Photomacrographic lenses, part 6
Like grandfather, like grandson?
Zeiss Luminar 63 mm f/4.5, version 1 versus version 3
The Carl Zeiss Luminar 63 mm f/4.5 is one of the best photomacrographic lenses, and my earlier tests (here, here, here, here and here) confirmed this. Among the photomacrographic lenses I reviewed on this site (by now, at least 22 different models, not including some I tested but did not publish about), only the Macro-Nikkor 65 mm f/4.5 performs better, and by just a tiny, and even questionable, margin. Exotic lenses like the Micro-Nikkor 70 mm f/5 and EL-Nikkor 68 mm f/3.5 also perform exceptionally well, but their problem is that they simply are not available on the second-hand market.
The Carl Zeiss Luminar series was produced in at least four successive versions, or generations. Among these models, the 40 mm was obviously re-designed in the process (including its optical formula), while the others had their mechanical parts updated, while their optical design may or may not have been updated (and Zeiss provided no information on these matters). The lens coatings do look slightly different between the older and newer versions, and likely were changed as well.
All my earlier tests involved a version 1 Luminar 63 mm (i.e., the first version produced after Zeiss acquired Winkel, thoroughly incorporated it, and started producing Luminars marked only Zeiss instead of the transitional Winkel-Zeiss or the pre-merger Winkel logo). It stands to reason to think that subsequent versions may have been slightly improved. Therefore, when looking for a specimen of these lenses meant to be used in photography, it should be desirable to try and obtain one from a later version. On the other hand, in the 1980s and 1990s the production of many brands of camera lenses and microscope equipment started concentrating on mass production at low costs, rather than durability and top quality. A worrying sign that Zeiss may have been following this trend is seen in several of their microscope objectives, which were marked with a silk-screen process instead of being engraved, and lost their markings after prolonged use. This is also the case with version 3 Luminars, which have engraved aperture scales (which would make a lens unusable if lost) but silk-screened specifications and model markings. Zeiss microscope equipment of this period is also plagued by an inferior glue used to cement optical elements, which separates easily and cannot be repaired (unlike elements cemented with the traditional Canada balsam). Fortunately, most Luminar lenses, including the 63 mm, do not use cemented elements and therefore are immune to this problem.
Since I had a chance to add a version 3 Luminar 63 mm to my lens inventory, I was able to carry out a direct comparison and answer the question of whether it is really better to choose an earlier or later version of this lens - or if it just does not matter from the point of view of image quality. This is a relevant question, since a version 3 Luminar 63 mm in like-new condition may cost twice as much as a version 1. The 63 mm also seems to be the most sought-after Luminar model. While the 25 mm is frequently available and the old 40 mm f/4.5 is fairly common, a version 3 of the 63 mm may not come up on eBay more often than once or twice a year.
The two lenses look a bit different. A side-by-side comparison shows also that the barrel design is not the only difference. In spite of its shorter barrel, the Version 3 has wider front and rear elements located at a higher distance from each other than the Version 1. Both versions use a Cooke triplet formula with the aperture diaphragm between the front and center elements.
In the case of the 40 mm, I still believe that a version 2 or 3 (with f/4 maximum aperture) is preferable to a version 1 (with f/4.5 aperture), but I don't have first-hand data on this. I should expect that the more expensive, redesigned optics were used in later versions for a good reason. The modern model is also slightly easier to focus, because of its wider aperture. I am not going to test this assumption any time soon, since I already have a Version 2 f/4 model.
An initial round of tests showed very similar results of these two lenses. The main problem, in fact, was to reliably distinguish between slight changes in illumination and focusing (caused by the setup) and real differences between these lenses. Changes in illumination are relatively easy to take care of, by attaching the SB-800 flash used for this test to a miniature Manfrotto Hydrostatic arm, so that it can be oriented with precision and solidly locked afterward. The two lenses were mounted on a microscope nosepiece attached at the front of Nikon PB-6 bellows. This allows the lenses to be switched quickly and with a minimal risk of disturbing the subject and framing. With fully extended bellows, this setup yields a 3.8x magnification, which is well within the optimal range for this lens (2x - 5x according to my tests and calculations, or 2x-10x according to Zeiss literature).
