Why Leica is staying at 37.5MP for the S (Typ 007)

[KAHA]

After careful contemplation I think it's the red dot Mojo that imbues that special 'Je ne sais quoi' or 'größten auf der Erde' Leica look, no matter what language you say it in 

but then again I'm a shameless Leica fan boy after all 

Hi,

Can you substantiate? Why?

• Is it the red dot?
• Is it the resolution Leica 37 MP vs . Phase One P45+mat 39 MP?
• Is it the microlenses?
• Is it the smaller pixels of the Leica, 6 micron vs 6.8 micron? Could be the case!
• Or are you just looking at the price tag?

My take is that a better lens would give higher edge contrast, but would not improve rendition as it is most limited by pixel level artefacts on the P45+ image.

Best regards
Erik

[ErikKaffehr]

Hi,

The information from Leica that I feel is grossly inaccurate is the advantage of the shallow well design of the CMOSIS sensor, when applied to the S-camera.

That design is very good for the large beam angles of wide angles for the M-series, but I would say that it is not needed for the S-cameras, as those are SLR designs having lenses with small beam angles.

There are also some claims on this thread that this design would eliminate vignetting on wide angles, which is also absolutely false. All wide angles vignette. It is a law of optics. Retrofocus lenses suffer a bit less from vignetting. Vignetting is in no way limited to digital sensors, centre spot filters were also used with film.

What the OP asks is why Leica is staying at 37.5 MP? The answers we get may be relevant for the Leica M, which has lens designs optimised for compactness, but are irrelevant for the S-series that are different designs intended for DSLRs.

On the other hand, Leica lenses are said to be extremely sharp. So we are combining sharp lenses with large pixels, which will not make the lenses justice and produce fake detail. Simply enough, small pixels are needed for good detail, especially with good lenses. The feather shots illustrate this quite clearly. Small pixels give better details. Another way to see it is that even 30 year old lenses outresolve large pixel sensors enough to yield artefacts.

So, for some reasons Leica decided to use an upscaled CMOSIS MAX sensor for the Leica S. Doing that they have given up some resolution compared to their competition from Phase One (IQ-250 -- IQ-280), Leaf, Hasselblad and Pentax.

Now, that resolution may not be the most important thing. But, the limited resolution will results in less smooth rendition of fine detail, more aliasing artefacts, with very little benefits. There would be benefits with a digital back based on the CMOSIS design, but such a back has not been presented.

BTW I think that in general this thread would benefit from less snarky remarks and more constructive discussion about the new S

Regarding the value of the Leica S, I would say that I love live view and a good high ISO capability is good for many situations. The reservations about the Leica I have are most related to the large pixels. Staying with 37.5 MP puts the Leica a bit behind the competition, will cause aliasing artefacts (but not more than on the existing Leicas).

With some probability we are going to see 46-54MP DSLRs from Sony and possibly Canon next year, and there are a few great lenses around. So competition heats up. All these needs to be taken into account, at least if the camera is intended as a device for making pictures.

Best regards
Erik

Erik, I'm a photographer and I'm not scared by your technical posts, I enjoy reading them. Whether I understand everything you write and whether that's necessary for me in the so-called 'real world' is another matter Anyway, I appreciate your effort.

Question: What exactly did Leica say where that is information of dubious value?

BTW I think that in general this thread would benefit from less snarky remarks and more constructive discussion about the new S.

[ErikKaffehr]

Absolutely OK!

but then again I'm a shameless Leica fan boy after all 

[ErikKaffehr]

Hi,

You also believe in flat earth theory? http://en.wikipedia.org/wiki/Flat_Earth

Best regards
Erik

I'm going to be really disappointed when Feathers R Us come looking for a photographer to produce their catalogue images with the strict stipulation that they must be shot at 3.8m and enlarged massively, my S will be straight in the bin!

This really is an absurd conversation, there is so much more to a system than picking one ridiculous test and then banging on about it ad neusem. Just forget it Eric, it's not a system you own, have used, will ever own or will ever understand, that's ok, it's not designed for you, same as many other systems out there. Keep taking test images as long as that makes you happy and let photographers work with the equipment that suits their style and more importantly their clients requirements.

[EricV]

If you prefer the image on the left, buy a 39 MP back like the P45+ or the Leica S2. In case you prefer the image on the right wait for a 150 MP digital back.  Both images shot a 3.8 m with comparable lenses at 150 mm. The significant difference is the sensor, 6.8 vs 3.8 micron pitch.  Simple, isn't it?

