Testing coverage of lenses

[Greg]

Was wondering how other people test their lenses. Did a search in the Forum and didn't come up with a thread on the different ways people use to test lenses (probably again just my not using the correct search words), so here goes... I invite others to describe their lens testing methods and procedures.

"Bench testing" shooting a commercial lens resolution chart I had done in the early 1980s, but found that all it told me was the resolution of the lens when shooting a flat surface at a relatively close distance of 8 feet away. Hardly at all like what I shot in the field.

So decided to shoot a subject at the typical distance of most of my LF photographs. For me here in New England, it was a few hundred feet away. Include subject matter that had enough detail for me to make subjective observations on the sharpness of the lens. Pick a covered shooting location that I could return to year round (not so simple here in New England, especially in the winter). Lighting from the sun optimal to be 90 degrees from the lens axis so as to give me shadows and highlights to indicate the contrast of the lens, flare, and more, so I only did my lens tests on sun lit days after shooting one lens test on a very overcast day and seeing that the negative didn't have enough contrast in it to make a valid determination of the sharpness of the optic.

What I chose was the front porch of our town's historical museum. First floor porch was elevated and rock solid. Second floor's porch and roof overhead provided me with protection from the weather and wind. Place was accessible to me 365 days a year. I was able to back up my SUV next to it. So close that I could work out of the back of the SUV being maybe 10-12 feet away from where I set up my tripod. Subject matter (attachment) was for me optimal in distance. Across the frame were many signs, license plates, road signs, telephone poles and wires, and more enabling me to evaluate the sharpness of the lens. Top quarter to third of the frame included the sky, which allowed me to see what the light falloff was for the lens. Also included in the background is the side of a hill, maybe a mile away. Tripod used is a Linhof Heavy Duty tripod with a Linhof heavy duty pan/tilt head... need I say no more, the camera was "rock solid".

Now for the actual testing my lenses. First off I shoot negatives of the next largest format: 8x10 for 4x5 and Full Plate lenses, and 11x14 for 8x10 lenses. For 11x14 just have to shoot 11x14. Usually shoot two exposures: one stopped down half the way, and the other stopped down all the way (I usually shoot at apertures between these). Only exceptions are for my Petzval and projector optics that I only shoot wide open... for these I introduce a 10 stop ND filter taped in front of the lens. Attached is a low res image file of the negative that I shot with my Darlot Portrait lens wide open on 8x10 film, actual image on negative is a whole lot sharper....

Results: Well it just works for me. Have easily been able to determine and measure the "sharp image" coverage and the "illumination" coverage of my lenses. Tried using X-Ray film but opted for using normal Ilford emulsions. Considering what I have paid for my lenses, shooting two or three sheets of conventional film represents a mere fraction of the purchase prices of the lenses.

OK.. what did I learn over the past few years and what would I change? Learned to absolutely be consistent and keep detailed notes when testing my lenses. Once the negatives are dried, I mark them with the lens and the f stop used using a Sharpie. Also when being approached by someone curious in what I am doing, I just stop my lens test and talk with them... multi-tasking, for me, usually ended up in my not recording or shooting my planned out exposures. Choosing a place to do repeated lens testing so important... previous place was on easily accessible property, but then all of a sudden was posted "No Trespassing".

[Mark Sawyer]

I'd consider shooting paper negatives rather than film, just to save on cost.

[Pere Casals]

Was wondering how other people test their lenses.

I do that with the help of 2 tools:

> A USAF 1951 chart, here you have group+element to Lp/mm https://en.wikipedia.org/wiki/1951_USAF_resolution_test_chart

> A DSLR, with high pixel/mm density, like Nikon D3200, placed in the back. Alternatively I use an eyepiece to see it directy.


I placed a Nikon F mount extension ring (those for macro) on a Sinar lensboard, to place the D3200 there(without the DSLR lens), then I place that lensboard in the back of the view camera.

By using the shift movement I can explore the circle boundary, note we can shift the front to left and the back to right to get greater off axis shift. Still it was not enough and I placed the F mount ring at far left of the lensboard.

The D3200 has some 260 pix/mm (130 pix/pairs, less bayer) on sensor, so it can be useful to see reasonably well a 50 Lp/mm performance from a 1951 target. You can make a USAF 1951 slide to test in high contrast conditions.



You also can use an eyepiece to be placed in the back of the view camera (also using a lensboard), and see it directly by naked eye, it is what I mostly use presently.

