"best" magnification for lens determination?

A forum to ask questions, post setups, and generally discuss anything having to do with photomacrography and photomicroscopy.

Moderators: rjlittlefield, ChrisR, Chris S., Pau

johan
Posts: 1005
Joined: Tue Sep 06, 2011 7:39 am
Contact:

"best" magnification for lens determination?

Post by johan »

I was wondering how this is determined - I often see that 'such and so lens is best at 3x', but I don't know how this determination is achieved. Is it through user testing or is there such a thing as a defined formula that can be used?

Thanks!
My extreme-macro.co.uk site, a learning site. Your comments and input there would be gratefully appreciated.

rjlittlefield
Site Admin
Posts: 23562
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

Usually that description comes from the manufacturer's literature and reflects the design specification.

But the lens is very likely to perform well and often even better at other magnifications, depending on use.

For example a microfiche lens might be optimized for 20X onto a large screen to be viewed directly by eye. From the designer's standpoint, that lens will be "best at 20X". But if you use it to project onto a DSLR sensor, that lens will be deep into diffraction territory at 20X, and you'll get a far better result by using it at 4X instead.

The same issue applies, to a lesser extent, when comparing full-frame and APS-sized sensors. A lens that is "best at 3X" on an APS sensor could well be "best at 5X" when used to image the same size subject on a full frame sensor.

--Rik

johan
Posts: 1005
Joined: Tue Sep 06, 2011 7:39 am
Contact:

Post by johan »

Thank you Rik
My extreme-macro.co.uk site, a learning site. Your comments and input there would be gratefully appreciated.

ray_parkhurst
Posts: 3416
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Post by ray_parkhurst »

Many lenses, especially those used for inspection purposes, have their optical formula optimized for a particular magnification. An excellent example of this is the 105mm Printing-Nikkor, which is optimized at 1:1. Mark Goodman did measurements on this lens and it was clear from the results that the designers really did optimize the formula at 1:1. Look at coinimaging.com, Macro Lens Tests to see these results. Nikon also made a 150mm Printing-Nikkor that had a ring the user could turn to optimize the optical formula for best sharpness and resolution across a wide range of magnifications. This concept was used in the Rodenstock MacroVaron inspection lens, and most recently in the Nikon Rayfact VL. These all seem to be "zoom" lenses but they are fixed focal length lenses with a variable optical system that optimizes the design for a given magnification.

rjlittlefield
Site Admin
Posts: 23562
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

Thanks for the additional discussion & info.

I did not mean to suggest that lens designers miss their targets. It's just that the conditions they're optimizing for may not be the ones that Johan cares about.

The case of the Printing Nikkor is interesting. Although the lens is optimized for 1:1, all the graphs at http://coinimaging.com/printing-nikkor_105.html show that it's actually sharper at m=0.66. Yes, the field gets flattest at 1:1, but if you're focus-stacking, then field curvature per se does not degrade the final image. As Mark points out, he even had to focus-stack his sample coin. It's not hard for me to argue without even playing devil's advocate that the Printing Nikkor is "best at m=0.66" among the tested conditions.

Adding to the confusion, getting the best picture is a system problem while "best at" is a lens specification. Taking that 105 Printing Nikkor as an example again, I'm pretty sure that given an APS sensor and a 10 mm subject, the PN will produce a better captured image at 2:1 than it will at 1:1 plus cropping, and also that the PN will outperform most lesser lenses that are "best at 2:1".

Putting it all together, I'm not quite sure what "best at" really means, other than general guidance about what ranges you might consider running tests in.

--Rik

johan
Posts: 1005
Joined: Tue Sep 06, 2011 7:39 am
Contact:

Post by johan »

To be honest, my question was really more about enlarging lenses and small JMLs and the like, ie opportunistic uses of lenses.

Ie a 28 Componon, being used in a reverse (a config unintended by designers) is 'optimal' magnification largely a matter for self discovery and needs rather than a formulaic fact. Sort of ties up with another question I had, ie is resolution through a lens the wrong way round the same as the normal way round. The normal way being a formulaic fact but how relevant is it really the other way round. I wonder if I'm making any sense =)
My extreme-macro.co.uk site, a learning site. Your comments and input there would be gratefully appreciated.

Harold Gough
Posts: 5786
Joined: Sun Mar 09, 2008 2:17 am
Location: Reading, Berkshire, England

Post by Harold Gough »

rjlittlefield wrote:The same issue applies, to a lesser extent, when comparing full-frame and APS-sized sensors. A lens that is "best at 3X" on an APS sensor could well be "best at 5X" when used to image the same size subject on a full frame sensor.
Rik,

I struggle with this.

