Archive | November 2013

All that science

20100807_154414

Sometimes, when I want to remember what the world used to be like, I go to a local used bookstore housed in a  former glove factory. There, amid the dusty wooden floors and endless miles of shelves lit by fluorescent lights on pull chains, I like to thumb through old photography books.

There are books with glossy slip covers, faded where they could see the light; books with red or green library bindings and crude spine-stamping; and yellowed pamphlets with holes for spiral binding.

There are volumes and volumes and volumes of data, compiled by innumerable Rochester research scientists, showing plots of spectral sensitivity, response curves, guide numbers, and filter factors.

There are manuals telling the average person how to poison himself performing color negative development at home, books explaining studio lighting, and various materials describing how to do processes that had to be done once and done right.

One of the most tragic things about the dismantling of film photography, and particularly the Kodak empire, is the loss of so much science. Years of people’s lives went into the products and quality control and testing and data booklets for products that in the past decade have become townsmen of a stiller town. Those scientists have long since moved on; the research labs have been razed; and there is no one left to make any new products.  Even for the handful of historic best sellers remaining in the current Kodak, Ilford and Fuji lines, it is questionable whether the people running those lines would have the capital or know-how to recover from a major equipment failure or change in environmental regulations. They are caretakers and sellers – but no longer creators.

Writing from the perspective of the twenty-first century, it seems strange that anyone would ever mix plastic, animal glue and silver halide crystals to do anything; it barely seems a step away from alchemy. Yet replicating this alchemy – the object of nearly every digital photography R&D exercise – has required gargantuan engineering efforts and massive computational power. This is a testament to the quality of the work that went into silver halide technology. Slide rules, graph paper and microscopes made it that good. But with such primitive tools came a primitive faith – that a seemingly opaque piece of film could be subjected to a liquid sacrifice, blessed by gestures, and yet result in a viewable image.

The fact that film is even available 18 years after the Quicktake 100 is somewhat remarkable. Film aficionados decry the demise of the medium as if this is an isolated occurrence. The reality is that everyday technology has changed more in the past 20 years than we realize.  Consider that if you graduated from high school in 1990, all of these were mainstream then but barely used (and in some cases illegal) today: answering machines; bias-ply tires; carbon paper; credit card imprinters; day planners, dial-up modems; dictation machines; dot-matrix printers; encyclopedias; fax machines; floppy disks; gummed postage stamps; home movies; hood ornaments; household fuses; hub caps; incandescent light bulbs; leaded gas; library card catalogs; LPs, cassettes, CDs, MiniDiscs, LaserDiscs, Beta, VHS, 8mm video; mercury thermometers; mimeograph machines; overhead projectors; pagers; paper calculators; paper maps; pay phones; standalone PDAs; phone books; rotary telephones; sea mail; shortwave radio; slide shows; telegrams, teletypes, and telex machines; tube televisions and CRTs; typewriters; UHF television; video rentals; and wind-up clocks. Although other eras have had rapid technological changes, they have resulted in a greater diversity of everyday objects; the late 20th and early 21st centuries primarily have been consolidations of function and extinction events.

Why does film seem like such a big deal amidst a sea change in everyday technology? Part of it may be that when people think of photography, they imagine artifacts of film: fading, sprocket holes, edge printing. There is some embedded belief that the fact of ephermality can only be expressed by a fixed, identifiable and conventional state of decay: thus Instagram.  Others seem to believe that  negatives and slides visible to humans are the only real method for long-term storage or “real” photography, notwithstanding the gradual disappearance of the means to turn them into finished images. But it may also be that various people have various points of nostalgia and that photographic enthusiasts are a self-selected set.

Even among the film faithful, there is little consciousness of how far they have drifted. The semi-digital workflow that makes it possible to use film and display its images given the pace of modern life is also the one that divorces film from its historic antecedents and all that underlying science: Frontier systems digitize film, correct away its original tonal response, and laser-print copies of images on silver paper. The essential use of the negative for projection printing has changed. There is little discussion of this, and seldom does Captain Kirk appear to tell the Yangs that their prayer is in fact the Constitution.

Things are in a constant state of evolution, and as Tancredi so aptly put it, «Se vogliamo che tutto rimanga com’è, bisogna che tutto cambi.»

