Images courtesy respective manufacturers, composited to roughly correct relative size – my samples had to return home before I got a chance to put them together in the studio for th usual product shoot, and I’m still awaiting delivery of my own personal lenses.
I’ve recently had a chance to shoot a) the best two wide angles available at the moment, and b) shoot them against each other on the same camera body. This is not a direct comparison. There are however limitations to the testing – very limited time* and no way to mount one without. Furthermore, the lenses were both final preproduction prototypes, which could mean they are either good samples because they’re hand adjusted…or there’s some variance, because…they’re hand adjusted. Tests were performed on a Sony A7RII body mounted on a Arca-Swiss P0 head and RRS24L tripod – i.e. sturdy – and released via IR remote. The adaptor used was a Metabones Nikon G-NEX model, tested and found to be good with various other lenses including the Zeiss 28 Otus. However, it’s worth noting that the shorter the focal length, the more sensitive a lens is to small skew because only very small movements are needed to change effective focusing distance. I’m sure many other limitations in methodology can be found, but remember we are aiming for the best we can do in field conditions without giving one lens or the other a sensor-based advantage. Observations must therefore be taken as preliminary.
*Literally, about an hour after dark during a recent visit to Sigma HQ in Aizu, Japan. Crops are 100% where stated; I will not be posting full size images because IP rights sadly don’t seem to mean a thing online.
First test scene. The links under each test scene are 100% crops of A-B comparisons between the lenses.
1. Overall angle of view comparison at f5.6
2. Centers at f5.6
3. Midzone at f5.6
4. Corner and foreground transition at f5.6
5. Centers, f1.4 and f2.8
6. Midzone, f1.4 and f2.8
7. Corner, f1.4 and f2.8
There is also the Nikon AFS 20/1.8G, the Leica 21/1.4 Summilux-M ASPH, and the Zeiss 2.8/21 Distagon ZF.2, both of which many will feel I should have included. I did not for the simple reasons that I did not have either accessible at the time, and in my experience of owning these – the 21 Distagon ZF.2 is superior to the Nikon and the Leica, and the 21 Loxia is superior to the ZF.2. It would not be a fair fight on price, either – the Nikon is the cheapest at $797, the Sigma is close at $899; the Zeiss is next at $1499, and the Leica – a hefty $6,732(!). Time permitting, I do however plan to repeat this comparison in the near future with the Nikon 20/1.8G once I have access to another 20 Art – since they are much more closely matched in price and specification.
However, I felt it was still an interesting comparison because what we do have here with the 20 Art and 21 Loxia are possibly the two best wide-angle options I’ve used in the last couple of years; both benefit from the latest optical design and computational advantages and represent two very different philosophies. The 21 Loxia is small, petite and manual focus; it pairs well size-wise with the Sony bodies. The 20 Art is a bit of a monster, but not much larger than any of the other Art series lenses and still considerably smaller than say the 28 Otus. It balances well on a D800-sized body and of course also has autofocus, with the ability for very precise fine tuning using the optional USB dock.
We know it is easier to design a short-flange wideangle than a longer one for reasons of telecentricity; however, telecentricity is still required when angle of incidence vs sensor considerations are taken into account. Once again, we see two different design philosophies at work: the Loxia is a 11/9 design and has a rear element that takes full advantage of the mount’s proximity to the sensor. The Sigma is a 15/11 design and has a bulbous front element that both necessitates a special deep front cap and precludes the use of filters. (I suspect the Lee filter kit for the Nikon 14-24 will fit, though.) The Loxia has a petite 52mm front thread and comes with the ability to declick the aperture for video use, as well as automatically trigger magnification when the focusing ring is turned.
Second test scene.
8. Center, maximum aperture
9. Edge 1, maximum aperture
10. Edge 2, maximum aperture
11. Corner, maximum aperture
Both lenses are well built, though the Sigma is not weather sealed (and the Loxia has limited sealing at the mount) – I would be somewhat careful with both when it comes to moisture. The Sigma is mixed metal-plastic, and the Zeiss appears to be mostly metal which also accounts for its density. There is of course a significant difference in weight: ~400g for the Loxia, ~950g for the Sigma. Part of this is due to the changes in optical design required to accommodate longer flange distances and the AF motor innards, but most of it is because there is a significant different in glass requirements to both collect light and correct for aberrations at f1.4 vs f2.8.
Let us first consider the 21 Loxia: it is already remarkably sharp in the center at maximum aperture, and taking into account some field curvature, the edges, too. By f5.6 it is very difficult to find fault with anywhere in the frame – longitudinal and lateral chromatic aberration are very well controlled and there is little evidence of smearing in the corners. I wouldn’t call it apochromatic, but it’s really not far off. Colors are deep and saturated, and in the typical Zeiss manner, microcontrast simply pops.
It is therefore more remarkable that the Sigma 20 comes very close to the Loxia in the center 50% of the frame at f1.4, a full two stops up – the corners also exhibit noticeable field curvature and softness, though this goes away soon and the two lenses are on par at f2.8. The Sigma also has excellent chromatic aberration control, but slightly more corner smearing than the Zeiss; the transitions aren’t quite as abrupt which makes for a smoother overall rendition but a little less bite. That said, the differences are pretty subtle here. By f5.6-8, the lenses are nearly indistinguishable.
Third test scene
12. Center 1, f2.8
13. Center 2, f2.8
14. Corner, f2.8
15. Spot highlight and CA handling, f2.8
Note transmission differences between corner and centre – some strange vignetting is going on…
There is only one trait with the 21 Loxia that I found undesirable compared to the Sigma: much heavier vignetting, despite the slower aperture. I suspect this was a consequence of keeping the overall size small and the front element relatively flat. The Sigma has much more even illumination but also slightly lower overall T-stop than the Zeiss, though this may be due to more elements/ coating differences, or the calibration of the Nikon G-adaptor. In practice, there is little to choose between them in terms of resolving power or distortion – though further testing is required to determine if field curvature is an issue or not, and to what degree.
