Fast M-series 21mm lens shootout review on Leica M9, NEX-7 and Ricoh GXR

Can the $1249 US Voigtlander Ultron Aspherical 21mm f/1.8 compare favorably to the $7000 US Leica Summilux-M 21mm f/1.4 ASPH.?

The contenders in this review, in order of maximum aperture, and as seen in the image above:

Leica Summilux-M 21mm f/1.4 ASPH.
Voigtländer Ultron Aspherical 21mm f/1.8
Zeiss Biogon T* 21mm f/2.8 ZM
Leica Super-Elmar-M 21mm f/3.4 ASPH.

To save you some reading, here is the executive summary:

Leica Summilux-M 21mm f/1.4 ASPH.:

Best used between f/1.4-2.0 or at f/11 for best across the frame sharpness, it’s centrally very sharp (with a touch of spherical aberration at f/1.4) with a flat (though probably wavy) plane of focus, but with a number of compromises due to the design (see in-depth comments below). This is the lens for you if shooting full frame and need f/1.4 light gathering as well as class-best subject/background separation and smoothness at equal shooting distances (compared to the CV21), especially towards the image periphery, thanks to the lens’s flat plane of focus. Finding Series VIII neutral density filters, which you will need for full sun wide open shooting, can be arduous. The lens also works as an all-rounder, but with some considerations required depending on your expectations.

Voigtländer Ultron Aspherical 21mm f/1.8:

If you can only get one of these four, shoot in a variety of conditions that will include low-light, high-ISO, want subject/background isolation, don’t mind the size and built-in hood, will also shoot at a variety of aperture settings for general (landscape) use and can accept stopping down to about f/5.6 for good across frame sharpness, this is the lens to get. If focused to 50cm compared to the 21 Lux’s 70cm, this lens will match or slightly surpass the Lux’s wide open ability to blur backgrounds. Probably also the best of the bunch on the NEX-7. It’s non-coded and may work best with one of Leica’s in-camera M9 28mm corrections, meaning EXIF will report the incorrect focal length. This has no impact on image quality but may complicate image sorting in catalogs reliant on EXIF tags. Works well adapted to the Sony NEX-7, with some purple edge color shift, and with best across the frame sharpness of all the 21mm lenses tested in this review once stopped down to around f/4. Also works well on the Ricoh GXR.

Zeiss Biogon T* 21mm f/2.8 ZM:

Very good central sharpness, improving across the frame past f/5.6, with virtually no chromatic aberration, Zeiss contrast/micro contrast/color characteristics and probably optimized for distance performance. A similar price to the CV21, but smaller and using Leica-standard 46mm filters, it’s a pretty close match for the 21 SEM once stopped down. It’s an older optical design not optimized for the design considerations required for best peripheral results on digital sensors, therefore may not work optimally adapted to many mirrorless systems.

Leica Super-Elmar-M 21mm f/3.4 ASPH.:

If you don’t need speed (though f/3.4 is not all that slow) or maximum background blurring ability, this lens offers the best, most consistent across the (full) frame performance from wide open, with excellent sharpness, micro contrast, and color in a very small package, at a reasonable-for-Leica price. Based on its MTF curves, as published by Leica, some speculate it rivals or exceeds the renowned Zeiss Distagon 21mm f/2.8 available to SLR users. Works great on the Ricoh GXR, but has stronger edge color shift on the NEX-7 than the CV21.

High resolution downloads of test images are available at the bottom of the page.

Buy from B&H Photovideo and support this review:

Leica Summilux-M 21mm f/1.4 ASPH.
Voigtländer Ultron Aspherical 21mm f/1.8 (In stock at time of this article!)
Zeiss Biogon T* 21mm f/2.8 ZM (black) or silver
Leica Super-Elmar-M 21mm f/3.4 ASPH.

Live in Canada and are looking for Zeiss lenses? Buy from Rob Skeoch at the Rangefinder Store or his sister site Big Camera Workshops, mention this site and receive a free B+W UV filter with your lens purchase, as well as free shipping (in Canada only).

Also thanks to the guys at Camtec Photo in Montreal for their great, fast service, getting the Voigtländer Ultron Aspherical 21mm f/1.8 to me promptly.

At Photokina 2012, Cosina, manufacturer of the Voigtländer line of lenses, announced what seemed like a curious and somewhat niche lens, the Ultron Aspherical 21mm f/1.8 in Leica M-mount with full frame sensor coverage and rangefinder coupling.

In the Leica M-system world, this is certainly a niche. But, in the world of ‘mirrorless’ compact system cameras, many with APS-C size sensors (resulting in a 1.5x crop compared to 24x36mm full frame), there is a gaping void for a fast lens of 28-35mm framing equivalence, which this lens nicely plugs. EDIT: I completely forgot about the Sony NEX 24mm f/1.8, of which I hope to obtain a loaner and compare against the 21mm f/1.8 on the NEX-7.

For the M-system, at $1249 US (at time of review), the Voigtländer Ultron Aspherical 21mm f/1.8 seems like a too good to be true bargain while in the mirrorless arena, where many are adapting inexpensive 1970s-1980s vintage SLR lenses, and are generally more price sensitive, it commands a fairly premium price. But for mirrorless users, there may not be much choice with very few, if any similar legacy system options (there is the Olympus OM 21mm f/2, which I believe is at a similar price point, and the slower, made for mirrorless, Sigma 19mm f/2.8) to offer a fast 28/35mm equivalent field of view.

