M8: the dust has settledM8 and magenta madnessMechatronics
M8: the dust has settled
M8 and magenta madness
Mechatronics
The first photographs with the digital M camera were made in 2006, 180 years after the taking of the first photograph ever to be taken with chemical means by Niepce in 1826. That is a time span of almost two centuries, in which we saw the birth of the first commercially available Leica camera in 1925 (81 years ago and 99 years after the making of the first photograph), the first M camera model in 1954 (52 years ago) and the acquisition of the Leica company by an Austrian investment firm. Any film-loading M camera can deliver first-rate state of the art pictures by using one of the current high-resolution emulsions and lenses, like the Spur Orthopan and the Maco Ortho 25 and the recently introduced Zeiss and Leica lenses. It is doubtful whether we will be able to use the M8 over half a century of use, but switching from solid chemical and mechanical tools to volatile digital and electronic components has a price. The M8 on the other hand embraces the technology of digital image capture with all advantages of direct viewing and immediate image processing. The camera combines exquisite engineering quality with simplicity of use to deliver a camera package that is not only a joy to handle, but also allows pleasure, or even passion to be part of the picture taking process. The M8 is the third camera design for digital capture originating from the Solms factory over a period of ten years. The Leica S1, now almost forgotten was the first digital product from Solms and was introduced about ten years ago. A period of a decade in the current time frame seems to cover more progress than the 180 years of photographic history did in the previous centuries. The S1 was in fact a mobile flatbed scanning device with a lens attached to it. It had some innovative features like the use of the Lab-colour space, but it did not look, nor did it handle like a camera in the Leica tradition. The DMR is the second digital product from the Solms factory. This product can be attached to the R8/9 camera, converting the film-loading SLR to a DSLR. The image quality that is being offered by this device is state of the art for small format digital imagery, but the solution still has to impress the photographic world. 'Small format' in this context refers to sensor sizes smaller than the classical area of 24x36mm. Digital backs are a fine solution for the medium format world where image files around 100 Mb are standard requirement (the S1 could cope with that demand), but the R8/9-DMR combo is too closely related to the mainstream DSLR to become convincing competition in the medium format market in transition from emulsion to sensor. For the photographer migrating from the 35mm reflex cameras to the digital APS sized DSLR, the DMR does not offer enough modern features to become a real alternative. Leica has in fact never been good at designing and marketing reflex cameras one could rave about. The Leitz dynasty in their time had never mentally and emotionally adapted themselves to the worldwide predominance of the slr-concept. The Leica SLR lineage from R3 to R9 has always stood in the shadow of that emotion. The optical designs for the reflex line are first class and it would be very important for the photographic world that this quality can be preserved and evolved to a higher level of performance. With the M8, the Leica company is in a safe area: the rangefinder concept is their core business, honed to perfection over a period of eighty years. With regular intervals, there is that prediction of a rangefinder renaissance, as it is currently en vogue to predict a revival or at least a continuation of the silver halide film emulsion business. This is wishful thinking at best. Silver halide recording and rangefinder cameras are and have always been specialized products, optimized tools for a small set of goals and ambitions. When there was no other choice, the products sold well, even very well, but as soon as better or more versatile equipment became available, the mass market deflected to the new products. It happened to the M3/M4 after the introduction of the Pentax and Nikon SLRs in the late fifties and to filmemulsions around the year 2000, when digital technology delivered superior quality at a lower cost and with greater manipulatory potential. The M rangefinder camera has from the start been designed as a specialized instrument, as the concept has its inherent limitations: macro photography, long telelenses, fast motordrives, accurate framing, focusing over the whole screen area have not been possible or with quite cumbersome solutions, like the Visoflex housing. Within its natural habitat, the Leica M could thrive with the most accurate rangefinding, the most intuitive and simple handling, the finest engineering and the best lenses in the world. The camera feels like a triumph of mechanical engineering and is indeed one of the best-built cameras. But one should not close one's eyes for the other side of the coin: the M camera has no waterproof sealing like some of the SLR competition. One should avoid using the camera for prolonged periods in heavy rain. In the past no one objected to the design constraints and worked with or around them. The currently available M8 has its share of design constraints, different from the past of course, but still there. I have reported on them in my series of articles about the M8. One could start a philosophical debate about the importance of the higher than normal infrared sensitivity of the Leica M solution. It certainly is a nuisance