A7r vs D800 vs M9 with top primes tested

[nugat]

I tested my three full frame cameras ( Nikon D800, Leica M9 and Sony A7r) with a variety of primes.
At 24mm I tried Nikkor 24/1.4 G ED on D800 and Summilux 24/1.4 both on M9 and A7r (via adapter).
At 35mm I put Sigma 35/1.4 on Nikon, Summilux 35/1.4 on Leica and Zeiss 35/2.8 FE on Sony. I also tried Sony with the 35mm Summilux.
For "normal" focal lengths I tested Nikkor 50/1.4 G on D800 and Konica Hexanon 60/1.2 on both Leica and Sony.
All lenses were tested wide open, at f2.8 and and f8. I set up a controlled natural scene (print, details etc etc) and spent today's morning comparing the central resolution/contrast delivered by the configurations. Not scientific by any measure, but I'll be interested to compare the results later with Imatest.
All raws were developed in LR 5.3 at standard setups. Also the bodies were with standard parameters on.

The 18Mpix sensor gives Leica M9 some trouble to compete with the 36Mpix rivals . Both Summiluxes and the Hexanon gave visibly more detailed photos at all apertures on Sony A7r. On Leica they were limited by the sensor which resulted occassionally with moire--a sign that the lens "outresolved" the sensor. The same glass performed beautifully with A7r and moire was observed only once: with the 24 mm summilux at f2.8. Of course the Nikon was protected from aliasing by AA filter, had it been D800E , moire might be present there too . But the same AA/OLPF cuts the Nikon's resolution quite significantly. Leica M9 with its superb diffraction limited lenses (and no anti aliasing/optical low pass filter) performs just 15% below its Nyquist limit (maximum theoretical resolution)-see Erwin Puts site imx.nl. Losses on the Nikon and it's lenses are bigger--up to 35% (maybe 25-30% on the "E" version). In fact M9 and its glass was almost there with D800 throughout the test. And with Hexanon it managed to be better than the D800+Nikkor 50/1.4 combo.
In fact A7r tells us what Leica could achieve with a 36Mpix sensor. A7r is also devoid of the OLPF.

D800+Nikkor 24/1.4 vs A7r + Summilux 24/1.4
The summilux wins hands down at 1.4 and from 2.8 it is better than the nikkor at f8.

D800 + Sigma 35/1.4 vs A7 + Zeiss 35/2.8 FE + Summilux 35/1.4
The Sigma is currently (DxO tests) the sharpest 35mm FF lens .
Wide open at f1.4 it is behind the Zeiss at f2.8 . From f2.8 both lenses/bodies go head to head. Sharpening in Nik Output Sharpener adds to the visual impact. D800 might reach the levels of D800E with some tweaking here. In general cameras without AA/OLPF filters (A7r and M9) do not gain as much from sharpening as the ones outfitted with an OLPF.
When I put the Summilux 35/1.4 on my Sony the combo showed some advantage over both 35mm rivals . It was better at f 1.4 than the Nikon/Sigma at 1.4 and slightly behind the wide open Sony/Zeiss (2.8). From f2 it was better than the other two lenses/bodies and from f2.8 it bettered their f8 perfromance.

D800+Nikkor 50/1.4 vs A7r+Hexanon 60mm/1.2
The cheap Nikkor is not known to be the sharpest of the "normals" crowd
(I wish I could try the Otus) but at f8 it is supposed to be nothing short of very good. Or it should be.
Hexanon on the other hand is a legend. It was manufactured by Konica in a limited run of 800 pieces. At f1.2 it is a bit soft, I love it so on M8 as a portrait glass. From f1.4 it is one of the sharpest lenses I have seen.
Mounted on A7r it demolished the Nikons at all apertures. At f1.4 it is equal to Nikon/Nikkor's f8. At f2 it trashes the rival set to f8 so badly, it hurts.

 

[D&A]

So much to contemplate and thank you for posting your in depth findings. It would also be of interest to learn of edge and corner performance of all these combinations, if it's possible to elaborate on.