I found myself coming away from the brief test extremely impressed by both lenses, but for different reasons. The Zeiss has the level of clarity we have come to expect from their best lenses, and is a perfect partner for the whole philosophy of ‘smaller, lighter, no optical compromises’ – it is a little jewel of a lens, and built to last a lifetime. It takes filters, and is my choice when working stopped down on a tripod or if weight is a priority. The Sigma comes very close to matching the performance of the Zeiss when stopped down, but its strength is really in offering two more very usable stops of shooting envelope, autofocus, and the possibility of tuning focus extremely precisely – all of which make it perfect for run-and-gun available light work where the Q isn’t wide enough. I do quite a bit of this kind of photography for corporate documentary and under dim conditions, which makes the Sigma very handy to have.
In hindsight, I think ‘battle’ may have been the wrong term to use: it really isn’t one. What we have are two of the best wide angle lenses I’ve yet seen, period. Both have tradeoffs, but they’re sensible; there is therefore no clear winner here – you may value the 21 Loxia’s size over the Sigma’s speed, or autofocus over the ability to take filters. Or field curvature over corner resolution – etc. Despite being similar in focal length, the 21 Loxia and 20 Art are an excellent example of where lenses are clearly built for very different application profiles – and still capable of achieving the same high standard of performance. There are situations in which I strongly prefer the Loxia over the Sigma, and vice versa – fortunately, these almost never overlap. For the first time in a comparison, my choice here is actually very simple: I have reasons enough to have both. MT
The Sigma 20 Art is available here in Nikon, Canon and Sigma mounts (B&H, Amazon) and the Zeiss 21 Loxia is available here in Sony FE mount (B&H).
__________________
Visit the Teaching Store to up your photographic game – including workshop and Photoshop Workflow videos and the customized Email School of Photography; or go mobile with the Photography Compendium for iPad. You can also get your gear from B&H and Amazon. Prices are the same as normal, however a small portion of your purchase value is referred back to me. Thanks!
Don’t forget to like us on Facebook and join the reader Flickr group!
Images and content copyright Ming Thein | mingthein.com 2012 onwards. All rights reserved
Ming Thein | Photographer 
I know you’ve moved on from Nikon mostly, but please share your thoughts on best-choice 20-ish mm for a Nikon D810. Based on the comparisons on your site and what you own, Sigma looks like a (the) winner (since the Loxia beats the Distagon and Sigma ~ matches the Loxia), but maybe there’s something you haven’t yet shared? I’m a bit concerned about sample variation on the Sigma; I don’t enjoy repeatedly returning samples.
On your images here, at the same aperture, I find myself preferring the Zeiss consistently, but I wonder if it’s only because the Zeiss is a little brighter? And in the 1.4/2.8 comparisons, the brightness seems similar. Your thoughts?
Nope, no secret sauce: I’m using the Sigma 20mm too. Not much sample variation that I can see.
Brightness should be similar since exposure was adjusted to accurately reflect aperture (otherwise my meter would be broken!) 🙂
Wait–that’s the Zeiss vignetting I’m seeing, isn’t it? Makes the center of the image seem brighter. I like it.
All modern Zeiss lenses prioritise resolution/microcontrast and aberration correction over vignetting – that’s the easiest to correct in post with the least effect on image quality…
Just discovered your site–wonderful! Thanks! I have two questions, only one of which is directly relevant to the review. I’ll start with that: Based on your experiences, would you expect the results to be essentially the same on a Nikon body? Speaking of the Sigma obviously.
Next: I love your first test scene. It has a painterly quality that I strive for and rarely achieve but greatly admire. Any thoughts, beyond what you’ve already said (tripod, head, remote) that you can share? Since you didn’t mention it, can I assume the lighting entirely ambient? Because somehow it seems perfect.
Thanks.
Results are actually a bit better on the Nikon because you don’t have adaptor planarity issues (magnified with wide angle lenses) or a sensor with filter array designed for much shorter back focus lenses (which can result in a whole load o other problems).
Lighting was ambient, yes. But exposure and post processing are pretty critical too, and I cover that in Outstanding Images Ep. 4/5 and Photoshop/LR Workflow III.
Hey man! I’ve been really enjoying your reviews recently (I just jumped back in the digital game and there is just so much out there!), so thank you for that.
I had one question in regards to the Loxia. From what I’ve read it does not have an auto-diaphragm, meaning that when you close down on the lens the aperture closes right away (not when the shutter is tripped). Doesn’t that make it much harder to quickly focus at, say f/11 or so as so much is in focus anyways that’s it’s hard to see where it falls off (am I making sense), and when it gets dark wouldn’t the EVF get very dim (especially if one would be using a flash and shooting at f/16 as I do quite often)?
Thank you!!
It isn’t auto diaphragm.
Is it actually not? I thought it was “diaphragm by wire” [and focus by wire] like the Fuji X’s?
No, you’re thinking of the Batis – those are all ‘by wire’. The Loxias are 100% mechanical.
As to the Zeiss Loxia 21mm lens NOT having an automatic diaphragm, I was very disappointed.
Since this lens was specifically designed for the mirrorless Sony cameras, I don’t understand Zeiss’s thinking!
(For a rangefinder camera < sure. No need for an automatic diaphragm.) Surely at a price of $1500 they weren't
worrying about keeping cost down!
Any idea about coma performance of the Loxia 21mm at f2.8 ?
If I buy one, and this seems close to me I would like to use it for astrophotography as well.
Didn’t find any comment about this in the review I encountered until now.
Thanks if any one could share these findings.
Not had a chance to use it for that purpose. In urban settings, there is some but it’s not too bad – Lloyd Chambers has tested for that though, I believe.
Hi Ming,
Maximum aperture (for the loxia) here ?
Are you sure ? Because there are still stars. Surprising was fully open ? No ?
Yes, f2.8 can still make stars. Aperture was fully open, I checked the EXIF data.
Ok, thank you for checking
Diaphragm may be a little off to f/2.8 ?
Nope, looks wide open at 2.8 to me.
OK
I thought it was amazing, because these are the blades that generate the stars normally.
There are still 10 points of the 10 blades.
This is an enigma 😉
After being delivered my copy… I confirm the presence of the diaphragm blades to f/2.8. Explanation of the starburst effect. 😉
Hi Ming
Thanx for all the hard work and write-ups, especially now that new gear has become a virtual avalanche.
Great lens for multi-day backpacking and landscape trips.
I would be interested in seeing the quality of the sun-stars you get from the loxia at f16 and f22. I occasionally make use of them in sunrise/set landscape images, so would be interested in this.
Thanx
Karl
Thanx Ming for all the ongoing hard work on these lens tests and write ups.
