Image from Photozone. Also the famous lens review website noticed the lens-cast issue on the [shoplink 10350 ebay]Zeiss 35mm f/2.8 ZM[/shoplink].
A week ago I got my NEX-7 and I am going to use third party lenses on it (by using appropriate adapters). But is it really so easy and flawless to use whatever lens you want on your NEX?
There are is one fundamental thing you have to watch out before to chose your lens. The “lens-cast” issue which creates color shift (mostly cyan shift) in the corners of your image. That problem appears or is stronger under following conditions:
1) Lenses with symmetrical design are not optimized for digital sensors, while lenses with tele-centric design do work well.
2) Wide angle lenses do accentuate the problem.
3) Problem gets worse on bigger sensor. Leica M9 shows more color shifting than the NEX-5n.
4) Problem gets more visible on cameras with smaller pixels. The NEX-5n has less problem with lens cast than the NEX-7.
Example: The Voigtländer 15mm f/4.5 M-mount lens ([shoplink 11043 ebay]Click here to see it on ebay[/shoplink]) does a wonderful job on film cameras but isn’t that good on the NEX-5n and completely sucks on the NEX-7. This because it has a symmetrical design and extreme wide angle. Does it mean you cannot use it on your NEX? There is a solution: The problem can be fixed with the use of the CornerFix software (there are both Windows and Mac OS X versions). It can correct Sony RAW files and transforms them in DNG files.
But if you like me don’t have time to convert every single file than you have to really watch out for the design of the lens you want.An alternative for the Voigtländer is the Zeiss 18mm f/4.0 lens ([shoplink 11044 ebay]Click here to see it on ebay[/shoplink]). Unlike the Voigtländer it has a tele-centric optical design. Those lenses are usually bigger than symmetrical lenses. That’s why the Zeiss is still bigger even if it has more or less the same aperture of the Voigländer.
In general wide angle Voigtländer lenses, Zeiss ZM Biogon lenses, Older Leica lenses come with symmetrical design. All new Leica lenses, most medium and tele lenses and soem Zeiss ZM lenses will work fine. With the help of our readers I will now collect info about a huge variety of lenses you could use on your NEX. It should become a sort of general guide to see how good a lens works on NEX cameras. I will post 1-2 guides per week starting from the next week.
To read more about the problem read that short text at LL (Click here), the guide at Digitalhapeman (Click here) and an M-mount index at Ken Rockwell (Click here).
62 Responses
The Zeiss is 18mm not 15, and it doesn’t work well on the NEX 7 either !
Sorry was a typo. Anyway, Ken Rockwell tested it on the M9 and said there is no such issue.
Cause Ken has active profile on that lens, if you activate 18mm f3.8 profile, it will remove color issues and vignetting. Pretty much all lens (at least from 15mm to 35mm) have color issues on M9, but Leica isnt stupid and they have those lens profiled so you can let camera remove color shift and vignetting.
You can do same, if you use Capture One RAW developer (cause most digital backs have similar issues). Or you can use simple CornerFix software, which works quite ok too.
I think it is Contax CY 18/4, it should work fine on nex 7 but not sure about M9.
There is Zeiss 18/4 for Leica M.
Otherwise C/Y lens wont work on M9. Simply cause you need rangefinder linked lens to focus. Tho with 18mm it would be possible, just set decent hyperfocal distance and shoot. But no reason, M mount will be probably better.
Hi, wonderful initiative.
In my 4000+ shots experience with the NEX-5N, *all* the lenses I have tested work well with no colour cast. The lenses are:
* Leica Elmarit-R 19mm/2.8 II
* Zeiss ZM Biogon 25/2.8 (this has the worst colour cast of all on test shots, but it is very slight and has never been an issue in real-life pictures)
* Leica Summicron-M 28mm/2 (a perfect combination with the NEX-5N)
* Voigtlander Colour-Skopar 35mm/2.5 PII
* Leica Summicron-R 50mm/2
* Zeiss Contact Planar T* 50/1.7
* Leica Summarit-M 75mm/2.5
* Leica Elmarit-M 90mm/2.8
* Zeiss Contax T* Sonnar 135/2.8
(complete test here, as you have already linked to : http://dearsusans.wordpress.com/2011/11/25/70-the-sony-nex-5n-and-exotic-lenses-leica-m-leica-r-zeiss-zm-zeiss-cy-panasonic/)
I hope to test Contax G lenses very soon and will report asap. For now, my greatest worry was the Biogon 25. But after shooting over 2000 frames with it, I can certify it makes a stunning combination with the NEX-5N. The Leica 28 Summicron-M has a tad more ‘organic’ rendition, but the biogon is as sharp, distortion free and vignette free. A winning combo.
