Comparing macro lenses using MTF - part III - 0.66x
Moderators: rjlittlefield, ChrisR, Chris S., Pau
Comparing macro lenses using MTF - part III - 0.66x
Some of you will comment right away „Why testing such small magnification?“. And yes, this kind of magnification can easily be reached by ordinary camera macro lenses which are frequently equipped with autofocus, stabilization and other bangs and whistles. However, those ordinary macro lenses frequently have problems with focusing plane flatness, they are heavy, clumsy and sometimes very expensive.
This magnification is interesting mostly for two purposes: photographing coins and digitizing FF negatives/slides. Although such modest magnification might look easy to achieve, it turned out that not many lenses are capable of doing it right. The only lens that provided very high resolution across the frame (including corners) and has very flat focusing plane is Rodenstock Apo Rodagon D 75mm f4.5.
At first, this lens was a huge dissapointment: I expected stellar performance at 2x but it delivered very average performance as you will see later. But for 0.66x this lens is second to none. It delivers 3433/3334/0.36, a fantastic edge to edge resolution superceeded only by incredible Fujinon 80mm f2.8 Macro. Some enlargement lenses did score higher central resolution but the edge resolution was much lower so those can't be called serious contenders for those two applications. I didn't test many lenses at 0.66x since Apo Rodagon D 2x was so good.
LAST MINUTE ADDITIONS: I was wondering about some kind of lens rating based on resolution. The good news is that MTF based testing is all about numbers. And Imatest has automatic weighted average resolution calculation where central resolution participates with 75% and the edge resolution with 25%. My addition is resolution rating where the average LW/PH value is divided by 4000 (sensor vertical resolution). It makes sense since the lens with the highest measured resolution (Fujinon XF 80mm f2.8 Macro) scored exactly 10 points at 1:1.
As always, there is one positive surprise; an affordable Rodenstock Rodagon 80mm f4 that scored highest central resolution at f4.8 and very high overall resolution at f5.6. Due to Imatest' weighting towards center, average resolution came out as the highest at this magnification. However, when digitizing slides or shooting coins, focusing plane flatness and equal edge to edge resolution is more important than extreme center resolution so my pick would be Apo Rodagon D 75mm f4.5 2x.
I didn't expect much better performance from 1x version of Apo Rodagon D since it's designed for 0.8x to 1.2x magnification. It's resolution falls considerably towards the edges at 0.66x. However, due to it's high central resolution it scored higher overall than it's 2x sibbling.
Even worse falloff happens with Schneider Componon-S 50mm f2.8 so this is definitely not a lens for this magnification.
If looking at score points only, Apo Rodagon N 50mm f2.8 seems very attractive solution but that lens has one serious drawback: serious focusing plane curvature. I have measured no less than 0.8mm edge bending 18mm from the object center.
If you consider Rodenstock Apo Rodagon D 75mm f4.5 keep in mind it must be extended from the APS-C senzor (Fuji/Nikon/Pentax/Sony size) 122 mm and the working distance is 160 mm for this magnification.
Hoping this first set of tests gives you some useful data. Next sequel will present even more lenses, this time at 1:1 magnification.
Miljenko
This magnification is interesting mostly for two purposes: photographing coins and digitizing FF negatives/slides. Although such modest magnification might look easy to achieve, it turned out that not many lenses are capable of doing it right. The only lens that provided very high resolution across the frame (including corners) and has very flat focusing plane is Rodenstock Apo Rodagon D 75mm f4.5.
At first, this lens was a huge dissapointment: I expected stellar performance at 2x but it delivered very average performance as you will see later. But for 0.66x this lens is second to none. It delivers 3433/3334/0.36, a fantastic edge to edge resolution superceeded only by incredible Fujinon 80mm f2.8 Macro. Some enlargement lenses did score higher central resolution but the edge resolution was much lower so those can't be called serious contenders for those two applications. I didn't test many lenses at 0.66x since Apo Rodagon D 2x was so good.
LAST MINUTE ADDITIONS: I was wondering about some kind of lens rating based on resolution. The good news is that MTF based testing is all about numbers. And Imatest has automatic weighted average resolution calculation where central resolution participates with 75% and the edge resolution with 25%. My addition is resolution rating where the average LW/PH value is divided by 4000 (sensor vertical resolution). It makes sense since the lens with the highest measured resolution (Fujinon XF 80mm f2.8 Macro) scored exactly 10 points at 1:1.
As always, there is one positive surprise; an affordable Rodenstock Rodagon 80mm f4 that scored highest central resolution at f4.8 and very high overall resolution at f5.6. Due to Imatest' weighting towards center, average resolution came out as the highest at this magnification. However, when digitizing slides or shooting coins, focusing plane flatness and equal edge to edge resolution is more important than extreme center resolution so my pick would be Apo Rodagon D 75mm f4.5 2x.
I didn't expect much better performance from 1x version of Apo Rodagon D since it's designed for 0.8x to 1.2x magnification. It's resolution falls considerably towards the edges at 0.66x. However, due to it's high central resolution it scored higher overall than it's 2x sibbling.
