How Important is Infinity Focus?

[glenerrolrd]

In my seemingly never ending quest to get my 50 summilux and 50 noctilux properly calibrated, a new issue has come up and I could use some help. The summilux after 3 round trips to Solms ..seems to focus accurately at 3ft ,8ft,20ft and probably 200yards. However the "collimation test" shows that it does not achieve infinity focus. This is a pretty absolute test done with a piece of test equipment and providing a displayed set of specifications. So its absolute ..not a judgement call . This is a test of the lens and does not involve an M8 rangefinder. I believe in another thread someone pointed out that not achieving infinity indicated some fairly significant problems with the lens. So that even if you could focus it ..it may not be up to its full potential. I don t know could someone provide better insights?

[cbretteville]

Of the first four lenses I sent in for coding (late Nov '07) three are back, one is still in Solms as it needed a CLA as well. Of the three my 50 Elmar-M no longer achieves infinity focus. That is unacceptable and it is going straight back.

- Carl

[glenerrolrd]

Carl How did you test for infinity focus? And ..how can you tell if its the lens or your M8? Does this show up in your images?

[cbretteville]

Its so bad the rangefinder patch dosen't show one image if the subject is a few hundred meters away and the lens is at infinity. It is the same on my M6 and it is the only lens that shows this so I don't think its the RF that is out of whack. It wasn't like this when I sent it in, I've used this lens a lot and noticed the change immediately.

I'll take it to the Norwegian distributor next week so they can have a look at it too. The coding was handled through them the last time so get to do it again.

- Carl

[dante]

What are you shooting at f/1.0 and longer distances?

From what I understand, the focus point shifts with all superspeed lenses - and in the Leica world (at least historically), the focus was set so that there was mild front focus wide open and so that as the lens stopped down and the focus point moved backward, the focus point was always well within the DOF of the lens.

I just went through the opposite problem with my 75 Summilux - collimated perfectly by Leica at infinity (actually, a little too good wide open given the above) and back focusing at every other distance (a little at f/1.4 and progressively more at smaller apertures).

I think if you achieved what you say you want, you wouldn't like the result every much.

[marknorton]

If you want to adjust or measure anything, you have to have some reliable reference - or proxy for it - to relate to and the closer that reference is to what you're measuring, the better the result and the higher your level of confidence. If you have a voltmeter, you rely on its readings because it was calibrated at some point against a dependable reference and you hope it hasn't drifted off since then. If you have a ruler, you rely on the people making it having done a good job in adding those markings.

We've discussed endlessly here how adjusting the camera and lens to work together might work out but might also mean you end up with a non-interchangeable lens camera - the body won't work with other lenses, the lens won't work on other bodies. Strictly, the lenses and cameras need to to be adjusted to separate, independent standards.

The advantage of an infinity test is that you can make it without letting that can of worms called the Leica M Rangefinder get in the way. Simply mount the lens, set to infinity and see if an object at infinity is in focus. The most basic focus test you can put an M body and lens to because it is made completely independently of the rangefinder.

Question is, how do you find an item at infinity? That's where the collimator comes in - it looks to the camera to be projecting an image from infinity, even though it's a short distance in front of the lens. A reflex collimator allows to you inspect the image projected back out by the camera for correct focus. Projected back out? Yes, any image relected off the film or sensor surface causes the camera to act as a weak projector and a reflex collimator allows you to inspect that return image.

So, I think an infinity test is important because it is independent of the rangefinder and establishes a reference for then calibrating it.

[Tony C.]

Originally Posted by glenerrolrd

In my seemingly never ending quest to get my 50 summilux and 50 noctilux properly calibrated, a new issue has come up and I could use some help. The summilux after 3 round trips to Solms ..seems to focus accurately at 3ft ,8ft,20ft and probably 200yards. However the "collimation test" shows that it does not achieve infinity focus. This is a pretty absolute test done with a piece of test equipment and providing a displayed set of specifications. So its absolute ..not a judgement call . This is a test of the lens and does not involve an M8 rangefinder. I believe in another thread someone pointed out that not achieving infinity indicated some fairly significant problems with the lens. So that even if you could focus it ..it may not be up to its full potential. I don t know could someone provide better insights?

