Testing lenses at home?

[dokko]

I find the most difficult thing in testing lenses on analog cameras is getting the focus 100% spot on, specially on a flat subject.
Henning Serger has a lot of experience in this and I agree with him that it takes many exposures to make sure that you have one with perfect focus.

it also means that we need very high resolution film (ie CMS 20 II) and look at the negative with a microscope or have a very high resolution scanner since otherwise we're measuring the total system resolution rather than lens performance.

with digital cameras it's much easier thanks to the live view zoom. but on the other hand the debayer algorithm comes into play. most modern algorithms will recognise straight pattern and try to improve them over the pure optical image.

all that said, often it's more worth while to test on real images including your preferred workflow. It might well be that for certain light-subject-film-printing combination, a softer lens is more to my taste than the sharpest lens.

for example, when shooting concerts on stage I prefer the rendering of older Nikkors over the more modern Zeiss ZF lenses. the Zeiss are clearly sharper, but I prefer the lower contrast and flare of the older lenses in most cases.

portraits is another case where I sometimes prefer older lenses over the latest sharpest.

 

[dokko]

If Trix resolves around 95LPM and Plus X at 125LMP what is the point of having a lens that resolve 320LPM." As news photographers you are going to be shooting TriX, GAF 500, HP4 and Plus X on occasion.

That's a valid point, but I often shoot Tri-X under very low light with the lens wide open, and many older lenses will be noticeably softer if shot at F1.4 than newer designs.
if we shoot around F4 the differences become much smaller.

the other thing to remember is if a lens resolves 100lp/mm and the film resolves 100lp/mm, the total resolution is significantly lower than 100lp/mm - it's around 50lp/mm (which isn't great)
the formula is:
1/Resolution(total) = 1/Resolution(lens) + 1/Resolution(film)

so if the lens resolves 300lp/mm and the film resolves 100lp/mm, the total resolution is around 75lp/mm.

if the lens resolves 800lp/mm and the film resolves 100lp/mm, the total resolution is around 90lp/mm.

with higher resolution films, the difference will be even larger, and we also have to factor in the resolution loss of the scanning system or the enlarger:

the weakest link will become the biggest factor. like if you scan on a Frontier system for a 5MB scan, then the resolution of the lens and the film hardly matter.
if you use a very bad lens, then the film choice becomes a small factor, etc.

for excellent total system resolution, all of the elements need to be optimised. some photos look better blurry though

 

[tjwspm]

Lens testing is really a complex system effort, especially taking into account of subject distance. I saw you mentioned 2m, while @xkaes mentioned 2.5m for 50mm lens. Is that a common practice vetted in mathematics, or industry tradition?

A distance of 50 x focal length in mm is often recommended because the lenses supposedly show their best performance there. Hence the specification of 2.5 m for 50 mm lenses from xkaes.

I have a different criterion for home use. Because I print the test charts myself with only 600 dpi, the center of the Siemens star and the very small elements of the USAF charts are not good enough for good lenses at close distances. But that is not a problem at all. I then increase the distance so far that the lens can only manage the larger rays and elements and thus get exact evaluations.

That's why I had to use a distance of 6.7 m (!) for the Sony alpha 6000 with macro lens in my example in the article. At 1 m, the camera-lens system would have been too good for the test chart and an evaluation would have been impossible.

 

[tjwspm]

the formula is:
1/Resolution(total) = 1/Resolution(lens) + 1/Resolution(film)

for excellent total system resolution, all of the elements need to be optimised. some photos look better blurry though

Thank you for all these hints. Do you have a source for the formula? That would be interesting. So far I've only seen it in connection with the calculation of electrical resistors.

Yes, you definitely have to pay attention to the entire process chain. I use the Adox cms 20 for the test photos. For digitizing I photograph the negatives with a Sony alpha 7 at 61 Mpixel. At the same time, I analyzed the negatives with a professional Wild-Heerbrugg 3z Combistereo microscope with a Leitz 25mm/2.8 lens and achieved similar results with both methods.

