Focus shift on new 35/1.4 FLE

[zlatkob]

Advertisement (gone after registration)

Do slr autofocus lenses show focus shift as well?

Yes, it's the same. Autofocus is achieved at widest aperture. The shift occurs during exposure, when the aperture is stopped down.

[Advertisement]

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

[horosu]

Possibly of interest to some. In his New Leica Compendium, Erwin Puts has a comparison of focus shift between this lens and the previous model.

The text is not entirely clear (to me) but he describes a shift to the front of 3cm with the new lens and 5cm with the older one (from f/1.4 to f/2.8) at a nominal distance which must be 1m (3% error mentioned).

 

I remember out of memory that Erwin Puts mentioned that the new lens still exhibits focus-shift but to a much lesser degree than the old one (found it: SX35FLE, part1 )

 

In any case, in my opinion, in practical shootings this focus-shift of the new lens is of little importance and would not compromise the expected results.

 

My surprise is the performance of the original double aspherical lens: focus is where it should be at all apertures. Combine that with the fact that its focusing-ring is serrated allowing it to be focused with two fingers (i hate focusing tabs ) and it is just perfect (for me).

 

Have a nice weekend, Horea

[01af]

Poor analogy. "Shooting 100 frames ..." is not useful. A rule of thumb must be useful ...

That's the point exactly. So it's a proper analogy.

 

In order to be useful, a "rule of thumb" doesn't need to be precise but it mustn't be deceiving.

 

The actual rule of thumb is this: The ratio of background DOF to foreground DOF varies from 1:1 at zero distance to inf:1 at hyperfocal distance and beyond. The narrower DOF is, the more symmetrical it is. And the wider, the more asymmetrical.

 

Sure—there are situations where the ratio really is 2:1 indeed. But there are more situations where it is not. For example, when shooting portraits at wide apertures where only portions of the subject's face are in focus, DOF will be nearly symmetrical. When shooting landscapes with a wide-angle lens at medium to small apertures, background DOF will be vastly wider than foreground DOF, not just twice as wide.

 

So please—simply forget the infamous (and incorrect) "2/3 to 1/3 rule," will you?

[wlaidlaw]

I think a lot of folks may be confused by the various terms and words appearing on these posts. "In front of", "behind", 'foreground", "background", etc, as they mean different things to different people. This ends up in people arguing at cross purposes when they both actually mean the same thing.

 

I think the easiest way is to give a numerical illustration:

 

A traditional depth of field calculation for a full frame 35mm camera, fitted with a 35mm f1,4 lens at f1,4 focused at 6 metres, will show that within reasonable standards, all should be in focus from 5.528 metres in front of the imaging plane (sensor or film) out to 6.599 meters in front of the imaging plane. In other words, the field of in focus further away from the point of focus (and the camera) at 0.599 metres is a little larger than the 0.472 metres field between the focus point and the camera (a 1 : 1.26 ratio)

 

Now back and front focus. Back and front do not apply to what you see or the image printed from your camera. They apply to where the image is formed inside your camera. In other words, back focus is where the sharpest image is formed behind the imaging plane inside your camera and front focus where the sharpest image is formed in front of the imaging plane. If a lens is said to be back focusing, when you print out the image, the point of sharpest focus will be further from the camera than the point at which you think you focused. To labour the point if a lens is said to be back focusing by 1.5cm and you focus at 6 meters, the point of sharpest focus when you print the image, will be at 6.015 meters from the imaging plane.

 

Aperture shift on closing down the aperture, normally increases back focus. Again if you are focussed at 6 meters, when you close down to f4, the point of sharpest focus may have moved to 6.200 metres. Now often this is disguised by the increase in depth of focus. To quote numbers at f4, the same combination of camera lens would in theory be in focus from 4.824 meters to 7.934 metres, when focused at 6 metres. You can see that the proportion of the focus field further away from the focus point at 1.934 metres has increased more than the amount between the camera and the focus point at 1.176 metres, so that the ratio is now 1:1.64. If the camera has an 0.2 metre focus shift, then all is in focus from 5.024 metres to 8.134 metres, so other than a slightly shorter amount of field in focus between the point at which you think you are focusing (6m) and the camera. You should therefore, barely notice the effect of aperture shift. It has to be really severe for it to affect images so you can spot it easily.

 

At a certain point in 2007/2008, 35/1.4 ASPH Summiluxes were coming out of the factory with static back focus and extreme aperture shift, even further backwards. To go back to my original example, if they were static back focused by 0,4 metres wide open at 6 metres, and back shifted another 0.6 metre when closed down to f4, then when you thought you were focused at 6 metres, at f1.4 you were actually in focus from 5.928 metres to 6.999 metres. At f4, when you thought you were in focus at 6 metres, you would actually be in focus from 5.824 metres to 8.934 metres. This is going to have a significant and easily visible effect on the appearance of your image.

