Putting Values against Performance of Various Lenses - Resolving Power

[bvf]

Lenses used in this test:

• Canon 28/2.8 IS USM
  
• Canon 50/1.8 EF STM
  
• Canon 200/2.8L EF
  
• EL-Nikkor 80/5.6 (metal version, enlarger)
  
• EL-Nikkor 105/5.6 (metal version, enlarger)
  
• Enna München 28/3.5 Lithagon
  
• Leitz Wetzlar 50/4.5 Focotar-2 (enlarger)
  
• Prinz Galaxy 135/3.5
  
• Samyang 14/2.8 ED AS IF UMC
  
• Sigma 24-105/4.0 DG OS HSM Art
  
• Sigma 105/2.8 Macro EX DG OS HSM
  
• Soligor 35/3.5 (enlarger)
  
• Steinheil 50/2.8 Cassar S
  
• Tamron 500/8.0 SP Macro Adaptall-2
  

See also: Putting Values against Performance of Various Lenses - Aberrations

---

I’ve been impressed by the work that Ulf is doing in formally measuring UV transmission of a large collection of lenses. And then there are all various scurrilous comments made about the quality of the Cassar S. So it seemed a good idea to try to evaluate the image quality of the various lenses in my possession and to try to get some numbers to support the discussion. So here are what I’ve been able to get using this lens testing chart:

 

 

This chart is designed to directly give resolution in lines/mm when photographed at 26x focal length of lens. As mentioned below, I had to use higher multiples of focal length.

 

I’ve split the results into two posts – this one on lens resolving power, and a second one on aberrations.

 

The table of results is at the end – but here are some comments and conclusions so you can avoid getting a headache looking at that wall of numbers.

 

Caveats and comments

 

1. Pixel density in the sensor has a major impact on the observed resolving power. The chart is printed for a maximum value of 80 lpm. This was designed for film, which has a pixel equivalent of 10-15 Mpixels per full 35mm frame. Using the full-frame Canon EOS 6D II (about 25 Mpixel in a full 35mm frame) most lenses were exceeding this. And on the Sony A6000 (also about 25 Mpixel but squeezed into an APS-C sized sensor) the observed resolution was even higher. As a result the lens-chart distance had to be increased with the lpm values scaled accordingly. When comparing UV/IR performance against visible, use the Visible APS-C columns, not the Full Frame columns.

 

2. There clearly are focusing errors in these result. For example, the real “Corner – Refocus” values cannot ever be worse than the “Corner” values, yet this happens quite frequently in my results.

 

3. Assume an error margin of +/- 10% in the numbers. A very slight focussing error could change the numbers, and it was sometimes hard to figure out what the actual resolution was. These difficulties are exacerbated by the high levels of aberration in some lenses – for example, with astigmatism bars printed vertically with higher lpm values might be resolved whereas bars printed horizontally with lower lpm values might not be.

 

4. There are two Igororiginals in the table. This is because the first one I received was clearly ill, with absolutely dreadful image quality. Igor sent a replacement (“IgorOriginal 1” in the table) but did not want the first one (“IgorOriginal 2” in the table) returned. Recently I dismantled the lens, and found that the problem was that the 2^nd element from the rear was reversed. So I had two samples of this lens – cosmetically different but probably similar under the covers.

 

5. All test images were created as 16-bit TIFF from the RAWs to avoid any sharpness artefacts created by going to JPEG.

 

6. All images made at ISO 100. Flash was used for all images except for the SP500 and Canon 200mm, which were made outdoors in sunlight.

 

7. The camera was moved for different focal lengths to keep the ratio of (lens-chart distance):(focal length) the same.

 

 

General observations

 

1. In general, the vintage UV-friendly lenses compete well with modern lenses in terms of resolving power in the centre of the image. However, they compare unfavourably at the edge of frame. The better enlarger lenses (Focotar-2, El Nikkor) do better in this respect.

