Shootout at 2.4x

[ray_parkhurst]

I asked in another thread here

http://www.photomacrography.net/forum/v ... 274#127274

what the best options are for 2x-3x magnification. I was only considering a few lenses, but the responses made me realize that I needed to open up the field to see what lenses I own worked best in that range. I decided to add several lenses suggested by folks (thank you!) that I already owned yet had not been considering.

Overall, the lenses are from 4 broad categories:
- Enlarger Lenses intended for wide mag range and tested in reverse
- Duplication Lenses optimized for narrow mag range
- Macro Lenses
- Microscope Objectives

In the last category, I was unable to test the Mitutoyo MPlan Apo 2x since I don't own or have access to one. What I did test is as follows:

Nikon 63mm EL-Nikkor in reverse orientation at f/4 (verified better than f/2.8 )
Rodenstock 67mm f/4.9 Scitex S3 in forward orientation (did not check reversed)
Nikon 50mm EL-Nikkor in reverse orientation at f/4 (verified better than f/2.8 )
Zeiss 74mm S-Planar at f/4
Rodenstock 75mm Apo-Rodagon D M=2 at f/4.5 in reverse orientation (verified better than forward)
Olympus MPlan 2.5x 0.05
Nikon 65mm Macro-Nikkor at f/4.5
Nikon 95mm Printing-Nikkor at f/3.3
Nikon 3x 0.09 MM / Measuring Microscope
Canon 35mm Macrophoto at f/4 (verified better than f/2.8 )

I used my Canon "HRT2i" camera (T2i with removed AA filter) for the tests. All lenses were adjusted for a ~10mm x-FOV for 2.4x magnification on Canon APS-C sensor. As usual, my subject is a Lincoln Cent, this time dated 1956-D. I centered the shots on the Mintmark.

I show one representative overall view below, followed by center and lower left corner crops from each of the lenses above. Here is the Overall shot:




Here are the Center Crops

Nikon 63EL


Nikon 50EL


Rodenstock 67S3


Rodenstock 75ARD2


Zeiss 74SP


Olympus MP2.5


Nikon 65MN


Nikon 95PN


Canon MP35


Nikon 3xMM



And here are the Corner Crops

Rodenstock 67S3


Nikon 63EL


Zeiss 74SP


Nikon 50EL


Rodenstock 75ARD2


Olympus MP2.5


Nikon 3xMM


Nikon 95PN


Nikon 65MN


Canon MP35



I put the lenses above in my opinion of worst to best for Center and Corner. Center and Corner were not consistent on any of the lenses. Here is my ranking by Center, Corner, and Overall Score:



Nikon 63mm EL-Nikkor................................ 10 / 9 / 19
Rodenstock 67mm Scitex S3 ...................... 8 / 10 / 18
Nikon 50mm EL-Nikkor................................ 9 / 7 / 16
Zeiss 74mm S-Planar.................................. 6 / 8 / 14
Rodenstock 75mm Apo-Rodagon D M=2..... 7 / 6 / 13
Olympus MPlan 2.5x 0.05........................... 5 / 5 / 10
Nikon 65mm Macro-Nikkor.......................... 4 / 2 / 6
Nikon 95mm Printing-Nikkor....................... 3 / 3 / 6
Nikon 3x 0.09 MM....................................... 1 / 4 / 5
Canon 35mm Macrophoto.......................... 2 / 1 / 3

The 95PN and 65MN both scored "6" overall but I judged the 95PN as slightly better since it was a bit more balanced across the field.

A couple of notes:

The 3xMM shows some extra illumination causing a "star" near the corners. I plan to re-do it to make sure the "star" goes away without the extra light. If not, it will drop to a lower ranking in the corners.

The 67S3 may need to be reversed. I will test it and see. The result is very disappointing and I am hoping it's just not optimized in forward orientation for mags above 1:1.

[rjlittlefield]

The 3xMM shows some extra illumination causing a "star" near the corners. I plan to re-do it to make sure the "star" goes away without the extra light. If not, it will drop to a lower ranking in the corners.

A star pattern in the corners, oriented with radial & tangential arms, is often due to astigmatism combined with stacking. Check the original frames at the "star" positions and I'll bet you see a radial streak shifting to a tangential streak as focus changes.

One question: since your goal is stitching for ultra-high resolution, did you choose the aperture to give best corner performance or perhaps most uniform performance, or did "best" mean center only?

