Legacy lenses for UV imaging:
Criteria for preliminary selection
My interest in legacy lenses is mainly in testing the most promising ones in digital near-UV imaging. There are hundreds, or very possibly thousands, of legacy camera lens models on the second-hand market. Only a minority have been tested in UV imaging, and the test results made public vary greatly in detail and reliability.
An additional problem specific to legacy lenses is that models with the same general specifications were often released in numerous variants, some of them differing in optical specifications in ways that directly affect their suitability for UV imaging. Quite often, test results are so hermetically, haiku-like summarized that it is not possible to decide which of the multiple lens variants was actually tested.
Criteria for selecting candidate lenses to test
Testing legacy lenses at random would be very time-consuming, and purchasing them at random for testing very expensive. Therefore, one needs general criteria to identify which lenses may be especially worth testing. Since these criteria are not always reliable, it is possible that this preliminary screening will exclude a minority of lenses that are suitable for NUV imaging in spite of failing one or more of these criteria. In the lack of reliable information from other lens testers, a preliminary choice among legacy lenses of similar focal length, speed and format coverage can be made according to the following criteria:
Single-coated lenses are preferable to multicoated.
Lenses with few optical elements are more likely to be usable in UV.
Thin optical elements are more likely to transmit UV than thick elements.
The adhesive used to cement optical elements together (traditionally, Canada balsam) usually absorbs UV. Therefore, lenses with few or no cemented elements are more likely to work acceptably well in UV than lenses with multiple cemented elements.
In relatively infrequent cases, the optical glass or lens cement may fluoresce under UV irradiation, causing a veiling glare in UV images.
Single-coated elements usually display a bluish surface reflection. Multi-coated elements can give green, red, magenta etc. reflections. Early multi-coated lenses often had a multi-coated front element and single-coated inner elements. This was probably done to reduce the cost of multi-coating, which was initially high. This cost became later so low that, in the 21st century, virtually all optical elements are multi-coated, regardless of actual need. Multi-coatings are generally optimized for the specific function of an optical element, and therefore, in modern lenses it is common to see different types of multi-coating on the different elements.
Without a sufficient UV transmission, a lens is not usable for UV imaging. However, just like knowledge is only the beginning of wisdom, UV transmission is only the initial criterion for choosing a legacy lens suitable for UV imaging. A lens that transmits UV well is by no means guaranteed to be suitable for UV imaging. This is discussed in the following sections.
Just how well a lens should transmit UV to be useful is debatable. Modern digital cameras with Bayer sensor, after being converted to "full spectrum", typically can record 340 nm radiation with one or two stops of attenuation with respect to 365 nm. They can still be used for imaging down to about 310-320 nm, although sensitivity is so reduced in this range that images are typically quite noisy or require a very powerful UV-enabled electronic studio flash. Lenses designed for UV imaging often (but not always) transmit sufficient amounts of UV to allow imaging at 320 nm (with up to a couple of stops of attenuation with respect to VIS). In general, legacy lenses can be used if they display a maximum of 3 to 4 stops of attenuation with respect to VIS at the shortest wavelength of interest. Many legacy lens models are usable only down to about 380 nm, a few down to 360 nm, and a handful down to 310-320 nm.
Chromatic aberration in UV
Whether a lens is suitable for NUV imaging depends also on its image-forming properties in the NUV. A lens may transmit sufficient amounts of NUV and produce a sharp image at a given NUV wavelength, for example around the 365 nm produced by typical UV LEDs. However, if this lens displays a massive amount of axial and/or lateral chromatic aberration across the NUV range, it will produce a fuzzy image when used with the broader-spectrum NUV from solar radiation or xenon flash tubes. I discovered that a few legacy lenses, which looked promising because of their decent NUV transmission, are affected by poor broadband-UV image resolution to such a massive degree to make them useless for NUV imaging.
For the above reason, tests with a monochromatic light source (e.g. a 365 nm LED torch) often give misleading results about the suitability of a lens for NUV imaging. In fact, I have seen tests with 365 nm LED torches that give virtually identical results with apochromatically corrected UV lenses and with cheap general-purpose lenses that perform poorly with a broadband UV source like sunlight. In conclusion, other than by testing image quality with a broad-spectrum NUV source, there is no practical way to detect whether a lens displays this problem.