And yes, one of my favorite photos is completely blurry

 

[dokko]

Thank you for all these hints. Do you have a source for the formula?

I'v seen it in several places, it's of course only an approximation since in reality it's an interacton of all the curves. but here is a paper by Fuji that lists it (page 64) along with some other interesting things:

https://asset.fujifilm.com/www/ca/files/2020-03/d52487c5c6f84e7f935c299491c5c1ff/ProfessionalFilmDataGuide.pdf

System Resolving Power :
The resolving power of a system can never exceed the lowest resolving power among the components that comprise the system. In fact, the system value is actually lower. For example, if a lens with a resolving power of 200 lines/mm were used with a film with a resolving power of 50, the combined system resolving power would be 40. To accurately determine the resolving power of a system, use the following equation :

1/R = 1/r1+1/r2+1/r3
(where "R" is the resolving power of the system and each "r" represents that of each component)

 

[tjwspm]

I'v seen it in several places, it's of course only an approximation since in reality it's an interacton of all the curves. but here is a paper by Fuji that lists it (page 64) along with some other interesting things:

https://asset.fujifilm.com/www/ca/files/2020-03/d52487c5c6f84e7f935c299491c5c1ff/ProfessionalFilmDataGuide.pdf

System Resolving Power :
The resolving power of a system can never exceed the lowest resolving power among the components that comprise the system. In fact, the system value is actually lower. For example, if a lens with a resolving power of 200 lines/mm were used with a film with a resolving power of 50, the combined system resolving power would be 40. To accurately determine the resolving power of a system, use the following equation :

1/R = 1/r1+1/r2+1/r3
(where "R" is the resolving power of the system and each "r" represents that of each component)

Thank you. This means that I can calculate the resolution of the lens if I know the resolution of the film.
So: I measure a resolution of the lens/film system (Minox AIIIs/Adox cms 20) of e.g. 218 LP/mm and estimate the resolution of the film very optimistically at 800 LP/mm.

Then I could calculate
rLens = 1 / ( 1/RSystem - 1/rFilm)
rlens = 1 / ( 1/218 - 1/800) = 300 LP/mm

Cool, even if it's just an estimate.

 

[dokko]

Thank you. This means that I can calculate the resolution of the lens if I know the resolution of the film.
So: I measure a resolution of the lens/film system (Minox AIIIs/Adox cms 20) of e.g. 218 LP/mm and estimate the resolution of the film very optimistically at 800 LP/mm.

Then I could calculate
rLens = 1 / ( 1/RSystem - 1/rFilm)
rlens = 1 / ( 1/218 - 1/800) = 300 LP/mm

Cool, even if it's just an estimate.

yes, as a rough estimate it's useful.
reality is a bit more complex, since even the term resolution is fuzzy. But if you're interested in a deeper dive, Zeiss has a paper which explains MTF, and some example of how lens and film interact (page 10 onwards, though it's good to read from the start):

https://lenspire.zeiss.com/photo/app/uploads/2022/02/technical-article-how-to-read-mtf-curves-01.pdf

there's a second one which is more related to lenses and digital sensors:

https://lenspire.zeiss.com/photo/app/uploads/2018/04/CLN_MTF_Kurven_2_en.pdf

 

[tjwspm]

reality is a bit more complex, since even the term resolution is fuzzy. But if you're interested in a deeper dive, Zeiss has a paper which explains MTF, and some example of how lens and film interact (page 10 onwards, though it's good to read from the start):

https://lenspire.zeiss.com/photo/app/uploads/2022/02/technical-article-how-to-read-mtf-curves-01.pdf

An article that leaves no question unanswered, thank you for that too. And it's also very heavy fare. Fortunately, the text is also available in German . Absolutely recommendable to get an impression of the complexity.

This is of course the professional league and Leitz naturally claims to publish correct absolute measurements.

My claim is a different one. It would be enough for me if I could compare my own cameras/objects to each other. This is much easier, because it can be relative comparisons. Of course, I can then also guarantee that I always measure under the same (simplified) conditions.