--Rik

[ray_parkhurst]

A star pattern in the corners, oriented with radial & tangential arms, is often due to astigmatism combined with stacking. Check the original frames at the "star" positions and I'll bet you see a radial streak shifting to a tangential streak as focus changes.

One question: since your goal is stitching for ultra-high resolution, did you choose the aperture to give best corner performance or perhaps most uniform performance, or did "best" mean center only?

--Rik

I'll check the 3xMM to see if this is happening.

For the lenses where I had a choice of aperture, I selected for best corner performance. This meant stopping down the MP35, but the 65MN and 75ARD2 were best wide open. The 95PN was the same in the corners at f/2.8 and f/3.3, but a bit better in the center at f/3.3. The 63EL and 50EL were best in both center and corners stopped down to f/4.

In reality there is very little aperture flexibility available to keep a sharp image at this magnification due to effective aperture reduction. If the lens has a lot of aberrations, you are forced stop down a bit, but having to go beyond f/4 would have made an obviously less sharp image and eliminates a lens requiring smaller aperture from consideration. So the real question ends up being "which lenses are sharpest at X magnification wide open?" The fact that the MP35 wins overall at f/4 shows there is still some room for improvement with a lens that is sharper at f/2.8 or f/3.3.

That said, why the 95PN did not do better in this test is unclear to me. It was flat across the field at f/2.8, and stopping down to f/3.3 only gave a slight improvement in the center. It "should" have shown the best performance but didn't due to other issues that apparently don't go away with smaller aperture. It did well, and I expect it to be the best for distortion, but its sharpness and contrast did not blow the others away.

[typestar]

The 95PN and 65MN both scored "6" overall but I judged the 95PN as slightly better since it was a bit more balanced across the field... The 67S3 may need to be reversed. I will test it and see.

Ray, can you show, how you will mount the 67S3 reversed, as it has not filterthread... I am suspicious about the further results -- as you appreciated the S3-series in lower magnifications...

In addition, that the Canon 35 mm macro performs that well is astonishing, according to the results i read on e. savazzis site:
http://savazzi.freehostia.com/photograp ... enses5.htm

where the canon had been described with very low contrast...

regards,
christian

[ray_parkhurst]

Ray, can you show, how you will mount the 67S3 reversed, as it has not filterthread... I am suspicious about the further results -- as you appreciated the S3-series in lower magnifications...

In addition, that the Canon 35 mm macro performs that well is astonishing, according to the results i read on e. savazzis site:
http://savazzi.freehostia.com/photograp ... enses5.htm

where the canon had been described with very low contrast...

regards,
christian

Christian...I'll probably do a "tape job" to install the 67S3 on the bellows. It has an almost snug fit inside a Canon FD-M52 reverse adapter I have, so a bit of tape will make an interference fit that is square to the camera. Should work fine for a temporary setup. Regarding the MP35, either I or someone asked Enrico about the contrast problem and he said that he intentionally set up the test to emphasize any contrast problems a lens may exhibit (my paraphrasing). I think as long as you are careful to minimize the light hitting the lens directly that flare and contrast reduction won't be a problem. Indeed, the MP35 was one of the highest contrast lenses in my 2.4x tests...Ray

[ray_parkhurst]

OK, I re-checked the 67S3 and the 3xMM. I found that the 67S3 was small enough to fit inside the Canon bellows, and would thread onto the backside of the FD-M42 adapter (with M39-M42 helicoil), so the mounting was quite easy. Here are the results:

67S3 reversed center


67S3 reversed corner


These results are much improved over previous ones. In my opinion, the center is now about equal to the Olympus MPlan objective, just slightly worse than the 65mm Macro Nikkor, while the corner improved even more and is just slightly better than the 65mm Macro Nikkor. What a big improvement! This lens gives a sharp result at M=0.8, so is probably optimized for M=0.5x forward and M=2x reverse (I am only supposing this based on my limited data).