and of a more ubiquitous nature than most observers and many reviewers are willing to accept. By incorporating these constraints into the design, there was room for optimizing the core values of the M system: its optical qualities and its inspirational ease of use. With some historical and evolutionary insights, the first reports about hands-on experience could have been more comprehensive of these roots and heritage of the M line. From the M3 onwards, every Leica M model has offered this blend of outstanding qualities mixed with limitations. The potential image quality is very high, but one should be careful to define 'quality'. In the area of resolution and sensitivity figures, the sensor size of the M8 sets clear limits compared to the best of the competition. If we add the classical silver halide characteristics into the quality definition, the M8 images show a clarity and crispness that is unequalled in the digital domain. The fact that the images recorded on silicon are as pure and unprocessed as they are on silver halide grains, is a decisive factor. If you develop in black and white (and print with one of the current excellent BW printers, like the Epson R2400 or R3800) the colour cast might even become an advantage. The role model for M photographers has always been HC-B, but I would like to draw attention to a German photographer, who did not work with a Leica, but whose pictures are better suited as a guide line for current digital M photographers: Albert Renger-Patsch. His pictures show a very keen perspective, coupled with a highly elaborated technique and mastering of the medium that is now so easy to learn with the M8. It is indeed this combination of learning the craft with the M8 (exploring the medium by shooting spontaneously without any cost and getting immediate response, however unflattering for the person who made the pictures) and transferring this workmanship to the M7/MP world that is such a pleasure to adopt. I am aware that most persons will not return to the wet darkroom, but I have to stress the fact that the results of meticulously processed (slow-speed) negatives are still unsurpassed in its look and feel. As a sideway I should express my unease with labels like professional and semi-professional and amateur that are now conveniently attached to persons and above all to cameras. I really do not have a clue what a semi-professional camera should be or do. Canon D30 or D5 are referred to as semi-professional, but what does it mean? Is a person who works to high-quality standards with a low cost camera but does not earn money with his pictures an amateur or a semi-professional or a professional? Perhaps we should evade these empty words and fall back on the only really important criterion: it the picture worth looking at. Using my recently acquired M8, I could do additional experiments. One of my favourite lenses is the Apo-Telyt-M 3.4/135mm and I put in on the M8: the results are excellent and focus is quite accurate (when using the additional 1.25 magnifier loupe). The beauty of the digital capture is the direct viewing possibility. Even without frames, you quickly learn to guess the picture area, by simply looking at the screen. It is quite illogical (Vulcan speak) for Leica to claim that the 0.72 finder on the M7/MP could focus the 135mm lens accurately without 1.25 magnifier, but the 0.68 finder on the M8 could not with that magnifier. The hidden argument is of course the 1.33 crop factor, which requires you to enlarge the image of the M8 by 1.33. The 1.25 plus does not compensate for the 1.33 minus: the net result being a finder of magnification 0.64 or so, adding in the Circle of Confusion equation. Anyway: my 135mm pictures with the M8 are quite good (technically speaking of course!). There is much discussion about the sensor size, not being equal to the classical 35mm Leica format. You need indeed to adapt to the narrower angle of view that the M8 offers when using the lenses with focal lengths that are meant to give a certain view with 35mm film area. But if you look at your pictures you will notice the fact that on many negatives you do not frame as tightly as is required for the composition. There is that old and almost forgotten remark for Leica users: close in on your subject and exploit the small negative size to the best you can or in other words: do not waste negative area. The same is true for the M8. You can compensate for the 1.33 crop factor and magnification factor by getting as close to your subject as is possible. Here I have to make a negative remark about the M8: the quite generous latitude in accuracy of the frame lines (especially the 50 and 75mm frames) is not helpful: luckily the screen view gives you a direct update of the scene selection. My rule: go closer than the frame lines suggest: picture quality will increase. I use the M8 with the magnifier attached as a standard. With this equipment I tested the accuracy of the range finder. I used the well-known Siemens star test chart and photographed the chart at a distance of 2.5 meters with the 1.4/50 wide open. On the chart I had drawn a vertical line, one millimetre wide, to assist the focussing. With that vertical line you can exploit the higher accuracy that the vernier acuity allows. The normal human eye can discern details when they are separated by an angle of 1/12 of a degree. I focused on the line and I also shifted focus by the width of that one millimetre line. In terms of viewing angle the difference is 1/40 of a degree, or more than three times as accurate as the human eye can focus. The spoke pattern of the Siemens star is clearly resolved