M8 focus shifts and "back focus"

[pklein]

Advertisement (gone after registration)

Folks: I've done some systematic checking of my current lenses on the M8, and I've had a bit of a surprise. The issues of focus shifts are very real, and lenses that seemed fine on film can appear slightly out of adjustment on the M8. The good news is that mostly the effect is very slight. You would see them for sure on a 16x20, but they might not be significant unless you "smelled the print" on 8x10 and smaller.

 

Is this a real problem or not? I don't know, and I'd like some advice.

 

I may just be seeing the difference between perfection and "within tolerances." Not to mention the results of being able to pixel peep at 1:1, on a digital sensor that doesn't have the fudge factor you get with film thickness. Oddly, I've checked the focus of my lenses on film in the past. I never saw anything like this before, with one exception.

 

For this test, I focused on a tape measure on the floor, using a point about 4 feet away, at an angle of about 40 degrees. I have done this both with a tripod and handheld sitting on the floor, and gotten similar results. Even handheld, the results are repeatable. If I do the test several times, I will get the same results the majority of the time. I have not tested at farther distances.

 

Let me tell you what I've found, and perhaps the more experienced M8 veterans can tell me what (if anything) I need to do. None of my lenses are coded or adjusted specifically for the M8. I've used them for years on two film M bodies with no focusing issues.

 

35/2 Summicron (Version 4, pre-ASPH, early 1980s): From f/2.8 - f/8, the lens focuses spot on. At f/2, it focuses about 2 inches closer than the point on which the RF is focused. So wide open, I need to focus on somebody's ears to get their eyes in focus. For the rest of the range, no adjustment is needed. This is the one lens that I think behaves similarly on film.

 

35/1.4 Summilux ASPH: At f/1.4 and f/2, the point I've focused on is within the zone of sharpest focus as seen when the file is viewed at 1:1. From f/2.8 - 5.6, it back focuses about two inches (the tape measure reading adjusted for the angle of view). The depth of field increases mostly to the rear as I stop down. So to get a person's eyes in focus, I should probably focus on the ears at f/2.8 and narrower.

 

50/2 Summicron (tabbed, current optical formula, early 1980s): Similar behavior to the 35 'Lux ASPH. Spot on at f/2, back focuses about 3/4 inch at f/2.8, and back focuses about 2 inches at f/4 - f/8.

 

50/1.5 VC Nokton: Spot on at f/1.5. At f/2 - f/8, back focuses about 1 inch.

 

I have three 90mm lenses, a 90/2 Summicron-M, a 1960s 90/2.8 Elmarit and a VC 90/3.5. They *all* appear significantly off at 4 feet--more then the 35 or 50mm lenses. I'll have to check them again, as I didn't write down which did what. Two back focused and one front focused. My VC 21/4 is *way* off on the M8, more than any other lens. But here's the funny part: I've never noticed any issues with them on film.

 

Once I figured out the approximate error of a particular lens and stop, I could nail the focus almost every time by compensating (usually by focusing and inch or two closer). But one of the reasons why I use Leica is so that what I've focused on is what the lens is actually focused on. So all this is a bit unsettling. Advice, anyone?

 

--Peter

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[lct]

...35/2 Summicron (Version 4, pre-ASPH, early 1980s): From f/2.8 - f/8, the lens focuses spot on. At f/2, it focuses about 2 inches closer than the point on which the RF is focused....50/2 Summicron (tabbed, current optical formula, early 1980s): Similar behavior to the 35 'Lux ASPH. Spot on at f/2, back focuses about 3/4 inch at f/2.8, and back focuses about 2 inches at f/4 - f/8...

I happen to use those same lenses with Epson R-D1 with no problem at all.

Sounds unbelievable given that the RF base length of the R-D1 is shorter than that of the M8.

May i ask you if your lenses have been CLAd recently?