Yes, I want more than just a feeling as to which lens is better, i.e. numerical values. But the measurement must be possible "at home" with on-board equipment. And it should be possible without studying physics and with little money and time. The article helped me a lot to check my text again for this balance.

 

[tjwspm]

To make use of this detailed excellent work, it really helps if you have some sort of intended use in mind.
For most pictorial work, the differences between the wide selection of available "good quality" lenses aren't particularly important - issues of condition aside.
If you need highly accurate copies of original documents, issues like distortion are very important, as are high line pair resolution numbers.
If you regularly make very large enlargements, high MTF numbers are important.
If your needs include exceptional low light capabilities, great results with the lens wide open are important.
A "needed results" oriented approach can make good use of objective tests.

Yes, the question of purpose is certainly a key issue.

My original reason for doing these tests was to find out how big the difference in image quality is between the Minox I (Riga Minox) and the Minox III. To my knowledge, no one has ever done this before. So it wasn't necessarily about absolute results, but about relative results.

During this work I also found out that my digital camera, which I had previously used to digitize the minox negatives, was much worse than I thought. I then did a few more tests with my digital cameras and even bought a new one to digitize the negatives. In this respect, the testing was worthwhile for me.

 

[blee1996]

My original reason for doing these tests was to find out how big the difference in image quality is between the Minox I (Riga Minox) and the Minox III. To my knowledge, no one has ever done this before. So it wasn't necessarily about absolute results, but about relative results.

So I am curious, what are your findings regarding these two Minox lenses?

 

[tjwspm]

So I am curious, what are your findings regarding these two Minox lenses?

The evaluation has not yet been completed. Correct exposure without blurring the Minox I is difficult because it has no tripod connection and no flash contact. I also had to take an additional series of exposures to check the distance setting.



As soon as I am ready, I will report in detail here.

 

[Crysist]

Who has ever tested lenses at home? What experiences have you had with this?
In order to test the resolution of cameras for which there were no evaluations yet, I was forced to carry out tests myself. But I had never done this seriously before, so I had to find out more first. There are many tips online, but they were either too technically complex or they were just the author's emotional assessments.

I was tired of hearing 'creamy bokeh' and 'sharp corners' in lens review videos. With a little effort, I could carry out meaningful lens tests at home. The results may not be as good as professional laboratory tests, but in most cases they were sufficient to get to know my own lenses better.

Do you have any ideas or comments on how to improve my process?
I've written down what I did here:

Testing lenses

This guide covers methods like using Siemens star and USAF 1951 test charts, measuring line pairs per millimeter, and practical field tests.

moments-of-now.com

Nice article! I haven't seen many others do structured resolution tests on film except for Henning Serger or Tim Parkin. I've seen others but they have a bunch of methodology issues. The main thing that Henning stresses is that scanners are the resolution killer. That's not to say your methods are insufficient, high resolution camera scanning is kind of the top-of-the line way to transfer film as digital cameras never stopped improving.

Regardless, when assessing lenses, maybe the best way is to do it entirely analog... and not even involve film. You can aim a microscope behind the lens and bring into focus where the aerial image from the lens comes into focus. That way you can visually check these features. Or do a hybrid approach and attach a camera and perform a slant-edge test. Also, remember that the light travels straight from the exit pupil to each point on the film plane, so you'll need to always look towards the exit pupil as you move away from the optical axis. I totally understood how this worked when I first tried holding loupes up to my lenses... *cough*

I first learned of this approach when looking into Tim Parkin's tests about film's latitude. I came across a nice page by Rik Littlefield, who is the current admin of photomacrography.net, where he shows off this setup at the end of an exposure latitude test to verify the lens isn't the limiting factor. It's also a rather nice test!

Another way is the inverse of that: lens projection (this might not be the right term?). You can put a slide/transparency of a test pattern at the focal plane of your test lens, shine light through the slide, and the lens will project that image onto a wall/screen. The fatal issue with this is you need a micrometer-scale pattern to project. I had the idea of using a cheap microscope reference slide, but I don't know if they come in the easy-to-measure patterns you mention. And you need to jerry-rig a projection apparatus. Proper resolution targets are sold as microscope slides by Thor labs and Edmund Optics, but they're etched in chromium at an extremely high resolution. They're quite expensive. But there's probably many more fine-patterned scales, etc that can be used.

