Focus shift question

[Guest #12]

Advertisement (gone after registration)

What do you mean by “zones”?

 

zone is distance from the lens axis

[Advertisement]

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

[mjh]

zone is distance from the lens axis

How does this make sense with regard to spherical aberration? Spherical aberration is the phenomenon that light rays that should intersect in a single point do not – the amount of refraction depends on their path through the lens. Correction of spherical aberration is about getting light rays passing through different zones to be refracted in the same way. Talking about different amounts of over- or under-correction in different zones makes no sense as the correction concerns rays traveling through different zones. Within a single zone there is nothing to correct.

[pico]

[...]If you "understand what is happening," you should be able to get a number for the focus shift at each aperture, right? How would you do it?

 

I don't know what you mean by "understanding" without seeing the outcome of pictures made with a particular lens. Simplified ray tracing does not work. It is easier to think that one "understands" the Petzval lens because at first glance it appears to be rather simple, but it is not really simple in quantitative terms. It took two people six months of clever hand calculation to ray-trace a Petzval design. (I write 'clever' because they worked in parallel, which was rather innovative.) That is far more than I can keep in my head. So it is outcomes by using the lens that really matters, unless one is only interested in how a lens design works _in principle_ only. Someone who describes how a lens behaves in principle, but has not used it extensively does not really understand the lens.

 

So, those pseudo ray-traces for simplified lenses only help illustrate how a lens behaves _in principle_.

 

I suppose quantifying focus shift could be helpful when using lenses that have significant focus shift at particular apertures. Is that what you are looking for?

[Guest #12]

I don't know what you mean by "understanding" without seeing the outcome of pictures made with a particular lens. Simplified ray tracing does not work. It is easier to think that one "understands" the Petzval lens because at first glance it appears to be rather simple, but it is not really simple in quantitative terms. It took two people six months of clever hand calculation to ray-trace a Petzval design. (I write 'clever' because they worked in parallel, which was rather innovative.) That is far more than I can keep in my head. So it is outcomes by using the lens that really matters, unless one is only interested in how a lens design works _in principle_ only. Someone who describes how a lens behaves in principle, but has not used it extensively does not really understand the lens.

 

So, those pseudo ray-traces for simplified lenses only help illustrate how a lens behaves _in principle_.

 

I suppose quantifying focus shift could be helpful when using lenses that have significant focus shift at particular apertures. Is that what you are looking for?

 

I would agree with most everything...photography and optics are two different things. I did not expect to learn anything practical. But I think the designer "understands" the lens in a meaningful way, too, even if he never uses it once.

 

As far as the Petzval lens, I just meant the spherical aberration is simple, not the ideas behind the lens.

[Guest #12]

...

 

Go back to Lars' lens...if you make a graph relating the height of an incoming ray with the point where it crosses the lens axis coming out, you see a nice decreasing function. But when you look at the same graph for an actual lens, like in a design handbook or wherever...it's almost always a hill. A little hill (that's good) or a big hill, depending on the design. For me a graph is concrete, but just saying "the lens is under-corrected so it focus shifts" does not explain anything (to me at least). For Lars' lens I can see if you progressively stop it down it will backfocus more and more. For a real lens, I still think it depends.

[mjh]

Go back to Lars' lens...if you make a graph relating the height of an incoming ray with the point where it crosses the lens axis coming out, you see a nice decreasing function. But when you look at the same graph for an actual lens, like in a design handbook or wherever...it's almost always a hill. A little hill (that's good) or a big hill, depending on the design.

I still don’t quite get it … are you talking about a ‘hill’ in the caustic, implying a secondary focal point? Perhaps you have a link handy to a graph showing such a ‘hill’? One can try to design the lens so the caustic gets a certain shape (long and thin, say), but I am not aware of hills in the caustic being a common phenomenon (or a design goal).

 

For Lars' lens I can see if you progressively stop it down it will backfocus more and more.

And it usually does. Of course it soon gets irrelevant as the growing depth-of-field masks this effect, but yes, in principle it backfocuses more and when you stop down. In practice the first couple of f-stops are the only ones you need to worry about; depth-of-field takes care of the backfocusing at small apertures.