Dave (D&A)

 

[ZoranC]

Let me just say the cliche :worthless:

 

[nugat]

My test was very simple and only for the central part of the frame. I am planning to repeat it with Imatest when I have the setup ready. It will have much more data and reliability then.
For now I attach several 100% crops of the very central 100% crop with a fine print sheet photographed with Nikkor 50mm on D800 and Hexanon 60mm on Leica M9 and Sony A7r. Apertures and camera/lens are in the file names.

 

[Jack]

One problem with drawing conclusions from only the center of the frame on a few images is that you can be severely misled about a single lens' performance if you mis-focus or the camera moved. To test objectively really requires a standardized "set" that allows you to evaluate performance in 3 dimensions at close, mid and far distances. It's not an easy task and requires shooting several -- at least 3 -- sets of images at each aperture and distance where you refocus for each shot. Then they need to be processed identically and best of each set compared. Again, it's a very arduous and unfun process...

The way you tested can give you hints, but more often than not when I've relied on quick, cursory evaluations, I've left an otherwise stellar performer in the closet for untold periods of time due to my own error.

 

[nugat]

I agree 100%.
Even though I bracket focussed manually, there are other factors that make proper lens testing seriously difficult. Even with Imatest.

 

[turtle]

'Real' lens testing is not something I'd recommend to anyone. I don't know how the real testers can bear it.

I shot one of my most successful images on a severely decentered CV 21 f4 P. I went out on a miserable grey drizzly day in Afghanistan to shoot a few frames with the then new to me (used) lens to see how it performed and check for obvious gremlins. All on film, so no instant results possible.

While testing (I was at one of my favourite haunts used for a long term project) I saw a great photo and so took it. Thankfully I was at f10 at the time and the layout of the scene meant that the subject was on the good side and the distant background on the 'bad' side. I did not even realise until shooting some time later how badly decentered the lens was.

I never used the lens again, but for all that I lost on this dud of a lens, I have half a dozen copies of the photo and have exhibited it in New York, London, France and Malaysia. I always remind myself of this when I check lenses out and get a bit too deep into the weeds!

 

[nugat]

I took some of my best pics with Olympus EP1 and Panasonic 20/1.7 lens.

 

[OliverM]

For my purpose, these tests are sufficient : all 3 cameras are very good and the sony sensor doesn't shake too much in front of a paper.

I think that the M9 and the Sony are primarily designed to make great travel & street photography, possibly compromising on the pure quality for the size. The good news is that the compromise is very low.
I can understand the need of some photographers to further test the limits of the technical capabilities of the 3 cameras. Something which is less documented and less easy to evaluate is the differences between their subjective characters :
1. great colors, rich & able to capture the nuances of transition of the skin for example
2. great transition between sharp zone & bokeh, 3D-look
3. atmosphere created by those aspects + possible fall-off

Said differently : why should I buy a used M9 and summicron today, apart from the dream of having & using a Leica (which is also OK) ? Will I get a unique character ? Will colors have a different impact ?

I found many great pictures with each camera, but with different light conditions, different set-up, different post-treatment, different selection/garbage rates.
I was equally impressed by landscapes with the 3 cameras. Highlights were possibly better captured with sony & nikon.
I was equally impressed by portraits by M9 & Sony ... with the exception that I was impressed by 200 pictures with the M9 and 10 pictures with the more recent sony. M9 could deliver a bit warmer tones, perfect for some atmospheres. But not all, and what was the post processing ?
In all cases, the result differs also a lot between the compressed image and the full-sized not compressed one. Sometimes you see a boring picture which proves to be great when properly displayed. And sometimes you are impressed at first look but very disappointed when viewing all full scale defaults.

Any input fot these subjective elements would be great !
... even more difficult than the protocole suggested by Jack ... but more fun, no ?

Thanks for sharing in any case !!

 

[turtle]

Yes, I agree. People are fixated with the innate imaging character of the equipment they can buy without perhaps thinking enough about the fingerprint they can put on the image that will render small technical hallmarks irrelevant.