Definitely Loxia for a backpacking landscape lens.
If you still have access to the 21 loxia, any chance of shooting sun stars at f16 and f22? For sunrise sun stars good spikes are important for me. I don’t often shoot sunrises like this, but the occasional shot always makes for a impactful image.
Karl
I have a lack of sun problem at the moment…but intense point sources are all similar. Here’s one I found in the archive:
Hi Ming
Thanx that will work as well. They look pretty good.
As a Nikon 1.4G series user, I have major sunstar envy!
I thought the 24 was okay?
It’s the best of the bunch (24/58/85), but not as satisfyingly pointy as that Loxia. I don’t like to stop down past f/11 with the D800/810, which doesn’t help. Ashamed to admit I once resorted to Photoshopping in a more pleasing sunstar from the Coolpix A in one shot that needed it :p
The compact leaf shutter cameras seem to make much nicer sunstars in general; no idea why that might be.
Nice article. For me it’s 100% Loxia, since I bought this camera for its transportability and handling in the first place. Loxia is the only line (IMO) that fits in that concept perfectly.
We need to be aware that, in your comparison picture of the 2 lenses, one in fact yet needs to add the big adapter to the Sigma lens.
Yes, and the Sigma is not really ideal for this platform. Yet we cannot compare it to the Loxia on anything else, because the Loxia will not mount on anything else…
Correct. In fact that’s why it’s Loxia for me. I have definitely chosen this platform, and choose the lenses that are most appropriate to go with it. By choosing this platform and limiting myself to it, I know that certain “jobs” will be less suitable for me. I’m in peace with that, not being in your situation. But once this decision is made, there is no reason anymore why not to go for dedicated FE-mount lenses. The fact that Loxia is mounting on this platform only, is no longer a con then.
Wow not the results I expected. I viewed the comparisons blind and was thinking all along “the Loxia is standing up pretty well against the Art lens it’s not that much worse, not bad for a mirrorless lens” but for most of them I had it the wrong way around and was actually preferring the Loxia.
Clearly at 2.8 (and 1.4 obviously) the Sigma lens has the advantage, but above that I picked the Loxia every time. Very interesting comparison and perhaps it’s time I took a more serious look at a mirrorless camera…
It’s actually easier to design a wide with a short back flange distance (i.e. no mirror to clear) – and one that’s slower, too. So I would view it as: impressive that the Loxia does that well, period, and more impressive that the Sigma matches it and is two stops faster. The only disadvantage of the Loxia is that you can only use it on a Sony body – the flange distance is so short it won’t mount to anything else even though it is fully mechanical. This may or may not be a longer term concern.
When you tested the Zeiss Distagon, you were already asked for a comparison with the Nikkor. Your reply: „No straightforward answer here, I’m afraid. Both lenses are excellent stopped down and VG-excellent wide open. The Zeiss is a bit more consistent cross-frame than the Nikon, but 1-1/3 stop slower and lacking AF. I do more documentary than landscape work with my wides, so I switched to the Nikon. I’m not sure I would have done if it was only deployed static on a tripod.”
Yes, and the Loxia is better than the ZF.2 Distagon. I’m not sure what the question was…?
I am sure it is a great lens but what will happen 5 years from now if Sony abadons the mount just like it did with the A mount. Pity all the outstanding A lenses such as the 135 f1.8. I am talking about mot using any type of adapters.
A very good question: and since the E mount is so short-flange, the lens cannot easily be adapted, either.
Sony makes an A-mount to E-mount AF adapter. Looks on the big and cumbersome side though. Rather unappealing.
I think that’s because the adaptors also include a translucent mirror and an AF sensor array, or at very least a mechanism to operate the auto-aperture?
It also has an AF drive motor in there for lenses like the 135 f1.8. So yeah. mirror, AF sensor, aperture lever, AF motor – no wonder it’s a bit chunky.
That would explain it. Aren’t there two versions though? One is smaller and I believe is missing some of the internal bits.
I’m not sure. That’s the only one I know of is a friend’s, who has one for his 135mm.
Thank you for this great comparison, I read it since I’m a big fan of the Loxia’s on the Sony A7 family. At 21mm I’m currently using a Zeiss ZE 21mm f/2.8 on both my Canon cameras and adapted to the A7RII. My current plan is to switch over to the Loxia for its smaller size but I was wondering if you could comment on how I might expect it to compare to the my current ZE distagon? Do you have any experience with either the ZE or ZF version to be able to comment?
The ZE and ZF 21 Distagons are the same; as I mentioned in the second paragraph, the Loxia is superior.
Realize that the ZE and ZF are the same. My poor reading missed your comment in the second paragraph where you think the Loxia is better than the ZF.2 version of the 21/2.8. Any ability to quantify the improvements of the Loxia image quality over the ZF.2. Size and features I understand. IQ is something that I would like the opinion of a better trained individual like yourself.
Thanks for the follow up and thanks in advance for taking the time to respond if you do.
No problem. I’m seeing the differences mainly in the corners – the field seems a bit flatter to me, distortion is a little lower, and there’s less smearing of detail. Lateral CA is also much better controlled.
Dear Ming,
Thanks for fantastic fast review,
i just my copy of the ART 20 mm , i was lucky this time since the lens is very good @ 1.4 and super sharp edge to edge @ 5.6.
AF USB calibration was super easy and fast . ( Values 10,11,11,12 on D810 )
Looking at your picture i have to agree with on line reviewers claiming performance of adopted lens on the Sony A7RMK II seems to suffer, i will be happy to see a second round of testing for the ART 20 mm once you get you copy on D810.
i am curious if you will get better results on Nikon mount.
One more issue i wanted to share , i owned all the ART series lens, sadly i ended up keep only the 35 mm & 50 mm ( and now the 20 mm).
I am a big fun of the Optical performance and value but not of the QC and QA.
let be clear , My D810 works perfect with Nikon glass yet no sigma was usable out of the box.
The buying of the ART lens is a long process since you end up doing the QC of the lens yourself and in most cases the image quality is far from perfect no matter what you do.
AF calibration has 50% chance of success. ( I am doing lab measurements all my life , so i know how to do it, the inconsistency in the lens).
I would really hope Sigma will improve there QC ,My guess this is not bug in the design but in the production.