 

2. There are a number of cases where the “Corner” resolution is higher than “Centre”. This is unlikely to be a focussing error as the measurements were taken from the same image. So it is likely to be a real effect. It happens predominantly with enlarging lenses – so perhaps it is a result of the lens designer trying to achieve flatness of field.

 

3. f/8 was chosen as the aperture to test at because all the testing I have done in the past, and resolution data I have picked up, always point at f/8 as being the sweet spot. However, there are a number of instances in the data here where the full-aperture performance is better than the f/8 performance. This only occurs in the IR, and with the Soligor enlarger lens (35mm f/3.5 – so the actual aperture is very small and more likely to produce diffraction) and with the Focotar-2 and El Nikkor enlarger lenses (where the max. aperture is not much larger than f/8). This must be because the effect of diffraction is greater at longer wavelengths, with the size of the Airy disk being roughly proportional to wavelength. So in the IR the optimum aperture will be larger than f/8. Conversely, in UV a smaller aperture would be optimum – if you’ve got enough UV light to take advantage of stopping down.

 

 

Comments on specific lenses.

 

1. It turns out that the much-maligned Cassar S has a lot to be maligned about! The image quality away from the centre was generally abysmal, although its IQ is less bad in UV than elsewhere. Who on earth was trying to tell us that this was a fine lens? Probably time to put my copy on ebay …

 

2. The Igororiginals were also pretty poor in terms of IQ – although again they performed better in the UV region – in fact IgorOriginal 1 came out best at the 345nm point. The two lenses performed fairly similarly except in the IR, where IgorOriginal 2 was significantly better – so there must be some differences in the glass or coatings.

 

3. The Focotar-2 was a real star, performing exceptionally well in the longer wavelength UV area and in IR.

 

4. The El Nikkors also performed creditably across the range.

 

5. The Lithagon generally did poorly – as I expected it to – but it did amazingly well through the Baader U when stopped down.

 

6. The Canon STM 50mm was a bit of a surprise. It was excellent in the visible range as I expected, but deteriorated in UV and IR.

 

The Numbers

 

[dabateman]

Bernard this is very interesting.

Also I think you will have fun playing here with the IR diffraction problem:

https://robertreiser.photography/diffraction-in-infrared-photography/

[ulf]

This is excellent and I really like what I have seen when looking at this very briefly.

Great job Bernhard

 

I think it is a good idea to have two different test distances and targets as some lenses, typically enlarger and dedicated macro-lenses, are designed for closeup work only.

It is still interesting to test those lenses at longer distances to find out if they still perform well there.

The problem is to create a suitable test target in a good size for that.

I would love to have something the size of the sensor. for 1:1

 

A big sized target can easily be printed with a quality printer.

 

This is an interesting site for test targets in general:

http://www.northligh...er-test-images/

I think this is a good target:

http://www.graphics..../res-chart.html

 

Norman Koren's work is also interesting:

http://www.normankor...rials/MTF5.html

 

If a visual assessment is the goal siemens-stars are very good as they combine several aspects.

[dabateman]

Ulf,

I have thought this way to test a lens is interesting. But possibly limited to infinity only and visible light only.

 

https://blog.kasson.com/the-last-word/testing-lenses-off-axis-with-small-field-targets/

 

[bvf]

Bernard this is very interesting.

Also I think you will have fun playing here with the IR diffraction problem:

https://robertreiser...ed-photography/

 

Thanks, David - that's a really interesting site. I ran the calculator for an A6500, which looks to be the same as the A6000 in the context of the calculator. This is what it gave me:

 

 

My rule of thumb of using f/8 in visible falls in between the two approaches used in the calculator. What the calculator says is to use about a 1.25 stops larger aperture in IR, which is consistent with what I saw. Of course, this is assuming an optically perfect lens, which the El Nikkors and Focotar-2 approximate to. If the lens is over-endowed with aberrations (no names!) then opening the aperture will swamp the improvement in diffraction effect with increased aberration effects.