Placing a film and developing would give real information, but also it is a mess. Here Pérez describes some methodology http://www.hevanet.com/cperez/testing.html.


To me a 20x eyepiece (I use a microscope Periplan) delivers enough information to know how it works, even it is not necessary to use the USAF 1951 resolving power target, just observing the 20x image of a landscape with the Periplan tells you if that image is good enough at that off axis shift.

Note that with the Periplan you observe the image in the air, so there is no projection on the GG that would degradate the image because the GG texture.


So short answer is place a 20x Periplan (or a telescope eyepiece) in the back of the viewcamera, use front and rear shifts to place it off axis. Add a resolving power target if you want numeric evaluations.

[Bob Salomon]

I do that with the help of 2 tools:

> A USAF 1951 chart, here you have group+element to Lp/mm https://en.wikipedia.org/wiki/1951_USAF_resolution_test_chart

> A DSLR, with high pixel/mm density, like Nikon D3200, placed in the back. Alternatively I use an eyepiece to see it directy.


I placed a Nikon F mount extension ring (those for macro) on a Sinar lensboard, to place the D3200 there(without the DSLR lens), then I place that lensboard in the back of the view camera.

By using the shift movement I can explore the circle boundary, note we can shift the front to left and the back to right to get greater off axis shift. Still it was not enough and I placed the F mount ring at far left of the lensboard.

The D3200 has some 260 pix/mm (130 pix/pairs, less bayer) on sensor, so it can be useful to see reasonably well a 50 Lp/mm performance from a 1951 target. You can make a USAF 1951 slide to test in high contrast conditions.



You also can use an eyepiece to be placed in the back of the view camera (also using a lensboard), and see it directly by naked eye, it is what I mostly use presently.

Placing a film and developing would give real information, but also it is a mess. Here Pérez describes some methodology http://www.hevanet.com/cperez/testing.html.


To me a 20x eyepiece (I use a microscope Periplan) delivers enough information to know how it works, even it is not necessary to use the USAF 1951 resolving power target, just observing the 20x image of a landscape with the Periplan tells you if that image is good enough at that off axis shift.

Note that with the Periplan you observe the image in the air, so there is no projection on the GG that would degradate the image because the GG texture.


So short answer is place a 20x Periplan (or a telescope eyepiece) in the back of the viewcamera, use front and rear shifts to place it off axis. Add a resolving power target if you want numeric evaluations.

Great test if you will be shooting flat art work with a lens corrected for it.
Terrible test if you are shooting people, products, landscapes, rooms, houses, etc. with lenses designed for general photography.

[Pere Casals]

Great test if you will be shooting flat art work with a lens corrected for it.
Terrible test if you are shooting people, products, landscapes, rooms, houses, etc. with lenses designed for general photography.

Hello Bob,

With the 20x periplan you also can observe how it results people, products or landscapes, you see it very enlarged and without the GG degradation, in real photographic conditions.

Or one can print a large USAF 1951 target and placing it 10m away.

Just tell me why observing "in the air" image circle boundary with a 20x periplan is not a good idea !!


Regards

[Bob Salomon]

Hello Bob,

With the 20x periplan you also can observe how it results people, products or landscapes, you see it very enlarged and without the GG degradation, in real photographic conditions.

Or one can print a large USAF 1951 target and placing it 10m away.

Just tell me why observing "in the air" image circle boundary with a 20x periplan is not a good idea !!


Regards

It is testing general purpose lenses with a flat chart at near distances that is the problem. Test your lens on real world subjects.

[xkaes]

"Coverage" covers a lot. Broken into two categories, there's "Light coverage" (LC) and "Resolution coverage" (RC). To a certain extent, they can be tested at the same time, in the same way, in the same negatives. While the manufacturer has its own standard for each of these (and most people just accept that) we can each have our own standards -- which can vary depending on the subject and the situation.

LC and RC both change as the aperture changes and what I consider acceptable probably won't be the same for you -- nor the method of "measurement".

LC can be examined by taking exposures of a scene or a gray card, etc. at all or selected f-stops, distances, lighting, etc. You can then simply eyeball the negatives or measure the density loss with a meter/densitometer from the center to the corner. This can tell you which f-stops to avoid (or use) depending on your needs. Another approach is to read the light on the ground glass if your meter has an accessory for that. The problem here is that the ground glass itself loses illumination from center to corner and needs to be subtracted out in the calculations.