As a denizen of a few Leica-centric photo forums, the popularity of lenses among rangefinder photographers seems to drop precipitously the farther one strays from the 28-35-50mm range, where the fast (and expensive) lenses in this group are all quite popular. There also seem to be a relatively fair number of Leica Summilux-M 24mm f/1.4 ASPH. users, which despite its $7000 US price, is understandable considering the popularity of 24mm f/1.4 lenses among SLR users and its roughly 35mm equivalence on the high-ISO challenged APS-H M8.

Then there are a few of us 21mm f/1.4 users… outliers, it seems. Niche in the 21mm niche.

What drew me to the 21mm f/1.4?

First of all, I love this focal length, and of my M9 images, about 30% have been done with this focal length, or wider. I find it extremely useful for people/events/reportage style work where one wishes to work close to the subject and simultaneously include the surrounding environment. This may occur in a range of lighting conditions, which in consideration of the M9′s relative high-ISO performance weakness compared to current DSLR options, makes f/1.4 extremely useful. It can make the difference between an adequate shutter speed of 1/60 at f/1.4 compared to 1/15 at f/2.8, where subject and camera movement will have considerably greater negative influence on image sharpness. And, until the new Voigtländer, the Leica was a very unique lens at this focal length, being the only ‘fast’ option when most others start at f/2.8 (the Sigma 20mm f/1.8, for SLRs, being an exception). What f/1.4 offers, if desired, is the rare ability among super-wide angle lenses to separate the subject from background distractions, which even at f/2.8 can be a difficult task.

Note: Hereafter the Leica Summilux-M 21mm f/1.4 ASPH. will be referred to as 21 Lux; the Voigtländer Ultron Aspherical 21mm f/1.8 as CV21 (CV for Cosina Voigtländer); the Zeiss Biogon T* 21mm f/2.8 ZM as ZM21 and the Leica Super-Elmar-M 21mm f/3.4 ASPH. as 21 SEM.

Therefore, a prime consideration for many interested in the CV21 will be subject isolation at f/1.8, which I compare in the samples below against the 21 Lux. And because of the wide expanses 21mm can capture, it’s also popular for landscape-type uses. Can the fast CV21 compare favorably to the more conservative ZM21 and 21 SEM?

The following are more in-depth summaries for each lens based on use with the Leica M9 camera. For results specific to the NEX-7 and Ricoh GXR, please see the section below these summaries.

Leica Summilux-M 21mm f/1.4 ASPH.

Having used the 21 Lux for about one year in a variety of situations, often as my ‘everyday’ 21mm lens, I feel I have a pretty good understanding of its capabilities. Prior to acquiring it, I was skeptical about the usefulness of f/1.4 for general use. Since then, I’ve found having the option available has made the lens a great all-rounder – one lens to carry for virtually any lighting condition. But as such, it’s a compromise lens.

First of all, for the M-series, it’s very large and heavy. But, compared to high-end SLR lenses, it’s still quite small. As someone who does not use an external finder with 21mm lenses, it blocks a fair amount of the lower right quarter of the viewfinder view, but of course, does not block the focusing patch.

As an aside, if you’re wondering how shooting 21mm without an external finder works, it’s very much a matter of familiarity with the focal length amassed over a couple decades. I found I frequently used my SLRs ‘blind’ with this focal length, basically pointing the lens at the scene in front of me while peering over the top of the camera in order to maintain better connection with the subject and peripheral vision while shooting/composing based on my feeling of how wide ~20mm should be. This approach was frequently used for very active event-type photography, such as wedding receptions. Coincidentally, this technique seems to be a pretty good ‘party trick’ when photographing people in a celebratory mood fueled by alcohol, often eliciting great reactions of surprise that work well in creating lively photos. It seems many people think you must look through a camera to take a picture and are caught by surprise and “are not ready for the photo” when it’s made while not looking through the viewfinder. Thus, expressions are more natural and less controlled when in this ‘unready’ state.

With the M9, I have a pretty good feeling of how wide 21mm is and find I can very effectively compose without a 21mm finder. The image above was done by guessing the framing – I also really like it for showcasing subject/background separation possible with this lens at f/1.4. If framing precision is important, and time allows, it’s a matter of shooting a number of frames while slightly changing the composition or reviewing the results on the camera’s display and re-shooting if necessary. It should of course be understood that the very nature of rangefinder cameras means non-TTL viewing results in framing inaccuracies. It seems on the forums many get hung up about this, especially when coming from SLRs. In my opinion, rangefinder shooting is not about perfection in framing and focusing, but the new Leica M’s live view ability will address this complaint.

Build quality is excellent – very solid and dense feeling, with very smooth focus throw and aperture movement (11-blade in half-stop increments, it’s the most circular aperture of this test group at middle settings). It’s what one would expect considering the price point. But the design is not perfect: the focus and aperture rings are too close together and too narrow. If shooting with gloves, it’s possible to inadvertently change the aperture setting while turning the focusing ring. Since I’m more often adjusting focus than aperture, I added the “Steer” ring from Leica Goodies. It’s a huge improvement. I highly recommend it.

