We often read that a given lens design is optimized for a particular distance or ratio ... How is that accomplished ?
For what are they optimized: color aberrations, distortion ?
We often read that a given lens design is optimized for a particular distance or ratio ... How is that accomplished ?
For what are they optimized: color aberrations, distortion ?
Last edited by Ken Lee; 31-May-2014 at 09:44.
Only a guess. Since everything in lens design requires compromises, if they do their calculations at a particular magnification, that's where it will be optimized. The real work more likely goes into getting good performance at other magnifications.
I read somewhere that some Apo-Ronars and other process lenses such as the Computar have a thick spacer ring that corrects the lens for 1:1 ratios. It is said that if you remove the ring, the lens will then have an optimization that is closer to infinity.
All aberrations are influenced by magnification, since a change in distance changes the angles of every ray hitting each lens surface (except for the central ray). Lens designers will use all parameters (glass types, radii, thicknesses/distances) for optimization. Only one of them is occasionally accessible to us, the cell distance in the case of some process lenses, as in some of the Docter-Apo-Germinar lenses which have a removable washer/ring. Afaik, Apo-Ronars never used such a ring but cell distances were optimized in such a way that the shuttered versions were usually optimized for 1:20 or so, and the ones in barrel for 1:1. Dialytes are also not super sensitive when it comes to cell spacing, that is why even a 1:1 optimized Apo-Ronar works perfectly fine at infinity and f/22. I would venture a guess that the more ray bending occurs in a given lens, the more sensitive it is to cell spacing and magnification changes. This would include all modern wide angles.Originally Posted by Ken Lee
Hi Ken !
There is are some very good explations in Rudolph Kingslake's "History of the photographic lens" so what I write here can anly be a poor paraphrase of this most useful book.
At the end of the XIX-st century, optical engineers have recognized that a perfectly symmetrical lens design used at the 1:1 ratio is able to deliver an image 100% free from coma, distorsion, and lateral color.
I remember a classical French textbook discussing about what could be a perfect optical system and the book explained that, yes a lens operating at 1:1 ratio could be nearly perfect, but "will be perfectly useless since the image is equal in size to the object".
People who use xerox copiers and scanners all day long will appreciate this enlighting remark from a wise theoretician of the past
Actually most of the lenses we use in LF, with the exception of tessar types and telephotos as well as some wide aperture planars and xenotars, are refinements of some basic symmetrical designs.
The tessar is as old as 1904 and is a remarkable non-symmetrical lens, hence it has to be optimized for some non symmetrical (object / images) distance configuration. Infinity-focus is in principle the basic starting point of optimization for a tessar. On the other end, the (4/4) dialyte like the apo ronar (the ancestor is the celor patented in 1902) is at the beginning optimized for the 1:1 ratio and is perfectly symmetric.
Hence at the beginning, the degree of symmetry / asymmetry plays a very important role in the best lens-to subject distances.
By playing with the lens prescription, tessars can be optimized to serve as an enlarging lens at their best at 1:5 ratio for example.
And it is rumored that Fuji-C lenses are optimized for large distances although the look very much similar to an apo ronar ..
My understanding is that modern (6/4) standard lenses coverning 70° to 75° are optimized not stricly at infinity but for some distance, about 11 times the focal length (1:10 ratio) but actually those lenses are useable from infinity to 1:2, restricting the use of real / true /symmetric macro lenses to a very limited range of applications.
For example the apo-rodagon D 1:1 evoked here for copying a film to a DSLR sensor is perfectly symmetric and works fine if you do not depart too much from the 1:1 ratio. There is a companion lens named apo rodagon D 2x which is asymmetric and is not at its best at 1:1 but as labeled at 2X or 1:2.
Another interesting lens optimization is the 5.6 / 135 mm Zeiss makro planar, a fixed-element lens (7/5) design, a compromise for a lens able to shoot from 1:1 to infinity. This lens now belongs to history, at the first glance, it is very similar to one of our LF lenses but bears more assymetry, though.
There is an interesting article on the Zeiss Camera Lens blog explaining how the use of floating elements yields higher performances for a lens operating from 1:1 to infinity. They do not say that the 135 Makro Planar is obsolete, but you can read it between the lines
Lenses departing from the symmetry principle are in fact the majority of lenses for 35 mm, with the exception of macro lenses.
The reason I see for this is that 35 mm photographic lenses developed mainly around the idea of reportage and hand-held photography, for subjects located at large distances and with wide apertures. If you need both a wide aperture and optimization at infinity, a symmetric lens design is certainly not the good choice.
Need a wide-angle lens on a reflex camera body ? You can't use LF WA lens symmetrical design, you have to use a retrofocus, which is higly asymmetric. Don't want a lens-to film distance as big as the focal length ? You have to use a telephoto design, again something asymmetric that will not be good at close distances
Apo-ronars were unknown in 35 mm photography, at least after WW-II !
If we look at recent view camera lens designs for digital photogrfaphy with a small view camera, they are definitely asymmetric but bear some similarity with our classical LF lenses.
Hence for LF lenses a short summary to the original question is that the optimization starts from some symmetrical design and departs from symmetric by changing lens surface curvatures and spacings, and changing the actual location of the iris.
But those degrees of freedom are somewhat limited and many modern lenses for small formats use floating elements to achieve the best image quality in a large range of lens to subject distances.
On the other hand, highly specialized lenses like those used for photo-lithography work only a one wavelength and one magnification with a fixed f-stop e.g. f/4. They are definitely asymmetric and lens designers cannot compromise anything outside the highly specialized technical goal they have to achieve, designing the the best of the best lenses for one single task, one single wavelength, one object field and one single magnification (1:5 for current wafer-steppers).
In our LF lenses optimization at one distance e.g. 1:1 or 1:10 is quite loose since those lenses are general-purpose lenses.
But we could dream of floating-element LF lenses that would outperform our classical "fixed" designs.
Suffice to look at the last generations Zeiss MF distagons with their floating element control ring, manually operated. This could perfectly be applied to our manually-operated LF lenses.
But since we operate on film on a non-reflex camera we do not need any retrofocus lens. The good old symmetric lenses are still so good !
Only the constraints of silicon sensors with micro-lenses unable to capture slanted rays pushes lens designers to refine some special retrofocus designs like the very last Zeiss 55 mm "Otus" design ... but this is another story 100% off-topic here
Lenses are optimized for their optical performance which in its turn is demanded by the optical requirements. The optical requirements are dictated by the intended use of the lens. As the use differs so do the optical requirements and the necessary optimization.
Each optical designer can decide what optical defect optimization is preferred at the expense of other defects.
Therefore your broad question has no single answer.
In terms of Large Format Photography and Photographic Enlarging, I'm familiar with fine-tuning flatness of field with front-rear cell spacing. I suspect there are also many considerations that would concern the lens designer.
Interesting points about the effects of changing cell spacing. One of the things it does is change focal length. That's what puzzles me about the fact (which I believe) that Rodenstock used a spacer to convert an Apo-Ronar optimized for 1:1 into an Apo-Ronar optimized for 1:20.
I have a Sinar branded version with a small normal shim and a thick removable shim which I believe to affect optimization. I have only seen this with the Sinar version... Also it is multicoated and a "CL" version of the lens... Very rare.
I believe the Red Dot Artar was a dialyte that was optimized for ~1:10-1:20 (depending on FL) while the Apo Artar was for 1:1.
Posting Permissions
Reply With Quote
Bookmarks