I would appreciate your comments , i am sure your experience is bigger than mine.
Thanks again for all the fantastic work you share with us.
Yair
There are always compromises with adaptors – you’re introducing an extra pair of mount surfaces which can affect centring, skew and overall planarity. Results will likely be better still on the D810, but we then cannot make a meaningful comparison with the 21 since it won’t mount on anything but the Sony, so this test would not be repeatable…
I have limited experience with the Art lenses since the other FLs are not so useful for me – I either have solutions already (24, 50) and don’t see the need to buy more lenses, or don’t have a use for them (24-35, 35).
Thanks for the fast reply.
So in this comparison the Loxia get an advantage over the ART which still it fights back not bad at all @ 900 $ compared to 1500$.
ATB
Yair
Great review Ming. Be interested to know your thoughts on the Batis 25mm in comparison to the Loxia if you’ve had a chance to test the former. Best, Nick
Only briefly, and I no longer have access to one. It also wouldn’t make sense as 25mm is a very different field of view to 21mm – and I wouldn’t interchange them in practice either.
Thanks Ming for great review. Although I am a loyal Nikon user with D750 and all nikon lenses, plus a back up Fujifilm x-e2 and a fuji lens for street photography bougt recently, I am watching the Sony alpha a7r series closely. And I am glad sony alpha system are filling with compact and quality lenses, so that in future if Sony a7r camera evolve in such a way, I am certainly buy one.
I always wish in future, the lense makers can produce the series of lenses of f2.8 prime (no need for f1.8 or f1.4) which are good right at its wide open, and light, compact like this lens, to match the body of mirrorless camera, being light and small for travel, hiking. Now the advantages off the body from MIC are nil by the hefty and heavy lenses which are of f1.4 but always been stopped down to f5.6 to achieve best performance.
I’ve been told again and again f2.8 is ‘not sexy enough for consumers’…or perhaps not sexy enough for the marketing people?
Ugh! That’s such a misunderstanding on their part. Quality is sexy at any speed. Today I’ve been printing some 3 year old images shot with the Leica 24mm/3.8 Elamar ASPH on my M9-P. That’s a sexy lens, small and super high quality. It misbehaved a bit with color casts on my M9-P but it’s correctable. On the M-P it’s even better. Now if Leica would only make me a 36-42MP M camera, that would be really camera-porn sexy. 😉
Ugh! I meant Elmar, Leica and their silly names + my butter fingers!
I suspect a lot of their lenses would not hold up at that level.
I’m sure that’s true of most of the 20th century Leica designs, though some of the newer ones like my 24mm/3.8 APH, 28mm/2.8 ASPH, and 75mm/2 ASPH, and some newer ones I don’t own, might benefit from more resolution. I think some ZM lenses like the Zeiss 35mm/2.8 and 35mm/1.4 would definitely work on higher res sensors, as we’re not talking about huge increases linear resolution when you go from 24MP to 36MP or 42MP, but 124% or 134% of the current Leica M’s resolution. Small, but significant, increases. But resolution is the least of the usability issues with the M 240 series. The live view and EV accessory are really not good enough, sluggish and blurry, and with magnification only available in the center. Not that good for tripod work. I like the RF and optical finder for fast work, though I’d prefer a wideangle 0.58X finder, like my M7 has, but that’s not available in their digital bodies.
I have no doubt we’ll be okay in the centre most of the time, but the edges and corners are likely to leave quite a lot to be desired. The thing is, Leica has been getting away with it for quite a while because their users are either a) centre focus and bokeh, and so corners don’t matter, or b) limited by shot discipline, or c) stopping down massively. There are lenses that work fine on the D750 (24MP) and start to show weaknesses on the D810 (36MP), so there’s definitely a threshold here…
To be fair, Leica has had a high resolution sensor available for the M mount for quite a while, namely ever since the first Monochrom was released…. and there was not a whole lot of complaint (from what I can recall – correct me if I am mistaken) over the quality of the modern Leica lenses (leaving out large aperture UWAs like the 21mm Summilux ASPH which is obviously not particularly high resolution, because how on earth could it be at that size), which is to say, with most of them, people saw a noticeable boost in resolution going from the M9 and M240 to the Monochrom.
Yes, the result is higher resolution, but it is not the same comparison either. The photosites are the same size as the donor Bayer sensor, but there’s no Bayer interpolation. It does not differentiate the lens’ ability to resolve smaller circles of confusion from higher pixel densities, which is what is in question here.
If Leica does ever make a 36MP sensor we can test our hypotheses. They do seem rather slow at implementing new technology in the M line, so it may be another few years, by which time who knows how much better Nikon, Canon and Sony sensors might become? I suppose Leica AG are afraid of another M5-type debacle, and alienating their fanboys, or a high res sensor showing how some of their vaunted lenses are not really as good as their myths proclaim (the overrated 35mm/2 ASPH comes to mind, from my experience, it wasn’t even that great on film, not bad, just not the ultimate lens that Erwin Puts and others proclaimed it to be, and that was a newish design when I used it). Not that Leica could ever please all of their users and collectors, especially those who actually take pictures, but the M240 is long in the tooth in digital product years, and no improved replacement is announced yet.
Why change something that’s selling when that change might upset the rest of the apple (lens) cart?
It’s like the Nuremberg effect for lenses, it’s very annoying. Film cameras had smaller well made primes in slower speeds that were well regarded not sure what changed. Also annoying is the insistence that small wide primes have zero field curvature at any distance at any apprature, not everything has to be Otus.
Yes – but even Otii are not completely immune from field curvature; it’s slight but present on the 55 with the 5DSR (not visible with the D810) and a little more prominent on the 28.
Dear Ming,
I am very fond of your test reports on cameras and lenses. You deserved a big THANK YOU!
Recently I have tested the Sigma 24/1.4 Art at full aperture on my D810 after fine-tuning the focus on the camera. On some of the pics the performance was excellent, however about half the time the focus was significantly off target. I felt, this lens was a poor sample and did the same test using the Sigma 50/1.4 Art which I own. This lens I had corrected by Sigma-software and the usb-adaptor. At close examination I found this lens also showed some focus-variation (e.g. plus or minus 5cm at a distance of 1.5m). In the past I did not notice this because I only used this lens at full aperture in disastrous light condition and low shutter-speeds. I was wondering if my D810 had a focusing-problem and tested the lens on my D4 – with simiar results!