Another quirk of this lens (and the 24 Lux) is use of the ancient non-threaded Series filter standard, in the case of the 21 Lux, Series VIII, of which one filter can be held in place by the lens hood. This might not seem like a serious issue until one considers that on a sunny day with the M9 set to base ISO of 160 and at the fastest 1/4000 shutter speed, the correct aperture is about f/2.8… Therefore, if one wishes to achieve the wide open shallow depth of field look on bright sunny days, a 2-stop or darker neutral density (ND) filter is required. The problem though, is that it’s extremely difficult to find Series VIII ND filters in stock. B+W is an option, though vendors I’ve checked with all list these as special order, which is understandable. Is there any other current lens that uses Series VIII filters? To compound matters, B+W seem to be discontinuing many of their Series filters. Instead of ND filters some use a polarizer… Well, none exist (to my knowledge) in Series format, and would be extremely difficult to rotate for the desired effect due to being fully enclosed by the lens hood. This is a matter where Leica should offer their own ND filters to ensure there remains at least one available option. Of course, such a solution would be expensive…

Leica states use of the included hood is mandatory because the front element protrudes slightly and thus offers protection against accidental damage if the lens is placed front down. The design of the hood is very nice. The rectangular shape is attractive and likely about as effective as can be achieved considering the lens’s angle of view. A flower petal style hood could have been an option, as seen with the CV21, but at the expense of additional length and likely greater diameter. Since the front of the lens is very wide, the inclination is to place the lens front down, which the hood allows due to its flat front.

While the 21 Lux is a compromise in physical proportions in order to deliver f/1.4, it’s also an optical compromise of sorts.

In my opinion, it’s a lens intended to be used at f/1.4, f/2 or f/8-11. It seems to be optimized for f/1.4-f/2.0, which makes sense considering one wouldn’t spend the money on this huge lens if it wasn’t good wide open.

At f/1.4, it’s sharp but with a bit of ‘Leica glow’ to take the edge off (compared to a super high micro contrast/sharpness lens like the 21 SEM). Sharpness is best in the center of the image but because this lens has an impressively flat plane of focus, one will find subject detail along the edge of the frame that is roughly in line with the central subject content is also impressively sharp. Vignetting is quite strong, thus favoring central compositions, but such vignetting is to be expected and can be an effective addition to images. Sharpness improves and much of the glow disappears at f/2.0. Contrast and central sharpness improves beyond f/2.0, but with a bit of a quirk.

This quirk can be seen in Leica’s MTF curves and in actual images.



There is a pretty substantial dip in the higher lp/mm curves that correspond to about the mid-zone area of a full frame image, as well as wide divergence of the sagittal and tangential curves, roughly where one would compose based on the rule of thirds. I’m taking a guess, but it might be a consequence of the design ‘forcing’ a relatively flat plane of focus, and this mid-zone dip appears to be a wave in, or a buckling of, the plane of focus. In the comparison images that can be downloaded in the links below, the infinity focus scene overlooking the city at dusk shows the consequence of this ‘dip’ quite clearly, where even at f/5.6 some of the building details in the mid-zone area have a ‘disturbed’ or ‘nervous’ quality to them. It’s not that it’s outright soft, but that the sharpness is unnatural.

The first crop is from the mid-zone area of an image shot at f/4.0:

And this one is from the edge of the frame, with better sharpness than the mid-zone:

In comparison, here’s the same crop from the f/4.0 CV21 frame:

When I first viewed these 21 Lux images at 100%, I thought I’d muffed the exposure by accidentally bumping the camera and inducing camera shake. But that wasn’t the case because the center of the images are sharp. Other than ‘landscapes,’ this becomes a problem with placing subjects off center at moderately wide apertures, such as a person in an environmental portrait.

Other compromises are wavy mustache-type distortion, very strong chromatic aberration (CA) and strong purple fringing in high contrast edge transitions around bright areas. But, all of these can be corrected in post. Adobe Lightroom version 4, which includes a profile for this lens, is effective at correcting all of these problems. Point light sources towards the edges will reveal strong coma smearing wide open, mostly gone by f/4 and appears to be gone around f/8.

And, minimum focusing distance of 70cm for a 21mm lens can be a limitation that results in quite loose framing of small objects. Basically, you’re not going to use this lens for small object close-up bokeh shots. Even a portrait will be a bit on the loose side for head and shoulders. But then, how close do you want to be with a 21mm for portraits due to the inevitable stretching of features?

This is about as close as you’ll get with this lens for portraits at roughly 70cm:

Returning briefly to the hood: While conducting tests, I found the Lux is fairly susceptible to flare, especially if the sun (and I suppose any other strong specular light source) is just outside the frame. Flare is definitely stronger with the hood removed. Of the four lenses in this review, the Lux will flare the most. In a year of practical use, I rarely found it to be an issue. Common situations I’ve encountered are strong backlighting, such as windows behind a subject, where the Lux performs quite well. It retains very good subject contrast and edge definition, though purple fringing will often be evident.

With all of these compromises, is the lens worth it? Well, yes, if you need the light gathering ability of f/1.4 and want the best-in-class subject/background separation and smoothness. The lens is surprisingly sharp wide open and has a really nice balance of smoothness and sharpness that is gentle and subtle, combined with really good color and tonality richness. As mentioned, subject isolation is the best of the lenses compared here, and noticeably better than the CV21 because of the Lux’s flat plane of focus, which results in greater background blur towards the edges of the frame. Combine this with the smooth (for 21mm), gaussian blur quality to background objects, similar to other Summilux lenses, and it all combines for more subject isolation. It can be used as an all-round lens if its optical compromises are acceptable to your needs and style of photography. If you’re a ‘landscape’ photographer looking for the best edge to edge natural sharpness quality, then consider the 21 SEM instead.