Well, I do not have any experience with "collimation" testing, but adjusting my M8 body to infinity the old fashioned way was absolutely crucial to getting my array of lenses to focus accurately. I would be very interested to hear the results if you were to check your M8 to infinity in that manner. Would the results be the same as those from the collimation test?

One final note: anyone doing the common infinity adjustment procedure should be sure to use a celestial object (i.e. a star), as that will produce much more accurate results than an object which is a few hundred yards away.

[rod99]

Are you sure it is the lenses that are not focusing to infinity, and not the M8 ?

I have an M8 on which I have tested all 4 of my lenses, none of which would
accurately focus to infinity. When set to infinity they all focused at anything
between 50 ft and 200ft (depending on focal length).

My M8 is therefore back in Solms for a check-up. I haven't had any response from
Leica yet.

Rod

[m0r0m0]

Originally Posted by marknorton

We've discussed endlessly here how adjusting the camera and lens to work together might work out but might also mean you end up with a non-interchangeable lens camera - the body won't work with other lenses, the lens won't work on other bodies.

If an authoritative treatment of all the considerations and methods involved in this important issue were on the forum's Wiki, then the need to have such endless repetitions would largely disappear, to be replaced by a simple reference to the Wiki article.

This observation applies, too, to quite a few other subjects I have seen discussed here. It would seem that the time spent up front to produce useful articles would be well rewarded by the time saved in not needing to reiterate a discussion. This is especially true for issues like this one for which the necessary information is too complex to explain in a simple statement or two without a trail of follow-ups.

Meanwhile the Wiki is largely a blank slate waiting to be filled with useful information.

Comments?

[glenerrolrd]

I am really just seeking to understand better "how to test a lens for proper calibration" and I may be confusing the test methods and equipment. I know from speaking with leica repair technicians both factory and independent that the starting point is typically a reference (calibrated M8) or preferably your M8. I understand and have observed the testing procedures which include reference targets at various distances. These are pretty easy to replicate and my "sloppy hand held tests" in general show the same results..that is my tests agree with the service departments tests. There is a piece of test equipment that they seem to mount the lens , which is independent of the body . This provides a digital read out of maybe half a dozen specifications ..which can be compared to the standards. This is what I have been assuming is "the collimation tester" ..it was from this test that one of my lenses was found to not focus to infinity. As Mark indicated this is independent of the rangefinder and serves as a point of reference. I do fully appreciate that every calibration is a balance of a number of trade offs and to a great extend a "tuning" operation requiring a skilled technician . The issue is the lens will not focus to infinty but otherwise seems to focus perfectly...is this an indication of other lens problems or just a balancing of the trade offs.

[marknorton]

You say your lens will focus to 200 yards (let's call it meters) but not to infinity. The question is the, how much out of adjustment is the lens?

To a first approximation, lenses follow the basic formula:

1/f = 1/u + 1/v

At infinity, the lens is exactly "f" in front of the focal plane, 50mm. At 200metres, it moves slightly further forwards to 50.0125mm, or by 1/80mm. To a first approximation then, your lens is sitting 1/80mm too far forwards.

The Noctilux focussing scale is 9cm long and in moving from one end to the other, the lens moves forward by about 3mm from its infinity position. That means your infinity stop position is off by about 9/(3/(1/80)) or 0.375mm. The 50/1.4 ASPH focussing scale is only half as long AND because the lens focusses closer, to 0.7m, the travel is 4mm; in that case, the infinity stop is off by a mere 0.015mm.

It may very well be that these are within the limits of how accurately the infinity stop can be set; in any event, even at f1, 200m is well inside the available DoF.