when focusing is accurate and quite soft when the focus is off by that 1/40 of a degree. This is an amazing result: the drop in contrast and definition of fine detail is quite visible and indicates that accurate focusing is really required to exploit the quality of the Leica lenses. We have to admit that a three dimensional object has a certain depth and that a small error in locating the focusing plane might not be visible. But the famous discussion around the bo-ke effects might be influenced by small but significant shifts in the plane of best sharpness. The M8 is a camera that fits into the entire M line: it is a camera with a strong personality and with clear advantages and clear disadvantages: that is the existential price you have to accept with every instrument that is designed to do some work with outstanding results and by implication cannot comply with all demands. The bottom line is that the M8 is a most pleasurable camera to use that can open your eyes to the basic understanding of the photographic process and can produce extremely high quality images as a result. Whether the M8 is a milestone camera is the topic of the next article. M8 and magenta madness In part two of my M8 series of reports I noted that the image quality of the M8 digital files is excellent and that the optical characteristics of the Leica M lenses are preserved. These special optical fingerprints are what puts the Leica Scuderia of lenses in a different league in this photographic world. This quality is the result of a number of conscious engineering decisions made by the Leica engineers. In part three (to be published soon) I will argue that this focus on optical properties asks for some compromises in other fields. The enhanced IR-sensitivity belongs in this category. It is a weakness, but one with which one can live in my view. I am not saying I am happy with it. But the same choice of priorities are being made with film: the Kodak Techpan has remarkably high rsolving power, but contrast is low and extended red sensitivy has to be counteractd with a cyan filter.
I am inclined to observe at this moment in time that the Leica Company and the Internet Community of Leica observers and opinion-forgers do the M8 a bad service to over-react to the IR-issue.
The way the involved persons handle the topic is in my view not the best approach to create a balanced view of the topics involved. Let me provide some historical context. Some years ago, Leica stated quite firmly that a digital M would not be feasible because of the characteristics of the M lenses, the most important being the position of the exit pupil (I discussed this theme in one of my columns) and the obliquity of the rays entering the film surface. I suppose this statement was made with the best engineering advice possible at that moment.
The squeeze on the Leica Company by Leica pressure groups and what I called the digital tsunami made the design of an digital M inevitable. Leica suddenly announced (after the introduction of the Epson RD-1 (a Bessa-based digital camera with M mount), that new developments made a digital M a feasible construct It does not work that way as we all know: engineering developments are not falling out of the blue sky.
Leica has always stressed the fact that a compromise solution was the best they could offer. I have repeatedly stated in my columns that the core competence of Leica is the engineering quality and the optical quality. In the digital arena their qualifications have to be earned yet. Especially since the DMR I have been sceptical about the capabilities of Leica to match the best Japanese companies in this respect. No surprise: the Japanese have thousands of man-years advantage in optimizing the imaging chain inside the camera (the color matrixing, the spatial filtering, the decompression and reconstruction algorithms).
The Japanese then designed the best pre-processing software the world has ever seen and corrected defects in the lens systems by software or not at all: Canon users have to accpt that the wide angle lenses for the full 24x36 sensor have lots of vignetting.
Leica followed a different path: as one engineer stated in an interview with a German magazine: "we adapted the electronics to our M lenses and not the other way around".
Every sensible person can add one and one: the first statement tells you that a digital M is not possible and the second statement tells you that the engineering of the digital components and the software is subordinate to the optical properties of the M lenses. One could have expected this if one would be prepared to accept my conclusions from the Epson DR-1 report.
The final result is now well-known: the Leica solution supports the exploitation of the Leica M lenses but in order to do so, some compromises are being made: the thin cover glass appears to allow IR waves around the critical 700nm threshold to be passed through. The Kodak literature referring to the KAF-10500 has a IR transmission graph that is not very clear in the range 670nm to 700nm.
It is evident that there are materials that radiate a high level of IR (a number of, but not all synthetic materials) and we know that some light sources also have a high level of IR radiation. But in many situations were one takes photos these levels of radiation do not occur. Of course there are a number of occasions (fashion or wedding or street photography) were the reproduction of some material may not be accurate. Then a workaround is needed. I did a simple test and took pictures of my line of black and darkblue business suits: out of five one was discoloured, the rest had the correct color.