Several Internet sources refer to "UV focus shift" as an undesirable property of lenses used for UV imaging. The general symptom of this UV focus shift is that focus must be readjusted when switching from a VIS-pass filter to a UV-pass filter. This type of focus shift is a manifestation of axial chromatic aberration.
Strictly speaking, UV focus shift has nothing to do with the generally accepted definition of focus shift, which manifests itself as a shift in the focus plane of a lens when its aperture is changed. "Proper" focus shift is mainly observed in really old lenses for large-format cameras, or in fast lenses once they are stopped down a couple of stops (especially at close focusing distances), and is mainly a consequence of uncorrected spherical aberration.
Lenses designed for exclusive use in the VIS range can be expected to display some UV focus shift, as they are not corrected for the UV part of the spectrum. However, it is essentially impossible to predict how much this will be a problem, other than by testing them with illumination by sunlight or a broad-UV-spectrum source like unfiltered xenon flash.
IR focus index
Lenses designed for use in the VIS often display an "IR focus shift" between the VIS versus NIR ranges. Many legacy lenses have a "red dot", "R" or "IR" engraving or a similar indicator some distance from their focus index. To use this red dot in NIR photography with NIR-sensitive film, one had to focus with the optical viewfinder of an SLR camera, then shift the focus ring by the distance between focus index and red dot. In the example above, for example, both lenses have been first focused at infinity in VIS, and their focus ring subsequently adjusted to their respective red dot.
Unfortunately, there is no predictable similarity between the direction and amount of IR focus shift versus UV focus shift. In achromatically corrected lenses, UV focus shift is typically in the same direction as IR focus shift, but the amounts of shift in the two spectra can be very different. In lenses that are apochromatically corrected with respect to axial chromatic aberration, IR and UV focus shift are typically in opposite directions with respect to the focus index in VIS. However, some "Apo" lenses are apochromatic with respect to lateral chromatic aberration, but only achromatic with respect to axial chromatic aberration (at least some of the Rodenstock Apo Rodagon lenses are examples). These lenses therefore behave like achromats with respect to the direction of NUV and NIR focus shift.
The large majority of digital cameras currently used for NUV and NIR imaging allow the use of live view for accurate focusing, so the problem of UV/IR focus shift, if moderate, is not as bad as it may seem, if a single image needs to be recorded in the NUV or NIR. Stopping down the lens to increase focus depth is a convenient way to reduce the effects of axial chromatic aberration.
A substantial amount of UV and/or IR focus shift, however, makes multi-spectral imaging difficult or impossible. Multi-spectral imaging typically requires one to combine images separately recorded in the UV, VIS and IR ranges into the color channels of a single composite image. If focus is not adjusted when changing spectral range, one or two of the channels may be blurred by unfocusing. Focusing before each individual image may change the image magnification, making the individual images difficult to correctly superpose to each other.
What to expect from a good UV lens
As mentioned above, if you are interested in recording UV false-color, you must use a broadband UV source like sunlight or electronic flash. You cannot use a UV LED torch equipped with a 365 nm LED for this, because at present UV LEDs only concentrate most of their emission in a band about 10-15 nm wide. In theory it is possible to combine multiple LEDs to obtain a broader emission band, but at present this is cost-effective only in a band between roughly 365 and 400 nm (and, more importantly, I am not aware of any commercial product that uses this principle). In theory it is also possible for the emission of a single 365 nm LED to be down-converted to multiple UV wavelengths by a mixture of UV-emitting phosphors, much like a white LED down-converts some of the emission of a blue or violet LED by a mixture of phosphors that emit in the green and red VIS regions. Mercury lamps are not suitable, either, because they emit only a few strong lines in the UV, instead of a more-or-less continuous spectrum. The large majority of fluorescent UV lamps are largely inadequate as well. Specialized UV broadband sources used in spectrophotometry are expensive and/or produce very low emission levels.