A further problem is that the two lenses are not exactly parfocal. The version 1 must be unscrewed by a couple of mm in order to focus on the same subject as the version 3 specimen. This results in a slightly lower magnification of the version 1 at the same bellows extension as the version 3, which is visible in the corners of the test images (see below).
Because of the lack of parfocality, re-focusing after changing lenses would slightly change the illumination angle, since the subject is placed on a precision focusing stage for this test but the light source is instead attached to the base of the stand. Although not important in a real shooting situation, the slight change in illumination angle alters the reflections from the subject, making the test pictures taken with different lenses difficult to compare. This problem was solved by using a custom-built stand with separate focusers for both the stage and bellows, and using the latter to refocus.
The continuous light sources (power LEDs) used for framing and focusing also had to be switched off during the actual exposure, because they were mixing with the flash light and slightly changing its color temperature to different amounts as the lens aperture was changed.
Focusing with precision was a much tougher problem. The custom-built photomacrographic stand used for this test has a total of three focusing controls (not including those of the Nikon PB-6 bellows, which are far too primitive to be useful in this context):
Coarse focus, moving the camera and bellows as a whole (from a Nikon stereomicroscope). This mechanism proved too coarse for accurate focusing, and was used only for initial positioning.
Coarse focus, moving the subject stage (from a Zeiss Photomicroscope). Also this mechanism is too coarse for precision adjustments.
Fine focus, built-in and coaxial with the preceding one. Paradoxically, this control is too fine. It is designed to provide fine focusing for a high-magnification compound microscope with a precision of less than a µm, and with a photomacrographic lens at about 5x it is necessary to move it close to one whole turn in order to see any change in focusing in the viewfinder. This is too slow for visual feedback, and ends up being distracting.
Finding a suitably coarse fine-focuser, or a suitably fine coarse-focuser, is going to take me some time. In addition, the viewfinder of a DSLR is simply not designed to allow manual focusing with a lens with an effective aperture of f/27 (given the magnification used in this test). In order to carry out this test, I resorted to a brute-force approach. Starting from a slightly but visibly out-of-focus position, I took a series of exposures at 20 µm focus intervals through the (apparently) in-focus zone and until the image seen through the viewfinder begun again to go out of focus. This was also made easier by selecting a very flat subject (an electronics circuit board), so that its entire surface is in focus at the same time. From each set of pictures, the one with the best in-focus detail of the dust particles on the green surface of the circuit board at the center of the frame was selected. Center and corner details were cut out of this picture, and are displayed below.
To make it simple, I can see very little significant difference between Version 1 and Version 3 of the Zeiss Luminar 63 mm f/4.5. Assuming that you could focus with the necessary precision when using this lens in real-world situations (which you can't), and spend the time to do some pixel-hunting (time that probably should be spent for a better purpose), the newer version will give you a slightly "richer" picture when seen at a large magnification. This effect is the result of several minor improvements. Resolution, contrast, flatness of field, chromatic aberration near the picture edges and color fidelity have all been improved from version 1 to version 3 of this lens. Together with the fact that the tested versions of this lens are not parfocal and the differences noted in the introduction, these results confirm that the optical formula of the Luminar 63 mm was indeed re-computed between Version 1 and version 3 (probably, between version 1 and version 2, although lens coatings may have improved between Version 2 and Version 3). Thus, the differences among versions of this lens are more than a cosmetic redesign of the lens barrel. In practical use, however, it is simply too difficult to focus these lenses with the precision required to take full advantage of the resolution of which they are capable, and the improvements in contrast and color fidelity are very hard to detect in practical photographs.
There might be other advantages in choosing a Version 2, 3 or 4 of this lens over a version 1. For instance, lubricants in a later version are newer and have had a lesser chance to stiffen and dry out, and the lens may have been subjected to less use and abuse. However, this depends in large part on the conditions and climate in which a particular lens specimen has been stored. The diaphragm blades in Versions 3-4 are matte black, while in Versions 1-2 they are very shiny, literally mirror-like, and may potentially create flare by reflecting incoming light back to the front lens elements. The contour of the newer aperture ring is also slightly more comfortable to operate. The newer versions provide a slightly higher working distance (in my opinion, this difference is not significant, but it might be if every millimeter counts). Finally, sporting a "blue dot" Version 3 or 4 Luminar may win you some status and envy among the couple of dozens of photomacrographers who really appreciate the scarcity of these lenses. In accurate tests, the Version 3 of this lens will give you a slightly better image quality than the Version 1, but I doubt that the difference is significant in practical use of this lens. If you need this lens and have a chance of acquiring a Version 2, 3or 4 (and can afford the price), by all means do, but Version 1 does not deserve to be frowned upon. The latter, however, is a slightly different lens.
