Delta grew out the Digital Enlarger Lens Test Archive (DELTA) previously located at 16:9 – my twenty year-old scratchpad for off-piste lens evaluations. Enlarger lenses have always been on the menu there, for professional shooting and casual experimentation, but never in their native habitat: joining industrial, pinhole and various vintage taking lenses in the digital kit bag when different looks were needed.
Delta therefore has a strong practical bias, and exists in its present forum for several reasons:
- I wanted to compare the rendering style and technical competence of ‘alt-lenses’ in a wide variety of applications on the level playing field offered by contemporary mirrorless cameras. Despite their age and relative obscurity, I’ve always felt they are relevant to photography now.
- While many sites offer guidance about the use of specialised lenses for macro and micro photography, outside that niche they have nothing to say about their ‘general’ use. Furthermore, some resources are based on APS-C sensors shooting circular objects, which barely examines more than Zone A. I wanted to know how these lenses behave at longer working distances, and in Zones B-D – in full frame, shifted full-frame, and medium-format applications – and also . . .
- . . . for videography, where there is strong interest in lenses with distinctive rendition. The DNA of some of the most desirable cine lenses overlaps considerably with enlarging optics, and the ‘rehousing’ process required to shoot them has much in common with adapting old lenses for modern cameras. Key elements of ‘the look’ achieved by highly-prized Cooke and Super Baltar lenses, for instance, is available with dozens of unregarded lenses in the Delta archive. Finding these became interesting.
- Building the archive has been an archaeological reconstruction: many fascinating stories have been buried. As we’ve searched, we’ve watched an edifice of expertise being eroded by the retirement – often, death – of those with first-hand knowledge. We’ve contacted almost every extant manufacturer and without exception they’ve all shrugged and looked the other way. With few exceptions, age has withered interest in and, correspondingly, valuations of these slow, peculiar little lenses. I felt the need to collect and collate this information before it disappeared entirely.
- In the case of enlarger lenses, John Jovic’s Big List at photocornucopia is tremendously useful, and a fascinating browse. However, at less than 400 lenses, it records only a fraction of those made and contains many gaps and guesses. In the case of cine/slide projector and industrial lenses, no ‘Big List’ existed. In both cases, there was a need to unravel the rebranding tangle and show where variously-named lenses are actually identical – and, conversely, to differentiate iterations of lenses with the same name but different optical formulae.
- Tangentially, it became evident that Delta’s short-range tests offered a new perspective on the long history of evaluating enlarger lenses as enlarging lenses. The forty year record of such reviews – culminating in the somewhat gnomic Popular Photography articles of 1997 – offers a confusing picture. Although we don’t test lenses on enlargers, Delta’s short-range Hall of Fame is the most wide-ranging survey of its type ever made, and its medal system appears to be a reliable guide for those seeking excellence in any application at that working distance: close-up photography, or film enlargement. It would be gratifying if Gold-Awarded lenses, for instance, live in posterity with deserved glory. Elite lenses like Apo-Rodagons need no further promotion. However, I have greatly enjoyed finding lenses I previously knew nothing of that matched or even exceeded them. It’s a special thrill to endorse them, and help restore a reputation they never acquired in their heyday, or somehow lost with the passage of time.
Hall of Fame Comparative Test Method and Medal Ranking
Writing gonzo lens reviews is fun. Properly testing lenses is not: it’s boring, repetitive, time-consuming and frequently frustrating. Fully evaluating a lens is also complex: the dataset can be summarised in different ways and filtered by different criteria – and will ultimately be judged by users according to different priorities.
Although ad-hoc evaluation of dozens of lenses has been conducted intermittently with several cameras over a ten year period, in 2021 we set a new objective for Delta: to deliver the simplest metric applicable to lenses of radically different types – a mark out of 10 for ‘sharpness’.
Because the aim is to sieve hundreds of lenses to find those of special interest, this mark doesn’t (significantly) comment on geometric distortion, global contrast, flare, coatings and internal reflections – or penalise a lens too hard for chromatic aberration. It also doesn’t comment on ‘vignetting’ – a concept best abolished and easily corrected in post. In the headline summary, where a lens doesn’t cover 35mm with a 43mm image circle, it is noted. This is important: a lens doesn’t ‘vignette’ – your sensor is just mismatched to its image circle!