For your Minox it'd probably be harder because I don't think you can remove the back? So there's no simple way to optically measure the lens without taking a picture. Unless you put tiny first-surface mirror inside at a 45 degree angle like an SLR would. Or salvage the lens from one.

I don't know the theoretical limitations of each approach. Bob Atkins mentions here that apparently there's an issue when interpreting the results of the aerial image near the diffraction limit.

Either way, this approach just takes the imaging medium out of the process.

When thinking about how sharp a lens is, the first question is what is the resolving power of the film? ADOX CMS 20 datasheet reports 800LPM, with your 16mm camera and lens I doubt that it will resolve 800LPM, I assume that a 61MP sensor can resolve more than the lens so you should be able to count LPM of the lens. With a AF test chart you sould be able to determine distortion. With a 16mm negative, the format is so small it will have great depth of field, would be interesting to see the bokha

I have a very relevant photo.net post to share, however! Last year I was heavily researching this topic and came across this interesting guy, Martin Tai, who made a post in 1998 about testing the lens on his Minox. Seems to be a giant Minox fan, he kept this thread updated until 2004 with his experiments, and he includes some photomicrographs (you need to use the wayback machine to access them).

He also shared it on this Minox user's group on usenet on which you can find an archive of these posts with better formatting. The broken formatting on photo.net hurts my eyes :c

Towards the end of the thread he explains the Minox's performance: as an example, shrinking a Tessar design down by 3x, making a smaller image circle, would make the focus circle 3x smaller. I had never considered that until then, but it makes sense. Especially considering APS-C lenses perform spectacularly... on a smaller image circle as well.

Here are a pair of his images from this post in case wayback isn't working for those links (for reference: the Gooderham Building in St. Lawrence, Toronto, Canada):

And the associated crop of the street sign:

As an epilogue to this inclusion, he is still sharing photos he's taken on Minox film cameras to this day, nearly 26 years later. He last shared some a few weeks ago. I admire this guy's tenacity and love of the format!

And that's about everything I know about this topic. Thought I'd dump it all here for anyone's benefit or for someone to fill in my gaps of knowledge.

 

[tjwspm]

Nice article! I haven't seen many others do structured resolution tests on film except for Henning Serger or Tim Parkin. I've seen others but they have a bunch of methodology issues. The main thing that Henning stresses is that scanners are the resolution killer. That's not to say your methods are insufficient, high resolution camera scanning is kind of the top-of-the line way to transfer film as digital cameras never stopped improving.

Regardless, when assessing lenses, maybe the best way is to do it entirely analog... and not even involve film. You can aim a microscope behind the lens and bring into focus where the aerial image from the lens comes into focus. That way you can visually check these features. Or do a hybrid approach and attach a camera and perform a slant-edge test. Also, remember that the light travels straight from the exit pupil to each point on the film plane, so you'll need to always look towards the exit pupil as you move away from the optical axis. I totally understood how this worked when I first tried holding loupes up to my lenses... *cough*

I first learned of this approach when looking into Tim Parkin's tests about film's latitude. I came across a nice page by Rik Littlefield, who is the current admin of photomacrography.net, where he shows off this setup at the end of an exposure latitude test to verify the lens isn't the limiting factor. It's also a rather nice test!

Another way is the inverse of that: lens projection (this might not be the right term?). You can put a slide/transparency of a test pattern at the focal plane of your test lens, shine light through the slide, and the lens will project that image onto a wall/screen. The fatal issue with this is you need a micrometer-scale pattern to project. I had the idea of using a cheap microscope reference slide, but I don't know if they come in the easy-to-measure patterns you mention. And you need to jerry-rig a projection apparatus. Proper resolution targets are sold as microscope slides by Thor labs and Edmund Optics, but they're etched in chromium at an extremely high resolution. They're quite expensive. But there's probably many more fine-patterned scales, etc that can be used.

