What the Otus gives - compared to a cheap and good Canon

[ErikKaffehr]

Hi,

Canon has a very good 85/1.8 lens at a very reasonable price. Let's check out sharpness compared to the famous Otus. The data here is coming from DxO and covers the field, that is center to corner. DxO is very careful in not describing what their figures measures, so just see this as an input. Theory says that when we stop down lenses at one stage they will be limited by diffraction.

Check the figures below:

At f/1.8 the Otus has a huge benefit and going to f/2.8 it is still there.

Going to f/5.6 the advantage of the Otus is very small.

Now stop down the Otus to f/8 and keep the Canon a f/5.6, now the Canon outperforms the Otus! That is because diffraction is taking it's toll on the Otus compared to the Canon.

Best regards
Erik

[ErikKaffehr]

Hi,

I wanted to keep MTF and "science" out of the first posting, but here is my interpretation of what happens.

A good way to see it is that we can look at MTF and break it up in two parts:

MTF_aberrations and MTF_diffraction.

• MTF_aberration increases monotonously when stopping down, but with diminishing returns at small apertures.
  MTF_diffraction decrases monotonously when stopping down

Now, MTF for a system is the product of the MTF of it's components. So MTF_system = MTF_aberrations * MTF_diffraction * MTF_other_factors.

So what we have seen in the first posting is that the Canon lens stopped down to f/5.6 actually outperforms the Otus at f/8. But, if we stop both lenses down to f/8 the Otus still holds a small advantage.

Best regards
Erik

[viewfinder]

Interesting,.....

What your posts really show is that nothing changes. especially the laws of physics.

Back in the 1960's we knew that for critical work lenses for 35mm needed to be stopped down two stops, and that maximum apertures were only for focussing for the photographer and marketing for the camera makers.   Many/most lens designs are the same now and are actually very old.   

Then, as now, Jap makers were concerned with cost and marketing and European makers could give performance at the cost of bulk and price.

One is reminded of the 'Olympic Sonnar' designed for the 1936 olympics to work at max aperture f2.8....a massive (and expensive) piece of glass......

[Paulo Bizarro]

Well, the Canon was released around 1991 or 1992? So not bad at all. This lens has always been good value for money.

[johnnycash]

Erik, please be more careful with overusing the word "outperforms".

First, there is no difference in your charts at higher f stops.

Second, the real world benefits of using lens X versus lens Y can't be measured in a lab, no matter what tests you perform, especially out of context.

I can assure you there will be a Canon DSLR 35mm sensor based limitation in any similar measuring (visible or invisible to human eyes) as Zeiss Otus lens can still, at the end of 2015, show how limited the resolving power of the best current Canon sensors is.

You are right that with the 5DSR, the diffraction kicks in at f/5.6 and beyond that, any lens theoretical testing of sharpness will equalise, that's just physics.

So it's nice to know that old lenses still perform well, no wonder, they were great years ago, there's no reason why they wouldn't perform well today. That can be the conclusion from your posting but nothing else.

As a Canon 5DSR user who has tried many L lenses - I think the oldest I used was the 1996 EF 135mm f/2L, through the popular primes up to the EF 800mm f/5.6L I would like to accent there is more to a lens than an MTF/etc. chart, esp. done by someone from DxO who are notoriously known for misrepresenting results, not including input parameters and providing skewed metrics based on unimportant criteria (I may prepare a dedicated post to this topic).

So if you are, at the end of the day, trying to figure out which lens are you packing with you on a photoshoot where you will use 5DSR with 85mm at f/8 and you are not concerned about any other quality parameter but general sharpness, you can take the 85mm prime, any 70-200 or any other non-Canon lens at given focal length.
However, being a landscape and a cityscape/architecture photographer, even the best Canon L lenses stopped down at the diffraction limit can never *outperform* my Otus, and you can tell pixel peeping using the camera LCD zoomed 16x in the field or better, later during the post on a big screen. Stop even further to f/8 - f/11 and you slightly lose sharpness but the overall quality is still present and Otus can still beat any L lens.

So it safe to say that shown results from DxO prove nothing worth considering in the real world photography and I claim their results are misleading, especially if used separately without a context.

[ErikKaffehr]

Hi,

The need of stopping down to stops is no law of physics. It is absolutely possible to make a lens that performs best at full aperture. But, such lenses will be expensive. Just as an example, microscope lenses can be very expensive. At large magnifications diffraction effects will be very significant, so microscope lenses are striving to very large apertures.

For any well corrected lens, diffraction will dominate over aberration at medium apertures. So stopping down to medium apertures will bring all decently well corrected lenses pretty close.

What I have noticed, essentially, that all my lenses perform best at around f/5.6 when tested on axis. Of axis testing is a bit difficult as you really want to have at least 50 times the focal length as a shooting distance, and that requires large targets.

I would also add that it is a myth that sensors outresolve lenses. Would that be the case we would never see moiré. A decently well corrected lens will outresolve camera sensors at optimal apertures, at least near the optical axis.