It seems to me that a lens, for a given distance from the subject, will give a certain resolution (in terms of fine detail) at the plane of focus. It is then a matter of the quality of the capture medium as to how that detail is recorded. Sensors of different sizes will cover different fields of view. Only if the quality across the image circle varies enough would there be any difference in the amount of image detail captured per linear width of the sensor, or per degree of effective angle of view, with different sensor sizes.

Or am I confused?

Harold
My images are a medium for sharing some of my experiences: they are not me.

rjlittlefield
Site Admin
Posts: 23562
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

johan wrote:To be honest, my question was really more about enlarging lenses and small JMLs and the like, ie opportunistic uses of lenses.

Ie a 28 Componon, being used in a reverse (a config unintended by designers) is 'optimal' magnification largely a matter for self discovery and needs rather than a formulaic fact. Sort of ties up with another question I had, ie is resolution through a lens the wrong way round the same as the normal way round. The normal way being a formulaic fact but how relevant is it really the other way round. I wonder if I'm making any sense =)
Good questions.

Let's consider that JML 21. Some good fellow (ah, it was you!) finally tracked down the spec sheet for that lens. According to that sheet, the lens is spec'd for "MAG: 0.0827X NOM" with field diameters of "object size" 216 mm and "image size" of 17.86 mm.

But we want to do macro, with m>1.

OK fine, reverse the lens. So then we're looking at a spec of m=12.09 with an object size of 17.86 mm and an image size of 216 mm. Excellent -- so it's just the thing for projecting onto 5"x7" sheet film.

But we're not using sheet film -- we have DSLRs. When run at m=12, that nominal f/3.5 turns into effective f/45 and hey presto, we're deep into diffraction territory.

At m=12 on an APS sensor, we'd be essentially wasting over 90% of the lens's detail-collecting capacity by using it to image a field that is far smaller than it's capable of doing well. On an APS sensor, the JML 21 does a far better job at lower magnification, despite that it's undoubtedly acquiring some aberrations and curvature of field in doing that.

You asked if resolution the wrong way round is the same as the normal way round. Well, sort of, but not as close as we'd like.

Say you want to run at m=5. So you haul out some MTF charts and look at the m=1/5 lines, thinking that m=1/5 normal is equivalent to m=5 reversed.

But the lines on the MTF chart are for something like 10, 20, and 40 lp per mm on the sensor. When you flip that lens around, those same lines mean 10, 20, and 40 lp per mm on the subject. At 5X, that's 2, 4, and 8 lp per mm on the sensor, and that level of detail is simply too coarse to be interesting.

At best, those 2, 4, 8 numbers are just surrogates for the much finer levels of detail that we really care about, which in these days of 4-micron sensors with pixel peeping are more like 50 and 100 lp per mm.

Problem is, they're not very good surrogates. It's completely possible for lens A to have better MTF than lens B at low frequencies (coarse detail), but to swap places at high frequencies (fine detail). Sigh...

If I'm frustrating you, I apologize. I find it frustrating too. I don't know a good way to predict what lenses will work well and what won't. Magnification and aperture size alone will tell you if you've gone too far into diffraction; huge deviations from design magnification are always a bad idea; using too little of the lens's angle of view suggests that you're wasting some capacity. But exactly how much any of these things matters seems to be a matter for experiment or divination rather than calculation. :(

--Rik

rjlittlefield
Site Admin
Posts: 23562
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

Harold Gough wrote:...per degree of effective angle of view...
Those words are key. To a coarse approximation, a lens will deliver so many line pairs per degree across its total angle of view (the angle of the high quality circle). So what you need to capture the most detail is to match the angle of view to the sensor size. The larger the sensor, the farther away it has to be from the lens, and the farther away, the greater the magnification. If we equate "most detail captured" with "performs best at", then the best magnification varies directly with sensor size. It is an additional refinement if the lens happens to have minimal aberrations at that same magnification.

--Rik

johan
Posts: 1005
Joined: Tue Sep 06, 2011 7:39 am
Contact:

Post by johan »

rjlittlefield wrote:If I'm frustrating you, I apologize. I find it frustrating too. I don't know a good way to predict what lenses will work well and what won't. Magnification and aperture size alone will tell you if you've gone too far into diffraction; huge deviations from design magnification are always a bad idea; using too little of the lens's angle of view suggests that you're wasting some capacity. But exactly how much any of these things matters seems to be a matter for experiment or divination rather than calculation. :(

--Rik
Not at all!

The maths you've outlined is (thankfully, phew) reasonably close to how I would assess an optic's potential when used in an unorthodox way like we do. And whilst this lack of predictability might occasionally frustrate and make our life harder, I also think that's part of the appeal (for me). In some senses, what we do is undiscovered territory, it's one of the few photography areas in which experimentation can still matter.