 

If you had a reasonable 35/1.4 ASPH Summilux which did not aperture shift too much, it was fairly easy to get them to work well by setting them up with a small amount of static front focus. My own 35 Lux is set to have its focus fields at around 4 metres set at .25 metres between the camera and focus point and .2 metres further away from the focus point. Then as you stop down, this front focus naturally corrects itself.

 

Wilson

[mjh]

If we put the 1/3rd 2/3rd rule in the context of Leica's core applications, that is reportage and street photography, it is a good rule of thumb. Indeed, most subjects are in general in the 1 to 3 meters range.

But that is the range where a much better rule of thumb would be to assume that DoF extends equally in front and behind the subject focused on. Not generally, of course, but for 35 and 50 mm and f-stops from 0.95 up to 2.8 this would roughly be true.

[mjh]

Do slr autofocus lenses show focus shift as well? Or is this problem peculiar to manual focus lenses, or the rangefinder system itself.

Both types of camera suffer from this issue. The rangefinder mechanism is typically calibrated to be spot on when the lens is used wide open and focusing precision is most critical. If the lens exhibits focus shift this will create issues when stopping down until the DoF gets large enough to mask this effect.

 

While SLRs do focus through the lens, the effective aperture of the autofocus is rarely identical to the aperture used for the eventual shot. One might think that the AF aperture would be equivalent to the lens speed since the phase-detection measurements are done with the aperture blades fully open. But the phase-detection sensor doesn’t capture all of the exit pupil so the effective aperture will be smaller – more like f5.6 usually. Hasselblad’s AF system (that does compensate for focus shift) has an effective aperture of f6.7, for example. The focus will shift on either side of that f-stop value and thus the effect will be most visible when the lens is used wide open. So in a way focus shift is even more of an issue for SLRs.

[Rick]

Advertisement (gone after registration)

Rule of Thumb: For most distances DOF is greatest behind the subject and increases with distance.

 

I've always leaned my DOF more for a weighting behind the subject. Thanks Michael for showing me that my own thinking was flawed in that, I thought the 1/3 and 2/3 was the rule.

 

It never ceases to amaze me how many things that people believe in turn out to be wrong. I see it in my area of business every day. People never cease to amaze me how they hold absolute certain beliefs about certain things that are wrong. I knew my belief about DOF was screwed up as soon as I saw Michael had posted. I watched my belief crumble right in front of my eyes. Now, I get to be the ignorant one. Reminds me to not get too set in what I believe.

[01af]

I knew my belief about DOF was screwed up as soon as I saw Michael had posted.

Well ... actually your belief in the stupid "2/3 to 1/3 rule" should have been shattered in the moment you learned about hyperfocal distance

[Rick]

Well ... actually your belief in the stupid "2/3 to 1/3 rule" should have been shattered in the moment you learned about hyperfocal distance

 

Ok, so maybe I'm missing something on that too that I didn't get the first several hundreds times through it. What would be your definition of it in a real technically pure sense?

 

ps. By the way, the hairs on the back of my neck started to hackle when I saw your first post, but I dismissed it. Then when Michael posted I had the "oh crap" moment.

[pico]

It never ceases to amaze me how many things that people believe in turn out to be wrong. I see it in my area of business every day. People never cease to amaze me how they hold absolute certain beliefs about certain things that are wrong. [...]

 

Humans choose to appear consistent in their reasoning, regardless of contradicting facts. It's a fundamental human characteristic. I've worked in academe for about forty years and find such behavior persists especially in the non-engineering 'intellectuals'. Me, I live in what I call the wisdom of insecurity - never certain, always learning and sometimes learning the wrong thing, so it's back to insecurity. I'm very comfortable in such a malleable mindset. We humans do not live long enough to become truly wise.

 

Regarding focus-shift, some would consider me a slob with 35mm. I do the best I can with focusing and if the picture is, IMHO, good, I let it be. However, with a 30" monitor and now using an M9 I can see how critical one can become - at the monitor. When printed at 240 to 360ppi, I am happier with the outcomes than with monitor displays at near 100%.

 

With large format, focus-shift with long lenses (I enjoy a 14.75" fast lens on 8x10"), refocusing after stopping down is critical. And easy to do with LF.

[01af]

Well ... actually your belief in the stupid "2/3 to 1/3 rule" should have been shattered in the moment you learned about hyperfocal distance.