For the high level of IR radiation there will be solutions and workarounds. The internet buzzes with proposals, some good, most not good. And Leica claims to be working on a more structural solution.
I am a simple person: I know that you have to accept compromises: to get the best M lens quality you have to pay twice what Zeiss asks and four times what Cosina asks. But the small differences are important to me and then I am willing to accept the price differential.
I know that with a Canon 5D I get outstandingly good overall image quality, but I do not get the fingerprint of Leica that I wish to see in my pictures. So a digital M is the premium choice even if I have to accept some compromises. I feel frustrated about the workaround but I can live with it. The folks at Leica should have said this when they discovered the IR-issue: we give you a digital M with the ability to exploit the characteristics of your Leica lenses, but we must also say that to offer you this we had to make some compromises: the following list gives the compromises and we give you advice for workarounds if you work in some special conditions.
But Leica did something else: they heightened the expectation level to quite high marks and stated that the digital M would redefine digital photography as it existed in the market. Well Barnack and Cartier-Bresson could redefine filmbased photography, but the digital M cannot not redefine digital photography, but it can offer some interesting alternatives. Period. The Leica reviewers in print and on the internet did not mention the IR issue, which is remarkable or they did mention it and overstated the case, which is also remarkable.
In the days of silver halide photography everyone knew and accepted compromises. Superior resolution implies low speed of the film and infrared film was not suitable for normal photography. We knew it and acted in an appropriate manner. The Leica M8 would not be different: good at this and not good at that. It is up to the user to make his decision to buy or not to buy. The new generation of reviewers followed the wrong approach by not mentioning the issue at all and creating a high level of expectation about the reults that could be optained with the camera. Or they felt inclined to push the issue into the foreground.
For me the more problematic areas are the noise at ISO values above 320 and the low performance of the automatic white balance.
I can live with the IR issue and I am sure most Leica users can. I can always find a IR blocking filter that will alleviate the weakness. But the main positive issue is that I can use the M8 as I use my M7 and can create the same image quality.
Of course you could say that any 200 Euro digital compact offers better Automatic White Balance than what you get with the M8. But the real argument would be: an M user can handle the quality of the light source manually because s/he is conscious of the photographic parameters.
The Leica M8 allows to exploit the lens characteristics of all current M lenses and brings a quality level that is as good as what we expect from the best Japanese companies. For a 50-year old camera design that is some performance. But we have to accept some compromises: for some that is acceptable, for some it is not.
That is the bottom line.
A prospective Leica M8 buyer has a simple option: to accept or not to accept the Leica way of defining digital M photography.
Mechatronics The quiet revolution at Leica Classical photography has its roots in mechanical engineering and in chemistry, the two main evolutionary domains of the 20th century. Mechanical engineering in its turn has strong relations with the precision engineering of the clockwork mechanism and the manufacturing requirements of the microscope. Chemistry provided the means and the science to master the capture of (light) electrons onto the silverhalide compounds on a microscopic level. The current Leica Mx series and the Nikon F6 are exciting examples of the Olympic heights that have been reached by generations of dedicated scientists and engineers, working in the great mechanical tradition. This however is a sunset industry, as it is called today: but a beautiful sunset as everyone knows will be followed by a deep black night. When the sun rises again, the consumer electronics industry is the main force for the 21st century. Here electronic components, governed by smart software provide the consumer/customer with features, tools and capabilities that are impossible to implement within the mechanical tradition. Canon may be cited as the leading force in this domain: it is Canon's stated goal to provide the customer with an intuitive interface that, supported by as much artificial intelligence as is possible, allows the user to freely develop his creative instincts. In the mechanical tradition the user had to adapt to the limits of the product. Now the product, being mainly software-based, can adapt to the demands of the user. This trend has been enhanced by the introduction of the solid-state image capture and the manipulation of information elements by the techniques of image and signal processing. The full imaging chain in current photography is an exercise in the mathematics of signal processing, constrained by the twin forces of visual perception and optical error reduction. The camera body, once a mechanical instrument has evolved into an interface to the image processing algorithms. I am not saying that this evolution is bad. On the contrary: visual expressions and experiences may be recorded and communicated with more ease and effectiveness than has ever been possible. And at the start of the imaging process, we still need optical means. What I am stating is that