As in, on some macro forum over the w/e I had some guy lose the plot on me IN CAPS, because 'omfg then HOW are we supposed to know then what a reversed lens is best at?!?!?!' My answer was... that's why we test. Which I suspect bothers some types because they want it laid out on a plate, an easy a,b,c, job done.

Not only that, but my own experimentation with various optics has thrown up interesting issues. Like the optic I bought that's stellar, best in class the right way round but absolutely disastrous when reversed. And the aged $30 plastic kit elcheapo that gets widely slammed for being terrible the right way round... but then turns round and produces insanely sharp results the wrong way round! Comical =)

Having said all that there is definite use for data on the optics the right way round though. Which is why I'm wading my way through Enrico's fine tome, with dictionary on hand, to get to grips with some of the advanced photography concepts underpinning it all. It's not a pulse racer, but between that and restudying some of the propellorhead threads here hopefully some light may emerge. But then, learning this stuff is fun for me, the true sign of an anorak =)
My extreme-macro.co.uk site, a learning site. Your comments and input there would be gratefully appreciated.

rjlittlefield
Site Admin
Posts: 23562
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

johan wrote:Like the optic I bought that's stellar, best in class the right way round but absolutely disastrous when reversed. And the aged $30 plastic kit elcheapo that gets widely slammed for being terrible the right way round... but then turns round and produces insanely sharp results the wrong way round!
I am intrigued. What lenses are these?

--Rik

Harold Gough
Posts: 5786
Joined: Sun Mar 09, 2008 2:17 am
Location: Reading, Berkshire, England

Post by Harold Gough »

There seems to be need for a FAQ, in which to list for various macro lenses, including enlarger lenses:

Useful range of magnifications
Optimum magnification (including manufacturers' specifications)
Recommmended apertures at various magnifications
Working distances
Mounts

Harold
My images are a medium for sharing some of my experiences: they are not me.

ray_parkhurst
Posts: 3416
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Post by ray_parkhurst »

rjlittlefield wrote:The case of the Printing Nikkor is interesting. Although the lens is optimized for 1:1, all the graphs at http://coinimaging.com/printing-nikkor_105.html show that it's actually sharper at m=0.66. Yes, the field gets flattest at 1:1, but if you're focus-stacking, then field curvature per se does not degrade the final image.
...
Putting it all together, I'm not quite sure what "best at" really means, other than general guidance about what ranges you might consider running tests in.

--Rik
You need to look back at the intended application to see why optimization was done. The Printing-Nikkors were designed with the express purpose of film duplication and format translations. Because the subject was flat within the mechanical tolerances of the duplication system, sharpness over the whole field was paramount. No concept of focus stacking existed in those days, so all had to be done optically.

The Printing-Nikkors were the first Nikon lenses to make extensive use of computers for optical design, and the designers went all-out. Completely new and highly complex optical configurations and secretive and exotic glass formulations were developed for this series of lenses. You can see from the measured results that field flatness wide open was the primary concern, and that a small amount of resolution and sharpness was sacrificed at 1:1 to achieve maximum flatness.

Of course flatness was also a primary design goal for enlarging lenses, likley at the expense of sharpness and resolution ala the PN's. But I don't believe any enlarging lenses were optimized for a particular magnification ratio. Instead they were designed for a range of magnification to allow the photographer to optically crop as desired.

We macro hobbyists are a small (and cheap) group that are using optics at the fringe of their application space, in ways never intended by the manufacturers. So to understand from manufacturer's perspective what "best at" means you need to go back to the original purpose.

Most of the optics being developed today are for high-speed line-scan inspection use. This is not much different optically versus film duplication, and in fact most of the used 105PN's on the market today came out of decommissioned automated inspection lines. Thus from manufacturer's perspective, the "best at" criteria has not changed much, and is still usually related to maximizing the resolution across a flat field.

Line-scan cameras these days are pushing 80mm with 5um pixels (16k) and this is the application that is driving a lot of the new lens designs. Resolving 5um across 80mm is a tough challenge optically, and as with the Printing-Nikkors requires optimization at a particular magnification. Series of lenses are available that are optimized for 1x, 0.5x, 0.33x, and so on, with resolution optimized over a flat field at that particular magnification, with the magnification determining the field of view for the inspection system.

Harold Gough
Posts: 5786
Joined: Sun Mar 09, 2008 2:17 am
Location: Reading, Berkshire, England

Post by Harold Gough »

Ray's post reminds me that the Olympus Zuiko 80mm bellows lens was principally a transparency copying lens.

Harold
My images are a medium for sharing some of my experiences: they are not me.

Post Reply Previous topicNext topic