Ok, so maybe I'm missing something on that too that I didn't get the first several hundreds times through it.

Isn't it obvious? When background DOF is infinite and foreground DOF is finite then the ratio cannot possibly be 2/3 to 1/3. Right?

 

And the ratio also cannot all of sudden jump from 2:1 to inf:1 at hyperfocal distance—nature just doesn't work that way. So if you know about hyperfocal distance and what it means then you simply cannot believe in the "2/3 to 1/3 rule" at the same time because the two, umm, "rules" are contradicting each other. So at least one must be wrong.

[ho_co]

Rick--from OED:

1957 Amos & Birkinshaw Television Engin. I. ix. 178: If a lens system is focused on infinity, the images are in focus‥for all objects lying between infinity and the hyperfocal distance. If, however, the lens system is focused on the hyperfocal distance, images are in focus for all objects lying between infinity and half the hyperfocal distance.

[luigi bertolotti]

Do slr autofocus lenses show focus shift as well? Or is this problem peculiar to manual focus lenses, or the rangefinder system itself.

Focus shift is intrinsic in lenses' design... and given that on most SLR one focuses wide open and the diaphragm closes when shooting, the problem may arise as in a Leica; you can, in theory, have a certain control on it working in "stop-down" if allowed... but not always one can enjoy the due control on a darkened view... expecially for is a phenomenom that has became "fashionable" only with digital, as Pico has correctly pointed.

[Rick]

Isn't it obvious? When background DOF is infinite and foreground DOF is finite then the ratio cannot possibly be 2/3 to 1/3. Right?

 

Yes. This right on paper and is very astute, but in only that way.

 

But, an optical system does have a practical point that can be taken as infinity. At some point we need to be able to talk in practical terms or we would never get our cameras out of the lab and off some theoretical piece of paper. So, I think we can make generalizations and therefore define 1/∞, in a practical sense.

 

I do appreciate your logic and it isn't lost on me, though. The theoretical is the check that keeps our rules of thumb from becoming too useless.

[Rick]

By the way, the practical definition of 1/∞ is zero. This is what I mean above, if you know what I mean.

[01af]

This is what I mean above, if you know what I mean.

Yes, I know what you mean ... but I don't understand why you're mentioning it here.

[colorflow]

Let me try to explain why the 1/3-2/3 rule is useful, at least to me. I know it is not precise but it is useful where I need a rule. May be its only me, but I don't need a rule when shooting close and I don't need a rule when shooting very far

But there are more situations where it is not. For example, when shooting portraits at wide apertures where only portions of the subject's face are in focus, DOF will be nearly symmetrical.

True, but when shooting portraits I don't care about foreground background dof and therefore don't need a rule of thumb. I just focus on the subject

When shooting landscapes with a wide-angle lens at medium to small apertures, background DOF will be vastly wider than foreground DOF, not just twice as wide. Here again I don't need a rule of thumb since I know the dof is very wide and I'll just focus on the object of interest closest to me.

 

Where I do need a rule of thumb is when shooting a scene, landscape or street, with say a 35or 50mm at f2.8 to f5.6 and there is a range of foreground to background, say 15 to 100 ft, that I want to be in "focus". I need to decide where in-between I should focus on. With this rule I would be focusing 1/3 of the way out from the foreground towards the background. Working out the numbers it would probably be close enough to 1/3-2/3 on Michael's chart(s). I won't be able to improve if I tried to be precise and memorize all the chart(s) since I am a poor estimator of distance and would not know which point to use in the graph.

 

Enough said.

[jaapv]

Whilst we are at it, shall we tke the discussion into this highly informative thread?

 

http://www.l-camera-forum.com/leica-forum/leica-m8-forum/23934-accuracy-depth-field-scale-lenses-m8.html

 

 

If it becomes an M9, or better still, a general DOF thread I'll happily move it

[Advertisement]

Hallo,

Look here: Leica M

[01af]

... or better still, a general DOF thread I'll happily move it.

Oh no! The last thing we need is another thread about depth-of-field. Those who want to learn about DOF should read Paul van Walree's page on depth-of-field. It's the only Internet resource about DOF I'm aware of that is a) complete and free of errors and misconceptions. Also see the links at that page's bottom (in particular, this one).

[jaapv]

Exactly - I don't like the DOF card being played in various threads either - so we should confine it to one thread if possible - and the debunking has taken place in this forum over the years as well... But misconceptions die hard.

 

 

 

Another source that might interest you - not only on DOF but the whole concept of blur and sharpness:

 

http://www.zeiss.com/c12567a8003b8b6f/embedtitelintern/cln_35_bokeh_en/$file/cln35_bokeh_en.pdf