———-

Voigtländer Ultron Aspherical 21mm f/1.8

I’ll be upfront, of the four lenses here, the CV21 is the best of the bunch when balancing cost, value, versatility, sheer optical performance and all-round usefulness, with fewer glaring compromises than the 21 Lux, but not matching the perfection of the 21 SEM.

As Cosina exemplified with the recent 35mm f/1.2 version II, released in the fall of 2011, it’s possible to spend ‘only’ about $1500 and get image quality that rivals Leica’s $5000 lenses.

And now with the 21mm f/1.8, Cosina appears to have released a near twin, both in terms of physical build and optical quality.

Place the 35mm f/1.2 II and the 21mm f/1.8 side by side and the similarities are obvious. The focusing ring looks to be the same on both; the design is the same; the anodized finish is the same… as is the overall length, nearly, when including the 21′s built-in, non-removable hood. Without the hood the 21 would have been a fair amount shorter, and since the optics are actually closer to the camera than with the 35mm f/1.2 II, the lens doesn’t feel as front heavy, and is definitely lighter in feel. But it’s still a fairly large lens for the M system, which may dissuade some.

Mechanical fit and finish is very good. Not quite as silky smooth as the 21 Lux, but still excellent. Better than the 21 Lux is the sufficiently wide focusing ring and the aperture ring that stands out enough for sure and separate grip, adjusting in somewhat clicky, slightly hollow sounding half-stop increments of the 10-bladed aperture all the way up to f/22 (vs. f/16 on the Leicas).

Of note, while f/22 is available (and also on the ZM21), the resulting images suffer from substantial diffraction degradation. In certain circumstances where very deep depth of field is required, f/22 does make a difference, and it may be possible to offset some of the diffraction with sharpening in post. But for most ‘normal’ uses, f/11 seems to be the point at which diffraction onset is noticeable, but still tolerable.

And as mentioned, this lens is rangefinder coupled. My copy seems to slightly back focus on my M9, though it could also be my camera’s rangefinder requiring recalibration, since a few of my other lenses are starting to creep towards back focus tendency. This seems to be most critical for wide open focus at farther distances, say 5m and more. And stopping down will quickly deepen depth of field to mask the issue.

One of the tradeoffs of non-Leica lenses on the M8 or M9 is the lack of support for edge color shifts and vignetting correction. Leica’s lenses are 6-bit coded in order to automatically apply lens profile correction for these issues at the time of exposure, if desired. Therefore the conundrum for users of lenses such as the CV21 is how to apply suitable corrections and which of the in-camera lens profiles to select as the best match.

Shot on the M9 without correction, the CV21 renders images with fairly strong vignetting (though offhand it didn’t seem worse than the 21 Lux with automatic correction) and some edge color shift, notably purple/magenta in the bottom left corner. In ‘normal’ use, I didn’t find the color shift to often be objectionable. One could use the lens uncorrected and instead correct on an as-needed basis in post with something like CornerFix or the Adobe Labs Flat Field plug-in for Lightroom. I used CornerFix prior to M9 firmware version 1.176, released in the summer of 2011, which effectively improved corrections for the lenses I frequently used. I haven’t yet tried the Flat Field plug-in, therefore cannot comment on it.

During initial tests, I ran through the M9′s in-camera profiles trying to find a suitable match. The 21mm Elmarit non-ASPH (11134) and 24mm Elmarit ASPH (11878/11898) profiles show promise, though the 21′s correction might be a touch too green/cyan in places, but offering the benefit of correct focal length information in EXIF. My preference is among the 28mm profiles. The 28 Cron profile corrects colour shift while retaining a fair amount of the lens’s vignetting characteristics – a nice option if so desired. For more vignetting removal, the 28 Elmarit (11804) is good, and probably what I would select for general use. The 28 Elmarit (11809) profile totally eliminates vignetting for the brightest (but perhaps too clean?) edge to edge look.

As many likely know, the downside to manually selecting in-camera profiles is forgetting to switch back to auto when using Leica’s coded lenses. The danger being shooting with an unsuitable focal length, such as 50mm or 90mm, while set to a 21mm profile. This typically results in odd (usually green) edge color shifts and too-bright image edges. So, for non-coded 3rd party lenses such as those from Voigtländer and Zeiss, many of the newer ones, including this CV21, come with a slight groove in the face of the lens mount to facilitate ‘hand coding’ with a permanent black marker, or paint. Initially I tried hand-coding as the 21 Lux, but for some reason the camera wouldn’t read the code. I later tried coding as the 28 Cron, which worked. I may have to revisit this, but for the time being, I’m happy with the look of the 28 Cron correction. For greater vignetting correction I’ll manually select the 28 Elmarit (11804) code when desired.

It will be interesting to see how much better, if at all, the new Leica M with its totally redesigned sensor will be with non-coded lenses.

As for lens performance… I was initially skeptical about the 1/3 stop difference between f/1.8 and f/2.0 because my experience with the Voigtländer 75mm f/1.8 was of no visible difference between the two settings, meaning it was effectively an f/2.0 lens. But with the 21mm f/1.8, there is a difference in central image brightness between the two settings. And the NEX-7, which increments shutter speeds in 1/3 stops, when set to Aperture priority, used 1/3-stop higher shutter speeds when the lens was set at f/1.8. That said, peripheral image brightness appears very similar between f/1.8 and f/2.0.

