minimum focus distance and lens design

[jbl]

Advertisement (gone after registration)

Hi,

 

I’ve been wondering about this and was just curious if anyone could explain it to me.

 

From what I can tell, it’s more difficult to design a lens that can focus at 0.7m rather than one that can only go to 1m. Take the 50 Summiluxes. I don’t know anything really about optics, but I’m curious why this is.

 

On a related note, is this the reason the Noctiluxes don’t focus closer than a meter?

 

-jbl

[adan]

The deal is that classic (non-floating-element, non-zoom, non-macro) lenses are optimized for infinity focus. I.E. a distance where the light rays coming from any one point in the subject matter are effectively parallel. That's the way the math works - it assumes parallel light rays (which is simpler math!).

 

If you are focused on something close, you are starting to capture light rays that are no longer effectively parallel. The light rays from one point of the subject will form a triangle, with the diameter of the lens aperture being the third side of the triangle.

 

With a large-aperture lens, the problem is worse. At a small aperture, the light rays that actually get through the aperture are still almost parallel, even when focused close, but if you are using a wide opening (60mm for the Noctilux), then obviously the light rays reaching the two sides of the lens opening are forming a fairly significant triangle with the source point - i.e. definitely NOT parallel (a triangle cannot have parallel sides).

 

And once the light rays you are capturing start to get significantly non-parallel, the math that is assuming they ARE parallel gets screwed up, the glass doesn't work as originally intended, and you get various type of aberrations (general fuzziness, coma shapes, "glows" around highlights, etc.)

 

For comparison, a true "macro" lens design does not use the assumption that the incoming rays are parallel. Instead, it is calculated for that triangular (divergent) light ray structure from the get-go (usually assuming a subject distance that will give about 1/5th life-size images on the film/sensor). It will often perform WORSE at long subject distances - but much better in the range from 1 meter down to life-size.

 

Floating elements (as in the most recent 50 Summilux-M ASPH and others) change the internal lens design slightly as the focus ring is turned, correcting for the change in the "math" of the light rays from close or far subjects.

 

So yes - the math is simpler if you assume parallel light rays, but the results degrade as you focus closer. And yes, that is why Leica put limits on close-focusing ability for some fast, older lens designs.

 

Howver, Leica also put a 1-meter close limit on lenses on the assumption that there would be poor framing (parallax between what the viewfinder sees, and what the lens sees from 2 inches down and to the right). Eventually, they relaxed that to .7 meters as photographers demanded it, and proved they could use the closer limit without cutting off people's heads.

[jbl]

Wow. Thanks for writing that. It completely makes sense. I've gotten interested in how these lenses are designed just because I'm seeing differences in them and am curious about how the designs have evolved over the years.

 

Would the issue with making a 0.7m Noctilux be that the math is just crazy?

[jbl]

And I suppose you'd probably lose the rangefinder patch too . In general is the Puts book a good one for information about this sort of thing?

[Michael Geschlecht]

Hello Everybody,

 

The more parameters/variables that there are to deal with the more complex the solution becomes.

 

Please see the the "tip of the iceberg" below.

 

Going to a larger F stop (ie: F2 to F1.4) means TWICE the surface area of EACH lens surface that light rays have to be traced thru & the lens has to be EQUALLY corrected for. As a lens widens in proportion to its depth the package of problems to solve changes & becomes more complex.

 

Focussing to a closer distance (ie: 1 meter to 0.7 meters) means TWICE the lens extension that the light rays have to be traced thru & the lens has to be EQUALLY corrected for. As the distance from the second nodal point of a lens to the film/sensor changes the package of problems to solve changes & becomes more complex.

 

Everything else equal: The more parameters, variables & problems there are to deal with: The more difficult it is to achieve the same degree of correction.

 

Sometimes the mathematics of Optics are like Life: To add something here you sometimes have to take away something there. Sometimes there are "bumps in the road" that were not expected.

 

Best Regards,

 

Michael

[adan]

I'll just add that my previous explanation, being a brief internet summary, was not "the whole truth."

 

Some lens makers use 100x the focal length instead of "infinity" in calculating a general-purpose lens - thus getting slightly reduced performance at both infinity and very close, but a better average performance over most of the range.

 

But the key point is that - over much of the history of lens design - time and money and available technology placed limits on how many calculations could be done. So the designer had to make "creative decisions" as to where to start and what to ignore.

 

- 100 years ago, lenses were calculated with pencil and paper.

- 80 years ago, they were calculated by 50 people in an office, each with an adding machine, and working 40 hours a week for 6 months just crunching numbers.

- 50-60 years ago (era of the 50 Summilux non-ASPH and the first Noctilux f/1.2) lens design and calculation was "computerized" - but that still meant running FORTRAN on Hollerith punch cards. NOT the same thing as using a chip CPU/desktop design program!: keypunch.

 

Michael mentions the "iceberg" of deep problems to be addressed in lens design - but there are also imponderables.