I then tested a new Sigma 24-35/2.0 Art with similar, yet better results of focussing consistency.
The tests were performed using both continuous focus or single focus, in the latter focussing was triggered three times before the shutter was released. Most of the tests were done in bright sunshine using a solid tripod, the central focussing point, “Mup” function and a cable release.
To conclude: it appear to me high-end Nikon cameras have a serious focussing-problem when using wide-aperture lenses.
It would be interesting to hear your experience and your views on this topic.
With kindest regards,
Felix
To be honest, I’m not surprised – in addition, my cameras also appear to ‘settle’ on different distances depending if you approach from infinity or the near limit. Fast wides are worse because the phase difference (‘transition’) is not as abrupt as with longer lenses, and so more difficult to discern in/out of focus. That said, there’s a remote (because it involves two cameras) chance something is not right with your camera bodies that can’t be solved by the USB tuning and occurs with different lenses on the same body.
However – not fully knowing the conditions not your test, it’s also possible that your focusing target might not cover the full AF area (which extends somewhat outside the boxes) and the camera is picking up something within the actual AF area box at a different distance, but not obviously visible?
Hi Ming,
thanks for your marvelous input on these gems! I need AF and slightly larger aperture 1.8 or 2 would be great….
Hopefully Zeiss is making a new Batis 18 or 20 1.8 or 2 with filter thread (no bulbous element!!!) would be awesome!
So again Ming its your time using your contacts/network tell this Zeiss PLEASE!!! I have already bombed them with mails/messages via social networks (also other friends/photographic academics and enthusiasts did so) in order to make one!!!
Finally, I am so excited to see the new FE Mount lens road map from Sony with 8 new Sony G lenses..hopefully still this year ;)! The Zeiss manufactured lenses are not included on this list, i expect 1 or 2 lens for the batis line and 1 new loxia at least in 2016!!
its Photokina year the camera / lens manufacturer have to show off a bit and impress us customers with new products ;)! They should….but I think Sony is the only one the new apple of photography gear…..innovations costs if a lot R&D is needed but it will make leaders on the longterm view at least….!!
Thanks a bunch & good to have you!! Take care!
Roman
I can tell Zeiss whatever I want til I’m blue in the face, but unfortunately they’ll make whatever they want to make…I doubt they’ll make something so close to an existing FL/aperture combination in a different line though. It would make more sense to see something to fill the gap either on the tele end of the Loxia or the midrange of the Batis so there’s at least one complete lens lineup.
I know it isn’t the point of the article- but 1500.00 is insanity. Take note: The 50mm can now be had for around 750-800 used and I’ve seen the 35mm Loxia go for 1k or slightly less. I don’t expect the 21mm to fall as fast due to lack of popularity. 1500.00 though is highway robbery. With that said, I can’t tell much from your pictures. I do appreciate your work though.
It’s upsetting because we have three lines and neither is complete. We’ve got Zeiss 55 and 35 then what? 28 Sony lens then two Batis lenses and now 3 loxia. It’s like 4 freaking lines of lenses. People want something uniform and if your presenting work to a client it certainly goes a long way.
Sony can learn a thing or two from Fuji and I don’t even shoot Fuji. They know how to build a lens system though.
There’s a big difference between $950 retail falling to $750-800 – 20% off – and $1500 going to the same level. It is significantly harder to design a wide than a normal lens, and I’m sure there are economies of scale at play too. You can tell plenty from the 100% crops; there’s a very high degree of correction at work in both lenses, with different tradeoffs and design philosophies. Sigma will be cheaper simply because it’s amortising the R&D cost across a far greater user base.
I agree Fuji did a good job with filling out the lens line…
Ming, another vote for the 21mm Leica Super Elmar M (SEM) being the best 21mm landscape lens by far here (it outperforms the summilux in every variable except maximum aperture). I agree it’s hard to use it via adaptor, but if the Loxia (or the Sigma) is optically superior to it I’ll eat my hat.
We’ll never know, because as I’ve already explained, there is no way to find out. No M-mount bodies have more than 24MP of resolution so we don’t know if it’s capable of resolving beyond this, and the lens is at a severe disadvantage on E mount especially in the corners because of the lack of microlenses on the sensor and filter stack thickness. Moreover, there’s no way to mount the Loxia to an M mount camera because of the flange distance. There is simply no way of doing a fair comparison nor is it a practical choice in the field.
Certainly there is a way to find out: A7r2 with Kolari mod vs Loxia on A7r2 without. The SEM 21 is very good on my A7.mod and of course it is fantastic on the m9/240.
Okay, send me an A7RII with the Kolari mod and the SEM 21, then.
Ming, I would love to 🙂 I’m not rich either! “We’ll never know”? There is every reason to expect the SEM 21, based on the performance of other modern Leica lenses, both in their charts and in field tests at multiple sites, may match or exceed the Loxia on 42mp. And yes we will know, since the A7r2 body will fall in value quickly and the mod will be done on many bodies.
Not trying to be annoying, and I understand your focus is not on testing, and this site is free, which we all appreciate. But this is a real option, today, why ignore it–or be annoyed if people bring it up? Would you not appreciate use of the small and excellent modern M lenses on a 42 MP sensor which was friendly to them? Just to be free from Sony copy variation seems worth the 500USD cost for the Kolari mod to me (and I am not paid by Kolari).
With respect,
Charlie
I have a better solution. Since Lloyd has both modified and unmodified bodies, and probably access to the 21 SEM lens too, why not suggest he performs the test?
I only get annoyed when people’s expectations become unreasonable. Spending a considerable amount of money on a maybe *does* start to fall into that category…especially when I can’t see the return both commercially and creatively. 🙂
I understand, and while I’d love to see you armed with a Kolari A7r2, it’s unreasonable to “expect” it. I comment only because many mention the SEM 21 vs the Loxia at 42mp and your response was “we will never know”. Usually you are outstanding at listing the various options we might consider, whether you have them in hand or not. Your expertise and real world experience is why your blog is popular, not to mention the fantastic photography. So please add the Kolari option to your “frame” of options today. Since the hue and cry about M lens performance on the A7 cameras has been loud and anguished since the introduction, it’s relevant to many of your fans, and I imagine you would be happy to try one yourself if you had a loaner. 🙂
Anyway it’s great to see there are finally serious UWA options for the A7 series, two years after it’s introduction, and as usual your blog is excellent and original on the topic, thank you 🙂
I’m glad you like sigma art at last, as you knew it, not many your readers can buy zeiss lenses.