But, I hear you ask, what is ‘sharpness’? What is ‘ten’? How are measurements and judgments made? And what about sample variation?
Working photographers seek reliable framewide ‘sharpness’. Interfering factors include poor resolution (sagittal and tangential MTF weakness) – particularly in outer zones – the veiling effect of spherical aberration and spurious artefacts of chromatic aberration. Sharpness derives from the absence of such flaws, combined with the ability of the lens to resolve fine detail down to the limit of the sensor. Although I have judged each lens by simply looking at the results (see below), and combined many empirically-measurable factors in a somewhat simplified fashion, sharpness is not a subjective criterion: it relates to repeatably rendering fine structures in a manner universally recognised as realistic or transparent.
The notional ideal of a perfect ten is defined for this purpose as fully resolving the test sensor without artefaction. From 2021 this was a ‘full-frame’ 47.3MP Panasonic Lumix S1R with a pixel pitch of 4.27µ and therefore demanding of a lens 105 lp/mm resolution* across the frame. To better differentiate lenses that seem to resolve > 100 lp/mm, a shift adaptor on a Lumix G9 subjected Zones 1-3 to the scrutiny of a 3.3µ pixel pitch sensor. Fully resolving a 20MP M4/3 camera requires a lens capable of projecting 136 lp/mm*.
Initially, lenses were referenced against the commonly-used benchmark of Sigma Art primes (50mm and 105mm). Somewhat arbitrarily, it was decided that the sharpest result achieved by these lenses – the 105/1.4 at f2.8 in the frame centre (Zone A) – would be graded 9.5. The first batch of lenses tested arranged themselves into an obvious hierarchy below the reference lenses. Marks were ‘sketched in’ to reflect levels of perceived sharpness. Several of the first batch were used as benchmarks when testing the next group and results were adjusted accordingly. At the time of writing, 67 overlapping group tests have so far been conducted with anything from three to twelve lenses per group.
Once approximately fifty lenses had been graded and regraded in progressively smaller increments, finding their proper place relative to each other, the scale stabilised. It became clear what a mark of ‘6.7’ or ‘8.5’ looked like and therefore asses new lenses accordingly. Christopher Alexander would call it ‘gradual stiffening’. After 200 lenses have been tested in this way, and with at least 80 lenses in stock for repeat comparison, the grading has become surprisingly swift and accurate, and revision of pre-existing grades is now rarely necessary.
Separate tests are made at 50-80cm distance and 5-15m distance. Each lens is cleaned and/or serviced before shooting. Naturally all focusing is manual. Each aperture is shot. Two or three captures are made focused on Zone A (centre), and a further set refocused on Zone C (full frame corner). The rear screen is set to display the target area with maximum magnification and examined with a high power loupe. Focus peaking is not absolutely relied on to find optimal focus, especially in Zone C where setting the focal plane for most favourable contrast doesn’t always yield the highest resolution, depending on the combination of aberrations present – particular when testing poorly corrected projector lenses.
The best captures are used for blind comparison: ISO 100 images are processed in DXO PhotoLab and exported with image names consisting of numbers only. Not knowing which lens produced which image, files are viewed at 200% and tabs arranged in descending order of sharpness. Files are then identified and new lenses are graded by cross-referencing against lenses with established grades present in the same batch. For instance, if a newly tested lens has Zone C sharpness falling midway between lenses with proven grades of 7.5 and 7.1, it is rated 7.3.
Six metrics were chosen to generate league tables of results: maximum aperture (near and far), f4-f8 averaged (near and far) and f5.6-f8 averaged (near and far) – all expressed as percentages based on ‘marks out of ten’ totals. The ‘Wide Open’ metric enables fixed aperture projector lenses to be compared directly to benchmark taking lenses and enlarger lenses – which are typically optimised for ‘two stops down’. The ‘f5.6-f8 averaged’ grade records the mean of Zone A (centre) and Zone C (full frame corner) sharpness at common working aperture where the lens is expected to perform at its peak. The ‘f4-f8 averaged’ grade is much more demanding for enlarger lenses – especially at long working distances. Very few score highly.