Another point is that large apertures is a selling point. So we have a lot of large aperture lenses having marginal performance fully open. With the Otus lenses Zeiss made a lot of effort to get optimal performance at full aperture. They also have a 135/2 APO Sonnar that performs at similar level, but at half the price. That lens is f/2 and not f/1.4, half the speed at half the price.

The main reason I did not buy the Zeiss 135/2 APO was that I don't feel at home with the 135 mm focal length. Initially I ordered the Zeiss Batis 85/1.8, but decided on the Sony 90/2.8G instead. That may have been a mistake, I don't know.

We cannot make educated choices of lenses unless we have the option to test a few samples of each possible candidates. I think that independent tests are a worthwhile resource to help in making educated decisions. The more well known test sites have some experience in testing lenses and they probable have far less bias than lens owners claiming the excellence of their choices.

Best regards
Erik

Interesting,.....

What your posts really show is that nothing changes. especially the laws of physics.

Back in the 1960's we knew that for critical work lenses for 35mm needed to be stopped down two stops, and that maximum apertures were only for focussing for the photographer and marketing for the camera makers.   Many/most lens designs are the same now and are actually very old.   

Then, as now, Jap makers were concerned with cost and marketing and European makers could give performance at the cost of bulk and price.

One is reminded of the 'Olympic Sonnar' designed for the 1936 olympics to work at max aperture f2.8....a massive (and expensive) piece of glass......

[kers]

I am sure the images made by the Otus will look better than the Canon lens even @5.6.
It is too easy to judge a lens on one parameter.
I have a Sigma ART 1.4 50mm and a nikkor 1.8 50mm AFS-
@ 5.6 the Sigma screams and the Nikkor is dull.

I would also add that it is a myth that sensors outresolve lenses. Would that be the case we would never see moiré. A decently well corrected lens will outresolve camera sensors at optimal apertures, at least near the optical axis.

I agree that the optical axe is able to resolve more than 100Mp with my best lenses- as i have tested it with my Nikon1 J5 camera.
; the sides and corners are usually not as sharp as 36MP, even @5.6; especially wide angle lenses have a problem- the corners are usually best @f11 and not 36MP.

[PeterAit]

Then, as now, Jap makers were concerned with cost and marketing and European makers could give performance at the cost of bulk and price.

Out of charity I will assume that your use of the derogatory and insulting term "Jap" is from ignorance and not malice. Am I correct?

[Rob C]

Out of charity I will assume that your use of the derogatory and insulting term "Jap" is from ignorance and not malice. Am I correct?

Yet another thread I may as well abandon right now. Jeeezus!

Rob C

[Slobodan Blagojevic]

...I have a Sigma ART 1.4 50mm and a nikkor 1.8 50mm AFS-
@ 5.6 the Sigma screams and the Nikkor is dull...

Do show!

[bjanes]

Out of charity I will assume that your use of the derogatory and insulting term "Jap" is from ignorance and not malice. Am I correct?

Here in the USA and after the Japanese attack on Pearl Harbor, "Jap" or "Nip" was used in a highly derogatory manner and admiral Halsey stated "The only good Jap is a dead Jap". The Pacific war was very brutal and American behavior was not exemplary. Today, Japan is one of our closest allies and the term "Jap" is seldom used in the USA by civilized persons. However, Erik is from Sweden and may not be aware of previous American usage.

Bill

[Slobodan Blagojevic]

... previous American usage.

Given that it is the previous usage that was intended as derogatory, one would hope that seventy years later it has lost its edge?

[PeterAit]

Given that it is the previous usage that was intended as derogatory, one would hope that seventy years later it has lost its edge?

Tell that to my wife, of Japanese descent, whose parents were interred in the camps during the war.

[PeterAit]

Here in the USA and after the Japanese attack on Pearl Harbor, "Jap" or "Nip" was used in a highly derogatory manner and admiral Halsey stated "The only good Jap is a dead Jap". The Pacific war was very brutal and American behavior was not exemplary. Today, Japan is one of our closest allies and the term "Jap" is seldom used in the USA by civilized persons. However, Erik is from Sweden and may not be aware of previous American usage.

Bill

Yes, and that is why I gave him the benefit of the doubt. I have no reason to think that he is a bigot, just a person who does not know what that word means to many people. Well, now he does know and I hope he takes my comment in the way I meant it.

[ErikKaffehr]

Hi,

This comparison was made between the Canon 85/1.8 and the Otus 85/1.4. The Canon was chosen as it is a very good lens. The short telephoto lenses are often very good, as the relatively narrow angle os view makes optical correction relatively easy.

The point I see is that correcting a lens fully at large apertures takes an immense effort and any good lens will be affected by diffraction when stopped down. Diffraction only depends on aperture.

One issue with DxO figures is that they don't describe their measurements very well, so when they talk about "acutance" it is not clear what they mean. It is obviously an MTF value but at which frequency?

Best regards
Erik

I am sure the images made by the Otus will look better than Canon lens even @5.6.
It is too easy to judge a lens on one parameter.
I have a Sigma ART 1.4 50mm and a nikkor 1.8 50mm AFS-
@ 5.6 the Sigma screams and the Nikkor is dull.

