4 lenses slightly above 10x

morfa · Post by **morfa** » Fri Nov 20, 2009 10:06 am

Downsized DX-sensor images:

This is of course NOT a scientific test by any means. It's merely a way for me to get some sense of these lenses. I've tried to be reasonably careful but I've come to the conclusion that it's better to do several slightly sloppy tests rather than one super precise test – at least for me this seems to be more effective.

I won't go into details about the imperfections of the testing procedure – just know that they're there and keep it in mind before making any conclusions.

All images stacked in Zerene Stacker using the PMAX method. 26 frames/stack (but Nikon Plan more like 40 frames and slightly more DOF)
After stacking, I've aligned the resulting images in Photoshop using auto-align layers to counter the slight alignment and magnification differencies between the stacks. Obviously this is not the "clean" way of doing it since you are pushing pixels a little differently in different images but since I failed to see any quality loss from this procedure (I inspected it closely) I chose to do it this way.

6MP files
Nikon: http://farm3.static.flickr.com/2571/411 ... ba3e_o.jpg
Tominon: http://farm3.static.flickr.com/2763/412 ... 103b_o.jpg
Componon: http://farm3.static.flickr.com/2660/411 ... 3848_o.jpg
Summar: http://farm3.static.flickr.com/2544/412 ... 1510_o.jpg

I have very little experience with magnifications above 10x so I'm not sure if these results are in line with what is to be expected.

/John

rjlittlefield · Post by **rjlittlefield** » Fri Nov 20, 2009 1:06 pm

Your procedures look fine and your results are definitely in line with what I would expect.

I have spent some time with those 3000x2007 pixel images that you linked to.

Even the sharpest of them seems to be adequately captured at this resolution. One of my standard tests for this is to crop an image to even dimensions (3000x2006 in this case), copy it, downsize the copy to 50%, then upsize the result to 200% so that it goes back to the original size, layer with the original, and flash back and forth so that I can see how much the image was softened by being crammed into 1/2 as many pixels on each axis. In this case, the answer is "hardly visible". What this means is that the images are oversampled enough at 3Kx2K that nothing significant would be lost in resampling them for alignment. So this is further confirmation or explanation of why you didn't see quality loss from your procedure.

Pixel-peeping at the large images, what I see is that the Tominon and the Nikon objective are about equal for resolution, with the Nikon showing some chromatic aberration that the Tominon does not. (Is the Nikon supposed to be a CF type?) The Componon is a little softer, particularly in the upper right corner where there seems to be a bit of smearing on the 10-o'clock/4-o'clock axis. The Summar is significantly softer, especially in the corners.

When I do layer-and-flash on all four images, it's apparent that there are significant differences in perspective and viewing angle between the four images, particularly between the Nikon image and the other three.

I suspect a combination of optical and digital effects here. The optical effects would be due to differences in the entrance pupil location and the appearance of OOF regions (bokeh). These optical effects feed into the camera's nonlinear response and the software's alignment algorithms to produce different reconstructions of the geometry. Essentially, one stack can end up getting skewed with respect to another, even though the physical geometries were identical.

If you did need to perform a really careful lens test with stacking, then it would best to be sure that the stacking setup is mechanically precise and turn off every alignment option that you can while still getting a good result. With a setup that is mechanically precise, you should be able to turn off rotation and both of the shifts (X and Y); at high magnifications, you can also turn off scale.

For most tests, however, it's perfectly adequate to leave all the alignment options turned on. The sharpnesses will rarely be affected, and if you know that geometry differences are normal you won't waste time agonizing over them.

I hope this helps. Thanks for the tests!

--Rik

morfa · Post by **morfa** » Fri Nov 20, 2009 2:07 pm

Many thanks for your informative feedback Rik!

That upsize-downsize routine is clever – I can see where that can come in handy. Thanks for the tip!

The Nikon objective is not a CF type – just a plain Plan. Still I expected it to outperform the Tominon resolution wise (after all it's at least a full stop brighter) but that wasn't the case.

I actually threw the Nikon in at the last minute and had already begun taking down the setup – that's why the geometry differs more in that shot than in the rest. This horizontal setup of mine is still very crude but at least now I've got a linear stage with a micrometer positioner to help with the stacking motion. The rest is hit and miss though and changing lenses always shifts the alignment of the optical axis in some way or another so that I have to nudge the subject a lot to get the framing right.