I like to start, however, with equipment I like and 'click' with. It does not have to be perfect, but it does have to 'fit'. Personally, I find the new A7 and A7R to be an absolutely brilliant new twist in the development of small, FF cameras. They just feel right and it doesn't always have to be more scientific than that.

 

[tashley]

Let me just say the cliche :worthless:

I completely disagree. It is useful and interesting to me. If people want to 'prove' their findings with pictures so be it, but it generally leads to other people questioning their methodology. So if someone wants to share some observations that they have made, that is fine by me with or without images. You can always ask politely to see some examples.

Very, very few of us here have access to the sorts of equipment that give definitive test results - and even those are limited in scope. Yet we rely on a community of opinions to form our own.

Frankly I think it is rude to use that Smilie unless it is really warranted. Sorry, but I said it.

 

[Jack]

Let's be even more direct -- lens testing is a total pain in the butt!

I used to obsess over finding perfect optical performance, built a stable of lenses that delivered it and went out and made technically perfect images that had a more or less sterile look. So I learned how to process for a better artistic rendering while retaining the optical excellence.

By contrast, now I look for specific character in lenses instead and am not overly concerned about corner performance or even optical excellence centrally -- as long as the lens draws in a way I find pleasing.

So I now own a set of look lenses and a set of render lenses, and by far my favorites to shoot with are the look lenses -- even ones with a slew of idosynchrosies.

Nowadays to test, all I do is shoot real images, usually a cityscape with the camera on a tripod and run the lens at each aperture from wide open to f11. With subject depth I can easily evaluate whether it has poor corner performance or just extreme field curvature, how it's signature changes -- or doesn't -- with aperture, and assuming some relatively straight vertical lines and a horizon, what kind of distortions it has -- in short, just about everything I need to know to determine whether I'll like the lens.

But that's me...

 

[cunim]

Let's be even more direct -- lens testing is a total pain in the butt!

I used to obsess over finding perfect optical performance, built a stable of lenses that delivered it and went out and made technically perfect images that had a more or less sterile look. So I learned how to process for a better artistic rendering while retaining the optical excellence.

By contrast, now I look for specific character in lenses instead and am not overly concerned about corner performance or even optical excellence centrally -- as long as the lens draws in a way I find pleasing.

So I now own a set of look lenses and a set of render lenses, and by far my favorites to shoot with are the look lenses -- even ones with a slew of idosynchrosies.

Nowadays to test, all I do is shoot real images, usually a cityscape with the camera on a tripod and run the lens at each aperture from wide open to f11. With subject depth I can easily evaluate whether it has poor corner performance or just extreme field curvature, how it's signature changes -- or doesn't -- with aperture, and assuming some relatively straight vertical lines and a horizon, what kind of distortions it has -- in short, just about everything I need to know to determine whether I'll like the lens.

But that's me...

OK, Jack. You know you are not going to get away without a short list of your favorite "look" lenses for the Sonys.

 

[Bryan Stephens]

"D800 + Sigma 35/1.4 vs A7 + Zeiss 35/2.8 FE + Summilux 35/1.4
The Sigma is currently (DxO tests) the sharpest 35mm FF lens ."


I was just curious as to why you didn't test the Zeiss 35/1.4 which I have an which I think is a terrific lens, beating my Nikon 35 1.4g on my D800e

 

[Guy Mancuso]

Its real simple all the ones I have he saw. LOL

Seriously not much has changed after having Nikon, Sigma, leica and Zeiss lenses. The only new ones in town are the Sony FE's. You can tell the look from anyone of the bodies. My glass looks the same if not better on the Sony A7r as it did with the D800e. The question mark is how they look on the Sony in regards to any issues that may come up like some of the M mounts other than that my glass looks the same as the D800e with maybe the only difference in how they render color , DR and such but thats a difference in bodies not the glass per say.

 

[Jack]

OK, Jack. You know you are not going to get away without a short list of your favorite "look" lenses for the Sonys.