While the Macro-Nikkor 65 mm f/4.5 has a slight edge on the Version 1 Luminar 63 mm, any difference between the Version 3-4 Luminar 63 mm and the Macro-Nikkor is probably too small to matter in practice.
Important update: After using the above lenses, as well as other top-of-the-line photomacrographic lenses, for the photography of actual subjects with a specially modified photomacroscope (which, among other things, makes the notorious difficulty of focusing at high magnification a thing of the past, and guarantees that these lenses can produce the best image quality of which they are capable), I found that this page needs a significant update. While the above discussion remains fully valid and does not need to be corrected, it does not tell the whole story.
It turned out that all photomacrographic lenses in my possession display variable but significant amounts of axial (not radial) chromatic aberration. Radial chromatic aberration results in colour fringes around part of the outline of small bright objects (typically, blue and magenta fringes on opposite sides of the same object). This is the type of chromatic aberration commonly observed in general photography, especially outdoors. Axial chromatic aberration, instead, causes light of different wavelengths to be focused on different focal planes, rather than at different positions on the same focal plane. Axial chromatic aberration is rarely seen in general photography, but becomes a special problem in photomacrography, because of the extreme reduction in depth-of-field. All good photomacrographic lenses correct both axial and radial chromatic aberrations in their plane of focus, and normally you see little or nothing of these aberrations in the well-focused areas of photomacrographic pictures. Things get out of control, instead, in areas that are substantially out of focus. In these unfocused areas, bright points of light (e.g., reflections from pin-prick mirror surfaces) become horrendous balls with multiple haloes of different colours, multiple refraction and diffraction fringes, and an overall appearance like nothing you see in ordinary photography.
It is simply impossible to correct all types of chromatic aberration at all focal planes, at least in a lens that contains a small number (3-6) of optical elements. This includes all photomacrographic lenses. The correction of aberrations is always a compromise among different aberrations, and correcting the axial and radial chromatic aberration in the focal plane actually enhances it outside this plane. If you look carefully, you can actually see this in the above test pictures, especially in the corners of the original image's field-of-view. The Version 1 Luminar produces a visible radial chromatic aberration (one side of the white dust particles is blue). The Version 3 Luminar performs better in this respect, but the centre of the larger dust particles (which is out of focus, i.e., above the plane of focus of the picture) with this lens takes a cyan tinge, caused by axial chromatic aberration (with this particular lens, the tinge would be magenta for objects located out of focus in the opposite direction, i.e., behind the plane of focus). On the other hand, the older Version 1 Luminar displays a lower amount of axial chromatic aberration, because it does not attempt an equally extreme correction of radial chromatic aberration in the plane of focus.There is only so much you can do to correct aberrations in a Cooke's triplet lens design, and the Zeiss designers had to choose what they regarded as the best compromise.
In practice, this means that the Version 1 Luminar 63 mm may perform better, with certain real subjects that are partly out of focus, than the re-calculated and better-corrected (in the focal plane) Version 3 Luminar of the same focal length. Whether, and to what extent, this is also true of other focal lengths and versions of Zeiss Luminar lenses, remains to be seen. A further possibility that I did not test, but is suggested by my practical results with these lenses, is that axial chromatic aberration is a lesser concern if the lens aperture is closed 1-2 stops beyond the point at which diffraction begins to visibly affect resolution. Probably, it is not by accident that Zeiss recommended a narrower lens aperture than I would like to use, since film photography for scientific use was not expected to provide the same level of detail that is possible with current digital-camera sensors. Therefore, a higher amount of diffraction was tolerated, mainly as a trade-off against a higher depth-of-field and lesser chromatic aberration.