So the sigma is a solution for unlucky amateurs, now it will balanced for your readers from now on, thanks.
No, you’re misinterpreting me. I never said I disliked the Sigmas: I said in direct comparison, they’re not as good as the Otuses (but still better than just about everything else). That has not changed. There is no Otus option in this focal length, or any other f1.4 options for that matter.
I was also under the impression that you disliked Sigma (up until your interview with their CEO) since you are about the only popular “review” blog that didn’t have a write-up about the 35, 50 or 24. I suppose the 20 1.4 is just too affordable and exotic an offering to overlook. Sigma is indeed appealing to hobbyists like many of us who visit your site, because we cannot afford to spend nor justify $4-5k on a single lens. These lenses are like a corvette: similar track performance to a ferrari, but without the refinement or cachet, and for a quarter the cost. It’s a compromise for the average hobbyist. Would be a delight to see your thoughts on the other Sigma Art lenses.
First misconception: I’m not a review blog. I’m a working commercial photographer who writes 99% about photography and 1% about the tools I use to make those photographs. Notice how there is no advertising or sponsorships here. What I do is on my own time, editorially independent, and free. This is NOT the same with sites that rely on traffic for profit.
Second misconception: Just because I don’t review something, it isn’t good. That springs from a) not having time or resources to review everything, because there’s frankly nothing in it for me, and b) in order to have a meaningful practical opinion, you actually have to use it to make photographs, not test chart reproductions. It therefore makes no sense to review anything I don’t regularly use for my work. 35 is a non-starter for me, 50 is already covered amply by the 55 Otus, as is 24 by the 24 PCE. There was and remains no use case for me.
Sure, but at least a third of your posts are about gear “battles” or “reviews” or “close but no cigar” cameras or cameras as luxury items. I think it’s fair to say that a lot of your content is about gear, and a lot is about photography as a whole. National geographic’s Proof photo blog is not a review blog (http://proof.nationalgeographic.com), and they do have some ads. But they don’t have picture of cameras or lenses. Even Shoot Tokyo features leica gear, but rarely has reviews or “battles”.
You know what, I’m quite happy to stop reviews entirely if they eliminate comments like this. Because I definitely get nothing out of them. Notice the lack of advertising, the absence of sponsored content, and frankly, what I get back in referrals doesn’t even cover the cost of hosting. I cannot help it if you only comment on gear posts and selectively ignore the rest. Photography is a technical pursuit to a large degree, and if you don’t control the tools, then you also cannot control the outcome.
I smiled when I realized even your hurried “test shots” are beautiful wonderfully composed photographs. 🙂
Thanks – hurried in total duration and scope of test, but not individual images. In any case, it doesn’t take much more effort, and if your images are going to be out there and associated with your name…they’d better be of the desired standard, right? Once online, everything is permanent… 🙂
MT – at the risk of being contrarian, I do not believe that ‘buy both’ is much of a solution for us amateurs: we still have to decide what to carry in the field! Sometimes it is a good thing to have clear preferences in rendering style or manufacturer because it frees us from analytical paralysis so we might get back to the actual photography.
I agree, but for me – tools are tools…
PL – at the risk of stating the obvious, but it is really up to yourself to decide which tools to choose for the job. If certain restraints (money being one of them) force you to work with just one tool, then you yourself need to make the decision as to which tool to choose. Only you know which tool properties are paramount for your photography. Or do you expect Ming to make that decision for you?
I’m not sure I read it that way, Peter. I think the suggestion of information paralysis to the less experienced is a very valid one, and how we land up with odd questions and inappropriate purchases. We can’t all try all options before committing; the internet takes over and uncaveated opinions become canon law – that’s the problem. I know I have the same information paralysis at times and that’s with the benefit of experience…
Fair enough. I may have been reaching. Still, I think people need to make their own purchase decisions. They obviously want to base them on expert opinions (which makes a lot of sense, especially when the expert is an independent one…), but the experts can’t know what is important for each and every individual. At most, an expert will point out the weak and strong points in a review for each separate piece of equipment. The public will have to decide for themselves.
Whether that leads to information paralysis (in the bigger scheme of things) is another thing. Personally, I think information paralysis is many times about that last 1 percent we’re nitpicking about. Information paralysis is the opposite of that same medal: we are way passed the point of sufficiency. Most of it is 99% good. Just roll with it 🙂
I think the confusion actually comes from having too much information of uncertain quality: which opinion has more scientific basis? Which opinion is more applicable to ‘me’? That’s very difficult to answer, because as you say everybody has different needs.
Hey Ming I visit your site daily and appreciate all your hard work and wonderful vision.
Too bad Sigma didn’t loan you their dp0 quattro camera (21mm equivalent) to compair as well. The lens is supposedly fantastic.
The dp2 quattro beat almost all the other cameras in dpreview studio tests for resolution < all the way out to the corners! Nobody seems to notice! (Too much controverse over the poor studio results with the new Sony RX1R II !)
Sigma should make dp series camera lenses in Sony FE mounts, tweeking them for the optical requirements of the Sony mirrorless line. (It seems not many people want to struggle with the computer processing problems from the fovean sensory.)
Keep up the good work
Thanks David. It would not have been a fair comparison at all because then everything would have been different – sensor and workflow too – and resolution differences cannot be solely attributed to lens performance (which is the purpose of this comparison). It would be another bayer vs Foveon test…
Thanks David. It would not have been a fair comparison at all because then everything would have been different – sensor and workflow too – and resolution differences cannot be solely attributed to lens performance (which is the purpose of this comparison). It would be another bayer vs Foveon test…
Ming > the other thought I had was perhaps the Sigma Quattro 0 could possibly give you the same or better results then your Pentax 645Z (with the 25mm lens). Since, as you said the camreas are tools, perhaps specific tools for specific jobs are appropriate.
It would be easier to carry around the Sigma 0 then the Pentax, although still a tripod camera but seems to be how you usually use the Pentax 645.
I sold the Pentax more than six months ago, actually.
The Quattros do not out resolve the 645, though the 25 is quite weak so it’s possible in that particular case. In general, they’re about on par with the D810.
