Focus problem with new S 100mm f2.

[v.tsivas]

Advertisement (gone after registration)

Hi everyone,

I just got the 007 with the 100mm but the lens is back focusing and it gets worse the farther I try to focus. In live view is perfect but in OVF is the issue.

The lens is a demo so like new, and the store recommended to send it over to leica for a service or to switch lenses. I can also get my money back but I do want the lens. 

I tried the 70mm on the 007 and work prefect so i guess the camera is fine..

Anyone had this issue before? What would you recommend I do?

[Stuart Richardson]

In my experience, if you can switch to another copy, I would do that. Occasionally when things are sent in to Leica they will just say that they are within spec and return them. Typically it only gets better if you send them both the camera and the lens, and even then, it is not certain. Rather than risk that, I would not get one until I was certain that it worked. The focus in the 100mm is quite difficult in the best of cases, so I would recommend trying another copy if you can.

[mgrayson3]

+1

An imperfectly focusing 100/2 will frustrate you into not using it. Get a different copy! Sometimes S lenses miss infinity. I always check the barrel distance markings to make sure it's at the right spot for that lens. But at normal distances? Don't!

[xiaubauu2009]

On 11/6/2021 at 8:55 PM, v.tsivas said:

Hi everyone,

I just got the 007 with the 100mm but the lens is back focusing and it gets worse the farther I try to focus. In live view is perfect but in OVF is the issue.

The lens is a demo so like new, and the store recommended to send it over to leica for a service or to switch lenses. I can also get my money back but I do want the lens. 

I tried the 70mm on the 007 and work prefect so i guess the camera is fine..

Anyone had this issue before? What would you recommend I do?

just part of the short coming of the OVF/Focus Module/Sensor system. Tolerance is off, and I am guessing it's the sensor distance is to the lens is different from the projected difference of the focus screen. You need to adjust either the sensor or the focus screen... don't know how, but maybe some shimming?

[mgrayson3]

5 hours ago, xiaubauu2009 said:

just part of the short coming of the OVF/Focus Module/Sensor system. Tolerance is off, and I am guessing it's the sensor distance is to the lens is different from the projected difference of the focus screen. You need to adjust either the sensor or the focus screen... don't know how, but maybe some shimming?

Not if a different lens focuses properly. There is one more step: the lens has to move its focus motor the correct amount, and if the lens is miscalibrated, that one lens will misfocus. Yes, open-loop systems are error prone.

[xiaubauu2009]

20 hours ago, mgrayson3 said:

Not if a different lens focuses properly. There is one more step: the lens has to move its focus motor the correct amount, and if the lens is miscalibrated, that one lens will misfocus. Yes, open-loop systems are error prone.

OP say it focus correct on the sensor live view, but look to have issue on the focus screen... so this must mean there's a misaligned mechanically somewhere between the focus screen to prism to lens or sensor to lens...

[mgrayson3]

Advertisement (gone after registration)

3 hours ago, xiaubauu2009 said:

OP say it focus correct on the sensor live view, but look to have issue on the focus screen... so this must mean there's a misaligned mechanically somewhere between the focus screen to prism to lens or sensor to lens...

Then how does it work well with the 70? Live view focusing is closed loop. The lens doesn’t have to know what distance it thinks it’s focusing at. It keeps moving until the sensor says it can stop. Phase detect sends a message to the lens and the lens has to interpret it properly. This lens doesn’t. The 70 does. 
 

Nowhere does the OP say that AFTER live view focusing it LOOKS out of focus on the ground glass. That would imply miscalibration of the Phase Detect system. 

[ZHNL]

100cron will have focus "problem" on any S camera depend on your expectation. I know this for sure from my personal experience on this lens with multiple trips to Wetzlar and using no less than 5 S cameras.  The lens doesn't have fully corrected spherical aberration  unlike any other S  lens within the system (though all slower than f2.5). This makes it very difficult for ancient PDAF S system to handle, especially at longer distance. 

First make sure testing this with well lighted defined target (I believe vertical line perform better with S but couldn't be sure anymore) at close distance to see if there is any issue. if not, sending it to Leica won't do anything. If it is already show focus error, it could be the lens and camera pair need sync/fix.     