 

Back in the days when (after the introduction of the M8) the focus shift craze started and everyone was testing their lenses for focus shift, the few lenses that actually exhibited focus shift did so in perfect agreement with the theory.

[Olsen]

Advertisement (gone after registration)

Focus shift is only a issue with the very large apertures. I have a Noctilux 50 mm 1,0 and an old Canon 50 mm 1,0L. The latter is impossible to focus manually or with AF with enough precision to be able to 'place' the DOF correctly at distances shorter than 2 meters. This is due to that neither the manual focus mechanism nor the AF is precise enough. The 'Ultrasonic AF/or manual focusing' makes 'jumps' moving the DOF a centimeter or two, making precise focusing a joke.

 

Canon has now taken the consequence of this, dropped the 1,0 version and launched a 1,2 version which produced a greater DOF, thus widened the tolerances considerably. (A Canon EF 50 mm 1,2L is for sale here in Norway at Foto.no - M The seller brags that it as been to service for 'focus calibration'. My guess: The seller should have his eyesight 'calibrated too'. You need an eyesight of an eagle to use such a lens).

 

While Leica has gone the other way and, freshly, launched a 50 mm Noctilux with even tighter DOF; the Noctilux 50 mm 0,95 version. Only a camera system with very tight tolerances can utilize such a lens. - Provided the photographer has an eyesight to match it. - And the knowledge of the 'focus shift' effect (... and the money to pay for it...).

 

A 50 mm 1,0 lens, regardless if it is Canon or a Leica has a DOF of about 2 cm at 1 meter focusing distance, - like shooting macro, at full aperture. Increasing the distance to 1,5 meter the DOF increases to 5 cm. Seems a lot, but it is a challenge for most people even with a good eyesight to 'hit' the perfect placement of the DOF. It is with tolerances of a cm or two that 'focus shift' plays an important role.

 

I had my Noctilux 50 mm 1,0 sent to Leica for 'adjustment' together with my M8 two times! Which tells of how uncertain I was about the performance of this lens. The last time I got it back with a comment hinting to that 'the owner should have his eyesight checked...'

 

It was about then I found out about 'focus shift'.

 

Imagine you have got, what you think is a perfect focusing hit by placing the DOF right over the eyes of a person 1,5 meter distance. Shoot, and you will see that it is only the nose tip that is sharp. Stop down to 1,4 and the DOF is widened to 7 cm and moved backwards with about 'well a cm' - to a perfect 'DOF hit'. If you don't compensate for this tiny focus shift at full aperture you will never shoot sharp at full aperture. If you have the lens calibrated to shoot 'straight' at 1,0 it will be off at all the other aperture settings.

 

Further: My eyesight is indeed failing me, I have to admit. I use a x1,4 Magnifier whenever I use my Noctilux. That helps a little. Whenever I don't manage to focus the Noctilux properly it is either due to my failing eyesight or my misjudgement of the focus shift.

[Guest #12]

...

Back in the days when (after the introduction of the M8) the focus shift craze started and everyone was testing their lenses for focus shift, the few lenses that actually exhibited focus shift did so in perfect agreement with the theory.

 

yeah, I was here ... I wouldn't read too much into user tests. But when I have checked lenses I have always found them to wander backwards.

 

I found a little focus shift in the 50 lux asph., a little more in the pre-asph, a whole lot more in the old Noctilux. And the same for additional copies.

[Guest #12]

I still don’t quite get it … are you talking about a ‘hill’ in the caustic, implying a secondary focal point? Perhaps you have a link handy to a graph showing such a ‘hill’? One can try to design the lens so the caustic gets a certain shape (long and thin, say), but I am not aware of hills in the caustic being a common phenomenon (or a design goal).

...