One last thing: it may be tempting to mix and match brands according to what you can find cheap on eBay. But having done so for a couple of months, let me tell you that changing lenses + adapters is not as pleasant (in fact, it’s a pain, unless you’re doing photography as a form of meditation ;o) as changing lenses while keeping the same adapter. As good as they are, all my Zeiss C/Y and Leica-R lenses are staying at home for that reason.
I would recommend sticking to one adaptor for maximum fun.
Cheers,
Pascal
Perhaps we should buy more adaptors so that the adaptor can stay on the lens, at least during a session.
But the Zeiss 50mm sonnar will work? I think so because it is not wide angle, it doesn’t have a symmetrical design and my nex 5 is 14 mpix on Aps-c so it wil get less conering issues then the 5n and 7?
I see there’s a rather loose used of telecentric in this item. The Zeiss 15mm is not a telecentric design, it’s a retrofocal design. A retrofocal lens has a long rear focal length than the front and is, effectively, an inverse of telephoto designs. A true telecentric design is one where the entrance or exit pupil is at infinity. In the latter (the important one for digital sensors) the light rays arrive perpendicular to the sensor surface. Retrofocal designs just reduce the range of angles, they aren’t truly telecentric. So the Zeiss (and all SLR wide-angle lenses) are retrofocal and rather a long way from being true telecentric designs.
The evidence appears to be that (for APS-C), rear focal lengths above about 35mm are enough to give tolerable results, and it might be expected that this would be closer to 50mm on full frame. That’s for moderately fast lenses wide open – stopping down will reduce the effect.
Modern CMOS sensors are highly dependent on optimised micro-lenses as the actual photodiodes are relatively small compared to CCD designs (what’s called the fill factor). That probably means that fall-off at the corners on a CMOS design will be worse than a CCD for any given rear focal length. It may well be a factor that lead Leica to choose CCD sensors on the M9. However, I don’t suppose this will last. Very little money is going into CCD development (and even less now Kodak is on the ropes), and there are many limitations. CCDs will continue on MF as they are much simpler to make and no company is going to invest the huge amount of money required to scale up CMOS to those sizes. However, future up to 35mm FF is surely with CMOS and I’ve not doubt sensor designs and compensation will be tweaked, but I’m not sure we’ll ever see those old, very compact, symmetrical lens designs being optimal on such cameras.
Note that this rather works against pancake lenses on short register mounts like the NEX’s. It means that (collapsible designs apart), it’s difficult to make very short lenses without compromising performance.
Steve,
can’t talk about the NEX-7, never having used one. But I can send in thousands of NEX-5N shots showing that the “tolerable” limit is well below 35mm. I am now using a second NEX-5N, not the first one I bought and am seeing the same results with that.
Unless you are shooting fashion or any other discipline that requires absolutely perfect colour, you can safely say that down to 25, colour casts are non issue with the NEX-5N down to at least 25mm. If anyone wants to lend me his Biogon 21 or Elmarit-M 24, I promise to treat it lovingly 😉
Perhaps I should have said “safe” in that some sensors are more tolerant that others. It might well be that the 16MP sensor is more tolerant as it might well be that it’s less dependent on accurate focussing of light by the microlenses as the photodiode will be bigger.
However, not all of the corner effects are down to that – some have been put down to longer path lengths of oblique rays through the IR and colour filter layers (as these all have significant thickness).
It’s also worth noting that this all depends on the exact design of the lenses as certainly many of the later ones have at least some retrofocal characteristics.
+1
Everyone needs to read the Zeiss wide angle paper by Dr. Nasse that came out a couple of months ago. Not only is lens color cast an issue, but more symmetrical lens designs also show poorer resolution on non-optimized sensors. The paper shows how much worse MTFs are for the ZM 21/2.8 on non-optimzed sensors. Actually, it isn’t the sensor so much as it is the sensor AA/IR filter pack. If Sony removed the AA all together, it would probably go a long way in improving symmetrical wide angle performance. As it stands now, the AA/IR filter pack is causing astigmatism that seems to be worse on the 7 when compared to the 5N.