Even worse falloff happens with Schneider Componon-S 50mm f2.8 so this is definitely not a lens for this magnification.
If looking at score points only, Apo Rodagon N 50mm f2.8 seems very attractive solution but that lens has one serious drawback: serious focusing plane curvature. I have measured no less than 0.8mm edge bending 18mm from the object center.
If you consider Rodenstock Apo Rodagon D 75mm f4.5 keep in mind it must be extended from the APS-C senzor (Fuji/Nikon/Pentax/Sony size) 122 mm and the working distance is 160 mm for this magnification.
Hoping this first set of tests gives you some useful data. Next sequel will present even more lenses, this time at 1:1 magnification.
Miljenko
Last edited by Miljenko on Wed Sep 12, 2018 1:43 pm, edited 1 time in total.
All things are number - Pythagoras
-
- Posts: 3439
- Joined: Sat Nov 20, 2010 10:40 am
- Location: Santa Clara, CA, USA
- Contact:
Nice to see results at 0.66:1. I do most of my shooting at around this magnification, typically 0.7:1, since that mag fills 90% of the vertical space of Canon APS-C sensor with a 19mm Lincoln Cent.
Before I changed over to the 85MV, my go-to lens for Cents was the 75ARD1. It's sharper at the center than the 75ARD2, and while it falls off on the edges and corners more, since coins are round and don't get anywhere near the short edges nor corners, the 75ARD1 is the better choice. But if shooting coin details, stamps, bugs, etc which go across the whole frame, the 75ARD2 gives a more uniform sharpness.
There are 2 pieces of test data that don't agree with my own tests. First, the Agfa Repromaster was the standout winner in my tests for center performance. It was not so great at the edges (I tested at top center edge), so did not score well overall. The difference may be copy-copy variation.
Although I didn't publish any CA results, I've always found the 80mm Rodagon to have fairly high CAs, unlike your results, which are very low. In this case I don't think copy-copy variation is the explanation since every one I've owned has fairly high CAs. Are you testing LoCA or LaCA or a combo?
Edited to add: my tests are empirical/subjective/non-quantitative but IMO still give a good compo between the lenses. Not sure if you have seen my 80mm shootout, see it here:
http://www.macrocoins.com/80mm-lens-shootout.html
Before I changed over to the 85MV, my go-to lens for Cents was the 75ARD1. It's sharper at the center than the 75ARD2, and while it falls off on the edges and corners more, since coins are round and don't get anywhere near the short edges nor corners, the 75ARD1 is the better choice. But if shooting coin details, stamps, bugs, etc which go across the whole frame, the 75ARD2 gives a more uniform sharpness.
There are 2 pieces of test data that don't agree with my own tests. First, the Agfa Repromaster was the standout winner in my tests for center performance. It was not so great at the edges (I tested at top center edge), so did not score well overall. The difference may be copy-copy variation.
Although I didn't publish any CA results, I've always found the 80mm Rodagon to have fairly high CAs, unlike your results, which are very low. In this case I don't think copy-copy variation is the explanation since every one I've owned has fairly high CAs. Are you testing LoCA or LaCA or a combo?
Edited to add: my tests are empirical/subjective/non-quantitative but IMO still give a good compo between the lenses. Not sure if you have seen my 80mm shootout, see it here:
http://www.macrocoins.com/80mm-lens-shootout.html
Ray, Imatest can only measure lateral CA. It may be that 80mm Rodagon have above average longitudinal CA. Or your copy might be badly adjusted. Rodenstock very often adjusts distance between lens groups by inserting shims of different but exact thickness. Loosing those shims afects LoCA the most.ray_parkhurst wrote: In this case I don't think copy-copy variation is the explanation since every one I've owned has fairly high CAs. Are you testing LoCA or LaCA or a combo?
Yes I've seen your test long time ago and I'm very thankful for it. It made me quit doing visual tests for good. I was never succesful at staring at wing scales, processor waffers and coins. I started to use dials and scales as a young kid and maybe believe them to much. But I believe every phisical characteristic can be quantified, measured and compared by numbers. Provided you use propper method and appropriate tools. After all, that's what manufacturers do internally.ray_parkhurst wrote: Not sure if you have seen my 80mm shootout
All things are number - Pythagoras
-
- Posts: 3439
- Joined: Sat Nov 20, 2010 10:40 am
- Location: Santa Clara, CA, USA
- Contact:
-
- Posts: 3439
- Joined: Sat Nov 20, 2010 10:40 am
- Location: Santa Clara, CA, USA
- Contact:
Yes, comparing graphs is preferred to pixel peeping. In some cases though the underlying reason for the reduced performance is important. Two lenses that measure about the same may look different in actual use on real subjects, and it takes an imaging test to differentiate their performance.Miljenko wrote:Not at all, Ray, your tests are very precisely prepared and executed. It's just me who hardly sees those fine differences. Comparing numbers is simply much easier!
