Front focusing errors with the 24-90mm?

[jrp]

Advertisement (gone after registration)

Diglloyd, who has a track record of finding oddities that are hidden in plain sight, says that he has found that "the Leica SL autofocus system was a disaster, degrading the majority of my work. I am sure that many users do not see the issue, if for no other reason than shooting at f/5.6 or f/8 or f/11, which often masks it enough to generate a (just) acceptable image. Also, overall contrast of the 24-90mm zoom is superb, which on the whole makes for an image with “pop”. But it does not mean sharp detail, ...".

 

There are some earlier threads here identifying focusing errors (normally with the spot on which the camera focuses, and mainly with the 90-280mm).

 

I have occasionally found misfocusing, but have just passed it off as user error or camera shake.

 

Do others have similar experiences?

 

I can imagine that the optical image stabilization might cause some image degradation or misfocusing.  It is also possible that some focus settings work better than others:

 

* AFs v AFc (seems to set whether the camera must have found focus before it will release)

* static v dynamic/tracking focus v face detection

* point v small area v larger area focus

 

Is there any combination that is more reliable (for static subjects, say) than the others?

[AlanJW]

I use single focus point, with 24-90, generally without tracking.   Using multiple focus points is not something I do on this or any other camera because I want to focus on something of my choice.  For static subjects, I would always use one focus point (easily moved with the joystick) and static AF.  

 

IMHO, tracking on the SL is not up to the level of what I can achieve on my Nikon D800e, but it is possible this is because I haven't used the SL as much.  The time I tried was with a rented 90-280 for an outdoor sports event so the f/stop was rarely wide open.  No front focus issues there but I had several missed focuses using multiple points and went back to single point.

 

I have never seen front focusing with a static one point approach on the 24-90, including low light, wide open.  

[Advertisement]

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

[LocalHero1953]

http://www.l-camera-forum.com/topic/255186-24-90mm-focus-shift-diglloyd/

 

The only problems I have found are when using a tripod and not turning off OIS.

 

And no, I don't shoot stopped down. 

Edited July 26, 2016 by LocalHero1953

[thighslapper]

Diglloyd, who has a track record of finding oddities that are hidden in plain sight, says that he has found that "the Leica SL autofocus system was a disaster, degrading the majority of my work. I am sure that many users do not see the issue, if for no other reason than shooting at f/5.6 or f/8 or f/11, which often masks it enough to generate a (just) acceptable image. Also, overall contrast of the 24-90mm zoom is superb, which on the whole makes for an image with “pop”. But it does not mean sharp detail, ...".

 

There are some earlier threads here identifying focusing errors (normally with the spot on which the camera focuses, and mainly with the 90-280mm).

 

I have occasionally found misfocusing, but have just passed it off as user error or camera shake.

 

Do others have similar experiences?

 

I can imagine that the optical image stabilization might cause some image degradation or misfocusing.  It is also possible that some focus settings work better than others:

 

 

 

 

This is another Digilloyd 'revelation' that has vanished into the mists of time. 

 

Many thousands of owners and not a mention of focussing issues , so either Digilloyd was wrong, or what he spotted was of no practical significance. 

 

I posted some tests that suggested the spot focussing cross was located a tiny bit higher than the actual focussing point ...... but the last firmware update made some rather cryptic allusions to 'improved AF accuracy' and increases in the number of AF points (although I am not sure anyone actually knows what that really means). 

 

AF on the 24-90 and 90-280 is reliable and fast ..... and less easily fooled than many othet AF cameras I have used.

 

Like many, I tend to stick to one point AF and either lock and re-compose or set position with the joystick.  All the other offered methods strike me as inherently unreliable

Edited July 27, 2016 by thighslapper

[jrp]

I must say that I had completely forgotten that Digillioyd had "discovered" this issue in January.  I have been using the system happily since the end of last year.  He does claim that it is still present in firmware 2.0, on his front page, and cites a "Bob H" making similar findings.  (He also claims that the S has similar issues.)

 

Anyway, I don't want to create problems where, as in my case, there are none, I just wanted to check whether there were settings that should be avoided, or whether I was missing something.