I'm trying to decide if I should get a more solid and precise setup for tasks like this one. It's not apparent since at this point I don't aspire to do this kind of photography for any other reason than to have something to go on when the bug season starts again and the inevitable "which lenses to bring" question starts popping up. On the other hand, even if I don't need more precision, a better setup might be simpler and less time consuming to work with and in that case it would be tempting.

BTW, I did expect more from the Summar but I suspect on a DX sensor it does better at 8x or so. Also, I have a feeling it would benefit from stopping down half a stop but in this case I felt reluctant to do so since (with the diffusor I was using) it would force me to fire the flash at full power which in my experience gives less accuracy in the output.

/John

Planapo · Post by **Planapo** » Fri Nov 20, 2009 3:27 pm

John, Always interesting to see such comparisons. Thanks for sharing!

One question, though I think you did it, but as you didn't mention it, and one never knows, does one?! So, what was my question again ... ah, yes...

You did use the Componon reversed, didn't you?! (Provided that it was the enlarging lens version, of course.)

And which method have you chosen to mount the Tominon with its odd M40x0.75 thread?

--Betty

rjlittlefield · Post by **rjlittlefield** » Fri Nov 20, 2009 3:36 pm

Thanks for the additional information, John.

I confess, I was surprised by how much different the geometry of the Nikon shot was, so I'm glad to hear that it was physically repositioned also.

The alignment issues that I mentioned are real, though. Several years ago, when I first made Photoshop layers of output from Helicon Focus, CombineZ, and Panorama Tools, it took me completely by surprise to discover that the output images could not be perfectly registered against each other. Sometimes even the apparent perspective would change, with for example fly eyes bulging more in one output than another, or an antenna appearing to shift across the head behind it. It took me a long time to figure out what was going on, and even longer to realize that there was an opportunity (synthetic stereo/rocking) lurking behind the problem.

--Rik

morfa · Post by **morfa** » Fri Nov 20, 2009 3:50 pm

Yes, the Componon was reversed and I should've mentioned that. Especially since I tried it non-reversed as well and the results were dreadful (worse than the Summar).

I've temporarily attached the Tominon to a M42->F adapter by adding a couple of layers of tape onto the filter threads of the Tominon and then screwing the adapter onto that – I've done this before with a Tominon 75mm and works surprisingly well.

/John

g4lab · Post by **g4lab** » Fri Nov 20, 2009 4:04 pm

Thank you for this nice test and also thank you to Rik as usual.

I have seen a lot of old Summars around. Was this an elderly one? Is its interior still spotless and water clear? I usually never got too excited over the ones I actually held in my hands because they were too old. Some uncoated and in serious need of cleaning. Could this have contributed to the softness as well as the mag. range??

morfa · Post by **morfa** » Fri Nov 20, 2009 4:31 pm

g4lab> My 24/4.5 Summar looks old judging from the barrel but it's coated and appears optically flawless. I haven't had time to use it much but my limited experience with it up until this test had led me to expect it to fare better here.

I'll definitely have to play with it some more before passing judgement on it!

ChrisR · Post by **ChrisR** » Fri Nov 20, 2009 6:46 pm

I believe the Summars were the middle quality Leitz macro lenses, with Milar below and Photar above?

g4lab · Post by **g4lab** » Fri Nov 20, 2009 9:44 pm

I never really had that straight. I would have thought it more a chronological thing Milars, Summars then Photars but that is just an impression based on my own chronology which started paying attention to such things approximately 1970

I think the summar did overlap both.
Paer?? Tardigrade??

morfa · Post by **morfa** » Sat Nov 21, 2009 12:44 am

Klaus says here that Summars are followers to the cheaper Milars.

BTW, I got mine from someone who's family had been in business with the Leica company a long time ago. According to him, this lens as well as some other Leitz-lenses I got from him, came into his possession directly from Oskar Barnacks office where they had been used as test lenses.

From top left: Leitz Summar 42mm f4.5,Leitz Milar 65mm f4.5, Zeiss Luminar 63mm f4.5,
Leitz Milar 40mm f4.5, Leitz Milar 30mm f4.5, Leitz Summar 24mm f4.5