[xiaubauu2009]

17 hours ago, mgrayson3 said:

Then how does it work well with the 70? Live view focusing is closed loop. The lens doesn’t have to know what distance it thinks it’s focusing at. It keeps moving until the sensor says it can stop. Phase detect sends a message to the lens and the lens has to interpret it properly. This lens doesn’t. The 70 does. 
 

Nowhere does the OP say that AFTER live view focusing it LOOKS out of focus on the ground glass. That would imply miscalibration of the Phase Detect system. 

Oh, I thought OVF is the issue means ground glass is the issue.

But I do find 100mm extremely hard get good focus. for me success rate is only about 60%

Edited November 11, 2021 by xiaubauu2009

[BernardC]

On 11/6/2021 at 8:55 AM, v.tsivas said:

I just got the 007 with the 100mm but the lens is back focusing and it gets worse the farther I try to focus. In live view is perfect but in OVF is the issue.

Does this happen only at f:2.0 (wide-open), or at all apertures?

AF through the viewfinder uses a phase-detect sensor at the bottom of the mirror box, just like any other AF SLR since the original Minolta Maxxum. There's a focus target in the middle of the screen, but it doesn't correspond to the actual focus point.

I presume you are using some kind of focus target, either a commercial one or a flat surface at an angle. Try taking test images at 4 different camera orientations: normal (landscape), 90 degree right/left (portrait), and upside down. If your images switch from front-focus to back-focus, this means that the focus point isn't where you expect it to be. This could be something as simple as a focusing screen that isn't seated correctly.

Also, the 100 is the widest-aperture lens in the S set, so it won't focus as precisely as the other lenses do. That's a characteristic of phase-detect AF, it can only "see" a certain maximum aperture, no matter how fast the lens is. Given that almost all S lenses are in the 2.5-2.8 range, it makes sense that the PDAF sensor would be optimized to those apertures. What that means is that focus won't be perfect at f:2.0, but it should be fine when stopped-down to 2.5 or 2.8. As noted by others, Leica can calibrate a lens and camera, which is a process that improves the camera's ability to guess the real focus plane. Something to consider if the lens is OK stopped-down, but slightly off wide-open.

[chrismuc]

It’s beyond my understanding why Leica did not integrate AF mircoadjustment, to the S cameras which is absolutely necessary for any digital SLR (and therefore basically every but Leica S has it). I would move on to a MF mirror less system, especially for longer focal lengths. Fuji GFX focuses 100% reliable with their 110f2 and 250f4 lenses (btw. with Steel AF adapter and Sigma Art 85f1.4 or 135f1.8 or … too!) and the GF lenses are tack sharp from open aperture with neglectable optical aberrations. 

[mgrayson3]

40 minutes ago, chrismuc said:

It’s beyond my understanding why Leica did not integrate AF mircoadjustment, to the S cameras which is absolutely necessary for any digital SLR (and therefore basically every but Leica S has it). I would move on to a MF mirror less system, especially for longer focal lengths. Fuji GFX focuses 100% reliable with their 110f2 and 250f4 lenses (btw. with Steel AF adapter and Sigma Art 85f1.4 or 135f1.8 or … too!) and the GF lenses are tack sharp from open aperture with neglectable optical aberrations. 

All that you say is true. And yet after a year, I sold my rather complete GFX system and moved back to the S. (I *do* miss the 250/4 - marvelous lens). As for AF adjustments, I think that that's a matter of pride. Leica just won't admit that their tolerances aren't perfect.

[BernardC]

3 hours ago, chrismuc said:

It’s beyond my understanding why Leica did not integrate AF mircoadjustment, to the S cameras which is absolutely necessary for any digital SLR (and therefore basically every but Leica S has it).

That has to do with how phase-detect sensors work. They are conceptually very similar to split-prism focusing screens that were common in the film era. Do you remember how some screens would completely black-out with slower lenses? And how some manufacturers offered screens that were optimized for fast lenses? That's because phase sensors and split prisms are optimized to see at a specific numerical aperture.

Let's say that a sensor is optimized for F:2.8 (a likely number with the S system). Faster lenses will be focused with an accuracy that corresponds to 2.8, and slow lenses might not be focused at all. Cameras compensate for the lack of accuracy with fast lenses by using look-up tables. Basically, they say things like "focus 2/3 toward the near side at 1m", "focus 3/4 toward the distant side at infinity." Micro-adjustments allow you to update that table, or enter values for a lens is unknown to a particular camera.