For your Minox it'd probably be harder because I don't think you can remove the back? So there's no simple way to optically measure the lens without taking a picture. Unless you put tiny first-surface mirror inside at a 45 degree angle like an SLR would. Or salvage the lens from one.

I don't know the theoretical limitations of each approach. Bob Atkins mentions here that apparently there's an issue when interpreting the results of the aerial image near the diffraction limit.

Either way, this approach just takes the imaging medium out of the process.



I have a very relevant photo.net post to share, however! Last year I was heavily researching this topic and came across this interesting guy, Martin Tai, who made a post in 1998 about testing the lens on his Minox. Seems to be a giant Minox fan, he kept this thread updated until 2004 with his experiments, and he includes some photomicrographs (you need to use the wayback machine to access them).

He also shared it on this Minox user's group on usenet on which you can find an archive of these posts with better formatting. The broken formatting on photo.net hurts my eyes :c

Towards the end of the thread he explains the Minox's performance: as an example, shrinking a Tessar design down by 3x, making a smaller image circle, would make the focus circle 3x smaller. I had never considered that until then, but it makes sense. Especially considering APS-C lenses perform spectacularly... on a smaller image circle as well.

Here are a pair of his images from this post in case wayback isn't working for those links (for reference: the Gooderham Building in St. Lawrence, Toronto, Canada):

View attachment 383261

And the associated crop of the street sign:
View attachment 383260

As an epilogue to this inclusion, he is still sharing photos he's taken on Minox film cameras to this day, nearly 26 years later. He last shared some a few weeks ago. I admire this guy's tenacity and love of the format!

And that's about everything I know about this topic. Thought I'd dump it all here for anyone's benefit or for someone to fill in my gaps of knowledge.

Thank you very much for all the information!

I was aware of the tests by Henning Serger, but I will certainly follow up the links to Tim Parkin and Rik Littelfield.

I really appreciate Martin Tai's work on Minox lenses and his photos. His photo “Wuahn East Lake” is one of the best Minox photos I know. By the way, because it's not that sharp. He gave me permission to use it on my site:

The Minox Paradox

When it comes to camera technology, to me applies: less is more. Little technology means that light falls briefly through a lens onto a film.

moments-of-now.com

 

[Crysist]

Thank you very much for all the information!

I was aware of the tests by Henning Serger, but I will certainly follow up the links to Tim Parkin and Rik Littelfield.

Of course! For all the time I spent researching this, I might as well share all the other important figures and their articles that I've come across!

I really appreciate Martin Tai's work on Minox lenses and his photos. His photo “Wuahn East Lake” is one of the best Minox photos I know. By the way, because it's not that sharp. He gave me permission to use it on my site:

The Minox Paradox

When it comes to camera technology, to me applies: less is more. Little technology means that light falls briefly through a lens onto a film.

moments-of-now.com

Oh that's awesome! I hadn't ventured too deep into his gallery, but stuff like that has a lot of charm! Also, messaging someone like him who has so much experience with this line of cameras, I'd feel like I'd stumbled upon a master of Minox.

Anyway, taking great photos like that seems fun to the Minox. I once held an early model at a camera shop, the one without the meter, crazy how small it is and how useable it is despite its size! The collapsing mechanism also doubling as the film advance is something I really liked as well! I would need to gain more courage to make that jump, though!

If I might add another idea: if you like films like CMS 20, a film that seems to have similar traits but is still produced would be Fuji HR-21 in 35mm. The neat thing for the Minox, however, is the film is sprocketless. You could slit twice as much as with normal 35mm!

 

[tjwspm]

So I am curious, what are your findings regarding these two Minox lenses?

As promised, here is my comprehensive comparison test. I have opened a new thread for this:

Myth Busted: Minox I Holds Its Own Against Minox III

Have you ever wondered how the image quality of the very first Minox camera, the Riga Minox (Minox I), compares to its successor, the Minox A (Minox II/III)? During the last months I dived deep into a fascinating investigation to answer exactly that question! To do this, I directly compared...

www.photrio.com