Unfortunately Cosina never publishes MTF curves, so it’s difficult to get a feel for potential lens performance without actually buying/trying their lenses (maybe this is intentional?).

This lens is centrally very sharp wide open at nearer distances, such as might typically be used for reportage style photography, where this lens might be seen as an obvious choice… And by those looking to benefit from subject/background separation by shooting near subjects. I would peg it and the 21 Lux as pretty much even at f/2. The CV is still very good at farther distances though seems to degrade at infinity, where the 21 Lux betters it (see the cityscape at dusk images). But, stopping down past f/2.8 revives sharpness at far distances.

Field curvature exists. It seems pretty good through the center and gradually curves away from the camera approaching the edges. A consequence perhaps is that distortion appears to be uncomplicated and barrel-shaped in nature. Due to the field curvature, frame-filling flat objects will be soft towards the edges at wider apertures. I’ll say by about f/5.6 the edges are excellent for distant scenes. This lens seems to be slightly narrower in field of view than the two Leica 21s tested here and may be another consequence of the lens’s distortion and field curvature characteristics, when compared horizontally through the center of the frame. Corner to corner comparisons are close to the framing of the Leicas, indicating that the CV’s barrel distortion pushes some image detail out of view at the central edges. Point light sources towards the edges will reveal strong coma smearing wide open, mostly gone by f/4 and appears to be gone around f/8.

Another consequence of this lens’s field curvature is the look of background blur. Here some might be disappointed if they were hoping for all of the 21 Lux goodness at 1/5 the price. At equal shooting distances, background blur is a bit higher from the 21 Lux in the center of the frame, though not a huge difference. But, the blur difference progressively increases in favor of the 21 Lux into the corners. Even at f/2, where one would expect the lenses to be equal, the Lux shows greater background blur around the image periphery because of its flatter plane of focus. And the nature of the blur is different, with the Lux being more gaussian while the CV21′s blur has stronger object edge definition and nervousness/busyness.

But the CV21 has a ‘trick’ up its sleeve: non-rangefinder coupled 50cm minimum focusing distance. If you can move in close to the subject and shoot at 50cm, the background blur will match, or possibly slightly exceed, that of the 21 Lux from 70cm. It will be a great solution with live view cameras.

While it may be difficult to see at web-rez, some samples below to illustrate the difference. Please download the high-rez files at the bottom of the page for closer inspection.

CV21:

21 Lux:

And from the same position, the CV21 focused to 50cm:

I’m not going to get too deep into lens rendering characteristics, as it seems to be quite subjective.

That said, there appear to be some Cosina/Voigtländer familial qualities in this lens. Sharpness is good, if perhaps micro contrast isn’t as strong as Zeiss and some of the newer Leica lenses. Combined with seemingly good, if perhaps slightly moderate contrast, the feeling I have from this lens is that of being somewhat gentle. Colors appear to be pleasing, with good richness (as much as can be gleaned from outdoor winter testing).

Without a doubt, this lens compares favorably to the others tested here, in terms of overall look. I was not left with any obvious impression of image quality faults.

Where I’m perhaps least happy with it is the lens hood design. I’m not a fan of ‘large’ permanent hoods, and Cosina seems to be enamored with these, first appearing in the M-mount versions of the 12mm and 15mm lenses. The permanent hood on the 12mm makes it huge compared to the previous thread mount version. With the 21, the flower petal-style hood is very deep on the top and bottom, I suppose ideal for high-noon tropical sunlight photography with strong reflections off the water and white sand beaches underfoot. However, the extremely shallow side petals are virtually useless for low winter sunlight in more extreme latitudes. Perhaps the greatest benefit of the hood is keeping fingers off the front element while prying the lens out of one’s camera bag. The front of the hood petals are flat enough to allow placing the lens face down on a table, but it does feel precarious.

Users of mirrorless cameras will not like the hood if intending to shoot landscapes with graduated filters. Obviously, the hood prevents the use of such filter systems, without permanent surgery.

The relatively common 58mm filter size means finding suitable ND filters, which will be needed if wishing to shoot wide open in full sunlight, won’t be an expensive or arduous task. It’s worth noting that one of the few items in the instructions is a tip for how to remove a stuck filter, because the hood petals prevent a good grasp on the filter ring. For this lens then, if a filter is desired, it’s likely a good idea to invest in a brass ringed filter (such as from B+W) to prevent potential filter thread binding.

As with the Lux, the CV21 is challenged by purple fringing at wide apertures in areas with sharp transitions from high brightness and also exhibits a fair amount of CA, though not as severe as the Lux.

While creating the CV21 product photos shown here, I was impressed by how well the front element’s lens coating minimizes light reflection. And the lens appears to be somewhat less susceptible to strong in-frame point light lens flare. While I haven’t used it enough, and I’m not entirely happy with my lens flare testing (most of which isn’t published here), my impression is that it’s better than the 21 Lux at most flare suppression and likely also better than the ZM21. In certain situations it might also be better than the 21 SEM.