RC can be tested in the same way, at the same time, by adding in a resolution chart in the center of the image and another at the corner. My tests are taken at 50 times the focal length of the lens. So my charts and gray cards for a 150mm lens are 25 feet away (150mm X 50 = 7500mm). That's not close-up, and if you will be using the lens(es) at a different distance, you can set that. You can add in Kodak or other color and density patches into the scene as well. After delevopment, you can use a high powered loupe, reversed enlarging lens, etc. or make enlargements to check the resolution.

But what resolution and light fall-off is acceptable is your call. With LC you have the options of: f-stop adjustment, center filter use, and/or burning in on the easel. With RC your only option is f-stop adjustment.

I've only gotten rid of one lens -- because of resolution. It wasn't bad, but it wasn't a match for my other lenses. Who knows, I might have just had a "bad" lens. The whole point is that you need to decide what is "acceptable" to YOU -- and know that YOU and "acceptable" can change over time.

Here are some of my results:

http://www.subclub.org/library/page2.html

[Pere Casals]

It is testing general purpose lenses with a flat chart at near distances that is the problem. Test your lens on real world subjects.


Bob, a bit I'm a green learner, you know how green, but let me point what I've been discovering about amateur lens testing.

About target distance, for example Sironar-N was optimized for 1:10 to infinity, so for a Sironar-N 210mm placing the resolving target at some 5m it is way enough to not have a worse image because a too near target. (Sironar-S is optimized for 1:5 to infinity, correct me if I'm mistaken).


Of course. When testing lenses one has to know what he is doing, and also knowing the lab vs field difference.

So measuring 1:x factor after bellows extension it can be important for the test.


IMHO testing lenses with amateur gear it can give impressive good results, for example one can measure easily chromatic aberration importance by simply illumitaing the target with a monochromatic (well, narrow band) green LED light. Flare is also very easy to measure with a luxometer (0.01 lux type are today near for free). But one has to know how to do it.

Another important thing is target contrast, a paper target may deliver 1:100 contrast as much, but an slide can deliver much more, and resolving power depends on target contrast, in special if film is involved.

To me the straightest way is an eyepiece in the back of the camera, as one can see through the lens with x20 magnification, with own eyes, so one can see targets or real subjects.


And I agree, testing on real world subjects it is also necessary, for example the lower Sironar S flare may make a great difference with fine textures, as film gains resolving power with increased microcontrast, (TMX can resolve 4x more lines with 1:1000 contrast vs 1.6:1 contrant) so scene flare may change the result.


But, for OP original question, IMHO observing the circle boundary with a 20x periplan it would be an straight solution.

[Alan Gales]

It is testing general purpose lenses with a flat chart at near distances that is the problem. Test your lens on real world subjects.


+1

That's what I do. When I buy a lens for any format camera I just go out and shoot normal subjects with it. If I don't like the results I get with it then I sell it.

It may not be scientific but I learn if I like a lens or not.

[Ken Lee]

About target distance, for example Sironar-N was optimized for 1:10 to infinity... Sironar-S is optimized for 1:5 to infinity, correct me if I'm mistaken

Sironar S lenses are optimized for 1:10. See http://www.arca-swiss-magasin.com/contents/fr/rodenstock_apo-sironar-s.pdf


http://www.kennethleegallery.com/images/forum/SironarSOpt.png

Sironar N lenses are optimized for 1:20. See http://www.linos.com/pages/mediabase/original/rodenstock_apo-sironar-n_e_2474.pdf


http://www.kennethleegallery.com/images/forum/SironarNOpt.png

Perhaps you're thinking of Fujinon A lenses: they are optimized for 1:5.

[xkaes]

Perhaps you're thinking of Fujinon A lenses: they are optimized for 1:5.

Slight correction. The A lenses are optimized for 1:1 to 1:5, but I've used my A 180mm f9 from 2X to infinity with results satisfactory to me.

[Ken Lee]

Excellent - Thank you !

[Pere Casals]

Sironar S lenses are optimized for 1:10.

Sironar N lenses are optimized for 1:20.



http://www.kennethleegallery.com/images/forum/SironarNOpt.png

Perhaps you're thinking of Fujinon A lenses: they are optimized for 1:5.

Thanks for pointing it.

So for a 1:20 magnification (Sironar-N210mm) we need little some 4.5m distance from lens to the target. It would be interesting to see how many Lp/mm are lost as focus is closer.