the gap between the mechanical and the mathematic traditions in photography is rather huge. Leica, deeply entrenched in the mechanical tradition, is struggling to narrow this gap. It is evident for everyone that the small company, located in Solms in the state of Hessen (Germany) has not the means to fully close the gap and become that challenge to Canon and Nikon that the new Sony Alpha might evolve into. Leica's core competences are the craft and science of mechanical precision engineering and optical design and construction. Mechanical engineering is a natural companion to the science of information technology as both are being used to transmit information, the former by physical movement and the latter by software algorithms. There is even a whole field of study devoted to this topic: mechatronics or "synergistic combination of precision engineering, electronic control and systems thinking in the design of products and manufacturing processes." Leica is evidently in the process of a technological evolution from mechanical engineering to mechatronics with a clear emphasis on the mechanical precision component of the synergy. Mechatronics as such seems to have no special relevance for the current change in photography from silver halide capture to solid-state capture and digital signal processing (DSP). But DSP and digital electronics and digital photography are merging into one single domain. It is advantageous to optimize the whole system of image processing and that is what Canon is doing. Their stated goal (see as example ) is to use as much electronics and software as possible to relieve the photographer from manipulating the photographic parameters in order to focus on user creativity. Intelligent analysis of scene brightness distribution, the relation between ambient and flash light, the automatic combination of AF focus points with local exposure optimization are all deployed to relieve the user from making technical decisions. Leica on the other hand firmly believes that engaging the photographer in the photographic decision process will unleash the true potential of the medium. That is why Leica insists on a manual operation of the aperture ring, even in the forthcoming highly automated 4/3 camera, a derivative of the Panasonic camera, itself a derivative of the Olympus product. Leica merges electronics and automation with mechanics when the advantages fit into their view of and philosophy of photography. The first example of this new approach (mechatronics) in a genuine Leica camera was and is the R9/DMR combo. In my view this is a rather crude implementation of the new approach, but it does show the direction. Remember the Leica 1? This was not the final product, and a full generation of engineering was needed to bring the camera to the high level of sophistication of the M3. In those days progress was measured in decades where it is now counted in years. The low pass filter as example of the Leica way of approaching photographic signal processing Signal processing may be a relatively new buzzword in photographic parlance, it is a standard element in electrical and audiovisual media, in particular TV and digital Video. It should be really clear now that digital photography is in essence the domain of signal capture, signal sampling and signal reconstruction. But this is not new. Film-based photography also captured the image as a system of signals of light rays. And the modern analysis of optical performance (the MTF analysis) used the sinusoidal waveform as input for the aberration analysis. When the image is captured and fixed in a pattern of silver halide grains, the sampling rate is continuous and the transformation/transmission of the information content (or frequency response) is done by chemical interaction. A lens works as a low pass filter for the imaging system. This statement may come as a surprise. But with some thought it is logical. The image of physical reality is being transmitted by a lens (or optical system). Every lens, due to aberrations, can only transmit a limited bandwidth of frequencies. Recall that we have been used the limiting frequency of the lens as a performance parameter: a lens can resolve 150 lp/mm as example. This implies that the higher frequencies are effectively not transmitted, because the lens cannot resolve these frequencies. This is exactly how a low pass filter works in electrical engineering and in DSP. The main difference between a low pass filter in analogue and in digital processing is the possibility of reconstruction of the input signal. A signal transmits information and we want the information content to be preserved accurately when transmitting or transforming the signal. Signals are continuous phenomena and in order to manipulate them we need the signal to be sampled with a certain frequency. The sampling theorem tells you what to do. Frequently this is called the Shannon sampling theorem, or the Nyquist sampling theorem, after the authors of 1940s papers on the topic. The sampling theorem indicates that a continuous signal can be properly sampled, only if it does not contain frequency components above one-half of the sampling rate. For instance, a sampling rate of 2,000 samples/second requires the analog signal to be composed of frequencies below 1000 cycles/second. If frequencies above this limit are present in the signal, they will be aliased to frequencies between 0 and 1000 cycles/second, combining with whatever information that was legitimately there. Aliasing then results in a distorted version of the original signal in the case of DSP. But with film, the situation is different: here we have a range of frequencies that can be recorded and a range of higher