As can be seen below, the CV21 performs very well with a strong light source in the frame. In order of appearance: CV21, 21 Lux, ZM21 and 21SEM. I should note I’m not 100% sure the flare in the ZM and SEM images isn’t from dust on each lens…

———-

Zeiss Biogon T* 21mm f/2.8 ZM

This was one of my first lenses for the M9 when I tentatively started with the system back in late 2010. Therefore, while I did use this lens in all of the four-way comparison test shots, I’m also relying on a fair amount of practical experience in my commentary.

It’s fairly compact, if a bit long, compared to ‘typical’ M-mount lenses, with good build quality, focus ring feel and very positive (probably the best) aperture click-stops, in 1/3 increments moving a 10-blade aperture. Coming from Canon EOS SLRs and DSLRs, I was quite enamored with the 1/3 f-stop increments, having relied extensively on this when shooting transparency film prior to digital. That said, on the M9 body, with half-stop shutter speed increments, 1/3 steps add a slightly confusing additional layer that in practicality is unnecessary. The M9′s DNG files are quite robust against underexposure (at least at lower ISOs), where a half-stop adjustment in post is not an issue.

Like the CV21, this lens is ‘non-coded’ and seems to work pretty well with the 21mm Elmarit non-ASPH (11134) code, if only to retain the correct focal length in EXIF. It’s possible one of the 28mm options that worked well for the CV21 will also apply here, but I haven’t tested it.

Optically, this lens is quite sharp centrally from wide open with overall high contrast and micro contrast. There is some field curvature away from the camera at the edges, meaning across-frame sharpness of flat subjects requires stopping down past f/5.6 for best results. My feeling is it’s better optimized for far distance use, but is still fine for sub-2m distances. Point light sources towards the edges will reveal strong coma smearing wide open and appears to be gone by f/5.6.

While conducting the tests for this review, I noticed that my copy’s images along the left side are not quite as sharp as the right side until past f/5.6. I borrowed a demo copy from Rob Skeoch, a Canadian Zeiss dealer at the Rangefinder Store, and it appeared to have better wide open edge performance, but at the consequence of poorer central sharpness. This could also be a recent rangefinder optimization consideration by Zeiss to moderate the effects of field curvature, but is only a guess. It would be necessary to try a few more copies to get a better feel for this.

At a similar price point to the CV21, which is the better choice? If it was 2010 again and both were available when I was starting with the M9, I likely would have chosen the CV because of the extra stop and the benefit that brings to available light, indoor photography. But the Zeiss is a more compact, lighter lens, with those subjective Zeiss image characteristics desired by some. I frequently used it for indoor work wide open, typically where across the frame perfection was not needed, as well as for outdoor ‘landscape’ type use, often stopped down to f/8 or f/11, and was very happy with the results.

As can be seen in the comparison samples, its angle of view is slightly narrower than all the other lenses. So perhaps it’s really 22mm? And oddly, the demo copy seems to exhibit even slightly narrower angle of view than my copy.

Of all the lenses here, it has the least CA; practically non-existent in normal use. There is some slight purple fringing wide open around very bright light sources from time to time, but nothing like the 21 Lux or CV21. Distortion is also of the wavy mustache-type, though not as strong as the 21 Lux. Again, in many practical use applications, it was not an issue.

Flare suppression is good with the sun in the frame, but can be an issue with the sun out of the frame, resulting at times in a massive red splotch in the central image area (see above). The hood is optional, in the $100 range, and considering how shallow it is, of questionable value. The demo lens I tested came with the hood and I’m not a fan. At very close focusing distance it seems to interfere with the M9′s rangefinder. At 1.2m focusing distance the edge of the hood begins to creep into the rangefinder spot and only gets worse at closer distances. I’m guessing Zeiss/Cosina designed the hood with the Zeiss Ikon camera in consideration rather than Leica Ms.

Speaking of close focusing, as with the CV21, the ZM will focus without rangefinder coupling down to 50cm.

Looking at the optical design, it’s the only one that appears to be a hold-over from the film era where challenges specific to digital sensors and how they respond to steep oblique light rays towards the image circle’s periphery were not at the top of the design considerations list. As a result, prior to M9 firmware 1.176, images from this lens required CornerFix correction, meaning an additional post processing stage. This aspect is also a factor if used on APS-C mirrorless cameras with thick sensor toppings (such as anti-aliasing and IR blocking filters), the consequences of which can be seen as purple fringing and poorer edge detail resolution in the NEX-7 test images.

One ‘feature’ of this lens that irks me a lot is the silver front filter/hood mounting ring. While cosmetically it looks nice, it’s a major irritant if shooting through glass. Not that it happens often… While the sample image below was shot with a Zeiss ZM 35mm f/2.8 lens, the same problem has arisen with the ZM21:

My solution is to mount a cheap, black, 46-49mm step up ring. I suppose a filter would also resolve this issue, though I prefer to avoid filters unless necessary.

So who should buy it? It’s roughly tied with the CV in price, though the optional hood adds a bit to the total. I guess if one desires its size, price and prefers Zeiss image characteristics. It’s a versatile lens, and very good across the frame past f/5.6, rivaling the 21 SEM, even if I’m not ravingly enthusiastic about it here.

Leica Super-Elmar-M 21mm f/3.4 ASPH.

As Zeissy a Leica lens can be, while still being a Leica?

Typical top-notch design and handling with image quality to match. This is probably the lens with the least image quality compromises of the group, if not looking for low light use or trying for the greatest possible subject/background isolation.

