By Ctein
In the many online discussions, here and abroad, about Michael Reichmann's letter to Leica, a point oft brought up (or tacitly assumed) is that there is nothing wrong with existing focusing systems and therefore Michael's exhortation to Leica to "think differently" is inherently misguided. I'd like to correct that.
I have no horse in this race. If Leica builds a super-cool camera, that's nice, but I probably won't buy it. I've used about every kind of focusing system at one time or another, and they all seem to work well for me, just with different strengths and weaknesses. One of my favorite cameras for many years was my Canonet G-II QL 17 (I wasn't as fond of Leica rangefinders, but that's purely a case of one shutterbug's ergonomic ceiling being another's floor; it's no criticism of the design. Just didn't work well for me).
First, understand the difference between precision and accuracy. If you have a wooden meter stick marked off in millimeter increments, you can measure things to a precision of about 1/2 mm. Want to bet that that meter stick is actually one meter long? The Leica website talks about the high precision of their rangefinder designs. Their engineers are admirably careful in their use of technical language. They don't talk about how accurate they are. That's because they're not, particularly; I've tested them.
Rangefinder cameras, whether they have interchangeable lenses or not, are heir to all sorts of tolerance problems in the mechanical linkages that degrade the match between indicated focus in the viewfinder and real focus in the film/sensor plane. A very long baseline lets you set the focus back to precisely the same point, time after time, very quickly. It may not be the point you actually want to be focused on.
I have never, ever tested a rangefinder camera, fixed- or interchageable-lens, which was particularly accurate over the entire range of distances. None of them were better than your typical SLR; many of them were substantially worse. This design simply doesn't lend itself well to extreme accuracy, except at fabulous expense, even by Leica standards.
SLRs must be aligned too
So, SLRs are the way to go? Unfortunately, no. In SLRs, there's one big problem: the distance from the lens to the focusing screen/focusing sensor usually isn't exactly the same as the distance to the the film/sensor. Now, there are easy ways to adjust that in an SLR; there are setscrews that determine the mirror rest position. You can bring the focusing plane and the film plane into near perfect alignment. I routinely do that.
Unfortunately, because a flapping mirror is involved, a tuned-up camera will not stay tuned up. One important difference I've found between robustly built cameras and "amateur" cameras is how fast they'll go out of alignment. In some cameras, a heavy day's work is enough to shift the focus measurably.
I have never seen a camera come from the factory in critically sharp alignment for use with a reasonably fast prime lens. Common wisdom is that most fast lenses are not very sharp at maximum aperture in the center of the field (hardly any lenses are uniformly sharp at any aperture over the entire field). That mis-impression comes from the fact that hardly any fast lenses get accurately focused! I've found that tuning up any camera I used will improve wide-open resolution by 50–100%. Right out of the box, cameras can be annoyingly inaccurate.
There's a third problem that all focusing systems face. Most work best at large lens apertures, but any lens that doesn't have perfectly corrected spherical aberration will undergo a focus shift as it's stopped down. Often the shift is so small, in comparison to the increased depth of focus, that it doesn't produce any important loss of sharpness. Sometimes it does.
One crude idea
This is all stuff that Michael knows (and didn't think needed elaboration), which is why he encourged Leica to try thinking outside of that box, figuratively and literally. Could focusing systems be substantially improved with innovative approaches? Unquestionably! Here's a crude idea I came up with after just 10 minutes thought:
Take a traditional coincident-image rangefinder. Add a piezo actuator that can shift the sensor in the z-axis (image stabilization systems use actuators to shift the sensor on pitch and yaw; we're just adding a third axis of control). Add another processor chip that looks only at the pixels in the sensor that match the coincidence window in the rangefinder viewfinder (limiting the data crunching to a small number of pixels means the device can operate in real time). The photographer focuses manually, or so they think. But once the focus gets close, the processor takes over and starts shifting the sensor until it actually is in exact focus. Then it flashes a little signal in the viewfinder to tell the photographer that the subject is in focus. The photographer thinks they're focusing manually, but what they're really doing is the rough focus and then handing it over to an autofocus system. Heck, with improvements in sensor sensitivity and signal processing, you could even have such a system work with the lens stopped down, eliminating spherical aberration focus shift.
Sure, this idea's less than half baked, and I have no doubt that folks can poke major holes in it. But if I can think up something like this so quickly, imagine what proper camera engineers could do putting in some serious work time on the effort. It's a proof of principle.
Your existing focusing systems are lousy. They can be improved.
Send this post to a friend
Note: Links in this post may be to our affiliates; sales through affiliate links may benefit this site. More...