[Andy Perrin]

Very nice -- as the maker of scurrilous comments about Steinheil creations, I wish to admit that I painted them with too broad a brush. The Cassar S seems to be quite a lovely lens, but don't bother with an Edixa Auto-Cassaron!

[bvf]

I think it is a good idea to have two different test distances and targets as some lenses, typically enlarger and dedicated macro-lenses, are designed for closeup work only.

It is still interesting to test those lenses at longer distances to find out if they still perform well there.

The problem is to create a suitable test target in a good size for that.

I would love to have something the size of the sensor. for 1:1

 

A big sized target can easily be printed with a quality printer.

 

 

Most of my tests were made at around 50x focal length - so 2.5 metres for a 50mm lens, 4 metres for 80mm, and 5.25 metres for 105mm. So in fact not ideal for any of the lenses (except the Focotar-2) - too close for the 35mm & 50mm camera lernses, and too far for the El Nikkor enlarger lenses.

 

My test target is 72cm across. I have an A3+ printer, but this won't print a bigger test target. Although I guess I could print it in sections and stitch the parts together to make a bigger chart. But in UV most of my work is close-up or macro, so that's what most interests me.

 

I may have a solution for a test target for 1:1 magnification (at least for full-frame) ... The test chart I used was a type I had about 40 years ago, but threw it out some time ago. For this exercise, I bought one on ebay, but that semed to get lost in the post. So I bought a second one - and then the first one arrived! But the second one was also shipped with something that might be of interest to you: some 35mm transparencies made by Leitz to test projectors. There are no measuring mechanisms, but they look good to assess image quality. Let me know if they would do what you want. Here's a photo so you can get an idea of what they are:

 

[dabateman]

 

 

Thanks, David - that's a really interesting site. I ran the calculator for an A6500, which looks to be the same as the A6000 in the context of the calculator. This is what it gave me:

 

 

My rule of thumb of using f/8 in visible falls in between the two approaches used in the calculator. What the calculator says is to use about a 1.25 stops larger aperture in IR, which is consistent with what I saw. Of course, this is assuming an optically perfect lens, which the El Nikkors and Focotar-2 approximate to. If the lens is over-endowed with aberrations (no names!) then opening the aperture will swamp the improvement in diffraction effect with increased aberration effects.

 

Yes you can manually type in the wavelength and watch it drop to F/3.2 for pixel peepers or f/6.3.

The manual range is only 590 to 990, so you can't get UV from this site.

 

[StephanN]

Wow, Bernard, great job.

 

So, the Cassar-S is quite acceptable in the centre, but crap in the corner, so it's useful for portraits

[Andrea B.]

Bernard:....absolutely dreadful image quality....the 2^nd element from the rear was reversed

 

I had to laugh because I've done that - reversed an element after cleaning up a lens. The funny thing is that the lens with reversed element produced such an interesting "abstract" effect that both Birna and I loved it. So I never have fixed that lens and enjoy its weird results. :lol:

---

 

Bernard, thank you for this excellent chart of lens resolution results. This must have been very time consuming, but we are very happy to have it. As mentioned in your other topic, I'm trying to decide how to collect all these lens results so that the references are not lost in the general forum topics.

[ulf]

I may have a solution for a test target for 1:1 magnification (at least for full-frame) ... The test chart I used was a type I had about 40 years ago, but threw it out some time ago. For this exercise, I bought one on ebay, but that semed to get lost in the post. So I bought a second one - and then the first one arrived! But the second one was also shipped with something that might be of interest to you: some 35mm transparencies made by Leitz to test projectors. There are no measuring mechanisms, but they look good to assess image quality. Let me know if they would do what you want. Here's a photo so you can get an idea of what they are:

 

 

They look very interesting.

Please tell me from where I can get that type of targets.