-
- Posts: 3439
- Joined: Sat Nov 20, 2010 10:40 am
- Location: Santa Clara, CA, USA
- Contact:
I remember seeing the MTF, etc curves as published by Rodenstock but for some reason can't find them now. Thought I saved them but no record of it. The online catalog today only has the 75ARD1, not 75ARD2 nor 120ARD. Does anyone have a link to the published MTF curves, or perhaps a saved copy? That info would complement the test data nicely. I also seem to remember the 75ARD2 curves actually showed that it was apochromatic, where the 75ARD1 curves show it's a well-corrected achromat, not apochromat.
Pau, of course it was used the propper way for all the tests. Still, it is not that good at 2x as you will see next week. OTOH, it works great at 0.66x. At 0.5x resolution at center raises as expected (to 3578 LW/PH) but edge resolution drops to 2897 LW/PH. Obviously, 0.5x is on design edge for this lens.Pau wrote:The AR 75/4.5 2× specification is misleading: it must be reversed for 2X, at normal position its best is 1/2×
Not surprising that it works well at 0.66X
You can find it in Rodenstock literature IF you read it carefully because it is confusing expressed .
All things are number - Pythagoras
-
- Posts: 3439
- Joined: Sat Nov 20, 2010 10:40 am
- Location: Santa Clara, CA, USA
- Contact:
I checked the Qioptiq website and in their Rodagon brochure they show the MTF curves for the 75ARD2. See page 5 at this link:
http://www.qioptiq.com/download/Rodagon ... -08_-w.pdf
However, they don't show the LoCA curve.
The Rodenstock archive has this file, which shows the MTF curves and the LoCA curve for the 75ARD1 on page 19:
http://www.rodenstock-photo.com/Archiv/ ... __8230.pdf
I seem to remember reading somewhere that these curves were computed, not measured. Anyone know?
http://www.qioptiq.com/download/Rodagon ... -08_-w.pdf
However, they don't show the LoCA curve.
The Rodenstock archive has this file, which shows the MTF curves and the LoCA curve for the 75ARD1 on page 19:
http://www.rodenstock-photo.com/Archiv/ ... __8230.pdf
I seem to remember reading somewhere that these curves were computed, not measured. Anyone know?
Hi Ray!ray_parkhurst wrote:I remember seeing the MTF, etc curves as published by Rodenstock but for some reason can't find them now.... Does anyone have ... perhaps a saved copy?
I just have checked my saved copy from Rodenstock -- The German and English language catalog has first been published in May 2009 (as the original PDF / Indesign properties shows) -- I saved it in 2015 -- it seems not to be edited since - and my safed copy is identical to your shown URL -- and also only shows the MTF for the 75ARD1 - as you have found, too.
Interesting to see the new Apo-Rodagon-HR 5.6/75 0.5x on the Qioptiq page...
Best wishes,
typestar
-
- Posts: 2627
- Joined: Thu Jan 17, 2013 9:34 pm
- Location: United States
- Contact:
Hi Ray,ray_parkhurst wrote:I checked the Qioptiq website and in their Rodagon brochure they show the MTF curves for the 75ARD2. See page 5 at this link:
http://www.qioptiq.com/download/Rodagon ... -08_-w.pdf
However, they don't show the LoCA curve.
The Rodenstock archive has this file, which shows the MTF curves and the LoCA curve for the 75ARD1 on page 19:
http://www.rodenstock-photo.com/Archiv/ ... __8230.pdf
I seem to remember reading somewhere that these curves were computed, not measured. Anyone know?
Rodenstock MTF data is computer generated, similar to Nikon, Canon, etc. This is from Rodenstock USA importer years ago.
-
- Posts: 2627
- Joined: Thu Jan 17, 2013 9:34 pm
- Location: United States
- Contact:
Hi Ray, I think you switched the lenses by mistake, the 2x is not APO, at least my two samples were not even close, the 1:1 is APO at 1x from everything I have seen first hand.ray_parkhurst wrote:I remember seeing the MTF, etc curves as published by Rodenstock but for some reason can't find them now. Thought I saved them but no record of it. The online catalog today only has the 75ARD1, not 75ARD2 nor 120ARD. Does anyone have a link to the published MTF curves, or perhaps a saved copy? That info would complement the test data nicely. I also seem to remember the 75ARD2 curves actually showed that it was apochromatic, where the 75ARD1 curves show it's a well-corrected achromat, not apochromat.
I could have had two bad 75ARD2 units of course, that would explain the worst CAs of any lens in my 2x test ( and it was the only lens labeled APO).
Robert
-
- Posts: 3439
- Joined: Sat Nov 20, 2010 10:40 am
- Location: Santa Clara, CA, USA
- Contact:
I was not going by test results, but by the published curves. The 75ARD1 has two crossings in the focus vs wavelength curve, so by definition is not apochromatic. I can't actually remember if the apochromatic curve was for the 75ARD2, or the 120ARD. Wish I could find those curves!RobertOToole wrote: Hi Ray, I think you switched the lenses by mistake, the 2x is not APO, at least my two samples were not even close, the 1:1 is APO at 1x from everything I have seen first hand.
I could have had two bad 75ARD2 units of course, that would explain the worst CAs of any lens in my 2x test ( and it was the only lens labeled APO).
Robert