[robsteve]

Roger:

Is your 50mm Summilux the ASPH version or the older version? Did Leica not offer to adjust this lens?

[glenerrolrd]

Rob/Mark They are taking care of the lenses and that is not an issue . I expect that the M8 and future versions will be my primary system....and I am convinced that periodic calibration will be required (if for no other reason than my lenses will change). My interest is in better understanding how the collimation test fits in (is it really telling us anything different from the "star test" using the body ). The difference would appear to be that on the M8 testing always has some "maybe small" variance due to the range finder. The second question is surely at somepoint subjective....Marks analysis would indicate that if you have acceptable results at 200 yards then infinity should be acceptable (if not perfect). At the shorter distances ..I do as quick test on a wall chart....if the lens backfocuses(wide open) by more than 6 inches at 2M , I know it will show up in my images. The infinity test, other than being an important baseline, is more difficult for me to evaluate (how it impacts image quality). . Now a casual reader may ask.....who cares? But if you follow sean reid s testing of the rangefinder lenses..you can see that his testing methodology includes taking multiple images and picking the best This is to avoid the small focusing errors made by the user. These errors may be large enough to make the differences in sharpness(between your lens choices) irrelevant in practical photography. So you could argue that having a calibrated system is as important as buying the best glass. Since almost every lens in the M system is pretty darn good.

[marknorton]

Originally Posted by glenerrolrd

Rob/Mark They are taking care of the lenses and that is not an issue . I expect that the M8 and future versions will be my primary system....and I am convinced that periodic calibration will be required (if for no other reason than my lenses will change). My interest is in better understanding how the collimation test fits in (is it really telling us anything different from the "star test" using the body ). The difference would appear to be that on the M8 testing always has some "maybe small" variance due to the range finder. The second question is surely at somepoint subjective....Marks analysis would indicate that if you have acceptable results at 200 yards then infinity should be acceptable (if not perfect). At the shorter distances ..I do as quick test on a wall chart....if the lens backfocuses(wide open) by more than 6 inches at 2M , I know it will show up in my images. The infinity test, other than being an important baseline, is more difficult for me to evaluate (how it impacts image quality). . Now a casual reader may ask.....who cares? But if you follow sean reid s testing of the rangefinder lenses..you can see that his testing methodology includes taking multiple images and picking the best This is to avoid the small focusing errors made by the user. These errors may be large enough to make the differences in sharpness(between your lens choices) irrelevant in practical photography. So you could argue that having a calibrated system is as important as buying the best glass. Since almost every lens in the M system is pretty darn good.

Roger, the purpose of my analysis was to show you that what you seeing as significant issues are down to tiny adjustments. Even if the lens is reset to work on a particular body, the rangefinder can and does go out of alignment which renders any lens fine tuning, if not pointless, then at least compromised.

Sean focus brackets to obtain the best image the lens is capable of providing and in doing so, takes the rangefinder out as well as human error (such as it might exist) out of the equation. Add the rangefinder back in and any focussing error it introduces will affect image quality.

My own view is that the optical quality of Leica M lenses exceeds the ability of the Leica M rangefinder to focus them accurately, which is why I have proposed a scheme of user lens fine-tuning.

Nikon have recently added a fine tuning focus menu to the D3. I have found this is mainly useful for dialling in a degree of front focus when the lens is focussed wide-open so that focus shift back as the lens is stopped down remains within the depth of field. I mainly use it with the fastest glass, for example, the 85mm f1.4.

[MichelB]

Originally Posted by marknorton

To a first approximation, lenses follow the basic formula:

1/f = 1/u + 1/v

At infinity, the lens is exactly "f" in front of the focal plane, 50mm. At 200metres, it moves slightly further forwards to 50.0125mm, or by 1/80mm. To a first approximation then, your lens is sitting 1/80mm too far forwards.