To achieve a Gold award in any category, a lens must average > 90%. Silver-rated lenses are considered recommendable for professional use and score 80-90%. Bronze-rated lenses have obvious technical flaws, scoring 70-80%, and lenses failing to reach even this standard are classified ‘Soft’. Individual category awards are more useful than the overall Medal rating: in order to arrive at a single sharpness grade these are averaged again. There are several cases – particularly at the border between Silver and Gold – when the numbers don’t place the lens clearly in one category or the other. The Schneider Apo-Componon 90/4.5 and Meogon 50/5.6, for instance, have averaged results that sit right on the line between Silver and Gold. The Apo-Componon just falls short of the highest standards of resolution, but performs consistently well in all applications, whereas the Meogon 50/5.6 is capable of extreme sharpness in some context, but not in others. For different reasons, it’s hard to categorically pin the right medal on these lenses: are they Gold with caveats? Or Silver and robbed?
Clearly there’s a need to do more than apply a label to a lens: no single number can summarise or do justice to the many aspects of lens performance. The house reviews offer the necessary deeper dive into properties beyond sharpness. It is my intention to extend these reviews with more gallery images, and report on their behaviour as video and macro lenses at a later time.
Modern lenses tend to suffer sample variation because they’re highly-strung marvels of nano-tolerance manufacturing. Enlarger and projector lenses vary because they’re old and beaten up: they worked hard and are now being eaten by fungus; they operated in damp environments and weren’t moisture-sealed. Many are easy to take apart, and so were – and not always properly reassembled. The majority of Delta’s tests hinge on results from a single clean sample. Because it’s never possible to guarantee that a given sample sits in the middle of the bell curve without having many copies of a given lens, there will always be a degree of uncertainty over results conducted on lenses of this age. This statistical noise inevitably compromises analysis. However, given that none of these lenses are still in production, the somewhat anecdotal nature of our survey fittingly reflects the real-world lottery of using vintage enlarger lenses. In 1997, multiple samples could be requested from a manufacturer. In 2022, we’re all dependent on whatever the used market throws up. Nonetheless, in many cases we have tested two, three, or even four examples and included only the best. More often than not, significantly different results depend on condition rather than centering or other manufacturing defects. With Fuji, Minolta and Nikon lenses, differences between tested samples in similar condition have so far been minor – not affecting a grade by more than 1-2%.
At peak apertures, unless they are ancient or broken, most lenses deliver ‘acceptably sharp’ results centre frame (Zone A) but fully resolving a sensor at near and far distance in Zones B and C is rarely possible. Unlike ‘portrait’ lenses of their day, industrial, enlarging and projection lenses were designed for flat-field, corner-to-corner consistency (though not usually at 100 lp/mm) and this, as much as the absence of a focusing mechanism, distinguishes lenses in Delta from conventional taking lenses. However, the purpose of Delta is more than the creation of ‘sharpness league tables’: many overlooked gems in the archive offer unique ‘looks’ despite lacking the Nth degree of per-pixel crispness – a quality many shooters (especially videographers) claim is over-rated.
The 2022 edition was made possible with the assistance of Mr.Cad – Europe’s largest dealer of used cameras. For all things analogue film photographic, get in touch with Alex Falks, whose “time in the business began roughly when Elvis joined the army”. Mr.Cad carries permanent stock of enlarger lenses and offers knowledgeable and approachable service. We would also like to acknowledge the pioneering ninety-sample enlarger lens test program of Ctein in the late 1990s and John Jovic’s articles and 472-entry Big Enlarger Lens List at photocornucopia, as well as the invaluable editorial assistance of Johannes and others.
Using and Understanding Delta
Each column heading (Manufacturer, Focal Length (mm), Minimum Aperture (f), etc.) can be selected to sort in ascending or descending order and combined with keyword search for more targeted results. For instance, you may want to ask the database to show all lenses with circular apertures, or display all Componons. You may only be interested in lenses produced in the 1950s, or those faster than f4. Perhaps most usefully, selecting the column headings for sharpness creates a league table of test results for averaged performance across the frame, and (on scrolling right) specific performance at each aperture and part of the frame (see below).
There is a broad correlation between the number of Elements and image quality. Generally (with noted exceptions), triplets (three element lenses) are cheap, low-quality optics that don’t resolve well in their outer image circle. Not every four and six element lens performs to a high standard, but all the best enlarger optics have a minimum of five elements.