Advertisements

Leica M Typ 240: Part 5 (Scoring against the Fuji X-Pro1)

leicafuji

This never happens to the other guy.

Introduction. The M typ 240, or M10 as it was previously called (why not MX?!) took so long to get to market that in the meantime, Fujifilm released the X100 and X-Pro1, both of which are highly competitive with the M9. The X-Pro1 in particular was to the Leica M9 what the Contax G and Hexar RF were to the M6 – a shot across the bow. A year after the X-Pro1, the typ 240 made its debut. There have been quite a few comparisons, but few of them were in any depth. Let’s see if we can fix that.

Glass Viewfinder. We can start here, because the defining characteristic of a rangefinder or similarly styled camera is an eye-to-glass interface. At the time it came out, the X-Pro1’s hybrid viewfinder looked like something from the future. It is relatively undistorted, has two zoom settings, shows pretty much any information you want as an overlay on a normal view, and has infinitely variable framelines that account for subject distance and field size. It can be switched to EVF instantly. Both modes can show composition grids and the built-in electronic level.  The Fuji viewfinder accepts 19mm eyepiece elements such as diopters (common to Cosina/Voigtlander RF cameras, Fuji GL, GM, GS, GW, GSW, GA, and GF series rangefinders, the Nikon FE/FM/FA series, and the F3 non high-eyepoint. As far as go non-SLR, non-prism rangefinder viewing systems, this one wins.

Like its predecessors running back 60 years now, the Leica viewfinder is a very bright, uncomplicated-looking optical viewfinder/rangefinder unit with new, unobtrusive LED-illuminated framelines.  In low light, they dim automatically to avoid blinding the user. The view is undistorted, and it features an LED readout developed when Atari 800s still roamed the earth. But other than that, the display shows shutter speed and flash status (in A mode) and the over/under/correct exposure indicator in manual mode. Very little has changed since the M7. Leica’s diopters are far more expensive.

On balance, despite the significant added functionality of the Fuji OVF/EVF, the nod goes to Leica on viewfinders. The Fuji’s Achilles’ heel is parallax error. When the camera is auto-focused, the framelines and focus target jump to the corrected positions instantly. This can often surprise the user with sudden changes to composition. Moreover, the focusing targets can at best show the infinity setting and the nearest distance; unfortunately, the latest Fuji firmware upgrade can – with some lenses – make the spread between the focus point for infinity and the point for 0.1 meters. This is a huge difference, and Fuji could stand to put some intermediate boxes in between, particularly at the 1 meter mark (which previously was described by Fuji as the reasonable near limit for an optical view). By contrast, both the Leica framelines and focusing rectangle move continuously with the focusing of a coupled RF lens. This makes it easier to adjust composition in real-time while focusing.

Electronic Viewfinder. Both cameras lag behind current ideas of EVF technology (especially compared to cameras dedicated to EVF use). The Fuji wins this one from the standpoint of eye-level operation; the user can switch modes viewfinder modes instantly using a front-mounted switch and viewfinder magnifications (for manual focus) using the rear thumbwheel/switch. It also shows composition and leveling aids. The Leica, though, has two advantages of its own. One is that its EVF can be pivoted for low-angle shots (the Fuji does not have any capability to add a positionable EVF). The other is that the Leica can automatically engage magnification by a sensor mounted to the RF pivot. In other words, it knows when you are focusing and amps up the view.  Using the EVF does not add any lag to the X-Pro1 – since it is reading off the sensor at all times anyway. It does drag down the responsiveness of the M to something like a Pentax 6×7.

Ergonomics. The M’s ergonomics are greatly improved from the M8/M9 in a number of ways. The relocation of the thumb wheel to a more traditional position (under the thumb!) makes it much easier to cycle through settings. The inclusion of dedicated ISO button makes it easy to adjust sensitivity quickly. And the front button (“focus”) provides press-and-turn functionality with the thumbwheel to provide ISO compensation without taking one’s eyes off the viewfinder. That said, the M also exhibits some longstanding Leica blunders: a shutter speed dial without a stop at either end, worthless half-stop shutter clicks, putting the M (video) button where the focus button should be, and controls generally too far apart to be operated effectively by normal fingers,

The Fuji is slightly better but offers its own set of problems. The first plus is that it has a little finger grip to enhance traction. And although its shutter dial has full-stop clicks and locks on A, and even though its function button is in the same place as the Leica’s M button, it poses some difficulties as well. First, the Q button (which overrides the optical finder with EVF and presents a grid of settings) is very easily engaged by accident while shooting – because it shares a thumb rest with the AF/AE lock button. The Fuji gets points for a top-deck exposure compensation dial but then loses some points because it is easy to move by accident.