With a narrow-band UV source like a UV LED that emits at nominal 365 nm, a legacy lens that transmits sufficient amounts of UV at this wavelength will provide much the same results produced by the very best, and most expensive, lenses designed for UV imaging. Therefore, testing lenses for UV imaging with a 365 nm LED torch will tell you very little about the lens itself, and the results will be largely misleading for readers who need to choose a lens for general-purpose UV imaging.
To obtain results in your test images that are easily comparable to images commonly available on the web, you should also use one of the most common and proven UV-pass filters. At present, this is the Baader U. This is the filter that I recommend for this purpose, in spite of its relatively high price. There are other filters for UV imaging that produce different results. There are also several UV-pass filters that have specific quirks or limitations in their use. I do not recommend any of them for initial explorations in UV imaging, unless you already own, or have access to, one of these less desirable filters.
Finally, using a daylight or flash white balance for UV imaging produces largely monochromatic images with violent, typically unpleasant color casts. There is no "correct" white balance for UV imaging, but most (albeit not all) cameras can record a custom white balance in UV that produces relatively neutral results that are easier to compare. See white balance in UV imaging for more information.
In order to give an idea of what can be expected from lenses that perform well in UV imaging, but are not specifically designed for UV imaging, I took pictures of the same subject with four lenses. All UV images were taken with Baader U filter mounted at the front of the lens. The images are cropped to present the subject at roughly the same size, in spite of changing focal length. All images were shot ay f/11 with electronic flash (Bowens 1500 Pro with Bowens non-coated tube and shield) at the same power setting. The flowers in the picture were chosen as typical examples of a flower that displays UV-yellow petals (with UV-black center markings), and UV-blue petals, respectively.
I used the following lenses, from top left to bottom right:
CoastalOpt 60 mm Apo f/4, designed for multi-spectral photography.
Legacy Kyoei-made 35 mm f/3.5.
Legacy Enna Lithagon 28 mm f/3.5.
Legacy El-Nikkor 80 mm f/5.6 (early variant, in metal barrel with chrome-plated attachment). Because of the relatively long focal length, I was forced to move the camera tripod backward, which caused a change in the angle of view of the subject.
The Lithagon 28 mm produced a slightly "veiled" image of lower contrast, likely caused by its relatively large number of single-coated elements. It would be easy to correct this by slightly increasing contrast in post-production. Also, the focal length of this lens is too short to work well with dielectric-coated filters like the Baader U. I did not notice in the test images the tell-tale shifts of color between central and peripheral regions that usually appear in these cases, but it is possible that the loss of contrast is, at least in part, also caused by the latter problem.
Aside for this, the results with all tested lenses are remarkably similar. The CoastalOpt 60 mm produces an image with a slight green tinge in the UV-yellow areas, consistent with the fact that wavelengths lower than about 330 nm are typically recorded as UV-green. If one should use a bandpass UV filter that lets through only wavelengths below 340 nm, substantial differences among these lenses would become visible. With this subject, filter and illumination, however, the results are virtually impossible to tell apart. This is all the more remarkable, since the prices of these lenses are spread across a roughly 200:1 interval.
The CoastalOpt 60 mm still remains a defendable choice because of its top performance between 300 and 1,000 nm, which makes it my reference lens to test other lenses and to make sure that there are no problems with UV cameras, light sources and filters. Without a reference lens, identifying problems with the UV equipment and imaging protocols would become problematic. For practical use, however, and especially if the budget for NUV imaging is limited, several other lenses are also excellent choices.
In the lack of reliable and detailed tests by other photographers, only general criteria of limited reliability can help to identify which lenses are potentially useful for UV imaging. After identifying a number of candidate lenses, only testing can verify which candidates are truly useful in UV imaging.
The amount of UV transmission of a lens is only one of the several criteria that must be satisfied for a lens to be suitable for UV imaging.
In controlled conditions that reflect a typical scenario in NUV imaging, a few carefully selected legacy lenses can produce results virtually indistinguishable from very expensive lenses specifically designed for this purpose.