The Aperture Type column shows both the number and geometry of aperture blades, which influences the lens’ drawing style. Circular apertures (common to most enlarger lenses made prior to 1980) tend to produce smoother bokeh and a more pleasing focus transition, and render circular ‘bokeh balls’ at all stops. The shape of the aperture (stopped down) naturally dictates the shape of the bokeh balls: straight-sided hexagonal apertures render straight-sided hexagonal bokeh balls. Enlarger lenses have rather less predictable aperture shapes than normal taking lenses – sometimes changing shape unexpectedly as you stop down. Where this is notable, we have noted it.
The shape of the aperture also influences when (or whether) the lens produces Stars: spokes of light radiating from in-focus point light sources. Some photographers find these attractive, so we have rated and noted each tested lens’ ability to produce them, awarding high marks (out of 10) for a lens that can generate stars at wide apertures. We have defined ‘good’ stars as having minimal flare and maximal definition, without reference to the number of light-spokes produced.
When we have tested a lens it receives a Sharpness percentage rating for Near Distance (typically 40-80cm) and Far Distance (10m-infinity). The ratings appearing in columns 7 and 8 are averaged sharpness across the frame at apertures f5.6 and f8. By comparison with reference lenses (Sigma Art DG primes) you will note that most enlarger lenses suffer performance lag at longer distances. However, at the short working distances typical of copy work, product and food photography (for instance), they match or exceed the average resolution of conventional taking lenses. Delta’s reviews do not delve into the 1:2-and-above realm of macro photography, for which use-case enlarger lenses are well documented elsewhere. If you scroll right, at the end of the table you will find the full breakdown of each tested lens’ performance from apertures f2.8 to f11. in Zone 1 (centre frame) and Zone 3 (outer image circle), and another Sharpness rating that averages performance across the frame using a more demanding benchmark of f4 to f8.
Don’t take for granted that a good enlarger lens will be sharp at every setting: commonly they are narrowly optimised for a specific aperture (typically two stops down) and the widest aperture intended only for focusing – which always was a questionable practice because many have quite pronounced focus shift. Meopta Meogon lenses, for instance, are relatively poor performers wide open, but generate some of the highest resolution figures at appropriate apertures. Conversely, Schneider Apo-Componon lenses and certain orthometars like the EL Nikkor 80/5.6 N give optimal Zone 1 rendition wide open, and the only purpose of stopping down is to raise Zone 3 to same level.
As touched on above, enlarger lens Bokeh is a contentious topic, and these lenses vary greatly. The rating awarded is inevitably somewhat subjective, but we have aimed for an empirically sound comparison based on prioritising smoothness, consistency fore and aft of the focal plane, and absence of distracting artefacts. We also give kudos to a lens’ power to defocus, privileging longer and faster designs. The bokeh rating specifically penalises the manifestation of soap bubbles, onion rings, cat’s eyes and swirling; and marks down ‘edginess’ or ‘vibration’ – noted in Bokeh Characteristics.
However, projector lenses in particular are now sought-after for exactly these ‘technically incorrect’ qualities – particularly those that generate ringed bubbles from defocused specular highlights. Note that the same rendition is seen wide open in many triplet enlarger lenses, too. At wide apertures most lenses oblate these bokeh balls into rugby/football or cat’s eye shapes that vary in distortion and orientation with distance from the centre of the image circle. This effect tends to go hand-in-glove with bokeh swirl, which can be effective when framing a central subject. Because both effects are proportional to the size of the image circle, be aware that the best way to achieve pronounced swirling is to choose a lens whose image circle only just covers your sensor. For instance, a 100-120mm projector lens may generate strong swirl on a medium format sensor in parts of the image circle that aren’t visible to a 35mm sensor. Shifting such a lens on a 35mm body will show swirling with apparent radial asymmetry. To achieve frame-filling swirling on a smaller sensor requires a smaller image circle – typically that of a 60-80mm lens on 35mm and 40-50mm on APS-C or M4/3.
Marks and notes for Colour, Contrast, Focus Shift, Weight (in grams), Used Value (in £GBP), Front / Rear Threads, Years of Production and observed ranges of Serial Numbers (from lowest to highest) are self-explanatory.