I also re-checked the 3xMM and saw the same star pattern in the stack on point highlights. I looked at the source files and found that when the lens is OOF high, points spread tangentially. They do this only in the extreme corners. This lens is pretty much telecentric and that might exacerbate the problem as the OOF tangential aberrations are not spread out radially as they are with a lens that changes magnification with focus. You can see this characteristic in the corner photos by looking at the locus of the single red pixel in lower right of the corner images. The 3xMM locus is almost a single point, indicating no change in mag across the stack. This "focuses" the aberrations...Anyway, I took a new stack at the rated mag of 3x (the earlier stack was at 2.4x) and the corner aberrations are much improved. The result now looks as sharp as the MP35 at the corner, and sharper at the center, though the comparison is not fair since the mags are not the same. Here are the resulting images:

3xMM at 3x mag, center


3xMM at 3x mag, corner

[CaptainFwiffo]

Hmm, the 3xMM seems to be a pretty clear winner at the center, but the astigmatism artifacts in the corners are pretty bad. I wonder if it would be possible to get a good result by combining the output of two different lenses into a single stack (i.e. take all the frames with the lens that is sharpest in the center, and one that is sharpest in the corners, and then interleave them in the stacking software)... OK, now that I think about, it sounds dubious...

Interesting that you've got a red hot-spot on your sensor in the corner that lets you see the movement of your setup during the stack.

[ray_parkhurst]

...Interesting that you've got a red hot-spot on your sensor in the corner that lets you see the movement of your setup during the stack.

The locus of the hot-spot may indicate setup movement, but I've repeated the stack 3 or 4 times with the 3xMM at 2.4x and each time all the red pixels from every frame line up such that they look almost like a single dot in the final image. The other lenses and objectives all show a locus to the points, not always straight, but repeatable. The MP35 shows the straightest locus, such that it is just changing in magnification during focusing without any other distortions showing up. Interestingly, the 3xMM at 3x showed a slight movement to the hot-spot, and this may be due to setup movement, but the large shifts showing up in the other images are likely mostly due to mag changes.

[ray_parkhurst]

Continuing with the testing...

Peter De Smidt graciously loaned me his Mitutoyo M Plan Apo 2 0.055 infinite objective to test for this shootout. I only have a 180mm tube lens, so the test was done at 1.8x versus the objective's optimum 2x and the rest of the shootout's 2.4x (except 3xMM). Here are the resulting images:





Judging is more difficult due to the magnifications not being matched, but in my opinion the center performance is very close to that of the 74SP, while corner performance is very close to the OM2.5.

To add a bit more to this, I think it's important to see where each of the lenses/objectives are in terms of effective aperture for this test. Here is the list again with f/eff noted, and a visual representation to help sort them:

Nikon 63mm EL-Nikkor................................ f/13.6
Rodenstock 67mm Scitex S3 ................................. f/16.6
Nikon 50mm EL-Nikkor................................ f/13.6
Zeiss 74mm S-Planar.................................. f/13.6
Rodenstock 75mm Apo-Rodagon D M=2..... f/13.6
Olympus MPlan 2.5x 0.05............................................. f/24
Nikon 65mm Macro-Nikkor................................ f/15.3
Nikon 95mm Printing-Nikkor............... f/11.2
Nikon 3x 0.09 MM........ .......................... f/13.3
Canon 35mm Macrophoto........................... f/13.6

[ray_parkhurst]

I've been commenting on another thread, using the 2x Mitty to show how my achromat is mounted in the Canon bellows, and some interesting results regarding extension of the achromat vs sensor. It appears the achromat I am using, a 180mm Meopta that has 52mm diameter which fits inside the Canon bellows nicely, was reversed. I'm seeing slightly better performance at the corners in the other orientation. I'm also seeing slightly better performance with the achromat focused BEYOND infinity, with about 160mm extension instead of 180mm. For completeness, I wanted to reference the other thread here:

http://www.photomacrography.net/forum/v ... hp?t=20206

I also wanted to publish in this thread the latest 2x Mitty results showing performance vs achromat extension. These are small crops from the lower left corner, with 140, 160, 180, and 200mm extensions.

.........140.....................160....................180...................200..........

[ray_parkhurst]

In another thread, Chris R suggested using a 2x teleconverter to improve the quality of crops that have "run out of pixels". This gave me the idea to try using a 2x teleconverter on the 105PN to see how it would hold up in the 2.4x shootout. The 105PN won't do well at the corners at 2.4x on its own since it's highly optimized around 1:1, but at 1.2:1 followed by a teleconverter it should give a good result. I tried it at f/2.8 and f/3.3, and the f/3.3 was slightly superior. At this aperture, effective aperture is about f/15, so it should be strongly diffraction-limited. Here are the images:

Center


Corner


The result does not beat the MP35, though it is similar in the center, and is about the same as the 65MN in the corner.

Edited to add:

I think my effective aperture calculation is correct...

I used the magnification of the lens w/o teleconverter, then multiplied the result by teleconverter factor.