 

I literally mean plot "the height of incidence of a ray as it enters the lens versus the position of the point at which that ray crosses the lens axis on emergence." One is just worrying about the longitudinal spherical aberration. The graph is called the "spherical aberration curve" or "longitudinal spherical aberration curve" or "focus shift vs. aperture" or "variation in focus with zone of lens" or some such. You can find curves for various lenses in most any handbook for lens design (I will point out, I haven't read any handbooks for lens design). Some curves I have seen: One curve for an f/3 lens wandered way forward from f/3 to f/4 and then wandered backwards from f/4 on down. One curve wandered a little forward from f/2 to f/2.5 and then wandered backwards from f/2.5 on down. The Petzval curve just wandered backwards from f/3.5 on down, and just a little bit at that. Now apparently you can get from here to the residual aberration or the amount of focus shift you will actually measure in the lens. You can find some other curves for various designs in Ch. 7 of Jaap's book (haven't read it).

[mjh]

While Leica has gone the other way and, freshly, launched a 50 mm Noctilux with even tighter DOF; the Noctilux 50 mm 0,95 version. Only a camera system with very tight tolerances can utilize such a lens. - Provided the photographer has an eyesight to match it. - And the knowledge of the 'focus shift' effect (... and the money to pay for it...).

While in Leica’s portfolio, the old Noctilux was probably the worst offender with regard to focus shift, spherical aberration in the new Noctilux is much improved. Focus shift isn’t an issue anymore. Some would say that this was to the detriment of bokeh, but then you cannot get the ultra-smooth bokeh typical for under-corrected spherical aberration without also accepting some focus shift.

[mjh]

yeah, I was here ...

As a guest or under a different identity? Since as #12 you have been participating for less than a year. Just wondering …

 

I wouldn't read too much into user tests. But when I have checked lenses I have always found them to wander backwards.

So what are you worrying about? Focus shift ceases to be an issue at smaller apertures because depth-of-field is sufficient to cover it; what would it matter if the focal point should move in the other direction again?

[Guest #12]

guest

 

No it wouldn't matter if the focus moved forwards a little then the other direction. I was worrying about academic, not lens accuracy. Thanks for the comments and I see I misused the term under-corrected.

[pico]

While in Leica’s portfolio, the old Noctilux was probably the worst offender with regard to focus shift, spherical aberration in the new Noctilux is much improved. Focus shift isn’t an issue anymore. Some would say that this was to the detriment of bokeh, but then you cannot get the ultra-smooth bokeh typical for under-corrected spherical aberration without also accepting some focus shift.

 

Do you, or anyone, have pointers to pictures with this over-corrected spherical aberration? It seems a curios idea - an over-corrected aberration. Does it create another kind of aberration, perhaps one not yet named? Anti-bokeh?

[pico]

Oh, this is my idea of BOKEH!

 

On Black: Contes d'upstair #7 by chachahavana [Large]

 

(not mine)

[lars_bergquist]

Do you, or anyone, have pointers to pictures with this over-corrected spherical aberration? It seems a curios idea - an over-corrected aberration. Does it create another kind of aberration, perhaps one not yet named? Anti-bokeh?

 

This is clear as water if you know what spherical aberration is.

 

Spherical aberration means that peripheral rays come to a focus at a point in front of the point-of-focus of the axial/paraxial rays. Under-corrected s.a. means that peripheral rays still focus ahead of the axial ones, but closer. Fully corrected s.a. means that peripheral and paraxial rays focus to the same point. With over-corrected s.a. the peripheral rays focus behind the paraxial rays. What's curious about that?

 

Bokeh: This is easiest to see with the small bright discs formed by out-of-focus point sources of light. With fully corrected s.a., these are evenly bright ceneter to periphery. With under-corrected s.a. discs on the far side of the plane of focus are brighter in the center than on the periphery. This makes for super-smooth bokeh (unless as in your example, astigmatism messes it up). With over-correction, the discs take on the famous 'rolled-up-condom' look with a bright outer ring and a dimmer interior.

 

But that was behind the plane of best focus in the subject space. On the near or camera side, the effect is the opposite!

 

In the picture below, look at the out-of-focus halogen lights under the ceiling. They show no ring effects – s.a. is well corrected. The lens used was a v.4 35mm Summicron, often known as 'the bokeh king' – though my Summilux ASPH is even better.

 

The just slightly spherical old man

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!