Screw mount fittings have varied frustratingly over the century+ spanned by the Delta project. To simplify as far as possible, all 39mm fittings are here treated equally and marked M39 (yes, even the Imperial-thread lenses), as they are practically interchangeable – unless they aren’t, in which case it is noted. For the record, Leica’s hybrid metrimperial Screw Thread (LST, also known as Leica Thread Mount (LTM) and L39) is an Imperial pitch (26 turns per inch). It is not identical to M39, which is 1 turn per 1mm. However, Leica’s Imperial 1 turn per 0.971mm is usually close enough for the handful of turns required by 39 > 42mm adaptors – depending on the machining accuracy of the adaptor in question.
Delta also includes true Imperial thread lenses such as C-Mount (1″), RMS (Whitworth 0.8″ x 36, Wollensak (1 5/8″) and Wray (1 1/4″) mounts.
Finally, columns marked FFD (recording the focal-flange distance of the lens) and, to avoid confusion, Extension (the distance between the lens flange and camera body flange (AKA flange distance) when focused at infinity) aid in finding the correct helicoid and/or adaptor for your lens. Extension is the minimum helicoid you need to achieve infinity focus (minus the depth of the mount adaptor).
Note that Extension figures are calculated for Canon RF and L-Mount cameras that have a body focal-flange distance (the distance between sensor and lens mount) of 20mm. For other cameras, the following adjustments are necessary: Nikon Z: +4mm; FujiFilm X: +2.3mm; Fujifilm GFX: -7mm; Sony E: +2mm; Canon EF-M: +2mm; Micro 4/3: +0.5mm; Canon EF: -24mm; and Nikon F: -26.5mm
To give a worked example: a Nikon EL-Nikkor 50mm f2.8 N has a 43mm FFD. According to the chart, achieving infinity focus on a Panasonic S1 requires a helical/adaptor extension of 23mm. This lens therefore hits infinity at 27mm extension on a Nikon Z6 (+4mm); 25.3mm on a Fuji X-Pro 3 (+2.3mm); 25mm on a Sony A7 (+2mm), etc. Infinity focus isn’t possible when mounting the 50/2.8 on an old-school DSLR camera like a Canon 5D: that would require a helicoid/adaptor of -1mm depth.
M42 helicoids are commonly available in extensions of 10-15mm, 12-17mm, 17-31mm, 22-55mm and 35-87mm. M42 mount adaptors are around 1-2mm thick. Tilt and shift mechanisms tend to be 25-35mm deep. In practice, then, a lens must have at least 42mm extension to give infinity focus and movements on Canon RF/Panasonic S bodies. Typically, most 75mm+ lenses fit the bill. Nikon Z users have wider options: a slim adaptor and the smallest helicoid enable a user to deploy a lens with a FFD of just 29mm.
Let’s also consider the case of adaptation to Fujifilm GFX, with its 27mm body FFD. The thinnest available helicoid (10-15mm) and adaptor has a real-world extension of 12mm. Therefore a lens with a FFD of 39mm will just achieve infinity focus with the helicoid racked to its shortest position. Such a lens will have 5mm of outward travel to reach its position of minimum focus – which is likely to be around 1m. The shortest (in terms of FFD, not focal length) lens of note you might want to adpat to a GFX body is the Leica Focotar 40mm f2.8, which (just) reaches infinity focus with a 10-15mm helicoid. For coverage details of this and other lenses, please see the Reviews tab of individual product pages.
Naturally, longer outward travel from infinity focus allows more versatility in obtaining close-ups. One particularly uncomfortable, but fortunately unique, case (given the specification of available helicoids) is if your body/lens combo calls for 15-16mm extension – for instance, mounting a Rodenstock Rodagon 35mm f4 on a Panasonic L mount or Canon RF body. In such a case, a 12-17mm helicoid provides infinity focus but dictates a minimum working distance of many metres, having only 1.5mm of extension before the helicoid maxes out –whereas a 17-31mm helicoid allows for great close-up functionality, but will not achieve infinity focus.
The F-FD of a lens is usually similar to its focal length but, as Delta shows, there is considerable variation: among 50mm lenses we find early Rodagons hitting infinity focus with 28mm extension, and a late Vega 11U with 18mm. Rear element protrusion also varies much more than the tidy backsides of conventional lenses. And with regard to fitting lens mounts to projector barrels (which are either smooth or feature wide proprietary helical threads) the extension will vary like a trombone, depending on placement of the adaptor.