[wlaidlaw]

The focussing was much improved by FW 2.0. I am having to do a lot less joystick twiddling than with the earlier FW. However I am having focussing difficulties with the 90-280 on distant subjects, particularly when they are of lowish contrast. I quite often see the red box which means insufficient contrast detected for accurate focus. Also the 90-280's point of sharpest focus is not at the centre of the green box but very much towards the bottom of the box or even slightly below it. Others have also noted this. The focus difficulties in low contrast are hardly surprising, since the whole system is based on contrast detection. You then have to switch over to MF, which in spite of some improvement on FW 2.0, I still don't like. I would prefer the fast movement focus to be "higher geared". I have never found tracking worked particularly well and tend to have it off all the time and just use the joystick to perform the same function, which works perfectly. I do a lot of my photography in low light, early morning or evening, so a lot of the time, the lens is fully open. I rarely get a mis-focus with the 24-90. The focus accuracy on the 90-280 definitely needs some improvement, hopefully in FW 3.0 this will be implemented and Leica can also move the point of sharpest focus into the centre of the green box. 

 

Wilson

 

 

[jrp]

Advertisement (gone after registration)

Does the focus improve if you switch off image stabilisation?

 

For my own part, I have occasionally found focus on the long lens to be very slightly off, for no apparent reason, but I put it down to camera shake or image stabilisation not having settled or using AFc inadvertently.

 

I also have been known to use the small area, rather than the point focus, because it is easier to see in the viewfinder.

[LocalHero1953]

It depends on how unstable your hand is!

If your grip is rock steady, you don't need IS and you should get better results without.

Solvitur ambulando - i.e. try it yourself.

I haven't noticed it to be a problem, so I haven't investigated it myself.

[wlaidlaw]

Age, arthritis and possibly alcohol has made OIS on longer lenses essential for me unless I am using a tripod or monopod. So turning OIS off might make the lens more in focus but it would be difficult to tell due to the amount of movement blur  

 

The OIS on the 90-280 is not as good at present I think, as the 5 axis IBIS on my little Olympus EP-5 with the 75-300 lens (EFOV 150-600). I suspect on the basis of how much the IBIS has improved on the EP-5 with FW updates, that focus and OIS on the 90-280 is work in progress. I wonder if we will have to mount the lens and download a separate FW update for that, like you do with Olympus bodies and lenses. The main problem on the Olympus was that the IBIS was "strobing" with the shutter and at certain shutter speeds it actually made the image worse not better. I wonder if something similar is occurring on the SL. 

 

Wilson

[ramarren]

I have no experience with the SL90-280 lens' AF or OIS, and the Olympus E-M1's IBIS is indeed incredibly solid. It must be said that IBIS with a FourThirds sized sensor is a heck of a lot easier to make proficient than it might be with a 35mm-FF sized sensor. It has to be, purely from the standpoint of the masses involved and the accelerative forces that need to be overcome. OIS should be about the same difficulty in implementation with either system, dependent mostly upon focal length and lens speed, because the masses of the parts involved can be nearly the same. 

 

All hardware/software systems are a work in progress. The goal is to get the system to a point of satisfactory operation to meet the design goals, not to seek "perfection" which doesn't exist. Those goals change over time as user input fluctuates, so there's really no definitive end-point for the endeavor, just a long series of new goals and requirements, incrementally develop to achieve, then new goals and new development, etc., over and over again, until the product or project is abandoned and a replacement is instantiated. Very little is generated from a blank slate: it's simply too inefficient and too costly to do so. 

Edited July 27, 2016 by ramarren

[tom0511]

Age, arthritis and possibly alcohol has made OIS on longer lenses essential for me unless I am using a tripod or monopod. So turning OIS off might make the lens more in focus but it would be difficult to tell due to the amount of movement blur  

 

The OIS on the 90-280 is not as good at present I think, as the 5 axis IBIS on my little Olympus EP-5 with the 75-300 lens (EFOV 150-600). I suspect on the basis of how much the IBIS has improved on the EP-5 with FW updates, that focus and OIS on the 90-280 is work in progress. I wonder if we will have to mount the lens and download a separate FW update for that, like you do with Olympus bodies and lenses. The main problem on the Olympus was that the IBIS was "strobing" with the shutter and at certain shutter speeds it actually made the image worse not better. I wonder if something similar is occurring on the SL. 

 

Wilson

how much better is os with the 75-300 compared to is of the 90-280 in your opinion?