Bulk/weight. The Fuji’s body is an amazingly lightweight box made of aluminum and magnesium parts. Its proportions and size are almost exactly that of a film M. The Leica body is so dense that it feels like it is machined out of a solid block of depleted uranium. It’s also fairly chunky in terms of depth from front to back. So which one is more compact?

Trick question! The Fuji gets its thinness from moving the imager plane to 17,7mm from the mount. This allows plenty of space behind the sensor for the sensor board, screen electronics, buttons, etc.  Because it is one of the shortest registration distances anywhere, it also allows almost any kind of lens to be mounted to the Fuji. The Leica has a registration distance of 27.95mm (or thereabouts – no one has published the exact figure for the digital Ms), which is almost half an inch more. Because digital camera electronics are similarly sized, this makes the Leica body “fatter.”

But that said, Fuji lenses are huge. Consider that many Fuji lenses are comparable in size to – or creeping up on – the equivalent FX DSLR lenses – particularly the 14mm f/2.8, 23nm f/1.4,  35mm f/1.4, and the planned 56mm.  A 58-62mm filter size is something that in the Leica M world is reserved for superspeed or exotic glass. Most Leica lenses clock in at 39mm-46mm (and that is pretty much the entire diameter of the lens. Although that diameter difference does not look like much on paper, it is in real life because the volume of a lens is πr² (cross-sectional area) times the length. So going from 46mm to 62mm at the front end almost doubles the volume of the lens – even before you consider that the hyper-corrected Fuji lenses are longer to begin with and often bulge beyond the filter size at the focusing ring.

That is not to say the system is without tradeoffs. The Leica’s sensor design allows a variety of small lenses with a variety of ray spreads to be used. This will provide good performance with smaller lenses that are more symmetrical and sit close to the sensor and really good performance with more modern designs (often admittedly larger – cf. the difference between the 35/1.4 ASPH and pre-ASPH). The Fuji relies on purpose-built lenses that drill rays straight into the sensor – and it is much easier to engineer really good lenses (the 14mm and 35mm are spectacularly good; the 18-55mm is remarkably good as well) if you can dedicate them to a single sensor and design from a clean sheet. So it should not be surprising that Fuji’s image quality can be competitive with Leica’s.

Responsiveness. A major difference between the two cameras is that the Fuji’s main imaging sensor is active most of the time – because it is used for viewfinding (with the EVF), focusing (both modes), exposure measurement, and the exposure itself. One key aspect of this is the necessity of closing and then reopening the shutter to shoot (so as to black it out). This means that the Fuji is fractionally slower than the Leica, which fires from a closed shutter. The Leica also never feels compelled for perfect focus to be achieved – which means maximizing the moment without a guarantee that things will really be in focus. With the X-Pro1 in AF mode, especially in low light, there can be some hunting around. In reasonable to good light, with a stepper-motor lens like the 18-55, the AF is very fast and reliable.

The Fuji’s processing of images also comes off as slower. The camera can shoot continuously at 6fps (with no between-shots focusing at all), but there comes a point where either (a) the buffer is full and it needs to finish writing or (b) operations like bracketing cause the camera to stop after each group of shots to render and write all the JPGs. Simple continuous shooting is not a problem, though the writing process can interfere with playback.

The Leica writes its files as it goes, and because its forms of bracketing are very simple (exposure only), the camera never finds itself bogged down in reprocessing the same RAW file repeatedly (such as for film type bracketing on the X-Pro1). The M also shoots three frames per second, which does not threaten the buffer as much. The Leica also obeys the command to take the shot, which doesn’t always happen immediately with the X-Pro in AF mode (if you switch to manual focus and use the AE/AF lock button to focus on demand, there is very little difference between the cameras save for simple shutter lag).

At the end of the day, these are both modern cameras – and using the M for a while can be a revelation if your frame of reference is the ker-clunkiness of the M8 and M9, with their thumb-bruising magnifcation wheels and sluggish playback.