Undertaking a wide-ranging survey of screw mount lenses deployed via M42 helicoids or bellows, it was tempting to include the spectrum of Leica LTM rangefinder lenses, large format lenses in Copal shutters (to which many of these lenses are closely related) industrial lenses, projector lenses, microscope lenses, scanner lenses, et al – all of which are all viable in the same rig and share many of the drawbacks and benefits of enlarger lenses. However, ultimately we made four rules for inclusion in Delta:
1) No internal focusing mechanisms
2) Adaptable to M42
3) Image circle > 25mm at infinity focus (to be useable on mirrorless cameras with a body flange-focal distance of c20mm)
4) No large format taking lenses (that would otherwise qualify under rules 1-3)
These rules allow us to include some scanner and microscope optics via RMS > M42 adaptors; V-Mount lenses, and many interesting bellows, projector, X-ray, repro and process lenses that are worthy of consideration – and not just for macro applications. I’m waiting for the first digital indie film to be shot entirely on Printing Nikkors in gratifying coming-full-circle homage to Stanley Kubrick (who insisted on using these lenses for film duplication at Pinewood).
Inevitably many lenses are absent simply because we haven’t located them yet. Such a catalogue can never be complete. However, if you spot something we’ve missed, or can send samples or examples (lenses or photographs), please get in touch – especially if you can cast any light on lesser-documented Russian, Indian, and Chinese designs.
Image Circle Considerations
Not all enlarger lenses (particularly below 40mm) have image circles large enough (> 43mm diameter) for full frame digital cameras at infinity focus. Such lenses are included with a caveat specifying their image circles. However, lenses with a 30mm image circle are still fully useful on cameras with sensors smaller than APS-H, and stopped down, or shooting certain aspect ratios, you may even get some utility from them on Super 35. Lenses with a 25mm image circle can only be used with Micro 4/3 and smaller sensors.
Having said that, some 12-35mm enlarger lenses with < 25mm image circles can be used in the macro range where the increased distance from the sensor naturally widens their image circle. Some also throw larger image circles when reversed – the natural orientation for certain applications.
Ultimately, a key benefit of so many of these ‘alt’ beauties is that they throw big image circles. Broadly speaking: the longer the focal length, the bigger the image circle. Most enlarging lenses cover a full-frame sensor with room for shift and tilt movements (at infinity focus) by 60-75mm. Put another way, most 60mm enlarging lenses cover medium format sensors by less than 75mm, and by 100mm most enlarging lenses cast a big enough puddle for medium format sensors to shift about in. Fast projector lenses (very cheaply) provide medium format users with shallow DOF unattainable by native or conventional lenses, but be aware that they tend to have more constraining FFDs: the shortest focal length projector lens that covers a 645 film frame properly is around 110mm, and the move up to 6×7 requires around 140mm, depending on the required presence or absence of vignetting and corner aberrations.
Within this archive is a vast playground of ‘looks’: novel ways of shooting stills and moving images; powerful macro lenses and every flavour of bokeh. All these worlds are yours (except Europa). Use them wisely. Use them in peace.
- The Nyquist limits for sensors with 4.27µ and 3.3µ pixel pitches is theoretically 117.1 and 151.5 lp/mm. However, real-world limits are dictated by other issues – here, primarily, that photosites are not contiguous. The Kell Factor (originally calculated in 1934 at 0.64, commonly averaged at 0.7, and continuously upwardly revised) is therefore applied when converting microns to line pairs. For the Lumix S1R a Kell Factor of 0.7 gives: 1/0.00427/2*0.7 = 81.97 lp/mm. But the Kell Factor is guesswork, flagging a need for specific information – of particular relevance here: gaps between photosites. Testing lenses that reportedly resolve 80 lp/mm against lenses that demonstrably outperformed them, setting an information ceiling of 81.97 lp/mm for the S1R just feels wrong. Indeed for CMOS sensors, the appropriate factor is typically 0.9, and rising as sensor technology improves. For the purpose of the article we have therefore assumed a Kell Factor of 0.9.