[rvaubel]

Folks: I've done some systematic checking of my current lenses on the M8, and I've had a bit of a surprise. The issues of focus shifts are very real, and lenses that seemed fine on film can appear slightly out of adjustment on the M8. The good news is that mostly the effect is very slight. You would see them for sure on a 16x20, but they might not be significant unless you "smelled the print" on 8x10 and smaller.

 

Is this a real problem or not? ...........My VC 21/4 is *way* off on the M8, more than any other lens. But here's the funny part: I've never noticed any issues with them on film.

 

. But one of the reasons why I use Leica is so that what I've focused on is what the lens is actually focused on. So all this is a bit unsettling. Advice, anyone?

 

--Peter

 

Peter

 

When I first boRD1 I spent a lot of time checking the rangefinder mechanizime against all my various lenses. It turned out I learned more about my lenses That 100% is very revealing about issues of backfocus.

 

My ultra wide 21mm Kobalux was the worst offender but it is very hard to detect on a wide angle because of the huge DOF. The long and the short of it is, yes lenses whether Leica or not can have back focus problems. For the most part, other factors are more significant.

[pklein]

LCT: No CLAs recently, but the lenses don't do this on either of my film bodies. And I just shot several rolls of film with a couple of the lenses mentioned.

 

I don't think this is a base length issue. I think it has to do with film having a thickness, and therefore silver "grains" that can respond most strongly to the in-focus rays regardless where in the film they fall. The active sensor plane is flat, so there's no "slop factor" as there is with film.

 

Remember, the lens focuses correctly wide open or nearly so, it focuses back as you stop down. The effects are fairly subtle. On the 35 and 50mm lenses I've checked, the point on which I've focused is just in front of the zone of focus as observed at 1:1 pixel-peeping., in At 2:1, you barely notice it, though the focused point appears on the near edge of the acceptable zone, or just barely outside it.

 

Others have discussed these focusing issues here and on other forums, so I don't think I'm the only one experiencing this. We've also heard from someone who says Leica told that the 35 'Lux ASPH does indeed behave this way.

 

--Peter

[pklein]

Peter

 

When I first boRD1 I spent a lot of time checking the rangefinder mechanizime against all my various lenses. It turned out I learned more about my lenses That 100% is very revealing about issues of backfocus.

 

My ultra wide 21mm Kobalux was the worst offender but it is very hard to detect on a wide angle because of the huge DOF. The long and the short of it is, yes lenses whether Leica or not can have back focus problems. For the most part, other factors are more significant.

 

Thanks, Rex. Yes, when I first looked at the files at 1:1, it was like discovering I'd been living in blissful ignorance all my life! We now can measure things that were beyond our ability to see before. The question is whether it matters in real photography, and at what print size. I'm certainly going to take real pictures and print them at my normal sizes before I decide that.

 

--Peter

[rvaubel]

Thanks, Rex. Yes, when I first looked at the files at 1:1, it was like discovering I'd been living in blissful ignorance all my life! We now can measure things that were beyond our ability to see before. The question is whether it matters in real photography, and at what print size. I'm certainly going to take real pictures and print them at my normal sizes before I decide that.

 

--Peter

 

Yes, I was pretty upset when I first discovered, thru the wonders of pixel peeping, that my optics didn't focus exactly to the point I had imagined. Part of that problem, was of course, the fault of the rangefinder mechanizm in the RD1. But open a lot more testing, I finally came to the conclusion that simple factors like head bobbing and body sway were more significient factors than focus shift or even rangefinder accuracy. That is, if you talking about a few inches of focus shift, the error has about an equal chance of making the focus BETTER than worse. Think about it...focusing without a tripod and with all the other uncontrolled variables is pretty much a crap shot anyway. What practical difference does an inch or two of back focus make anyway?