 

I dont see any problems with 90-280 is so far.

[wlaidlaw]

how much better is os with the 75-300 compared to is of the 90-280 in your opinion?

 

I dont see any problems with 90-280 is so far.

 

It is difficult to put a number on it and it might be the focus fussiness of the SL and weight of the 90-280 rather than the IBIS v OIS but with the Olympus you can just point and shoot. With the 90-280 you often have to look around for an object with higher contrast for it to pick up focus. Then I then find I need to have a shutter speed of close to twice the focal length to get a really sharp image whereas with the Olympus 1 x focal length (real not EFOV) will do. I am ignoring EFOV as that is just like cropping an SL image. 

 

Wilson

[wlaidlaw]

It may be my imagination but the latest FW seems to have cured the low focus point problem with the 90-280. I have just taken a few test shots and the point of sharpest focus now seems to me to be the middle of the green box. 

 

Wilson

[douglas ball]

I bought the SL and 24-90 in February in preparation for a trip to Chile. I was well  aware of Diglloyd's comments about the problems he had encountered with the 24-90 but proceeded to test the lens extensively on my own because of his tendency to rant over anything Leica.

First of all I want to make clear that I have always had a slight mistrust of AF from my Canon days but acknowledged the fact that it is useful and should speed things up. It's certainly nice to have if you can depend on it.

In my early tests I found on occasion definite front focus problems and I eventually concluded that Diglloyd was right. There were however two major problems, not one.  

The lens was often not focusing accurately to start with always (front focus problems) and  after when stopping down there was even more shift towards the front. So much so that often the initial target  on occasion became even less sharp. 

 

The solution that seemed to work was to focus well to the rear where you wanted sharpness and stop down until the foreground was in focus.

This was very much like the R 28-90 I had. If you were shooting a room interior for example, you focussed on the rear wall or furthest corner and stopped down until the foreground was sharp. 

I had always thought, in general, you focussed a third of the distance into the setup and by stopping down the increased DOF sharpness would extend to both the rear and to the front. This was not the case with the R 28-90 and now the 24-90. By f8 or f11 the DOF  has extended only to the rear ever so slightly and the sharpness has rushed forward.

 

Once you know this the lens becomes very easy to control and focus. With our trip to the Atacama and Patagonia I had absolutely no focus problems. One shot was badly out of focus but this was because the focus point had drifted down to the bottom of the screen.

 

I should point out that in most but not all cases I used the toggle button to focus on MF, used single point, and when the picture was important, as in most cases, I magnified the image to fine tune. I always focused on the most distant part of the picture where sharpness was needed.

This could have been a distant volcano, but by f11 (at 90 mm ) the foreground was sharp, sharp. This resulted in some excellent results that I was happy with.

 

This was all with the original firmware naturally. Diglloyd discovered that the focus point changed when you magnified the image. It became lower. I think that this lower spot was the true location of the focus point. This explained that when I focused on the bottom of a tree, the camera was focusing maybe 5-10 meters to the front.

With firmware 2.0 this problem was corrected and now that leaves only the focus shift problem which to my mind is very manageable and can work to your advantage even.

 

 

I was very interested this week to see Diglloyd go on yet another rant about the SL and the 24-90 focussing issues. This was with firmware 2.0 which surprised me. After studying his pictures I came to a simple conclusion. 

He is sometimes focusing 1/3 or 2/3s into the picture and after stopping down, the rear does not get much sharper. Even when he focuses on a flat surface, such as a barn, the sharpest picture is most likely the wide open shot. 

When he focused on a log, about 10 feet away, and stopped down, the log got even less sharp he said. To be truthful, on my screen I could not always see what he was claiming but i assume he is seeing something that I cannot see.

 

One of the things that I really like about the SL is, when on manual focus, you can get the a very exact measurement to the focused area, to the rear, and to the closest point. If, in your testing, you know the exact distance of a certain target area, you can then compare this to what your camera is showing on an ongoing series of tests. 

 

I downloaded firmware 2.1 yesterday and ran another series of tests with the 24-90 and the 90-280 lens. The good thing is that the focus point still stays in the same place after magnification, as it did with F 2.0. 

The 24-90, as expected, still behaves the same meaning focus to the back of the scene if greater DOF is needed. If shooting wide open, the AF is generally good enough now without magnification. If critical, I magnify.