Mark:
1. Can you explain the basic formula? What is "u", "v"
2. How can you find 200 m and 0.0125 mm?
Thank you in advance.
Michel

[marknorton]

Michel

This is a formula which applies to a basic lens where U and V are the subject and image distances and F is the focal length. If U is infinity, 1/V = 1/F, so the image distance is the focal length behind it - this is the definition of focal length.

If, instead, U is 200 metres, 200000 mm, 1/V = 1/50 - 1/200000 = 1/0.019995 and V is 50.01250313mm; in other words, for the lens to focus on an object 200m away, it has to move forwards 0.0125mm from the infinity position, which is what the focussing mount does. As you focus an M lens closer and closer, the lens barrel moves forward from the focal plane.

This is a very basic formula; it falls apart with fast lenses, with compound lenses, with close focus, but to illustrate the tiny tolerances at play here, it's good enough.

[Added from a later post to keep them together....]

The next step is to think that the focussing mount is a simple helical screw - as you turn the focussing ring, the lens moves in and out. By the time you are focussing a Noctilux at 1m, the formula predicts that the lens will be 2.632mm out from its infinity position. The length of the Noctilux focussing scale is 9cm, so I wanted to see how well the formula can predict where Leica would put the focus distance markings on the focussing ring - 1, 1.2, 1.5, 2, 3, 5, 10 metres and infinity.

Turns out very accurately, the position of the mid-point of the digits measured with a rule away from the infinity mark is within 1% of where the formula predicts them to be.

Interestingly, it's not so close on the 50mm f1.4 ASPH at close focus - the focus scale length is shorter and the lens has close focus correction.

It does show though that the 1/u + 1/v = 1/f formula is accurate enough to approximate the characteristics of the lens.

[glenerrolrd]

Mark There is no question that rangefinder and lens fine tuning can hit a point of diminishing returns. Further accuracy being insignificant considering the other factors you mentioned. Many users have tried the "focustestchart.com" by Tim Jackson to determine if their M8/lens is focusing correctly. When I see a lens backfocus by more than 20mm and certainly by 40mm ..using the chart and Tim s process , then it shows up in consistent back focus errors in my images. Thats between 1-2 inches at 3 ft . When the system is adjusted to be "dead on " if there is such a thing , then my percentage goes way up . Sure I miss a lot like anyone that shoots handheld and wide open ..but when I can shoot a series of 3-5 images ....I need to get one critically sharp. Don t think this is possible...see attached chart...this is my 75 Summilux after "tuning" in NJ. This was about 2 meters away. The technician and I both shot this chart 3 times each and we agreed 5 out of 6 that the center 8 was the sharpest. Sure we were off a little each way but you would have to agree thats pretty close. Before the adjustment ..the 75 focus point would have been back to the next figure 8. So simple tests exist that can be translated into a relevant (shows up in images) evaluation of the lens/M8 calibration. The 2nd image is the same 75 Lux in action. . My question has always been about infinity testing. . . When the first rangefinder adjustment theads and everyone had an allen wrench out whacking on the infinity adjustment, a really big deal was made about ..wow ..get the infinity right first or you will never find the sweet spot. My questions where ..OK ...the M8 s have been calibrated and work flawlessly with 8-10 lenses..so I am not touching the rangefinder unless it drifts out by itself. Lets say I get a new lens and it focuses perfectly at 3 ft and 2M and seems fine out to maybe 100 ft but at infinity ..the rangefinder will not reach infinity. I have no relevant test to determine if this will impact image quality. I guess maybe the star test might work compared to a reference lens. I did find when the rangefinders were being adjusted that shots at infinity before and after adjustment did show up in image sharpness. I know what to expect at the closer distances and how to translate it to a real photograph but not so with the infinity test.

[robsteve]

Roger:

I guess the only time infinity focus is important is if you shoot a lot a scenic photos where the lenses are shot at infinity. I tend to do this type of photography with my DMR, but I have tested all my M lenses and they are also sharp at infinity.