It’s definitely the AA filter and the 5N is not immune from this problem either. That very ZM 21mm f/2.8 referenced in the Zeiss paper is what I tested on the 5N and it is quite bad until stopped down past f/8, due to severe astigmatism. The ZM 35mm f/2.8, another near symmetrical design also exhibits nervous corner detail rendition. In contrast, the Voigtlander 35mm f/1.2 II, which is a more retrofocus design, looks very smooth. If anyone is interested in seeing actual high resolution files straight from the 5N with these and a few other lenses, I’ve posted them in a write up on my site. http://www.ronscheffler.com/techtalk
And I disagree in the original post at top that the Voigtlander 15mm f/4.5 doesn’t work well on the 5N. It does, corner to corner, wide open.
I’m sticking with stock lenses bc I don’t have the cycles to spare to devote to this tech. I kinda wish I did but I don’t.
Just representing. (:-{)}
Peace,
Bill
Without abusing the terms “symmetric” and “telecentric” and all that-
my recommendation based on actual usage is to get Olympus pen F lenses. All perform well. Besides..
1. Ergonomically, they all are a perfect match for the NEX-7.
2. Even the colors of the camera and the lenses match!
Forget C-V, Leica, Zeiss etc lenses in m- or screw mounts and leave them to the range finder users.
Oh..let us campaign for compact decent performers from Sony and also for fair pricing of the 24/1.8 E-Sonnar. 🙂
Forget Cosina/Voigtlander?
(*sips coffee… spurts coofee onto screen*)
I don’t leave home without my Nokton’s! The best glass you can buy for shallow DoF on APSc!
Ergonomics shmergonomics – the set up is still smaller and ligter than my Nikon DSLRs with any lens. And makes better images too.
If you want ultra-wideangle for the Nex get a C/V 12mm 5.6 Aspherical II. There’s a good review on photozone.de.
The olive green special edition Bessa cam won’t let you forget C-V lenses and you need not be constrained by the APS-C. Enjoy! 😉
Not wide enough (20mm and up). The same is true about M4/3 lenses (except the expensive Panasonic/Olympus 7-14mm f/4). The Nex needs native ultra wides.
Yes! Native E-mount lenses, the compact variety. 🙂
Sony did open up the mount to other manufacturers, so there’s hope…
…or just buy a m43 camera which has proper wide angle lenses. 🙂
Never 😀
I just sold my GF1, the sensor is really not good at high ISO, as well as for DR, I constantly got burned skies.
I just came back from a trip to Spain. I used the Voigt 15mm/4.5 extensively on the Sony NEX-5n and did not see any issues whatsoever in any of the shots that I took.
Please note that:
1. LL talks about the 12mm and not the 15mm
2. Steve Huff tested the 15mm on the NEX-5n and NEX-7 and shows good results there with the 15mm too.
Regards
AG
Great idea admin. thanks
I just ordered from eBay an OM mount Olympus lens for my NEX 5N.
@ Vivek
I tested some Pen-F lenses myself too but I have to say that the
1:3,5 f = 20 mm and
1:2.8 f = 25 mm
were really bad on my Nex-5
The 1: 1,8 f = 38 mm was fine, no problem.
The wideangles had VERY soft corners which were also blurred.
I sold them really fast 🙁
Right now my workhorses are Contax Gs and Leica Ms.
Greets !
Fotoingo
Ingo,
I have a NEX-5. It has not been used for about a year now.
I find the NEX-7 a *real* camera.
I tried the following pen F lenses (with their proper hoods):
1. 25/2.8 (fabulous @f/5.6 for infinity shots, no need to stop down any further!).
2. 38/1.8 not that contrasty wide open but still better than many other brands.
3. 38/2.8 pancake. Simply awesome for street shooting!
4. 38/2.8 (rarer tessar clone). OK.
5. 38/3.5 Macro. Superb!
6. 40/1.4 The best among the pen F!
7. 42/1.2 Dreamy bokeh.