Interesting read! How do you test if you have a good copy? I imagine the store owner getting mad if I say I want to try 8 different copies of the same lens, and even then I doubt I could see a difference on the camera lcd.
I’m looking for symmetry in the corners, sharpness in the middle, and performance generally commensurate with one’s expectations of price (i.e. not expecting much from an 18-55 kit zoom, but an Otus is something else…)
Let’s just say I have a very good relationship with my dealer. 😛
What a wonderful time it is to be a gear-head. I use the Loxia on my A7RMKII and the Siggy on my D810. Easy peazy… 🙂
Indeed – we are spoiled for choice lens-wise, which is a very good thing.
Thank you for this article. I have been very curious about the Sigma art series – they seem to hold their own. Bascially all about personal preference as well as needs/usage.
Yes – the Otii are challenging under documentary situations, but really quite superb for static work. And I wouldn’t want to carry more than one of either Otus or Art for extended periods…
I compared the 20mm 1.4 Art to the Nikon 20mm 1.8G and i prefer the Nikon.
The corners on my 20mm 1.4 Art were just too bad and the coma correction isn’t great either (to say the least…).
Sharpness in the center: not much of a difference worth even talking about (on a D800).
I took the 85mm 1.4G over the 85mm 1.8G and the 50mm 1.4 Art over any Nikon 50mm option but here i will stick to the Nikon 20mm 1.8G.
Sample variation may have a role to play here. For complex ultra wide designs, a small amount of out of tolerance can make a massive difference in performance. I tried eight copies of the Nikon 20 and did note some significant variation, but even the best copy is not as good as the Sigma at large apertures. By f8, it’s pretty equal across the board of course…
Nikon’s 50mm options are disastrous.
At 85mm, the 1.8G has less LCA and LOCA than the 1.4G, but much more flare and lower contrast – so it’s a case of picking your poison, I think. Personally, I’m now finding the Contax Zeiss 2.8/85 Sonnar to be the best of everything – and about the same price as the 1.8G, too!
At ~85mm it is a mix of 90mm 2.0 Summicron pre-Asph and 85mm 1.4G for me
(but my favourite portrait lens in my arsenal is actually the 200mm 2.0).
But back to topic:
Regarding the 20mm 1.4 Art and the 20mm 1.8G i expected what you described.
But at f1.8 i already prefered the corners of the Nikon, the Sigma hat too much astigmatism (“glow”) there.
Another minus for the corner on the Sigma: the plain with lowest coma was not the plane with the best sharpness.
I also tested the lens for field curvature, i did not detect that in a noticeable amount, until f4 the corners just weren’t up to my standards. Can just speak of my sample of course.
I may also upload the corresponding .NEF files for you and your readers, but not until wednesday (on assignment until then).
What i can already do is show the coma comparion (100% crop from top left corner on D800), the low quality is due to the restrictions from a board were i posted these, but i think you can still get the point:
This is the picture as a whole:
Thanks for the crops – those look quite a bit worse than the sample I used, but mine was not perfectly symmetric – I don’t know if this was the lens or the adaptor, though. I have to say, buying a ‘good copy’ has become a lot more difficult with the current generation of cameras…
Unfortunately: i have to second that 😦
With the 20mm 1.4 Art i may gave up to early, but for my astro work i’ll stick to 14-24mm 2.8@ 14mm 2.8 and 20mm 1.8G @ 1.8 for now:
The 20mm 1.8G balances also nicely on the A7 series and it still takes “ordinary” 4×4/4×6 Lee Filters.
Can’t say that about the Art 🙂
Ah yes – if you need filters at all, the Art is a no go – it’s one of the reasons I keep the Loxia around too; that and the fact that using filters tends to also mean you’re not needing fast apertures. Conversely, I don’t use filters for documentary work because I need all the light I can get, and the subjects tend to be mostly central anyway.
BastianK, thanks for the Nikon vs Sigma samples. I tried and sent back 2 copies of the Nikon due to disappointing coma up to f/2.8. I have a Sigma on its way and due here Friday. Your results are worrying to say the least. I guess we’ll see. Perhaps there’s some copy variation and I’ll get lucky with a good Sigma. I plan on using it on both a D810 and A7R2 for auroras and landscapes. I currently use a 14-24, ZF.2 21, Samy/Rok 24/1.4, Samy/Rok 14/2.8 and even the Sony/Zeiss 28/2 and want something faster at 20-21mm. Who knows, maybe I’ll end up back with a Nikon 20/1.8 🙂 I did appreciate its size and ability to take filters and its overall image quality when stopped down a bit.
MT, thanks for your thoughtful musings on gear and photography in general. I enjoy your site.
I think if you want filters, the 20 Art is out. Even with a Lee attachment, you’ll have to do something to compensate for corner shading which makes the whole thing unwieldy.
Is your 85 mm Contax mmj or what? Also I would be interested in your feelings about the other 2.8s (or any aperture) in the Contax line. The Contax’ are great all-rounders if you are doing cine and stills.
It’s a MMG. Other Contaxes – honestly, a bit hit and miss. Video OK, stills not so; most just lack sufficient CA/LOCA correction for the current generation of stills cameras. The 35PC and 100-300 are pretty good though, as is the 100/3.5. The rest are not so exciting (and 21, 28, 85/1.4 are the same as the ZF.2s.)
Sorry if I’m saying something stupid but, could you not fine tune the Sigma’s focus using the Sigma’s software ?
You can, but I think the problem he is having isn’t one of focusing point/plane – it looks like some sort of coma or element misalignment…and that cannot be fixed with the software (which is for fine tuning AF).
Thank you Ming.
Would like to see the ZF.2 21 Distagon against the Loxia 21.
Hard to believe the Loxia is superior. 🙂
Put it this way: I no longer have the 2.8/21 ZF.2 in any form. 😉
I’ve tested both (several copies) and found the ZF.2 far better than the Loxia. Not really close at all so it was an easy decision to send the Loxia back.
Practical question, how did you manage to buy (and return) a Loxia 21 when they’re not shipping yet? 🙂
Consider the Leica 21/3.4 for future comparison along with the f1.4. The f3.4 is the sharpest Leica 21mm, the f1.4 is made for speed and although fair to compare it to the Sigma the limitations of M lens size puts it at a disadvantage, the Sigma being much larger and able to overcome issues that are not easy in a smaller optic.