 

In gross design (and available technology) the 75 Summilux does not appear all that different from the 50 Summilux of the same era - yet it focuses down to .75 meters without nearly as much "glow" and visible aberrations as the 50 at .7 meters. It loses some quality - but not as much. What lucky guess did Walter Mandler make with that lens (which he believed to the best he ever designed) in choosing how to optimize that design?

 

As to a Noctilux (f/0.95) at 0.7 meters:

 

Not likely to block the rf patch any more than the 75 f/1.4 does (i.e. not at all).

 

The "lever arm" of all that heavy glass pushed forward another few mm might require a heavier barrel (although, again, the 75 Summilux handles it). Additionally, moving its floating element enough to compensate for 0.7-meter focusing might just make the whole thing too cumbersome for an M body. or Leica may simply have decided that anyone who really wants to focus close can get a 75 Summicron or 50 Summilux instead.

 

Puts' book(s) are good at giving the background on the "art" of optical design. Probably not the only ones, but the Compendium is generally a good "popular" dissection of the science and practicalities of lens design.

[pgk]

Advertisement (gone after registration)

What lucky guess did Walter Mandler make with that lens (which he believed to the best he ever designed) in choosing how to optimize that design?

I'm not so sure that he made any 'lucky' guess:). If you read the article "Double Gauss lens design: a review of some classics" by Jonas and Thorpe of ELCAN Optical Technologies (formerly Leitz Canada), (although it refers to the 35 Summilux, it is still relevant) they state:

 

"This lens [the 35mm f/1.4 pre-Aspheric Summilux], designed in 1958 was ray traced at a time when only a few manual design changes a day could

be evaluated. 48 years later, we thought it would be interesting to take a second look at the design. Walter Mandler had a much greater knowledge and experience of the design of camera lenses, but we have much superior computing power and can evaluate the effects of a change in the aberration balance in a few seconds. Can the design be improved without changing the form? ..... It appears that this simple double Gauss form cannot give improved performance

given the field and aperture constraints even with additional glass choices."

 

Or in other words, Mandler understood exactly what he was doing and was constrained by the glass technologies of the day. They go on to state that the successor to the pre-Aspheric departed from the Double-Gauss design in order to improve performance and recomputing a Double-Gause design does not improve performance even when new glass technology is included in the computations.

 

Perhaps their last statement is most telling (quite an accolade coming from two lens designers):

 

"Several designs produced by Walter Mandler enjoyed long production lifetimes and became benchmarks against which others were judged. This was due not only to his experience, skill and intuition as an optical designer, but also because of his understanding of the optics manufacturing process. His name will always be synonymous with the best in photographic double Gauss lenses."

[jbl]

Where was this article written? I'd love to read it.

[adan]

Hi, Paul!

 

Well, Mandler achieved something with the 75 Summilux double-gauss that he was not able to achieve with the 35 and 50 Summiluxes. The ability to focus closer than 1 meter with acceptable (to Leitz) performance.

 

Luck? Well, perhaps it was luck in the Arnold Palmer sense: "It's a funny thing. The more I practice, the luckier I get!" He'd had another 19 years of practice by the time he got around to the 75.

[pgk]

Well, Mandler achieved something with the 75 Summilux double-gauss that he was not able to achieve with the 35 and 50 Summiluxes. The ability to focus closer than 1 meter with acceptable (to Leitz) performance.

Hi Adan,

 

I'm told, by a lens designing friend, that the design of lenses of ~70-90mm is very well understood and has been for a long time. Consequently, many are exceptionally good designs, so I would expect no less from a Mandler design:). As focal length decreases, the close-focus performance drops unless optically corrected by use of floating elements or such-like, which was not the case in Mandler era lenses. The 75 and 80 Summiluxes may well sit in a 'sweet spot' of lens design in which 'conventional' designs can achieve very high performance even if large apertures are being catered for.

 

One thing which I feel is under-appreciated in many Mandler designs is that mid aperture performance displays extremely smooth tonality, and as mid apertures are often in use, this makes sense to me.

[Michael Geschlecht]

Hello Everybody,

 

First a correction to my Post #5 above:

 

Where I wrote that the focus extension from 1 meter to 0,7 meters was TWICE the extension of the focus travel to 1 meter: I should have written that the extension was + 50% of the focus travel to 1 meter. That is: If the focus travel from Infinity to 1 meter was 10 units, then the focus travel from Infinity to 0.7 meters would be 15 units.

 

Everything else that I said in Post #5 above is accurate.

 

As to the 75mm Summilux being superior in correction to its 35mm & 50mm counterparts:

 

An item of consideration is the angle of coverage of the 3 lenses involved. Double Gauss lenses are a design which was developed in order to extend a higher degree of correction to the edges of the format. Sometimes at the expense of the quality of the image at the center of the format. Because the angle of coverage of a 75mm lens is less than the angles of coverage of a 50mm lens or a 35mm lens: There are less variables to solve for.

 

Less variables to solve for means a greater likelyhood of a better solution.

 

Best Regards,

 

Michael