 

The 90-280 has less focus shift and the increased sharpness is more evenly distributed front and rear. If anything I have found the AF on this lens to be very good indeed. It does make me believe that Leica did not on purpose design in the shift I encountered on the R 28-90 and the 24-90. I guess it was just a mistake. However, now that I know, I can live with it very well.

[wlaidlaw]

I suspect that just like with the 35 ASPH Summilux, there is some sample variation between lenses. Given the complexity and number of elements in both the 24-90 and 90-280, this should not be a great surprise. The factor which varied most on the 35 ASPH Summilux was the aperture shift. There was a forum member who was having problems with a number of samples of this lens to the extent of their being unusable from aperture shift. He refused to believe that there were any good ones. He ran his aperture shift tests on my 35 Summilux and was rather upset to find that whereas there was some aperture shift, it was minimal and always stayed well within the DOF for that aperture. 

 

The aperture shift on my example of the 24-90 is sufficiently minimal that it can in effect be ignored, as it is well within the DOF. I have taken around 1000+ images since FW2.0 (I had taken around 4000+ with FW 1.2) and my missed focus shots have dropped to almost nil in comparison to around 2 to 3% on the earlier FW. The FW 2.1 seems to have cured the problem of point of sharpest focus being below the green box on the 90-280, so now I am completely happy with the focus on both lenses. 

 

Wilson

[FlashGordonPhotography]

I though focus shift, by design moved the focus point rearward only. I've never heard of the focus point moving forward before when stopping down.

 

Gordon

[FlashGordonPhotography]

I suspect that just like with the 35 ASPH Summilux, there is some sample variation between lenses. Given the complexity and number of elements in both the 24-90 and 90-280, this should not be a great surprise. The factor which varied most on the 35 ASPH Summilux was the aperture shift. There was a forum member who was having problems with a number of samples of this lens to the extent of their being unusable from aperture shift. He refused to believe that there were any good ones. He ran his aperture shift tests on my 35 Summilux and was rather upset to find that whereas there was some aperture shift, it was minimal and always stayed well within the DOF for that aperture. 

 

The aperture shift on my example of the 24-90 is sufficiently minimal that it can in effect be ignored, as it is well within the DOF. I have taken around 1000+ images since FW2.0 (I had taken around 4000+ with FW 1.2) and my missed focus shots have dropped to almost nil in comparison to around 2 to 3% on the earlier FW. The FW 2.1 seems to have cured the problem of point of sharpest focus being below the green box on the 90-280, so now I am completely happy with the focus on both lenses. 

 

Wilson

 

I had one of those 35 luxes (pre FLE ASPH) as well. On mine wide open it ever so slightly front focused and then the focus shift moved the focus back as you stopped down. It was perfect under testing at f2. The effect was that in normal shooting it focused really well as normal subjects almost always fell within DOF. I could make it show the shift but only in a testing environment at close distances. I bought the lens used and someone suggested that others had had their lenses recalibrated so f2 was perfect. I don't know if this was true or not.

 

Gordon

[kiubabar]

I followed the Diglloyd blog for several years, I think the article make sense. He can't explained why the sharpness change so obvious between M lens and Elmarit SL 24-90. In my experience, focus errors happened especially in AFc mode no matter the shuttle speed or aperture you used. The frequency of focusing error seems higher than I normally expected. 

Edited July 30, 2016 by kiubabar

[kikouyou]

Just got my 24-90, it seems really good. No focusing issues. Indeed, AFM with joystic. No tracking, one point. Will try the focusing range on the lcd soon with tripod. But so far infinite on background only works well (on lcd).

[Advertisement]

Hallo,

Look here: Leica L-Mount

[wlaidlaw]

I though focus shift, by design moved the focus point rearward only. I've never heard of the focus point moving forward before when stopping down.

 

Gordon

 

Gordon, 

 

This argument has come up a number of times on the forum. Front focus is where the sharpest point of focus on the object is behind the object. This is because "front" refers to the image in the camera not the object itself i.e., the sharpest point of the image inside the camera, lies in "front" of the imaging plane. Correspondingly "back" focus is where the sharpest point of the image inside the camera lies behind the imaging plane. Outside a camera with "back" focus, the point of sharpest focus would lie in front of the object. Confusing isn't it? 

 

Wilson