35mm SLR systems had to deal with a wide variety of lenses (often f:1.2 to f:5.6 maximum aperture), and a lot of different camera models. Some expensive cameras had multiple phase sensors (slow/fast), and some didn't. Most cameras could focus slower lenses only with their center points, but not outside of center.

One interesting aspect of the S system is that most lenses are f:2.5 to 3.5 (7 of 10 lenses). There's one 100mm f:2.0 lens, one manual focus lens (120TS), and one zoom which is 3.5 at the wide end. That's a very narrow aperture range, therefore we can assume that the phase-detect sensor in the mirror box is designed to work optimally in the 2.8-3.5 range, and sub-optimally down to 5.6. The 100 f:2.0 is the outlier, it has the widest aperture and the narrowest depth of field.

Should Leica have implemented user-accessed AF micro-adjustment? You can see why they didn't: it would have only been useful with a single lens, at a single f:-stop; that lens is already micro-adjusted at the factory (which isn't true of most competing SLR lenses); micro-adjustment failures are indicative of bigger issues which can only be remedied by Leica service.

The 100 f:2.0 is never going to be an easy lens to focus. The depth-of-field is similar to the 50mm f:0.95 Noctilux, and S users are more demanding overall (or they make bigger prints). That's a lot to ask of an AF system, especially one which is limited to a single point at image center. That means you should use the (excellent) optical viewfinder to confirm focus, especially when shooting wide-open.

That's not to say that there isn't anything wrong with the original poster's lens or camera. As others have said, that person can try a different 100mm (or a different camera), or send the camera and lens to Leica. One or both may have been dropped, or they could be at opposite ends of the tolerance range, or maybe that behaviour is normal for the 100 at f:2.0.

[mgrayson3]

18 minutes ago, BernardC said:

That has to do with how phase-detect sensors work. They are conceptually very similar to split-prism focusing screens that were common in the film era. Do you remember how some screens would completely black-out with slower lenses? And how some manufacturers offered screens that were optimized for fast lenses? That's because phase sensors and split prisms are optimized to see at a specific numerical aperture.

Let's say that a sensor is optimized for F:2.8 (a likely number with the S system). Faster lenses will be focused with an accuracy that corresponds to 2.8, and slow lenses might not be focused at all. Cameras compensate for the lack of accuracy with fast lenses by using look-up tables. Basically, they say things like "focus 2/3 toward the near side at 1m", "focus 3/4 toward the distant side at infinity." Micro-adjustments allow you to update that table, or enter values for a lens is unknown to a particular camera.

35mm SLR systems had to deal with a wide variety of lenses (often f:1.2 to f:5.6 maximum aperture), and a lot of different camera models. Some expensive cameras had multiple phase sensors (slow/fast), and some didn't. Most cameras could focus slower lenses only with their center points, but not outside of center.

One interesting aspect of the S system is that most lenses are f:2.5 to 3.5 (7 of 10 lenses). There's one 100mm f:2.0 lens, one manual focus lens (120TS), and one zoom which is 3.5 at the wide end. That's a very narrow aperture range, therefore we can assume that the phase-detect sensor in the mirror box is designed to work optimally in the 2.8-3.5 range, and sub-optimally down to 5.6. The 100 f:2.0 is the outlier, it has the widest aperture and the narrowest depth of field.

Should Leica have implemented user-accessed AF micro-adjustment? You can see why they didn't: it would have only been useful with a single lens, at a single f:-stop; that lens is already micro-adjusted at the factory (which isn't true of most competing SLR lenses); micro-adjustment failures are indicative of bigger issues which can only be remedied by Leica service.

The 100 f:2.0 is never going to be an easy lens to focus. The depth-of-field is similar to the 50mm f:0.95 Noctilux, and S users are more demanding overall (or they make bigger prints). That's a lot to ask of an AF system, especially one which is limited to a single point at image center. That means you should use the (excellent) optical viewfinder to confirm focus, especially when shooting wide-open.

That's not to say that there isn't anything wrong with the original poster's lens or camera. As others have said, that person can try a different 100mm (or a different camera), or send the camera and lens to Leica. One or both may have been dropped, or they could be at opposite ends of the tolerance range, or maybe that behaviour is normal for the 100 at f:2.0.