 

Rex

[colorflow]

Advertisement (gone after registration)

Peter:

 

I did exactly the same as you. I posted it a while ago titled (if I remember correctly) "Semi-rigorous focus test...". My results were also the same, i.e. all my lenses shift focus back as aperture is stopped down. I tested Cron 28, Cron 50, Lux 35, Lux 50, Lux 50asph, lux 75, cron 90, cron apo 90, and elm 90. Some back focus and some front focus a bit wide open. None is spot on. The Lux 50 asph backfocus 3 or 4 inches at 6 ft so it is back to NJ. The Lux 50 vers 2 front focus by 6 inches and is being adjusted by Focus Point Inc.

 

I also ran same test with M6 and ISO 400 film I had handy. Had developed and scanned at Cosco. Results agree with M8.

 

Alan

[colorflow]

Yes, I was pretty upset when I first discovered, thru the wonders of pixel peeping, that my optics didn't focus exactly to the point I had imagined. Part of that problem, was of course, the fault of the rangefinder mechanizm in the RD1. But open a lot more testing, I finally came to the conclusion that simple factors like head bobbing and body sway were more significient factors than focus shift or even rangefinder accuracy. That is, if you talking about a few inches of focus shift, the error has about an equal chance of making the focus BETTER than worse. Think about it...focusing without a tripod and with all the other uncontrolled variables is pretty much a crap shot anyway. What practical difference does an inch or two of back focus make anyway?

 

Rex

 

Rex:

 

In most cases not much. But for portraits with the Cron apo90 and lux 75 it is a big difference.

 

Alan

[jsjxyz]

Peter:

 

I did exactly the same as you. I posted it a while ago titled (if I remember correctly) "Semi-rigorous focus test...". My results were also the same, i.e. all my lenses shift focus back as aperture is stopped down. I tested Cron 28, Cron 50, Lux 35, Lux 50, Lux 50asph, lux 75, cron 90, cron apo 90, and elm 90. Some back focus and some front focus a bit wide open. None is spot on. The Lux 50 asph backfocus 3 or 4 inches at 6 ft so it is back to NJ. The Lux 50 vers 2 front focus by 6 inches and is being adjusted by Focus Point Inc.

 

I also ran same test with M6 and ISO 400 film I had handy. Had developed and scanned at Cosco. Results agree with M8.

 

Alan

Does it mean actually they have a problem too with our film M series but didnot discover it because we sheldom print bigger than 8X10?

 

Jerry

[colorflow]

Does it mean actually they have a problem too with our film M series but didnot discover it because we sheldom print bigger than 8X10?

 

Jerry

 

I think so based on my film test, but will let the experts conclude.

Alan

[Guest guy_mancuso]

Does it mean actually they have a problem too with our film M series but didnot discover it because we sheldom print bigger than 8X10?

 

Jerry

 

When was the last time you blew a neg up 100 times and put a lupe to it, never i am sure. The best we had was a 8 x lupe looking at a 24x36 neg or chrome, certainly could have missed a lot and did not realize it

[ho_co]

Is this a real problem or not? I don't know, and I'd like some advice.

Peter--

I think you've hit the point. Tim Ashley and some others have had some specific lenses that exhibit problems far greater than you've discovered. Solms responded to Tim's queries that (for the most part) the behavior you describe is normal.

 

Generally speaking, lens and rangefinder should match wide open. As you stop down, point of best focus will move away from the camera but should stay within the depth of field, which is also growing as you stop down. (The focus shift is caused by spherical aberration; the more you stop down, the more of the aberrant rays you mask.)

 

See also E Puts' article at Testreports.

 

Basically, we've never been able to take a picture before and see it 30 seconds later at 1:1 or 10:1 or greater. What you're becoming aware of is what has made these lenses so insanely good for years. (Some people have some focal lengths that exhibit these tendencies in very unsubtle and exaggerated ways, but your description makes it sound as if things are actually working correctly.)

 

And if I'm interpreting Alan's results correctly, after his surprise results with the M8, his film tests show that the behavior has always been there, but unnoticed.

 

The current issue of LFI was to have had a look at the topic according to issue 3/2007, but the magazine seems to have thought better of publishing and attracting any more attention to the matter. (That's my interpretation of why the article isn't there; maybe the very long and excellent, sometimes exciting interview with Steven K Lee simply didn't leave them the space for the promised article.)