Sharpness is really impressive from wide open, across the frame, at all distances, with high micro contrast but not too excessive global contrast. Colors are also excellent in richness. As per Leica’s marketing literature, stopping down only really serves to increase depth of field and reduces vignetting. Diffraction induced image degradation starts to become evident past f/5.6, though really not a serious issue (on the M9) until f/16. Of all the lenses here, it has the flattest plane of focus with the best sharpness across the plane. The challenges of placing a subject off center and getting good sharpness with the 21 Lux are not an issue with this lens.

There are some tradeoffs, including wavy mustache-type distortion, though not as strong as the 21 Lux. There is also a touch of visible CA, if one really looks for it. Both clean up in post, if desired. Perhaps another consequence of the impressive sharpness and possibly the very flat plane of focus, is that focus transition from sharp to soft can sometimes feel a bit… unnatural or nervous?

Flare resistance is good with the sun in or just outside the frame, but there can be some flare and will depend on the circumstances. The included hood is also a square design, like the 21 Lux, and seems a bit deeper, perhaps due to the smaller front element (Leica-standard 46mm filter), and is effective at blocking direct out of frame light coming at a steep angle. The aperture, with half-stop steps to f/16 is 9-blade, the lowest of the bunch, but also in a lens that likely will have the lowest user expectation for rendering of out of focus background elements. Point light sources towards the edges will reveal very weak coma smearing wide open and appears to be gone a half-stop down at f/4.0.

Likely the closest rangefinder equivalent for optical performance, and with much lower distortion, is the Zeiss C Biogon T* 21mm f/4.5 ZM, which unfortunately wasn’t available from the Rangefinder Store for comparison. It’s also a lens that has severe edge color shift and vignetting issues on the M9, but perhaps there is hope on the new M?

Buy the 21 SEM if you want absolute image quality through the full aperture range and don’t need faster than f/3.4.


Note, distortion correction was applied for the above image in Lightroom 4 using the included profile for this lens.

———-

Full-rez M9 downloads:

Click on each image to download the zipped folder of images. Most sets include images for the full aperture range from wide open through f/11, in one-stop increments. Beyond f/11 there really is no improvement in image quality, and in fact, image degrading diffraction effects become quite noticeable. The file names indicate the lens used, as well as the f/stop. Because the M9 does not know the actual aperture setting used due to not having any connection to the lens that reports this information, the embedded EXIF aperture value is unreliable – it’s an educated guess by the camera. All image ranges are in the order of CV21, 21 Lux, ZM21 and 21 SEM, starting from wide open. If a lens starts with an in-between aperture value, such as f/1.8, the next image will be the next ‘full’ value, such as f/2.0, to facilitate direct comparison against other lenses with the same aperture value.

All images were shot in the camera’s RAW capture mode and converted with Adobe Lightroom 4, at default settings. In most cases, white balance was set to the same setting for all conversions of a given scene, meaning, there may be some tint shifts from lens to lens. In some cases files were processed to taste with some curves adjustments. But, in no case was clarity used on any of these downloads so as not to distort background/foreground rendering characteristics. No distortion or CA corrections were applied in order to show each lens’s true performance in given situations.

These files are hosted on Google Drive and will redirect to another page to begin the download process. Also note that some of these are quite large!

Due to changing winter weather conditions and the time of day some of these were made, it was impossible to avoid changes in light quality. Please try to ignore this when evaluating certain lens performance aspects.

A park scene with plenty of tree branches (though not back-lit) to challenge lens edge sharpness (~215MB):

Cityscape at dusk. All lenses were set at the infinity hard stop. Note also the coma characteristics of pin-point light sources throughout the aperture range of each lens, as well as point light starburst characteristics (~135MB):

Close up scene 1, at about 70cm shooting distance. Note, there is one wide open defocused image for both the CV21 and ZM21 to give an indication of how much more the background for this scene will be blurred if shot at 50cm instead (~113MB):

Close up scene 2, also at 70cm. Note, there is one wide open defocused image for both the CV21 and ZM21 to give an indication of how much more the background for this scene will be blurred if shot at 50cm instead (~85MB):

Close up scene 3, also at 70cm (~73MB):

———-

Performance on APS-C mirrorless Sony NEX-7 and Ricoh GXR with M-mount module

If the M9 tests these lenses for across the frame performance, then the Sony NEX-7, with its high pixel density 24MP sensor stresses these lenses for central sharpness. But by f/4, all of these lenses perform very well on the NEX. It’s at the image edges where the characteristics of each lens, coupled with the demands of an APS-C sensor with standard, non-rangefinder optimized sensor toppings become evident.

The best sharpness at farther distances appears to be the CV21, followed closely by the 21 SEM that just lags a bit at the edges (and might be a hair sharper centrally). The 21 Lux is a hair less sharp than the ZM21 centrally and is the worst, by far, at the edges, probably due to its unique FF mid-zone characteristics explained above, which happen to fall at about the edge-zone on APS-C. Past f/4, some of the lenses reveal slight diffraction degradation.

Where the NEX-7 seems to suffer the most is edge color shift issues. All of these lenses induce some degree of purple/magenta fringing, particularly with more distant scenes, where the rear lens element is closest to the sensor. The CV21 and 21 Lux appear to be the best, with the 21 SEM and ZM21 trailing.