[ulf]

Bernard:....absolutely dreadful image quality....the 2^nd element from the rear was reversed

 

I had to laugh because I've done that - reversed an element after cleaning up a lens. The funny thing is that the lens with reversed element produced such an interesting "abstract" effect that both Birna and I loved it. So I never have fixed that lens and enjoy its weird results. :lol:

I remember seeing images of that here on the forum, but cannot find them now

[bvf]

They look very interesting.

Please tell me from where I can get that type of targets.

 

Ulf - I meant that you can have mine - it is very unlikely that you will find these anywhere else as they must be very old. If you want them, message me your adress - I can send them in a few days after I've used them for my own lenses.

[ulf]

Thank you Bernhard,

 

I have sent you my address privately in a message

[Andrea B.]

Gamma Terragon 35/3.5

https://www.ultravio...__fromsearch__1

 

I thought I was the person who reversed the element in the Gamma Terragon, but it seems I got the lens like that. So above I must have been thinking about some other lens I cleaned and reversed an element in.

[Andrea B.]

Bernard, would you please verify the list of lenses which I added to the first post?

Some of the newer lenses have various models so I am not sure I got the correct model.

We must be specific so that any reader wanting to try out a lens you tested can obtain the correct version.

Thanks!

[bvf]

 

Bernard, would you please verify the list of lenses which I added to the first post?

 

 

I've made a few edits, Andrea.

[Andrea B.]

Thanks!! I will copy those elsewhere as needed.

[Andrea B.]

Question: Shouldn't the "München" remain in the Steinheil Cassar S name?

[bvf]

Question: Shouldn't the "München" remain in the Steinheil Cassar S name?

 

I don't think so, nor for Enna. It doesn't appear to be part of the company name, and is not required for disambiguation. (As opposed to Carl Zeiss Jena, where the Jena is part of the company name and was required for disambiguation from the West German CZ.)

[Andrea B.]

I usually try to put almost everything as seen on the lens label because if you don't, then you get questions from people who don't speak German and don't recognize that München = Munich, Germany who are wondering if the Steinheil München Cassar S they found on Ebay is the same as the Steinheil Cassar S mentioned in our lens review.

 

It's difficult listing lenses properly!! Best to err on the side of including as much as possible.

 

BTW, that "PRINZGALAXY" gets translated as Prinzgalaxy or Prinz Galaxy equally often. Also sometimes as Prinz-galaxy. Either of the first two is OK by me for this one. I suppose we should list it using all capitals, but then it appears as though the lens list is screaming. :lol:

 

Added: For our Technical Lens Data "Label" or "Manufacturer's Designation", I want to see the capitalization as written on the lens. Elsewhere, in topic titles or in lists such as those given for Bernard's work, we can subdue the capitalization.

[ulf]

I usually try to put almost everything as seen on the lens label because if you don't, then you get questions from people who don't speak German and don't recognize that München = Munich, Germany who are wondering if the Steinheil München Cassar S they found on Ebay is the same as the Steinheil Cassar S mentioned in our lens review.

 

It's difficult listing lenses properly!! Best to err on the side of including as much as possible.

 

BTW, that "PRINZGALAXY" gets translated as Prinzgalaxy or Prinz Galaxy equally often. Also sometimes as Prinz-galaxy. Either of the first two is OK by me for this one. I suppose we should list it using all capitals, but then it appears as though the lens list is screaming. :lol:

The best way to identify a lens is by pictures of it, from different angles.

 

I have seen different lenses with their front ring labels spelled both PRINZGALAXY and PRINZ GALAXY with capitals.

Lately in my lens topics I have tried to write the Lens label as it looks on the lens, but maybe the topic name should be less screaming. I am OK with that.

[Andrea B.]

Ulf, when writing the "Designation" or "Label", I would definitely use capitals as shown on the lens. We want to be as accurate as possible. I even added some red letters for those lenses which have them.

 

Elsewhere, I suggest not using all capitals -- in topic titles or lists such as those above.

I will make my preceding comment clearer.