[Pere Casals]

Sironar S lenses are optimized for 1:10. See http://www.arca-swiss-magasin.com/contents/fr/rodenstock_apo-sironar-s.pdf


http://www.kennethleegallery.com/images/forum/SironarSOpt.png

Sironar N lenses are optimized for 1:20. See http://www.linos.com/pages/mediabase/original/rodenstock_apo-sironar-n_e_2474.pdf


http://www.kennethleegallery.com/images/forum/SironarNOpt.png

Perhaps you're thinking of Fujinon A lenses: they are optimized for 1:5.


One contradictory thing is that Sironar-S graph looks worse than the Sironar-N:

166607

Sironar-S Modulation for 20 Lp/mm falls earlier (as off-axis distance grows) than Sironar-N . I guess this is because scale is not the same, 0.05 vs 0.1, so Sironar-S test is made in more a complicated situation for a lens that is also optimized for distant objects.

[Bob Salomon]

One contradictory thing is that Sironar-S graph looks worse than the Sironar-N:

166607

Sironar-S Modulation for 20 Lp/mm falls earlier (as off-axis distance grows) than Sironar-N . I guess this is because scale is not the same, 0.05 vs 0.1, so Sironar-S test is made in more a complicated situation for a lens that is also optimized for distant objects.

The curves on the N chart start their drop at 78.7mm of coverage, the S chart starts the drop 10 mm further out.

All of the first three S curves start above 90% they are higher then the N curves.

You have to realize that the bottom scale is for half of the coverage of the lens design so the S scale is expanded further out then the N scale since the S covers a larger circle.

[Pere Casals]

The curves on the N chart start their drop at 78.7mm of coverage, the S chart starts the drop 10 mm further out.

All of the first three S curves start above 90% they are higher then the N curves.

You have to realize that the bottom scale is for half of the coverage of the lens design so the S scale is expanded further out then the N scale since the S covers a larger circle.

Thanks for pointing it Bob, I had not seen that the abscissa had a different scale, related to a larger circle.

Anyway the 45 corners (153.7 mm Diagonal/ 2, so some 75mm is farthest point) are better for the N, 70% vs 60% modulation transfer for 20Lp/mm, if 45 sheet is well centered in the circle.

I ask myself what graph would it have the S at 0.1 scale...


What it is also true is that if the 45 film is not well centered in the image circle then with N corners go earlier to much lower performance.


Anyway I guess that S may have some ED elements (isn't it ?, I think I read that somewhere), and this is what may make the real difference, as secondary chromatic aberration may be lower, at the cost of larger thermal expansion of the glass, and probably more curved surfaces design because lower refraction index, so slightly more weight.


I guess that real practical difference S vs N is not well explained by the (20Lp/mm) graphs. Also those graphs don't speak about ultimate resolved Lp/mm(+70?), where secondary chromatic aberration may have a real effect on resolved lines amount, and microcontrast.

[Jac@stafford.net]

The problem here is that the ground glass itself loses illumination from center to corner and needs to be subtracted out in the calculations.

No. If you wish to measure corner illumination, please consider aerial focus throughout the the entire field.

But two points: Why are you asking a large format group about your testing of miniature formats? Scale does not follow ordinary arithmetic, and including enlarging lenses is irrelevant regardless of film size because it is an entirely separate empirical paradigm.

[xkaes]

No. If you wish to measure corner illumination, please consider aerial focus throughout the the entire field.

But two points: Why are you asking a large format group about your testing of miniature formats? Scale does not follow ordinary arithmetic, and including enlarging lenses is irrelevant regardless of film size because it is an entirely separate empirical paradigm.

Three points back at you:

#1. The instrument I have is designed to measure light on the ground glass, not aerial measurements. The original OP wanted to know how "we" measure things, so I offered my approach. You don't like it? The OP doesn't like it? Fine with me. No one is forcing anything on you.

#2. I am in no way asking anyone to compare smaller formats to larger formats for the purpose of this discussion. I am merely showing how I set up my tests. The original OP wanted to know how "we" measure things, so I offered my approach. You don't like it? The OP doesn't like it? Fine with me. No one is forcing anything on you.

#3. I may be in the minority, but I actually make prints. So I always include results of my -- lens, film, developer, etc. -- tests made through an, appropriate for the format, enlarging lens (which has also been tested for optimum results). But that's just how I do it. You don't like it? The OP doesn't like it? Fine with me. No one is forcing anything on you.