8. 60/1.5 Awesome lens! Can’t find a better portrait lens!
9. 70/2 Really nice lens (find a good sample!).
10. 100/3.5 Superb!
11. 150/4 Very nice compact tele!
12. 250/5 Again very nice compact tele!
I do not have the 20/3.5. I have not tested the 25/4 (will not be better than the 25/2.8 is my guess).
The comparison between the NEX-5 and the NEX-7 images are like night and day. The NEX-7 image quality is superb (besides it being the real camera in the NEX series)!
If you had sold them real fast, you have made the buyer (if they are using a NEX-7) very very happy! 🙂
I have to admit that the performance stopped down was “okay” but if I have a lens with a nice 2,8 aperture I want to use it and that was not possible on the Nex 5.
So for infinity shots stopped down the 20 and 25 may ! work.
Thanx for your nice reply.
Greets, Ingo
Cheers, Ingo!
Photos get tagged here (no other data):
http://www.flickr.com/photos/vivek-iyer/
I looked your photocollection it was really good.
daniel magnusson
I just posted a detailed comparison, with links to the corresponding RAW files, of the Carl Zeiss Biogon 35mm F2.0 ZM on both the NEX-5N and the NEX-7.
http://www.bmupix.com/journal/2012/1/7/nex-7-part-2-corner-performance-with-zeiss-biogon-t-35mm-f20.html
In short, while the extreme corners of the frame are softer on the NEX-7 at large apertures, the NEX-7 files show more detail over most of the rest of the frame. At F4.0, the NEX-7’s corners pretty much catch up with the 5N. At F5.6 and F8.0, the entire frame – including the corners – of the NEX-7 show more detail (as you would expect from the higher resolution sensor). Whether that’s important or not will depend on your individual needs.
Excellent test – and yes, there’s definitely a colour cast on the NEX-7 in the corner at f2.8. With the wider aperture some light rays will be arriving at a more oblique angle and presumable the effect is a complex combination of longer path lengths through the filter “sandwich” and (maybe) something to do with the microlenses (I think the microlenses are actually at the bottom of the filter “sandwich”.
It occurs to me that a cheaper version of the NEX-7 with a 16MP sensor would be the ideal camera for many people as I suspect it would offer sufficient resolution for most people and it seems to be more tolerant at the corner.
Steve; so it’s not my eyes deceiving me. I’ll try shooting a Cornerfix lens profile tomorrow.
Yes, a 16MP NEX-7 type body would have been interesting. Unfortunately though, if I had to choose between the better ergonomics and controls of the NEX-7 and the NEX-5N’s better corner performance with certain M-mount wides, I’d have to choose the former.
Thanks, Bjorn. This is my most used lens, and the results are interesting. I’d say the NEX-7 really doesn’t catch up in the corners until f5.6, and it still seems to have a little more CA at that aperture.
You’re welcome. I wanted to see whether and to what extent the 35mm F2.0 Biogon’s performance was worse in the corners on the NEX-7. I got the NEX-7 for its improved handling and was willing to compromise somewhat with the performance of this lens. For me the trade-off has been worth it since I don’t own any wider M lenses.
You’re right of course in saying that the NEX-7 doesn’t catch up in the corners until F5.6. I was looking at a larger corner area: at F4.0, you don’t have to go very far into the frame at all before things improve.
Yeah, what worries me is that the ZM 35/2 is among the best ZM and CV wides, as far as corner performance goes, so, if we seeing this much of a difference between the 5N and 7 with this lens, I can only imagine how bad other wide angle rangefinders will be in the 7’s corners.
Practically speaking, there aren’t many instances where you need wide apertures and corner resolution is critical. Typically wide apertures are used in circumstances where there’s a subject fairly central. There aren’t many such circumstances where there there will be anything in the corners which wouldm be in focus anyway.
Where edge-to-edge really matters is landscape, architectural, fine art and the like. Generally you’ll operate stopped down to get some DoF and be in the lens’s sweet spot. Longer exposures of course, but that’s what tripods are for.