Autonomy. Put simply, the Leica has longer battery life both because the camera both because the battery is bigger and because the camera is not normally running the sensor and processing images continuously. The Leica with its percentage remaining display (hit “Info”) gives you much better warning of impending failure than the Fuji’s four segments. On the other hand, Fuji batteries are about 1/5 the price, and chargers are similarly inexpensive.

Flash.  Both cameras can be used with any low-voltage-sync automatic flash in automatic mode. Who cares about that? Let’s talk about their TTL systems instead, using the EF-42 (the only serious flash made for the X series) and the Leica [Metz] SF 58 (the only flash made for serious Leica people).

  1. Both use a preflash system to determine exposure. The Fuji is not particularly fast; the Leica system seems to vary in speed with the aperture used (smaller apertures, less visible preflash).
  2. Both have flip-down wide-angle diffusers. The Leica has a bounce card; the Fuji does not.
  3. The Leica’s accessory diffuser is almost completely flat; for the Fuji, the only real option is a bulkier Sto-Fen.
  4. Both support bounce and swivel; however, the Leica actually locks in place.
  5. Both cameras transmit ISO information to the flash; however, the X cameras can also transmit selected aperture, where M cameras cannot. This bears on the usefulness of the distance scale – with the Leica, you must remember to set the right aperture.
  6. The Leica M, very significantly, can synch to the SF 58 at all shutter speeds. This, coupled with the massive output of the flash itself, makes the SF 58 a useful tool for fill flash outdoors.
  7. The SF 58 has a low-battery warning indicator that tells you when you should change the batteries. This is pretty crucial because with NiMH batteries, if one of those expensive AAs gets drained to zero, it can reverse polarity and be ruined. Alkalines often leak very quickly after they are exhausted. The EF-42 tends to run batteries all the way to zero and burns a considerable amount of power even when in standby. If you are not using it, turn it off!
  8. Due to the presence of an automatic mode and similar pin out to Nikon flashes, the Metz can be used on Nikon D-SLRs, Hexar AF, Hexar RF, and others in auto mode. The EF-42 has a Canon-style layout and lacks an automatic mode that can be used on other cameras.
  9. In a first ever (hath Hell frozen over), a [rebranded] Metz flash has an easier-to-understand control system than a [rebranded] Sunpak flash.

Of course, you would expect a Leica flash that costs 4x as much to be 4x as good as the Fuji flash – and it is. The M with SF 58 is a powerful, tightly integrated, elegant system that delivers good flash shots in all light without much heartache – as as noted, at least its usability on other cameras. But of the factors above, the biggest one is high-speed synch, which goes places (primarily backlit or sunlit ones) that the Fuji with its simple 1/180 sec synch cannot go (the X100 can because it has a leaf shutter).

Use of M lenses.  There is nothing such as a poor-man’s Leica, and by the same token, there is nothing poor about Fuji optics. Fuji lenses are so good at a fraction of the price that you should forget about buying an X-series camera and an adapter just to mount Leica lenses.  Most Leica-mount lenses made before the 1990s underperform the Fuji lenses, some don’t fit the Fuji adapter, and all of them are more of a pain to use than they are worth, even in super-wides (which don’t translate well to the Fuji anyway). Using an EVF to focus manual lenses, focus-peaking notwithstanding, is a slow and tedious exercise. If you need any more convincing on this point, read Ken Rockwell’s screed on lens hacking. We don’t need to reinvent that wheel here.

Lens selection. Part of why the X-Pro1 emerged early with an M adapter was the fact that it had but three lenses: the unloved 18/2.8, the superlative 35/1.4, and the glacial 60/2.4. The reasons to put up with adapted lenses for other uses began to diminish with the 14/2.8 and the 18-55mm f/2.8-4, which was a versatile, fast and compact stabilized zoom lens. Fuji has made good on its road map so far, subsequently releasing the 27/2.8 pancake, the 55-200/3.5-4.8 telephoto zoom, the 23/1.4, and soon, the 56/1.2. The 10-20/4 wide-angle zoom will certainly be interesting. There are also Zeiss 12mm and 35mm lenses. To get the kind of performance coming out of some of these Fujinons requires Leica glass costing two to four times as much money.