An excellent response - meaning I agree completely 😎.

(I always wondered what “phase” was being detected. I couldn’t see how knowing the polarity of the photons would help. Then I found out that it was just another term for maximizing autocorrelations of split prism measurements. It well explains misfocus due to contrasting elements near the sensor. And knowing exactly how much to move the lens to achieve that focus.)

[chrismuc]

7 hours ago, BernardC said:

That has to do with how phase-detect sensors work. They are conceptually very similar to split-prism focusing screens that were common in the film era. Do you remember how some screens would completely black-out with slower lenses? And how some manufacturers offered screens that were optimized for fast lenses? That's because phase sensors and split prisms are optimized to see at a specific numerical aperture.

Let's say that a sensor is optimized for F:2.8 (a likely number with the S system). Faster lenses will be focused with an accuracy that corresponds to 2.8, and slow lenses might not be focused at all. Cameras compensate for the lack of accuracy with fast lenses by using look-up tables. Basically, they say things like "focus 2/3 toward the near side at 1m", "focus 3/4 toward the distant side at infinity." Micro-adjustments allow you to update that table, or enter values for a lens is unknown to a particular camera.

35mm SLR systems had to deal with a wide variety of lenses (often f:1.2 to f:5.6 maximum aperture), and a lot of different camera models. Some expensive cameras had multiple phase sensors (slow/fast), and some didn't. Most cameras could focus slower lenses only with their center points, but not outside of center.

One interesting aspect of the S system is that most lenses are f:2.5 to 3.5 (7 of 10 lenses). There's one 100mm f:2.0 lens, one manual focus lens (120TS), and one zoom which is 3.5 at the wide end. That's a very narrow aperture range, therefore we can assume that the phase-detect sensor in the mirror box is designed to work optimally in the 2.8-3.5 range, and sub-optimally down to 5.6. The 100 f:2.0 is the outlier, it has the widest aperture and the narrowest depth of field.

Should Leica have implemented user-accessed AF micro-adjustment? You can see why they didn't: it would have only been useful with a single lens, at a single f:-stop; that lens is already micro-adjusted at the factory (which isn't true of most competing SLR lenses); micro-adjustment failures are indicative of bigger issues which can only be remedied by Leica service.

The 100 f:2.0 is never going to be an easy lens to focus. The depth-of-field is similar to the 50mm f:0.95 Noctilux, and S users are more demanding overall (or they make bigger prints). That's a lot to ask of an AF system, especially one which is limited to a single point at image center. That means you should use the (excellent) optical viewfinder to confirm focus, especially when shooting wide-open.

That's not to say that there isn't anything wrong with the original poster's lens or camera. As others have said, that person can try a different 100mm (or a different camera), or send the camera and lens to Leica. One or both may have been dropped, or they could be at opposite ends of the tolerance range, or maybe that behaviour is normal for the 100 at f:2.0.

Sorry but I have doubt that it’s correct what you say. The problem of every SLR is that the distance to the sensor (or film) , to the AF measuring device and to the split screen are different. They never will be the same because of lack of precision in manufacturing and assembly. That causes the small focusing error using either AF or MF. Shimming of the split screen can help a bit to improve the MF and AF micro adjustment compensates the AF measuring device misalignment and improves AF. Mirrorless cameras on the other hand measure the focus on the sensor - either by contrast detect or phase detect sensoring. So by definition and physics the focus is correct and non of these mirrorless have issues to focus correct at any aperture (apart from focus shift which is a lens issue, not a PDAF issue), using one of the two AF systems. 

[Stuart Richardson]

23 hours ago, chrismuc said:

It’s beyond my understanding why Leica did not integrate AF mircoadjustment, to the S cameras which is absolutely necessary for any digital SLR (and therefore basically every but Leica S has it). I would move on to a MF mirror less system, especially for longer focal lengths. Fuji GFX focuses 100% reliable with their 110f2 and 250f4 lenses (btw. with Steel AF adapter and Sigma Art 85f1.4 or 135f1.8 or … too!) and the GF lenses are tack sharp from open aperture with neglectable optical aberrations. 