Any M mount lens would be at a disadvantage because the sensor microlenses are integral to the optical formula and corner performance. We can’t put the Loxia on M mount because the flange distance is shorter, so the comparison would have to be done on Sony (and we would want to anyway because system resolution is higher). We haven’t talked about the price difference, either. I don’t think it’s a realistic comparison, to be honest…
Again I would just mention the comparison is entirely possible today, and would be of considerable interest to many. You would need a extra A7r2, and have Kolari replace the 1.9mm cover glass with their selected and tested .8mm, a service which costs 500USD, and is increasingly popular. SEM 21 is very good on the A7.mod, and I’d expect it’s better yet on the BSI configuration. So you would mount the SEM 21 on the A7r2.mod and the Loxia on the stock A7r2.
This is not far fetched at all, as many of us are fed up with the Sony natives copy issues and ergonomics, and far prefer the small M lenses with consistent focus throws, less decentering, etc. The only sacrifice is very slightly degraded performance with FE natives, but many of us don’t have any 🙂 We use the A7 as a platform for our favorite lenses, M, R, Nikkor, Canon, etc. The Kolari Mod improves performance with all of them as far as I can tell.
As to the demanding nature of 42mp and Leica’s possibly inability to perform well on it, this is pure speculation, and contrary to optical theory, according to some anyway, who say 42 is not enough to “outresolve” even a mediocre SLR lens. That’s not to say you can’t see differences, but they are the same as you would see on film: or so this line of think goes. I am not sure myself.
Anyway such a comparison would be of high interest to many landscape photographers, and others who are looking for the lightest, smallest FF system than gives great results. For them the extra 500 to mod the A7r2 is only the price of a inexpensive lens, and it opens the door to a wide world of great lenses 🙂
Both Reid and Lloyd have done extensive testing with Kolari mods on the A7x, and I hope someday you will do the same 🙂
Anyway, thanks again for you wonderful blog, which is one my favorites.
I’m definitely not buying another A7RII, nor paying extra to have it modified to test lenses I don’t currently own and don’t have a need for. Remember, this is a) a free site, and b) not focused on equipment testing. I comment on the gear I actually use, because that use gives me experience beyond just as review – not review everything…
36 is enough to out resolve a mediocre SLR lens, let alone 42. I see it all the time, and would be happy with the 28-300 for everything instead of having to chase Otus-level primes.
I have been looking forward to this review, limited though it may be: two ultra wide angles close enough in focal length for comparison but built for different application motives. I wonder if an ultra wide Otus will combine the strengths of both while minimizing their respective weaknesses.
“Let us first consider the 21 Loxia: it is already remarkably sharp in the center at maximum aperture, and taking into account some field curvature, the edges, too. By f5.6 it is very difficult to find fault with anywhere in the frame – longitudinal and lateral chromatic aberration are very well controlled and there is little evidence of smearing in the corners. I wouldn’t call it apochromatic, but it’s really not far off. Colors are deep and saturated, and in the typical Zeiss manner, microcontrast simply pops.”
Is the old Kodak formula applied to the 21 Loxia to be revised to:
“F/5.6 and be there!!!”?
I find f2.8 works just fine too if your subjects have a slight bowl shape towards the camera.
An ultra wide Otus is scary: if the 28mm already requires a 95mm front filter, I dread to think about 20/21mm…
Is Zeiss really committed to making every single Otus lens an F/1.4 optic? If they produce a short tele macro, whether 100mm or 120mm, wouldn’t it be much better to limit such a lens to F/2 or less? Too much curvature in the front element would adversely affect the corner-to-corner flat field characteristics a true macro should have. I also love your idea of adding a tilt (but no shift) feature to macros.
I do not see why an ultra wide has to be F/1.4 either. Olympus once produced a legacy 21mm F/2 ultra wide. As you indicated, F/2.8 is generally bright enough for such a short focal length. I can appreciate the 28mm F/1.4 for some photographic genres, such as real estate/architectural/interiors photography. However, it is a bear of a lens to transport for most photo genres.
I believe that is the intention, but I honestly hope not…we are already challenging the limits of practicality with the 28. I want to carry all three, but my back does not.
Just guessing, but I’d expect a 35mm/1.4, and either a 100/1.4, or 135mm/1.4 in the Otus line before Zeiss markets a 20mm/1.4. Not that I’s buy any of them, the current Oti being so heavy as to be unnatractive. Ming’s back may ache from them, but for me it’s the knees that complain when I walk around with too much weight. Speed madness has taken over the prime lens market. I wish we’s see some nice small f/2.8 and f/2 compact SLR lenses from Zeiss, as I’ve mentioned before.
Not just you, Steven. Lloyd and I have been sending the Zeiss people every single comment asking for smaller and slower 🙂
It is a lot more practical to expect Olympus to release high speed lenses for the smaller µ4:3 system. Olympus has patents for 12mm F/1.0 and 14mm F/1.0 lenses. The greater depth of field that the diminutive µ4:3 system possesses compared to the 135mm format has µ4:3 lenses sitting the diffraction wall around F/8. It makes much more sense for them to extend the aperture range by increasing the max aperture.
I can see Zeiss employing a max aperture with the more moderate lenses (normals, short-medium tele, moderate-medium WA’s). I would hope that when/if they develop ultra WA’s and super tele that the F/1.4 design goal be tamed!
Typo correction: In the third sentence of the first paragraph, “sitting the diffraction wall” should be, of course, “hitting the diffraction wall.”
There are manufacturing considerations, too: shorter real FLs mean smaller elements and faster apertures become feasible. A 28/1 for FX is a lot harder to make than a 14/1 for M4/3 because it would not only have to cover 4x the image circle, but you’d already need a geometric aperture twice the diameter…
I was also told that it’s usually a question of economics and price points more than feasibility though 🙂
Thanks Ming for testing these 2 lenses.
In your opinion which one is better suit for astro photography?
I don’t believe either has hard stop infinity?
Thx kai
Kai,
You might also consider the Rokinon 14mm as an affordable choice for astro photography.
Loxia has hard infinity stop, Sigma does not. I’d go with the Sigma as the light collection ability is 4x…
Thanks. I’ll check the Rokinon/Samyang, and probably also the Sigma.