In most of the last half of the 20th century, the weapon of choice for low-light candid photography was the 35mm rangefinder camera. There were lots of reasons.
- The cameras were small and light.
- The focal-plane shutters were quieter than SLR shutters (except for the tiny number of SLRs with pellicle mirrors), and not much noisier than the between-the-lens shutters of TLRs and 4x5s.
- Focusing was faster and more accurate in dim light than with an SLR or TLR.
- Fast lenses were available. Fifty millimeter lenses usually had f/1.4 or f/2 apertures, which was fast compared to the lenses on 2 1/4 square cameras, and really fast compared to those on 4x5s. f/1 lenses were available, although they were huge and expensive. To be fair, soon after 35mm SLRs became popular in the early 1960s, they got similar lens choices.
- Films such as Tri-X allowed high film speeds with acceptable grain. Actually, the use of these small negatives with high speed film produced prints with a pronounced grain structure, which, over time became acceptable to viewers, and even became a kind of advantage as the look grew to be associated with gritty, unposed images. Push-processing increased contrast and accentuated the raw look.
Ten years into the new millennium, things have changed. Digital sensors first improved to the point that they offered superior image quality to high-speed films at the same sensitivities. In the last few years, image quality has continued to improve, while usable sensitivities have skyrocketed. To my way of thinking, the breakthrough camera in this regard was the Nikon D3, which allowed creamy-smooth, high-dynamic range images at ISO 3200. The D3s has recently raised the bar even higher. Canon has recently introduced a camera for which they claim great low-light performance, although I haven’t used it and can’t verify their assertions.
In the eighties, autofocus cameras began to be popular. At first they weren’t so wonderful in dim light, but the technology available in professional-level cameras has advanced to the point that in all but the very darkest situations — so dark that you need a tripod — the camera can focus much faster and more accurately than you can do manually.
In the 1950s, the 35mm rangefinder was the mainstream camera. By the end of the 60s, it had been pushed aside by 35mm SLRs. At the end of the 1990s and the beginning of the following decade, as there got to be enough professional digital cameras for the concept of a mainstream to be meaningful, it was dominated with 35mm-sized SLRs, first with smaller than 35mm (24x36mm) sensors, with full-frame sensors growing in popularity.
Until recently, the 35mm-sized digital rangefinder camera was a mostly-unfilled niche. That all changed with the introduction of the Leica M8, with a sensor with a 1.3 focal length multiplier, and a few months ago with the M9, an otherwise-similar full frame camera. Both can use almost any M-series lens, many of which are truly remarkable. Both handle like a real 35mm Leica M-whatever, and both preserve most of the advantages and disadvantages of the legendary line.
However, for low-light photography, the world has changed since the 1970s, and the biggest changes served to improve the digital SLRs that are the M8 and M9’s competition. I mentioned low light autofocus before. In dim light it’s so difficult for me to focus an M8 with a 50mm f/1.4 lens that at a few feet only about a tenth of the images shot wide open are critically sharp. With a D3, achieving precise focus is easy. With the combination of continuous autofocus and movable aim points, it’s not hard to get crisp images of moving subjects, something I wouldn’t even attempt with an M8.
The D3 at ISO 4000 has about the same noise as the M8 at 400. That’s more than a three-stop advantage. Let’s say you were in a situation where you really needed to use the $10,000 f/0.95 Leica Noctilux at full aperture. Let’s be generous and say that, because of the absence of mirror slap, you could use a shutter speed a full stop longer than with the D3. Because of the sensitivity advantage of the Nikon, you could use an f/1.8 lens on the D3, and it would be stopped down two-thirds of a stop.
The Noctilux is a sharper lens than the Nikon lens (it should be; it costs eighty times as much), but the vastly improved odds that the Nikon pictures will be in focus more than nullifies that advantage. As a bonus, you’ll have a lot more depth of field at f/2.2 than at f/0.95. It looks like the M9 and the D3s have improved low light performance by about the same amount, so the comparison should hold with the newer cameras.
The big Noctilux cuts into the weight advantage a rangefinder camera traditionally enjoys: the M8 or M9 with the fast 50mm lens weighs three ounces less than the D3 or D3s with the f/1.8 lens.
What’s left as an advantage for the Leica in dim light? The shutter is quieter, you can see things outside the image area in the finder, and the M9 has higher resolution. The disadvantages are higher cost, inferior battery life, slower operation, the obscuring of the lower right corner of the finder by the lens, and less precision in framing because of the bright-line finder and finder parallax, in addition to the reduced depth of field and focusing difficulties previously discussed. Doesn’t seem like a difficult decision to me.