 

We've grown too picky for the most part. We finally have a sensor that shows exactly what the lenses are doing--we never had that ability before, and now we're surprised to find out just how the lenses perform their magic.

 

--HC

[wlaidlaw]

When was the last time you blew a neg up 100 times and put a lupe to it, never i am sure. The best we had was a 8 x lupe looking at a 24x36 neg or chrome, certainly could have missed a lot and did not realize it

 

The other factor on film is how good is the enlarger lens. I used to have access to a friend's darkroom before his divorce. He had eye-wateringly expensive Horseman and DV Gilde medium format cameras but was using 50 year old Wray lenses on his enlarger. When he bought a couple of modern Schneiders, his prints improved consderably. With digital you obviously don't have this fudge factor.

 

I have decided not to run any more focus tests on any of my lenses, getting more paranoid with each test, as I don't seem to be having problems in real life prints at A3+ apart from operator error and occasional camera shake. I have noticed with the Noctilux that if I come up from near focus rather than coming down from far focus, I get a better result at f1.0. I don't know whether this is the way my eyes like to work or all the tolerances/permitted inaccuracies in the rangefinder mechanism coming together.

 

Wilson

[lars_bergquist]

I have been told I am totally wrong, the 'back focus' we hear about is not focus shift caused by spherical aberration, but mechanical errors of focusing. So the phenomenon is not affected by stopping down. Stupid old man!

 

And still all four examples above (Peter's posting) speak of focus shift caused by stopping down. (The jury was still out on the three 90 mm lenses.) In fact, nearly all cases I have read about and which have been well reported, not just vague whining, have been about focus shift by stopping down. So who is right, I or the 'experts'?

 

Remains the question how a lens can have enough spherical to move the plane of best focus several inches simply by being stopped down a couple of stops, and still produce not only an acceptable but even a good image wide open. I do admit that this stumps me.

 

We were told that the problem would be treated of in the May issue of Leica Photographie International. This would appear on May 2nd, but I have not received my issue yet. Has anybody seen it?

 

The old man from the Age of Astigmatism

[ho_co]

@ Lars, as I mentioned above, that article isn't in the magazine.

The current issue of LFI was to have had a look at the topic according to issue 3/2007, but the magazine seems to have thought better of publishing and attracting any more attention to the matter.

 

Puts and Leica both say it's spherical aberration; why shouldn't you?

 

I don't think anyone is more expert than you.

 

Leica obviously uses the shift in focus, particularly in the aspherical lenses, to keep the performance wide open so good.

 

Gee, now you're confusing me. I thought this was long since recognized as spherical aberration tuned to make good pictures. Tim and a couple others have some deviant lenses, but the description here is exactly what Leica and EP and standard definitions of spherical aberration call for. Why are the M lenses so good? Because Leica can work with all the tools. Unless I'm more confused than I think, you couldn't design lenses for a reflex to work this way? Am I right, experts? Help!

 

 

 

@ Wilson--very good point that there's an extra optical element when enlarging from film. That's one more intermediate step that marred the response we could retrieve from our lenses. Again, our chain is getting shorter and we are seeing better how Leica has worked their magic!

 

I have noticed with the Noctilux that if I come up from near focus rather than coming down from far focus, I get a better result at f1.0.

I have no experience on the topic, but my understanding is that in motion picture production, focus is done by rotating only in one direction, and aperture is set by rotating only in one direction.

 

 

 

@Rex--also a very good point in your following post, that we can test our lenses all we want, but the deviation from our idea of perfection is as likely as not to work in our advantage just because both we and the subject aren't frozen!

 

--HC

[rvaubel]

 

I have decided not to run any more focus tests on any of my lenses, getting more paranoid with each test, as I don't seem to be having problems in real life prints at A3+ apart from operator error and occasional camera shake.