I agree that the optical axe is able to resolve more than 100Mp with my best lenses- as i have tested it with my Nikon1 J5 camera.
; the sides and corners are usually not as sharp as 36MP, even @5.6; especially wide angle lenses have a problem- the corners are usually best @f11 and not 36MP.

[gwhitf]

Can one still use the adage "a lens is sharpest two stops down" in general, with digital also? Or does every single lens have its own sweet spot? I would assume f8 in general. Would not think 5.6 or 4. Have always wondered this. I took at trip to California once and shot an H/p45+ at f22 the entire trip. On tripod. Was shocked how soft the files were. Why even have f22 or 16 on a lens if guaranteed soft?

[ErikKaffehr]

Hi Bill,

I have not used that word...

In general I don't it is meaningful to talk about German and Japanese lenses. Most of the Zeiss lenses are actually made in Japan at Cosina under Zeiss quality control. Cosina used to make affordable lenses, but obviously has the capability to make excellent lenses. Some of the  APO Lantar telephoto lenses used to be famous.

Zeiss lenses used to be very good, AFAIK, but that comes from being good design and careful assembly.

Personally, I have a bunch of older lenses from Zeiss for my Hasselblad, and I considered to buy the 135/2.0 APO Sonnar as it is a very good lens and fully corrected wide open, and it is half the price of the Otus. But the Otus has f/1.4 and correcting f/1.4 fully takes a lot of optical effort. As I normally use f/8, large apertures matter little to me.

I actually ordered the Zeiss Batis 85/1.8 for my Sony A7rII, but cancelled it when the 90/2.8G came out.

Best regards
Erik

Here in the USA and after the Japanese attack on Pearl Harbor, "Jap" or "Nip" was used in a highly derogatory manner and admiral Halsey stated "The only good Jap is a dead Jap". The Pacific war was very brutal and American behavior was not exemplary. Today, Japan is one of our closest allies and the term "Jap" is seldom used in the USA by civilized persons. However, Erik is from Sweden and may not be aware of previous American usage.

Bill

[ErikKaffehr]

Hi,

It would be very difficult to design a lens that is "diffraction limited" at a large aperture. In the long telephoto department there are quite a few lenses that perform near optimal at large apertures.

The reason it is difficult to make diffraction limited lenses at large apertures is that the diffraction limit is very high.Check the figures below:


So, a diffraction limited lens at f/4 would need to resolve 450 lp/mm while one that is diffraction limited at f/8 would resolve just 225 lines per mm.

In general, diffraction limits sharpness when a lens is stopped down. On axis, good lenses often reach maximum performance at around f/4 but they can often improve quite a bit off axis when stopped down somewhat more. So I would say that lenses are not sharpest two stops down, but it takes a lot of effort to correct a large aperture lens fully. I would say that the Otus is best at f/4 - f/5.6 when it is stopped down 3-4 stops. But the same Otus is good enough to show moiré at full aperture on a 36 MP sensor.

Although lenses loose a lot of MTF at small apertures it is still possible to restore sharpness fully, as long as resolution is not affected. The trick is to sharpen the image without increasing noise.

I sometimes make two exposures and combine one at small aperture for maximum DoF and one at medium aperture focused where I want maximum sharpness and combine both images in Photoshop. I actually often have 3-4 exposures with different focus and blend in Photoshop.

Best regards
Erik

Can one still use the adage "a lens is sharpest two stops down" in general, with digital also? Or does every single lens have its own sweet spot? I would assume f8 in general. Would not think 5.6 or 4. Have always wondered this. I took at trip to California once and shot an H/p45+ at f22 the entire trip. On tripod. Was shocked how soft the files were. Why even have f22 or 16 on a lens if guaranteed soft?

[johnnycash]

Personally, I have a bunch of older lenses from Zeiss for my Hasselblad, and I considered to buy the 135/2.0 APO Sonnar as it is a very good lens and fully corrected wide open, and it is half the price of the Otus. But the Otus has f/1.4 and correcting f/1.4 fully takes a lot of optical effort. As I normally use f/8, large apertures matter little to me.

Erik,
you are right the Zeiss Apo-Sonnar 135mm f/2 is almost as good as the Otus. Not as good but very close. What I don't understand is the flimsy lens cover and the silver ring in front, so overall the 135mm could employ a better design, I think about it as a predecessor to the Otus.
I think many people would prefer to buy a lighter Otus f/2 (24mm/35mm/55mm/85mm/100mm) if only such lenses would've been manufactured. I believe the additional f-stop to f/1.4 was a must to show Zeiss craftsmanship and it served them greatly for marketing purposes.

[ErikKaffehr]

Hi,

A good site for comparing lenses is www.the-digital-picture.com .

They have comparison images between different lenses, including both the Canon 85/1.8 and Otus 85/14.

In this images corners are definitively sharper on the Otus at f/8 compared to the Canon at f/5.6.

http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=106&Camera=453&Sample=0&FLI=0&API=4&LensComp=957&CameraComp=453&SampleComp=0&FLIComp=0&APIComp=6

Yes, I know that contradicts the DxO results. I could speculate to find some explanations, but with too many unknowns I don't want to that.

Best regards
Erik