Leica’s lens selection does not involve any continuous zooms or AF lenses, but it does include some of the world’s best optical and mechanical designs.  The design of the M sensor does allow the use of most Leica lenses going back to the 1950s, if not the 1930s (the Fuji technically does as well, but the focusing is considerably more cumbersome and the on-sensor performance highly variable), as well as lenses made for the two Leica mounts (thread and bayonet) by Canon, Cosina, Kobalux, Konica, KMZ, Minolta, Nikon, Ricoh, Rollei and Zeiss (among others). That said, the demands of the newer Leica sensors (Monochrom and M) mean that the best performers are often very expensive recent glass.

Both systems, in sum, present a good enough variety of glass to cover most applications – but bear in mind that once you have invested in a lot of glass, it’s expensive to change systems (not unlike owning an iPhone and wanting to switch to Android or switching from a PC to a Mac). Also worth noting is that the X lenses are all for APS-C, so it may not make sense to invest in them if FX lenses are coming with a new system in the future.

While we are on the subject of costs, it is worth noting that Leica bodies – like a lot of other high-end cameras – tend to decrease in value by about $1,000 a year until they bottom out around $1,800 (witness that the M8 – now 7 years old – is still selling for that much, having started at $4,500). Fuji – by aggressively discounting and constantly releasing new models, makes its camera bodies (and some of its lenses) depreciate by about 40% on the first day of ownership. This is the first time that someone has seriously challenged the auto industry in terms of motivating people to buy lightly used equipment instead of new.

Video. I have only heard of one person who likes the M’s jittery handheld video, and that is Thorsten Overgaard. Bear in mind that he smokes 60 packs of Camels per day, so maybe the the jittering while watching synchs with the jittering while shooting, making it all work out well. But seriously, video on the M is a fluid-head tripod exercise (and then it comes out well). The Fuji – on which video is similarly an afterthought – overcomes this handicap by having two stabilized lenses (the 18-55 and the 55-200). But with either, eye-level viewing is a – way to shoot (not in the least for the Leica because the framelines for stills do not reflect the HD video frame). Both cameras allow you to perfect your soccer mom style, but that is not terribly stable, either. The M allows you to shoot looking down (provided you buy the EVF), but that blocks access to external microphones (the Fuji can’t even take those). For video (only) cameras would benefit from tilt-up main screens and separate mic jacks. Whatever. If you want to shoot video with a big sensor, get a NEX.

Eyefi. Both cameras work well with Eyefi. The M now has a plastic insert in the baseplate that allows the cards to transmit out. The Mobi version seems to work a lot better than the X2; pick small JPGs that will transmit quickly into your direct-mode devicd (iPhone, etc.), buy the separate Eyefi reader (see Ebay for this) and use it to move the RAW files to your computer (for some reason, Eyefi doesn’t want you to know that the X2 readers will also work with Mobi cards).  Both the M and the X-Pro1 are respectful of the Eyefi card and will delay going into standby when files are being transferred.

Subject isolation/format considerations. Many people fetishize subject isolation in pictures. If you believe in shallow depth of field, you will get very close to the subject with a medium-to-long lens, shoot at a wide aperture, and probably pick an FX sensor. If your compositions involve near-far relationships (such as in landscape and architectural work), APS-C uses shorter lenses for any given angle of view and gets more depth of field out of a given f/stop. This author believes that the “full frame” vs. “APS-C” debate is overplayed in general.  As a general matter,

  • Using shorter lenses means that APS-C cameras are better suited to shoot sharp pictures in low light;
  • It is easier to manufacture smaller lens elements to high precision than large, leading to more better performance for the money;
  • The performance of APS-C lenses often exceeds sensor performance (where “full frame” rarely does); and
  • The APS-C form factor allows the inclusion of things like ultrasonic sensor glass cleaning and, potentially, sensor-based image stabilization.

