Fred Miranda
Offline
Admin
Upload & Sell: On
|
 p.77 #10 · Voigtlander 50mm f/2 APO-Lanthar Review | |
inksandpaper wrote:
In continuation from my CV 35/2 A-L depressive story from that respective thread, I would like to share a successful one:
Despite my disappointment at the store with the 35mms, my itchy fingers drew me to two new copies of Voigtlander 50mm F2 A-L on the shelf. I had tested four copies already last year and was pretty sure my copy was the best among them, despite its radially asymmetric astigmatism. The problem is subtle and far more difficult to decipher than I ever expected, causing a lot of problems with the initial shimming to correct it for swing. There are two distinct focal planes that are in focus at the right edge, which is flipped on the left, visible only depending on subject orientation. I could optimise for one of the two diverging planes, but it would make tangential detail worse. Or compromise, and live with an overall worse blur for all kinds of detail.
The third copy I tried was perfectly straight (first time I've ever seen one) but it was noticeably softer than mine, despite its issues. It's quite weird to see sharp-ish detail that is so clearly also blurred with astigmatism - makes me recall Fred's first samples of the CV 35/2 off-axis. I tried to convince myself that my copy was OK and I should stop looking, but I was never really satisfied. This lens is such an important and special one, I felt almost cheated to settle for not having a really great piece.
So the fifth copy was also perfectly straight but soft like the third. Copies like these are very hard to diagnose without a sharper copy on hand, since they look perfectly fine on their own, until you see side-by-side just how much sharper a sharp copy can be. Yes, they seem plenty sharp on their own even to fussy eyes-I'm human, not an MTF machine, my judgement is relative. To give an idea of what it's like, in comparison with a truly sharp copy, it's something like a subtle 0.3-0.5-px radius gaussian blur has been applied.
My experience now is the V50s vary in central sharpness by quite a lot, relatively speaking, since when it's good it's absolutely phenomenal and causes massive amounts of aliasing on 4-micron pixel sensors. But most are not this way. Cosina has modernised their assembly and are building these lenses on an MTF bench (I hear Leica is doing the same with the SL lenses, otherwise they will never live up to the published MTFs), which gives them on-the-fly readouts and thus allows them to optimise the optical alignment to a high degree. Roger Cicala revealed that the technique is not to achieve perfect centering of each element, but rather to put each one slightly off-center, so their errors are averaged/balanced for better radial symmetry. It could be that all these lesser V50s are already optimised and this is just the limit of Cosina's precision at this cost level. They make it known that they allow their final polished thickness to vary by 10 microns for each element, measuring them at 1-micron precision, sorting them and picking elements to ensure more consistent total thicknesses. This could be one of the reasons why the field curvature was seen to vary for the 35mm version.
Cosina's polishing is probably not as precise as Canon or Sony's, evidenced by how much more severe the onion ring effect is in the bokeh of their lenses compared to Sony GM. Sony has a published tolerance of 0.01 microns or 10nm. Canon is even better: "Canon has achieved an overwhelming level of machining precision, including the ability to grind a lens surface to a tolerance within 100 nm (nanometers), then polish it to an even higher degree of precision, reducing surface roughness reduced to within a 1 nm range." -https://global.canon/en/technology/support26.html
Such is the power of economies of scale allowing for greater precision at the same or lower cost. That's two(!) orders of magnitude better. I don't know how this bears on the precision of the spherical and aspherical surfaces which matter, but that could also explain some of this variation. Don't misinterpret me that this implies Canon lenses are 2 orders of magnitude better overall.
So I'm not surprised that Fred finds the CV 50/1.2 Nokton is sharper on-axis than the CV 50/2 A-L, and Lenstip's measurements suggest the same. Getting a really great V50 is just very rare. The design is optimised for off-axis resolution, which typically makes on-axis resolution drop. If the MTF-guided assembly also does off-axis readouts, then perhaps the optimisation routines do cause the on-axis resolution to vary further.
The sixth V50 copy I saw is modestly swung and mildly tilted and I nearly ignored it after noticing that. But by refocusing every time on each region of interest (all edges and corners) revealed that this is a wonderfully centered copy and only needs to be shimmed for its crookedness. I've never seen a V50 look as good as this one. It has even less axial CA than my own copy. PSF rendering off-axis is way superior to my copy owing to significantly less astigmatism. The resolution is far higher off-axis. Every comparison I've made so far on every corner or edge is a little to a lot better than my copy, even on-axis (but only just). What an exceptional unit! I bought it.
This is the first time I'm choosing a lens knowing it's crooked but perfectly centered, while rejecting another copy that is perfectly straight and perfectly centered but just has clearly lower resolution overall, both on and off-axis. I will be working on shimming it soon. This affair has caused me a little worry that I need to see a great number of CV 35/2 lenses before I can safely decide on purchasing a really great copy to end all 35mm lens purchases. Not so easy when you have to make a decision at the store rather than buying multiple copies, testing in the comfort of your home and then returning the reject copies. I should point out that I'm purchasing this lens specifically for making large, wide-field images of the starry night sky, hence the extreme lengths I've gone to ensure these tiny points of light remain both tiny and round for a highly realistic and pleasing result. If you're photographing portraits, don't bother with being this demanding....Show more →
My first CV 50/2 APO VM had some tilt + swing that quite noticeable when comparing it to the Leica 50/1.4 Lux.
My second copy was perfectly centered on my Leica M10-R. For landscapes and high magnification, one could detect the difference between the two 50/2 APO copies even at smal apertures. Here is the test with the latest copy:
https://www.fredmiranda.com/forum/topic/1707236/0#15636310
|