frequencies that are not recorded. This will diminish the amount of information and it will diminish the accuracy of information, but the signal is not distorted. In digital photography the Nyquist frequency is the limiting frequency of the detector that can be sampled properly. It is related to the size or diameter of the single pixel area. If the sensor of the camera has a pixel size of 10 micron, then we have a limiting frequency of 100 lines per mm or 50 linepairs/mm or cycles/mm. If we have an input signal that has higher frequencies (the lens can handle the high frequency content of the signal) then aliasing will occur. The most used solution is this: first do a lowpass filtering (to the specific Nyquist frequency as dictated by the pixel size) and then start the sampling of the signal. In such a way you can filter away all frequency components higher than the Nyquist limit. Two problems arise: the lowpass filter is never absolute, so some aliasing will always occur and some information content is removed from he signal. Highly complex algorithms are required to reconstruct the original signal. But the information that has been removed cannot be reconstructed fully. And the distortion is also present at the lower frequencies This balance between information content and aliasing has been troubling the industry for a long period. The original Canon D2000 offered a removable low pass filter so that the user could decide to use one of both situations depending on the requirements. The current Mamiya ZD offers the same options. Canon did drop that option in later generations and we may assume that their research indicated that for their purposes this option offered no added value. Leica wants to record as much information as the lens can transmit and relies on mechanical/physical constructs to handle the information flow. This is in line with the mechatronics approach, where synergy between mechanical and electronic components will be optimized. Canon and others are obviously betting on the DSP and image processing. With some simplification we may say that Leica has the best cards in hand at the input side of the imaging chain and Canon at the output side of the imaging chain. It is most intriguing to note that Olympus with its E-4/3-series also focussed on the input side, but choose a smaller format, limiting the potential of the camera to a print size of A4 and a bit more. The Barnack legacy. In this stage of the Leica evolution it is enlightening to look at the historical course of the company. When Barnack constructed the device, later known as the Leica camera, he had in mind a new kind of tool that allowed him to take pictures at moments and in situations where the then mainstream cameras could not be employed. Eye-level, real-time and spontaneous photography was born, now commonplace but in those days a novelty. It was not the main intention of Barnack to create better imagery than could be delivered by the photographic tools of the day: in fact he aimed to deliver he same quality, but on a much smaller format. The small format was one of the main pillars to fulfil the requirements of portability, handling and speed that the new camera should distinguish form the rest of the pack. The quality of the lens was defined as being able to give the same results as the medium format cameras at the same print size. The main advantage of the Leica lens was the greater depth of field and good imagery at a wider aperture. During a long period of time the optical, handling and engineering qualities of the Leica set it apart and above most of its contemporaries, but from the moment the Japanese came alongside in the areas of engineering and optical quality, the advantages of the Leica sharply diminished as can be seen from the reduction in market share since the nineteen fifties of the previous century. Only persons with a narrow Leica-centric view can ignore these facts and claim an elevated position for all Leica products at all times. Current film-loading rangefinder cameras (Mx, ZM, Bessa-x) are all capable of excellent results. Differences in imagery between lenses and bodies are often subtle and not so evident as in the past. And to be realistic: in sheer image quality the film-loading medium format cameras and the large size sensor-based DSLRs are at least the equal of the current M and R/DMR performances at large print sizes. Of course there are differences in recording capabilities and image representation and in some instances these differences may be become decisive for the choice between different tools and instruments. As with the classic Leica, the main features of the current M camera are portability, ease of handling, superior engineering and outstanding optical capabilities. The original vision of Barnack has lost nothing of its unique charm. The rangefinder concept may have lost its competitive edge as the most efficient instrument for distance detection when compared to the best of AF systems. But for manual focusing, which is one of the main characteristics of the M design, there is no better alternative. The facts that are publicly/officially known about the new M-like camera with onboard sensor would suggest that we are entering a new era for Leica engineers. The unique capabilities of Barnack (a combination of strong vision, engineering excellence, innovative modesty and outstanding grasp of production technology) are needed again. In a few months we know whether the new product will be approached with that same benign curiosity that once has been showered on the Ur-Leica or whether the camera will be recognized as an instant classic like the M3 with unique features that will allow a different kind of photography.