That is true, however what I’ve found with some lenses, such as the ZM 35mm f/2.8, is that even if the edge areas are out of focus wide open, they’re more nervous looking than another lens of the equivalent focal length, such as the Voigtlander 35mm f/1.2 II, with a different optical flavor (i.e. retrofocus rather than the Zeiss’s near symmetrical). I have a specific comparison of these two lenses with high res samples shot on the 5N here: http://www.ronscheffler.com/techtalk/?p=87
As you’ll see, the Voigtlander has a more pleasing out of focus blur than the Zeiss. Also I’m referring to and specifically comparing rangefinder lenses. Adapted SLR lenses should generally be problem free.
The problem is not really the color cast which can be easily fixed by Cornerfix. The problem could be the lack of sharpness in corner at wide aperture. Please, check the comparisons make by Rob, on http://www.flickr.com/photos/sportsphotorob/6231919025/in/photostream/
Check the High Rez pics to compare between wide and middle apertures.
Rob compared the different ZM and new Sony lenses on Nex-7.
As Bjon Utpott (see above) said with the ZM 35mm, starting f:5,6, ZM still have better resolution than other lenses. Don’t be affraid by Steve Huff (who likes to us “rought beginner style” images). color cast is not a big problem, and, as said in LL, the lack of sharpness in corners is not as bad as we fear. A good exposition, the use of a tripod if possible, a clean lens with an adapted hood (adpated for APS-C sensor I mean : check on http://www.bokkeh.com/about-bokkeh/itemlist/category/6-lens-hood), and an aperture adapted to the subject will provide an excellent resolution up to 16 in. enlargment. If you want more, go to an Alpa 12 !.. and pay for !
M-Mount lenses remain excellent choice if we understand that we will keep them for a long time, that the microlenses technology on 24 MP sensors is just a question of time. The symmetric design is in many ways much better than the retrofocal design (and not a telecentric design) : less distortions, better compactness, etc… Then VC and ZM,s still are excellent choices. I’m not ready to pay for a Leica new design, even it’s a retrofocal design with symmetric quality. to pay up to 400% more expensive for 10% more sharpness.
No experience to offer but the end results of the survey will be very useful to many people I am sure. An excellent idea.
+1. Same here.
Leica isnt just about sharpness, but truth is that sensors with AA simply cant show it much.
It doesnt have anything to do with size of pixels. NEX-5N is somehow optimized for rangefinder lens, while NEX-7 isnt. NEX-5 had bigger pixels, so issues should be less than NEX-7, same goes to NEX-3 or C-3, but in reality all of them have more issues than NEX-5N.
5N sensor assembly is simply different, I dont know why exactly but issue isnt in size of pixels. Btw. Ricoh GXR M-mount have zero issues due microlens. 🙂 Wait for 16 mpix model, if you must have M-mout lens on mirrorless.
Actually, it could be about the pixel size. It’s possible that the 5N and 7 have the same sensor toppings, but the smaller pixels and micro lenses are causing things to go wonky.
So why NEX-5, NEX-3 and C-3 have more issues with those lens than 5N? 😀 All those have less pixels than 5N (so they have bigger pixels which would mean they should have less issues not more, but its exactly opposite).
Sony could have adopted a new sensor filter pack with the C3 and 5N, and maybe it just doesn’t play well with the underlying NEX-7 silicon. Who knows? The point is, there are a number of things that could be the reason for the edge performance differences.
I have a Nex 5, a Nex 5n and a Nex 7.
I use a variety of Leica, Zeiss and Voigtländer rangefinder lenses.
On the Nex 5 and the Nex 7 all rangefinder lenses wider than 35 mm give a colorcast towards the edges.
The Nex 5n only shows a slight colorcast with the Voigtländer 15 mm and the Zeiss Biogon 25 mm. In practice correction is here hardly required.
The Nex 7 also shows a magenta colorcast over the entire frame with wider than 35 mm lenses and apertures larger than f/2.8. Upon stopping down the colorcast changes from magenta to green.
With NEX 7, theres actually magneta colorcast over whole frame, if you lift shadows by +2 eV in PP. Was it just me, or youve encountered same?
I think theres something wrong with RAW processing for NEX-7, I guess its more like RAW developer issue than NEX itself.
Oh and thats with 24mm f1.8 Zeiss.. not adapted lens.
In addition to my previous comment. Also very fast lenses of longer focal length can show a magenta to green colorcast over the entire frame with the Nex 7. Going from magenta to green upon stopping down.
