is not a still photographer who shoots large amounts of pictures at high speed and a still photographer is not a movie maker who selects the best image out of a sequence of a few hundred pictures taken. I am always amazed that even in these days there are persons who pride themselves on the fact that he is able to shoot ten thousand pictures in a short time. There are famous photographers who do not accomplish this number during their whole working life. Ten thousand pictures do not give you more insight that a few hundred carefully set-up images.
If we place the M in the heritage of the M3 and the M9/M Mono how does it perform? If you use the M as a digital CRF then the image quality compared to M Mono and M9 is not the overriding selling point. The propensity for higher under-exposure latitude is a very welcome feature, but one overshoots the issue when it is being claimed that this characteristic improves the core competence of the M as the essential digital rangefinder. The best and most lasting images have been made with film emulsions rated at ISO 400 or at best pushed to 1600 with large grain as a consequence. The essence of the CRF is not its ability to take pictures with high ISO speeds in dim or dark environments. This feature is however a welcome expansion of the sphere of activity of the CRF-photographer. In pictorial use a slight amount of grain/noise may even enhance the impact of the image. For me the charm of the M lies in two characteristics, one hidden and one tactile. Let me start with the last one. The handling has been improved by the clean layout of the buttons and the placement of the turning wheel at the location of the thumb is very clever. The camera is extremely intuitive in use, the shift between the various focus methods is seamless and all controls have that famous combination of smoothness and solidity that is Leica’s hallmark since the M3.
The shutter speed dial is no longer mechanically linked to the shutter mechanism but in order to provide weather sealing has a magnetic linkage. With sensitive fingers one can clearly note the smoother clicks when selecting a shutter speed on the dial. The shutter sound has been reduced to the mechanical short noise of the shutter braking movement, but the transport sound has been subdued totally. The overall sound impression is now comparable to that of the MP or M7, but does not yet equal or surpass it.
The layout of the menu options and the legibility of the information on the monitor is a pleasant surprise.The hidden characteristic is the major jump in engineering quality and accuracy of manufacture. The main tolerances for the performance-relevant components and adjustments are now on micron level which is a significant improvement compared to previous models. This alone is for me the reason to have ordered one. The M plus the Apo-Summicron-M 1:2/50 mm ASPH. lowers your bank account with Euro 13000 which is a fair amount of money, but what you get is an opto-mechanical system of the highest accuracy that is worthy of the designation of true precision engineering. Compared to the M, the original M3 is a crude instrument, but was in its day the best one could buy. As a coupled rangefinder camera the M can hardly be faulted, but the inclusion of features that enhance the versatility and broader the sphere of activity is a sensitive move. These features do not in any sense improve the capabilities of the camera as a CRF, but maneuver the camera in the mainstream of current photographic equipment.This move is required to broaden the sales base for future Leica products, because the CRF as a species has always been a bit of an outsider. Setting an ambitious growth model on one particular kind of camera is a dangerous strategy. Rangefinder photographers have always been a select group of persons who demanded special and specific cameras and lenses for their goals. Transpose an early user of the M3 to the modern times and give him/her an M without knowledge of all successive stages, He would be very amazed when someone would tell him that this is the descendant of the M3. The first M camera gave the Leica company a prestige on which it still capitalizes. But with every new model, the original features of essential minimalism were sacrificed to the demands of the time. The current M is on the brink of what a rangefinder should be. The shape is immediately recognizable, but the vital statistics are slightly elevated, like a coat that is one size bigger than you need. In its evolution the M has incorporated features that are there because everyone else has them too. The M is almost perfect, a well-constructed compromise, but such a compromise that it is in danger to loose its soul. The M Mono is, from this perspective the true heir to the digital CRF throne. The M however is the camera that has all important mainstream features and its success might be living proof that the classical rangefinder photographer does no longer exist. The future will tell! The next report will look at the performance in detail.
The Leica M is the most flexible, adaptable and user-friendly of the range of sensor-equipped M-bodies since the launch of the M8 in 2006. The major new introductions since that moment have been the M9 in 2009, the M Monochrom in 2012 and the M at Photokina 2012. Technically the changes are the switch to a 24 x36 mm sized CCD sensor in the M9, The omission of the Bayer RGB pattern in the M Monochrom and the change to a CMOS sensor with increased number of pixels (from 18 Mp to 24 Mp). Basically the same chassis has been used, but the shape of the top cover has seen a number of design variations, an indication that Leica has been wrestling with the right shape for the camera. Under the bonnet the construction has been totally redesigned for a more economical production process and for the incorporation of Live View features. One might say that the construction and design of the M8 and M9 were derived from the manufacturing techniques, pioneered for the MP and M7 bodies, and were in a sense backward-looking. The M is a forward-looking design that could become the platform for a range of cameras to be made in the new factory of Leica in Wetzlar, to be opened late in 2013. The important question to be answered is the performance issue. A more modern design and up-to-date features do not necessarily companions to more image quality. It makes sense to do a comparison analysis between M and M Monochrom to see which technology offers the best image quality. Image quality is a very elusive concept and treated very differently by different analysis programs, companies and individuals. The M Monochrom (MM) has been designed for a group of photographers who are accustomed to the handling, look and performance of silver-halide emulsions. A review or test of a sensor-equipped camera is nowadays a system analysis of the interaction between sensor performance and lens performance. For this review I used the Apo-Summicron-M 1:2/50 mm ASPH because this one and the new Zeiss Distagon 1:1.4/55 mm are arguably the best standard lenses in the world. Being a bit chauvinistic I have chosen the ASCR for this part. A review of the Distagon will follow in due time. As this is a camera comparison, the ASCR has been used with the optimum aperture of f/4 to extract the most quality from the sensor and associated in-camera processing. For the evaluation I used the software and test charts of Image Engineering.