This comes up from time to time, and Leica's response is that it was not necessary for the S system, as they have gone further than most manufacturers did at the time by embedding the exact focal length and specs of ever individual lens and body they make into the lens and camera electronics. So how it is supposed to work is that the lens tells the body its slight deviations from ideal, and the body factors in its own slight deviations, and the AF system compensates for them so that they focus together properly. Other companies (at least when the S system came out) did not do that, so that is why you need AF micro adjustment. Leica's take was that if it was not working, then there was something else going on, like a focusing screen out of adjustment, or the lens or body got knocked out of their initial measurements. 

In reality, AF systems are not always perfect, but Leica's thought was that they were solving the issue in a better way. The biggest problem with that is that when things are off, they need to return to the factory (both camera and lens). My experience has met that out. All the S lenses I had focused perfectly (as long as there was sufficient area for the AF patch), but on my very first S2, it had focusing problems, but it was the body itself and affected every lens. It was new enough that they just replaced the body.

[BernardC]

12 hours ago, chrismuc said:

Shimming of the split screen can help a bit to improve the MF and AF micro adjustment compensates the AF measuring device misalignment and improves AF.

In our example, the user has one lens that is mis-focusing, but others are OK. All lenses would misbehave if micro-adjustment was meant to correct an issue in the camera body.

 

[bsmith]

I have an S3 and a 45, 70 100 and 120 S lens. Only the 100S slightly mis-focuses. (at all distances) On my SL2, (with Leica adapter) the 100S focuses perfectly at all distances. 

If I send the 100S  lens AND S3 body to Leica to adjust, won't I risk my other S lenses not focusing properly ? Isn't it better to send only the 100S or is it a waste of time without the body?

[LeicaR10]

BSmith,  I sent you a PM.  r/ Mark

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[setuporg]

On 11/13/2021 at 7:43 AM, BernardC said:

That has to do with how phase-detect sensors work. They are conceptually very similar to split-prism focusing screens that were common in the film era. Do you remember how some screens would completely black-out with slower lenses? And how some manufacturers offered screens that were optimized for fast lenses? That's because phase sensors and split prisms are optimized to see at a specific numerical aperture.

Let's say that a sensor is optimized for F:2.8 (a likely number with the S system). Faster lenses will be focused with an accuracy that corresponds to 2.8, and slow lenses might not be focused at all. Cameras compensate for the lack of accuracy with fast lenses by using look-up tables. Basically, they say things like "focus 2/3 toward the near side at 1m", "focus 3/4 toward the distant side at infinity." Micro-adjustments allow you to update that table, or enter values for a lens is unknown to a particular camera.

35mm SLR systems had to deal with a wide variety of lenses (often f:1.2 to f:5.6 maximum aperture), and a lot of different camera models. Some expensive cameras had multiple phase sensors (slow/fast), and some didn't. Most cameras could focus slower lenses only with their center points, but not outside of center.

One interesting aspect of the S system is that most lenses are f:2.5 to 3.5 (7 of 10 lenses). There's one 100mm f:2.0 lens, one manual focus lens (120TS), and one zoom which is 3.5 at the wide end. That's a very narrow aperture range, therefore we can assume that the phase-detect sensor in the mirror box is designed to work optimally in the 2.8-3.5 range, and sub-optimally down to 5.6. The 100 f:2.0 is the outlier, it has the widest aperture and the narrowest depth of field.

Should Leica have implemented user-accessed AF micro-adjustment? You can see why they didn't: it would have only been useful with a single lens, at a single f:-stop; that lens is already micro-adjusted at the factory (which isn't true of most competing SLR lenses); micro-adjustment failures are indicative of bigger issues which can only be remedied by Leica service.

The 100 f:2.0 is never going to be an easy lens to focus. The depth-of-field is similar to the 50mm f:0.95 Noctilux, and S users are more demanding overall (or they make bigger prints). That's a lot to ask of an AF system, especially one which is limited to a single point at image center. That means you should use the (excellent) optical viewfinder to confirm focus, especially when shooting wide-open.

That's not to say that there isn't anything wrong with the original poster's lens or camera. As others have said, that person can try a different 100mm (or a different camera), or send the camera and lens to Leica. One or both may have been dropped, or they could be at opposite ends of the tolerance range, or maybe that behaviour is normal for the 100 at f:2.0.

Thank you Bernard, this is a stellar example of why it's so great to be on this forum -- one can always learn so many new things!