A lens like the Zeiss 18mm ZM Distagon is also showing edge colorcast (magenta/cyan) while it is not a symmetrical lens design.
However it has a very large ray angle where the light is hitting the filter/sensor from a large angle.
So looking for non-symmetrical designs is not s safe starting point.
The best bet is to use retrofocus dslr designs. But these are big and make hardly any sense on a small camera.
You’re right.
Therorically, the problem is not simply a difference between symmetrical and retrofocal lens. That depends of the flange distance. A retrofocal lens on a DSLR is not the same as a retrofocal on a range-finder, like the ZM Distagon. Optically, we could say that the range finder lens like the ZM Distagon is “less” retrofocal than the DSLR one. That’s the reason Leica can make retrofocal lenses with less distortions than Canon or Nikon. Longer the flange distance is, more difficult is the problem of distortions’ correction. Then, on Leica, the flange distance is short, even with the distagon, you can have color cast, not a lot, but you have it, since the rear lens is very close from the sensor, even for a retrofical lens.
We could have a long symmetrical lens, as we can find with the Alpa 12, we wouldn’t have any color cast.
That depend of what kind of use we expect from our lenses. As for me, I’m interested in the Nex-7 because I travel a lot in my business, and I want to have permanently a camera with me, without to have to hold a complete DSLR system. I make landscapes, architecture, etc, pictures. I always stop down to the optimal aperture, on monopod if needed, low ISO, then as for me, M-Mount lenses remain the best solution, even with the color cast which is easily removed with Cornerfix.
As for me, the solution of using DSLR retrofocal lens is not appropriate.
J-M,
Many wrong assertions.
1. There are very few “symmetrical” lenses for Leica (if in doubt check with Zeiss.). The Zeiss Topogon, for example, is a symmetrical lens. No Biogon is a symmetrical lens.
2. Whatever on the DSLR lenses and corrections- not correct.
3. Alpa 12 (lenses) have major problems though they themselves aren’t “symmetrical”.
Hi there, I’m using a voigtlander ultron 28mm f2 and I find it to be really good. Not color cast, sharp even in the corner wide open. Good duo with a Nex-3!
I use my 5N almost exclusively with Contax G 28, 45 and 90mm lenses. I can say there is no colour cast whatsoever as reported with earlier NEX cameras. Even wide open, corner to corner is super sharp. I wish someone would develop a adapter that can use the auto focus of the marvelous G lenses. I would pay for that adapter even if it costs more than my 5N 🙂
Considering electronics for AF are probably Sigma, it might be possible. Conurus is doing conversion of C/Y N lens for Canon AF (with AF!). Now he is working on Canon AF lens for NEX (no AF but auto-aperture).
You might try asking him.. I think that adapter for Contax G to NEX with at least auto-aperture shouldnt be impossible.
Thank you all for the intelligent discussion! It’s nice to see this kind of positive conversation for a change.
People, people!
Get out and SHOOT instead of worrying about the very corners of your images! 😛
My idea. It almost never happens to me that an important part of the subjects I shoot is in the corner of the image, so why bother? Though it’s nice when the lens manufacturer gives you what you paid for 🙂
Well, actually its really visible (color shift) on wide-angles. Its not “very corners”. Sharpness drop ofc isnt that much visible.. (tho if you compare that to Ricoh GXR M-mount than it IS visible:D).
Before there is a proper solution for the sensors, the only way to avoid this is use DSLR lenses or forget the wide angle lense from RF.
NEX-5N works much better with RF lenses than NEX-7
I’ve been shooting the Sony NEX 5N with my Zeiss ZM 35mm f2 Biogon and Leica 50mm f2 Summicron with image quality issues whatsoever, the Biogon is almost absent of any distortion and a great all round lens. The only issue I have with the 2 combos is the minimum focal distance with the lenses, being used to close focussing SLR lenses, this long MFD is really bugging me. But my solution is on its way in the form of Hawk’s helicoid adaptor, which is currently in transit on its way to me.
I’ve documented my thoughts and experiences shooting the 5N with the 2 above lenses here. http://handcarryonly.com/post/15218902174/6-days-in-bangkok-with-the-sony-nex-5n-leica-50mm
Your mileage may vary!
I meant, with NO image quality issues 🙂