The theoretical Nyquist frequency of the M is 1984 Line-pairs per picture-height (24 mm) or 83 linepairs/mm. For the MM the figures are 1736 LP/PH or 72 lp/mm. The difference is for most intents and purposes irrelevant. It really depends on the contrast that is retained at these high periodic structures.
The test chart is filled with nine Siemens stars with very fine structures that the system has to resolve. The finest structures however pose a problem for the software, a fact that is not often acknowledged or addressed. The focus with the M was the Live View, the focus with the MM was the best result of a series of focus steps. The M is capable of reproducing between 77% and 89% of the theoretical resolution, depending on the position of the individual star patterns. The M can resolve on average 1700 LP/PH. The MM is has definitely a better performance with more than 100% of the theoretical resolution. The program has trouble with these figures and produces a -100%, meaning that the theoretical resolution should be used. The MM generates consistently a result of more than 2100 LP/PH. That is a difference between the M and the MM of 71 lp/mm and 87.5 lp/mm. The capabilities of the ASCR50 are best enjoyed with the MM. The MTF at 80 lp/mm, as measured with the Zeiss K8 is very high: at f/4 the contrast transfer is above 40% for the full image area. This is a surprisingly high value and is a clear indication of the superior performance of the ASCR50. (see the lens report for more information).
Below: Resolution graph for the M
Below: Resolution graph for the MM
One has to be honest however and remark that this very high definition ( the combination of contrast transfer and line-pair frequency) can be achieved only under careful conditions (accurate focus and exposure and of course a tripod). Under normal shooting conditions (handheld and slightly inaccurate focus) one may be happy to get 20 linepairs/mm or 480 LP/PH! These figures are presented here to set a realistic target for digital camera performance. It is easy to cite resolution figures of 2000 LP/PH as the maximum achievement, but without additional information this number is without information content. From a historical viewpoint one might recall the senseless debates in the past about the maximum resolution of lenses, assuming that a lens that could resolve 100 lp/mm is always better than a lens that can resolve ‘only’ 80 lp/mm.
The edge contrast is a quite important parameter for the performance of a camera/lens combination. In this discipline both the M and MM excel. The MM has a slightly higher edge contrast that combines very well with the outstanding clarity of the ASCR50. The pixel width of the M covers a wider area than the MM does, an indication that the finest textures are somewhat less well recorded with the M than with the MM. The difference is however rather small and one can make too much of it. Below the edge contrast for four modulations: 100%, 80%, 60% and 40% for the M.
Below the same graph for the MM.
A detailed impression of the edge contrast can be seen here.Below: M
Below: MM
The dynamic range of the M has been checked with the OECF-ISO21550 chart. This one has a density range of 1:10000 or 13.3 stops. You could argue that the M does have a dynamic range of 13 stops, because one can see different numbers and therefore different densities in every single patch. But it is more realistic to note that the bottom three and the top three patches produce black and white in the print. In this case the useful dynamic range is closer to 8 or 9 stops, still a very good result. It is easy to become fooled by the numbers.
Note that the MM has better reproduction in the highlights and in the shadows. Noise is lower too and at higher ISO values (it might be preferable to refer to Exposure Index, as one is in fact under-exposing when using higher ‘ISO’ values) up to 3200 the MM score is better. In my view there is nothing wrong with a slightly noisy picture as this can enhance the viewing experience. The graphs below show the difference: a higher value means more noise. Below: noise graph for the M
Below: noise graph for the MM
On all counts (except of course the color reproduction) the MM is the better choice when one looks at the classical parameters of performance as one would do when analyzing AgX emulsions. The M is preferable when the Live View option is required. The LV is a must-have-feature for a scientific study of what happens on the sensor surface and when using wide-angle lenses (exact framing) and tele-lenses (easy focusing). There are many photographers who still own a number of R-lenses and for them the M is a godsend. One has to realize that many R-lenses do not bring optimum image quality to the M because of their age and different design parameters. I could attach the new Zeiss Distagon 1.4/55 mm to the M (using an adapter). In such situations the LV is a fine addition.
If you live in the best of all possible worlds, you might want to own both cameras as they offer a different spectrum of performance and photographic experience. Switching between the M, MM and M9P produces different sensations, experiences and results. One can observe a tendency in the Leica community to dismiss the M9 as no longer relevant (the same attitude to dismiss the M8 when the M9 came out), but that is not fair. No one dismissed the M6 when the M7 was announced! The M9 (or M-E) is a very pleasurable camera with an excellent performance that goes beyond what one needs in everyday picture taking. When using the M-E one should disregard the low score that DxO gives the M9 and M-E.