Maybe outfitting the Leica with the superfast lens is a little extreme. Going for an f/1.4 lens on the Leica instead of the f/0.95 saves a ton of money and a pound of weight. Now, for the same image noise, the Nikon lens needs to be stopped down a little further than f/2.8. That raises an interesting possibility. Now, maybe the Nikon lens for the comparison is the 24-70mm f/2.8. The combination of that lens and the D3 is definitely going to lose the size and weight contest, but it’s going to provide sharper real-world images in dim light, because of the focusing difficulties of the rangefinder camera under those conditions and the additional depth of field. The extra versatility of the zoom lens is a big bonus in many circumstances, although it goes is the wrong direction if you buy into the Leica minimalist approach.
If you go longer than 50mm, the difficulties in focusing the rangefinder lenses get even greater, so the results favor the D3 even more. But what about shorter? Many people would say the standard focal length for a Leica is 35mm for the M9, and maybe 24mm (translates to 32mm) for the M8. Leica offers an attractive array of fast lenses in the 21 to 35mm range. In the street, lenses 35mm and wider can be zone-focused, and you don’t even need to mess with the rangefinder — the generously sized depth of field markings on the Leica lenses make zone-focusing really easy. But when things get really dim, you’re a few feet away, and the aperture is set wider than f/2, you need the rangefinder. The percentage of in-focus shots is higher with the shorter lenses because there’s more depth of field, and parallax is less of a problem with the wider angle of view. All is not roses, however. You can only go as wide as 24mm on the M8 and 28mm on the M9 before you need to use an auxiliary viewfinder. At wide apertures, going to the shoe-mounted viewfinder creates problems beyond convenience; the time spent shifting from one finder to the other mean that small movements of the subject will mean fewer sharp pictures, and there’s that temptation while you’ve got the subject perfectly framed in the auxiliary finder to try to compensate for small motions by guessing what’s going on and moving the camera, a really tricky feat to pull off. In addition, the 24mm lens, when mounted on the M8, occludes the lower right corner of the frame in the finder.
When you get really wide, the Leica lenses lose their speed advantage over the Nikon ones. Indeed, the situation is reversed, with the Leica 18mm and the 16-21mm zoom (marketed as a triple-focal-length lens and not a zoom because the auxiliary “Frankenfinder” has only three focal length settings) both having apertures within a third of a stop of f/4, and the Nikon 14mm, 20mm, and 14-24mm zoom all having f/2.8 apertures.
So, in sum, the only range of focal lengths where the Leica rangefinder is even close to being as good a tool for dim light photography is between 21 and 35mm, and that’s only if you seriously value the Leica’s small size, light weight, and quiet shutter.
I can hear you thinking: “OK, so Nikon got the jump on Leica with their sensor technology. Surely Leica will catch up.” Well, they may make some progress, but they’ll not be playing on a level field because of decisions made by both Nikon and Leica in the 1950s. It’s amazing how decisions can have repercussions far beyond the horizons of the decision makers. Perhaps you know how the Roman Empire chariot wheel-to-wheel spacing determined the standard railway gauge two millennia later. If not, take a look here.
A crucial specification in camera design is the flange focal distance, or back distance, the separation between the image sensor plane and the flange that the interchangeable lens mounts against. For the Leica M mount, the standard used on the M8 and M9, the distance is 27.8mm. The Nikon F mount flange focal distance is 46.5mm. The reason the Nikon distance is greater is to allow room for the mirror to flip up. The designers of the M mount, because they had no mirror to worry about, picked a more compact configuration that gave them a smaller camera. In addition, SLR lenses (with a very few exceptions, like the original Nikkor 8mm fisheye, which required that the mirror be locked up before the lens was installed) are designed so that the lens elements don’t extend back into the camera body, to make sure that the mirror doesn’t hit the back of the lens. Lenses for rangefinder cameras are not limited in that way.
Because the M mount lenses sit closer to the focal plane than SLR lenses, as you move away from the center of the image, the light hits the image sensor at a greater angle in he rangefinder camera. With film sensors, this wasn’t such a big deal; maybe there was a little more light fall-off towards the edge of the image. With the exception of the Foveon sensors, digital cameras have an array of filters over the light-sensitive semiconductors to convert what is basically a monochromatic device into a color one. There are microlenses over the filters to make the sensor more efficient. These lenses don’t work right when the light comes from way off vertical. Leica’s solution is to offset the lenses as you go further away from the center of the image. It works pretty well, but it’s got to be a compromise since different lenses have different characteristics in this regard. Because Leica needs a sensor with special lenses, it can’t just employ the best device around.
Eric Hanson says
Thanks for this explanation. Jin. I feel I can explain the Leica M sensor better than before.