The first image below is the CV21 on the NEX-7; the second image is the CV21 on the GXR:

The picture is slightly different on the Ricoh GXR’s M-mount module. Its sensor is rangefinder lens optimized by omitting the anti-aliasing filter. This results in very sharp-looking images, but also a fair amount of jaggies along diagonal lines and false color artifacts. Edge color shift is not as noticeable as with the NEX-7, but it exists as a cool blue tint, which seems to better blend into most images, rather than the annoyingly obvious purple/magenta tinge in the NEX-7 images.

The best GXR performer, in my opinion, is the 21 SEM, which is sharp across the frame from wide open. The CV21 is pretty close, lagging just a touch at the edges at equivalent apertures until past f/4 where it’s pretty much a tie. The 21 Lux is the worst at the edges, needing past f/5.6.

An interesting comparison between the NEX and GXR is to resize the NEX files to the same pixel image size as the GXR. Depending on the degree of sharpening during conversion, the NEX files can really stand out with excellent sharpness in comparison, and with very smooth, natural looking diagonals compared to the GXR’s jaggies.

But for the most hassle-free use of rangefinder lenses on a mirrorless camera system, the GXR is tough to beat.

UPDATE:

I did a few close range shots with the CV21 on the NEX-7, downloads are below, since it’s a new lens to me and I was curious, as I’m sure many are about it’s close range performance. It seems to be pretty sharp centrally wide open at 50cm, though I have the feeling the images are a bit low in contrast. Processing to taste, with some contrast and clarity adjustments will compensate, though also affect overall rendering characteristics. My impression is also that field curvature is pretty strong at very close distances, where the corners start to come into focus in certain situations with near backgrounds – and this is on the smaller APS-C sensor too. I can only imagine what it will be like on a full frame sensor, such as the new Leica M. Where the subject is at 50cm and the background is 5-10m away, background blur is nice. If focus is in the 1-1.5m range, background definition is quite strong, meaning you’re able to pretty clearly see everything. Though it is out of focus, the feeling isn’t of really smooth blur. See the photo of the silver car where you can pretty clearly make out the car traveling through the intersection and the houses in the background. In that image you can also make out some purple shift in the tint of the car’s finish along the left edge of the frame. When or if I’m able to get ahold of the Sony/Zeiss 24mm f/1.8 for NEX, I will compare it against this lens. An aspect of interest to me is to see how the Sonnar design of the 24 compares against the 21/1.8.

Full-rez NEX-7 and Ricoh GXR downloads:

Click on each image to download the zipped folder of images. Most sets include images for the full aperture range from wide open through f/8, in one-stop increments. Most of these lenses peak in absolute sharpness between f/4 and f/5.6, at least on the NEX-7, and since there really is no overall improvement in image quality beyond f/8, samples from higher aperture values where image degrading diffraction effects become quite noticeable, are not included. The file names indicate the lens used, as well as the f/stop. Because the M9 does not know the actual aperture setting used due to not having any connection to the lens that reports this information, the embedded EXIF aperture value is unreliable – it’s an educated guess by the camera. All image ranges are in the order of CV21, 21 Lux, ZM21 and 21 SEM, starting from wide open. If a lens starts with an in-between aperture value, such as f/1.8, the next image will be the next ‘full’ value, such as f/2.0, to facilitate direct comparison against other lenses with the same aperture value.

All images were shot in the camera’s RAW capture mode and converted with Adobe Lightroom 4, at default settings. In most cases, white balance was set to the same setting for all conversions of a given scene, meaning, there may be some tint shifts from lens to lens. In some cases files were processed to taste with some curves adjustments. But, in no case was clarity used on any of these downloads so as not to distort background/foreground rendering characteristics. No distortion or CA corrections were applied in order to show each lens’s true performance in given situations.

These files are hosted on Google Drive and will redirect to another page to begin the download process. Also note that some of these are quite large!

NEX-7:

A park scene with plenty of tree branches (though not back-lit) to challenge lens edge sharpness (~189MB):

A light colored building shot wide open and at f/8 to show off some of the edge color shift, as well as across frame sharpness. Also included are shots wide open and at f/8 of a blank wall shot at the infinity hard stop, where edge color would be the worst (~40MB):

GXR:

A park scene with plenty of tree branches (though not back-lit) to challenge lens edge sharpness and a light colored building shot wide open and at f/8 to show off some of the edge color shift, as well as across frame sharpness. (~138MB):

———-

UPDATE: CV21 on NEX-7 close range photos, between 50cm and about 1.5m, all wide open at f/1.8

Download the high-rez files by clicking on the first image below:

Other images in the close up set:

Plus a few more included in the download not pictured above….

If you’ve found this information of use, please consider sending me a donation via PayPal:





Buy from B&H Photovideo and support this review:

Leica Summilux-M 21mm f/1.4 ASPH.
Voigtländer Ultron Aspherical 21mm f/1.8 (In stock at time of this article!)
Zeiss Biogon T* 21mm f/2.8 ZM (black) or silver
Leica Super-Elmar-M 21mm f/3.4 ASPH.

Or search B&H directly from this link:


B&H Photo - Video - Pro Audio








Live in Canada and are looking for Zeiss lenses? Buy from Rob Skeoch at the Rangefinder Store or his sister site Big Camera Workshops, mention this site and receive a free B+W UV filter with your lens purchase, as well as free shipping (in Canada only).


Copyright © 2014 TechTalk with Ron Scheffler. All Rights Reserved.
No computers were harmed in the 0.277 seconds it took to produce this page.

Designed/Developed by Lloyd Armbrust & hot, fresh, coffee.