 

Wilson

 

My earlier point is that in ordinary non-tripod sort of photos, your body and head sway is around +- an inch or two. Now combine that with subject movement, even a portriat, and the total deviance from what you would get from a tripod setup and a test target is more than any backfocus irregularities. On top of that, pixel peeping is a extremely deviant form of photography that in no way represents real life inspection of real prints. Not that I don't enjoy pixel peeping as much as any other form of self gratification. But really, although very interesting, and fodder for inumerable forum threads, the practical implication for all this angst is really minimal. Like I said before, the chances are almost 50- 50 that the backfocus of a lens might bring a real life subject into BETTER focus.

 

Rex

[lars_bergquist]

Howard, I missed your posting but I do mostly agree with you. There is always some residual aberrations, because beating one of them down to zero will actually increase some others. Every lens design is a compromise. So OK, residual spherical aberration does cause some focus shift in all lenses, especially fast wide-angles. The big question in my above posting does remain however. Can the shift really have been that big all along? Can it be that big, irrespective of the sensor (film or chip)?

 

I also agree with Rex. There are so many causes of error in normal, hand-held photography that introduce an element of chance. Just think of what happens when you focus in landscape format (as most of us do) and then turn the camera by 90° into portrait format? Can we do that without moving the camera along the optical axis? Can we stand, or even sit, absolutely immobile? Can a living subject do that, for that matter? Perhaps we should only photograph freeze-dried specimens. – Given this, we should perhaps be more free with the shutter release. We have a camera with a built-in winder. Memory cards are cheap. Bad frames can be erased into oblivion.

 

I was once told: Always fire two rounds at your adversary. One should connect. If the second does not hit, don't worry about the third, start on your prayers. Most of our subjects don't shoot back, so we can fire three times, or four.

 

Still, those reports of focus shifts measured in feet do worry me. They seem unreasonable. I still think that sloppy experimental procedure or plain pilot error must be behind many reports. The main culprit may be that monitor ...

 

The old man from the Age of Sharpshooting

[ho_co]

Lars--

I know, it's asking a lot to say that this focus shift is all intentional and used to our benefit. I think Tim reported that one of his lenses shifted focus away from the camera up to a certain point, and then the focus point started moving back toward the camera...

 

Is that superb use of complex aspheric design or urban legend?

 

Rex is right. We've got a lot of other things that can mess us up; discovering new optical terminology won't help with making pictures.

 

I think Peter describes an interesting but normal phenomenon. I'm sure the gnomes are laughing at us as we ponder classical optics and they're considering whether discovery of the Higgs boson might give them new design capabilities.

 

--HC

[Advertisement]

Hallo,

Look here: Leica M

[wlaidlaw]

Lars--

 

 

I think Peter describes an interesting but normal phenomenon. I'm sure the gnomes are laughing at us as we ponder classical optics and they're considering whether discovery of the Higgs boson might give them new design capabilities.

 

--HC

 

Howard,

 

This has already been covered in another thread. It was decided that the Higgs Bosons and other sub-atomic quarks were being contained by the small screw on the accessory shoe, which is why it is essential not to remove it.

 

Wilson

[carstenw]

Peter, I am surprised that no one has mentioned this so far, but two things are missing before conclusions are in order. First of all, is your rangefinder correctly adjusted? The close focus *seems* good, but how is infinity? It may be that one or both is out, causing the focus of the lenses wide open to be pushed in one direction compared to perfect, and that this is enough that the focus shift when you stop down jumps out of the proper range. Some of the lenses you list are known to have significant focus shift, but others are normally fine, leading me to question this. Secondly, when you continue stopping down, do you get the focus back? That should be the case, but if it is not, something is really out of whack, and you need a CLA for sure.

 

 

If I were you, I would send the 35 Lux Asph and the 90 Cron along with the M8 to Leica, and get them set up as perfectly as possible, since these are the "tough" lenses, depth of field-wise. After that, forget small errors visible in pixels, because pixels are printed at least 240 to the inch, and you simply won't see a *small* softness in print.