Despite the “small” sensor, the X-Pro1 delivers a value proposition in two and likely three lenses so far. First, the 14mm has very close to zero distortion. Nearly every 21mm Leica lens exhibits more distortion than that. Undistorting the Leica lenses (assuming it can be done – the Wide-Angle Tri-Elmar does not have a simple profile) erases any advantage in having 18 vs. 16mp (M9 vs. X-Pro1) and arguably erases some of the advantage of 24mp in the M. Second, the 35mm Fujinon beats the 35mm Summilux ASPH over the 14mm radius and equals the 50mm Summilux ASPH (it’s true angle-of-view equivalent) in across-the-frame performance. It also delivers 50% more depth of field than the 50mm Leica lens, meaning that it’s more likely that focusing errors will be obscured. Finally, the 23mm Fujinon – at least by the charts – will provide very stiff competiton for its equivalent 35/1.4 ASPH. All of this may or may not sway you from an M, but it may make you feel a lot better about using a Fuji.

Ultimate image quality. Which camera comes out on top depends on how you look a the files. At a pixel level, both camera exhibit nearly identical performance. This is not surprising, actually, since each uses a CMOS sensor. This requires a modern lens on the Leica and one of Fuji’s good lenses (a good reference is the 14mm).

The X-Pro1 no doubt excels because within the cloisters of its own lens line, every one is designed for an APS-C sensor and the short Fuji register distance. That said, the Fuji has some white and red smearing (most predictably where there is a white object on a blue background, overexposed – or when the red channel is nearly saturated. But for all of the mostly-thoretical arguments about the X-Trans sensor color layout, the camera does spectacularly well. The X-Pro1 does not do well at all with adapted wideangles designed for the Leica mount. This is not surprising, since its sensor was never designed to handle these lenses.

The Leica M does better with a large cross-section of historic lenses. So far, I have tried it with a 21-35mm Dual Hexanon, 28mm M-Hexanon, 35mm Summilux-M ASPH, 40mm M-Rokkor, 50mm M-Hexanon, 75mm Summarit, 90mm M-Hexanon, and a beater 105mm Nikkor adapted from Leica thread mount. This set represents the gamut of usable focal lengths with a rangefinder, as well as a number of different designs (retrofocus follow-focus zoom, retrofocus wides, traditional normal lenses, highly corrected “economy” lenses, Ernostar atavisms, and long-lens softballs). Leica’s new sensor design handles these much better than the M8/M9, with no visible color shifts and not a lot more than simple vignetting. It does occasionally manifest lateral color, but no more than the X-Pro1 exhibits bleed.

Low light performance is not materially different between the cameras.  At the end of the day, though, the Leica images can be downsampled to the same size as the X-Pro1 files and actually improve apparent performance. This also increases its apparent low-light performance, particularly when you downsample to a D700-sized file.

Productivity. One thing that never seems to get any attention in reviews or comparisons is how productive a camera is. That is not measurable in frames per second but rather in what can be done with the camera during down time on the road. In this aspect, there is a huge difference between the M typ 240 and the X-Pro1.

The X-Pro1 has RAW reconversion capability, which is not a small thing. That means that once a RAW file is generated, you can create additional JPGs from the same file – using different post-exposure compensation, saturation, sharpening, color balance, etc. From there, the Fuji can also re-crop the frame. In fact, the Fuji X series has about the best post-processing setup of any camera that exists – and it allows the user to get great results with a minimum of underlying RAW files.

The Leica does not have any such capability. As a result, whatever is generated as a JPG has to suffice unless and until the user gets to a computer or transfers the RAW file to another device (which for the iPad, unfortunately, there is no brand-name DNG converted). The workaround is to bracket more heavily, but because the bracketed shots are spread out over time, they may miss the moment. And the bracketing does not account for the other variables that on the Fuji can vary in its own version of post. One practical impact that may be obscure at first is diminished file space on SD cards. If you are shooting to get a good-looking shot for a direct transfer via Eyefi, you will end up shooting 3 to 5 times as many pictures – because the JPG that looks best on your portable device is not guaranteed to correspond to the best DNG file. You can’t selectively delete JPGs on the M (or anything else for that matter).

Conclusion. Since the time I was very young, I’ve heard people at local camera stores tell people that x, y or z was “as good as a Leica,” whether that be a film point-and-shoot, an early 1-Mp digital camera, or whatever. It is not easy to say whether an X-Pro1 is better or worse than an M typ 240; their feature sets overlap to a large degree and then go off in different directions. One thing is for certain – if you are mentally locked into rangefinders, buying an X-Pro1 will be a waste of time and money. And if you want more of a do-everything camera, buying an M might result in your buying a second camera. Those two things are about all you can say without making someone, somewhere go crazy.