Reality is it's the same as my current Nikon look lens list. (28/1.4 AF-D, 50/1.2 AIS, 85/1.4G, 105DC, and the 85mm Lomo Petzval.) But I'd probably have to add Guy's 19R or Woody's 18 SEM to the list -- both a little different from each other, but really nice. Woody won't like to hear this, but I think his 18 SEM blew his 21 SEM out of the water in look. The 19R gets the nod but only because I could get it adapted for use on my Nikon bodies. Finally, I'd add Bob's new 58/1.4G lens to the list -- his copy renders very nicely, a more subtle look than the 50/1.2 AIS -- I could (and am) making a case for owning both.

 

[nugat]

"D800 + Sigma 35/1.4 vs A7 + Zeiss 35/2.8 FE + Summilux 35/1.4
The Sigma is currently (DxO tests) the sharpest 35mm FF lens ."


I was just curious as to why you didn't test the Zeiss 35/1.4 which I have an which I think is a terrific lens, beating my Nikon 35 1.4g on my D800e

When I was looking for a 35mm lens for my D800 I thought it'd be nice to see most of that resolution. According to dxomark.com Sigma 35/1.4 is the sharpest lens of the three (Nikon 35/1.4, Zeiss 35/1.4) and of all current FF 35mm lenses. It also scores highest overall and costs much less. For the three (Sigma, Zeiss, Nikkor) the overall performance scores were respectively 39, 34, 33 and the resolution 23, 17, 17 perceptive Mpix. The last metric is interesting, it shows how the combo of body+lens performs together. So the Sigma gets on D800 23Mpix, or in the imaging chain the "losses" are 36%. For the other two the "loss" figure is 53%.
Zeiss Otus is scored 45 dxomarks overall and given 29 p-Mpix on the D800.
That's 80% "efficiency". On a D800E it might be at 85%. But that's a 55mm lens. I'd love to put it against my Konica hexanon 60mm mounted on Sony A7r.
PS."loss" and "efficiency" are shorthand here. The percentages are derived from area count and do not correspond to linear measurements like lp/mm (line pairs per milimeter)


http://www.dxomark.com/Reviews/Sigm...gma-35mm-f1.4-DG-HSM-Nikon-versus-competition

 

[Godfrey]

Let's be even more direct -- lens testing is a total pain in the butt!

I used to obsess over finding perfect optical performance, built a stable of lenses that delivered it and went out and made technically perfect images that had a more or less sterile look. So I learned how to process for a better artistic rendering while retaining the optical excellence.

By contrast, now I look for specific character in lenses instead and am not overly concerned about corner performance or even optical excellence centrally -- as long as the lens draws in a way I find pleasing.

So I now own a set of look lenses and a set of render lenses, and by far my favorites to shoot with are the look lenses -- even ones with a slew of idosynchrosies.

Nowadays to test, all I do is shoot real images, usually a cityscape with the camera on a tripod and run the lens at each aperture from wide open to f11. With subject depth I can easily evaluate whether it has poor corner performance or just extreme field curvature, how it's signature changes -- or doesn't -- with aperture, and assuming some relatively straight vertical lines and a horizon, what kind of distortions it has -- in short, just about everything I need to know to determine whether I'll like the lens.

But that's me...

+1

I'm with you, Jack. Actually, regards the A7, I'm practicing an even simpler Zen: I *have* the Leica R lenses already. They're what's going to be used. Far as I'm concerned, they're good enough for anything I'm going to need...

That puts the onus on me to actually do something interesting with the camera and these lenses... ;-)

G

 

[Shashin]

Let's be even more direct -- lens testing is a total pain in the butt!

Amen to that!

 

[bensonga]

I was thinking of doing a bit of informal "lens testing" with my Pentax 645D of the P645 45mm A lens, the P645 45-85mm FA lens and my P67 45mm lens.

After reading the above posts....I've reconsidered and will just spend my time shooting. I am even more appreciative for the time and effort that people like Dave (D&A) have put into testing a whole range of lenses.

Gary