Not quite so simple.  Consider two other solutions: 1) move closer (2.1m) or change lens focal length (270mm).  These solutions would provide the same increased density of pixels/feather, thus achieving the same improvement in image quality as your example.  The reason to prefer a 150MP sensor is to provide this improved image quality across a wide field of view.  Then there is stitching ....

[Joe Towner]

What's odd is the delay between the announcement and release of the 007 S - Q2 2015.  It's got advantages for anyone already owning HC or Contax AF lenses. 

Plus, by being based around something other than the Sony 50mp chip, there may be differences that appeal to certain shooters.  Time will tell, the more interest in the market, the better for everyone.

[eronald]

But you do know more than Leica engineers ...right? It's just that you have a better job than they can offer you if you apply to them, that keeps you from "teaching" them... right?

Theodoros,

 I would hope he is more careful than Leica engineers, because if reactors had the same rate of defects as Leica cameras, life near them would be dangerous - I had 4 M8 bodies!

Edmund

[ErikKaffehr]

Hi Joris,

The first objection I have is this statement:

Dr. Zimmer estimates that the sensor in the S will have a SNR of around 79 dB, which will probably translate into somewhere between 13 and 14 stops of usable DR.

It is technically correct, 79 db is 13.12 EV, as it is between 13 and 14 EV. But 13.12 EV is essentially around 13EV, the article implies another EV in DR. (*)

The original CMOSIS sensor has 76 dB of SNR which corresponds to 12.6 EV, see below. DxO-mark tested it 12.34EV (this is the lower, unnormalised value)
http://www.cmosis.com/news/press_releases/new_leica_m_uses_cmosis_24_mp_cmos_image_sensor
Pushing the design 3 dB from (76 dB to 79 dB) is an engineering feat.

The other major objection is this:

The result is that there is no sensor vignetting, no color shifts and no loss of sharpness in the corners. It’s easy to put this theory to the test. Take a Leica 18mm Super Elmar-M ASPH lens. Put it on the M 240 and take a shot. Then, mount the same lens on a Sony A7r. Yeah…. Oh and just to see that the advantage isn’t only at extremes, try a 35mm Summilux ASPH next. The Sony sensor in the A7r just isn’t adept at handling non-retrofocus lenses, with non-telecentric designs

The suggestion here is totally irrelevant to DSLRs which the Leica S (Typ 007 happens to be). It is quite clear that the Sony A7r has problems with Leica-M ultra wides. But the solution that Leica has to the problem goes far beyond the sensor, they also have very thin cover glass and lens cast correction in software based on the coding of the lenses. The S-series wide angles are retrofocus designs, as they need space behind the lens for the mirror box. So the CMOSIS sensor has advantages when used with rangefinder lenses designed for film, but not with SLR lenses on the S (Typ 007).  

The A7r that was taken as an example here in the original posting works well with all Leica R, Nikon F or Canon EF lenses. And you can obviously not put neither the 18mm Super Elmar or the 35mm Summilux on the S (Typ 007), as they would not focus at infinity.

So I feel it is a misrepresentation of the facts in the context of S (Typ 007), and that is the camera the article is about.

Best regards
Erik

(*) This is speculation but the original CMOSIS sensor was reported to have a full well capacity of 40000 electron charges. The Nikon D600 is said to have an FWC around 76400. So the Nikon D600 which has similar pixel size can collect almost twice the number of electrons, which would yield about 40% less noise. The relatively low FWC on the CMOSIS sensor may depend on the shallow design.

The FWC data for the Nikon comes from Sensorgen: http://www.sensorgen.info, I am a bit skeptical about the accuracy

Erik,

I just re-read David's article once more.  

A simple question.  Could you highlight exactly which statements in that article are factually incorrect?  

I have re-read your comments as well and quite frankly it is not entirely clear to me.

Thanks, Joris.

[JV]

Hi Joris,

The first objection I have is this statement:It is technically correct, 79 db is 13.12 EV, as it is between 13 and 14 EV. But 13.12 EV is essentially around 13EV.

The other major objection is this:The suggestion here is totally irrelevant to DSLRs which the Leica S (Typ 007 happens to be). It is quite clear that the Sony A7r has problems with Leica-M ultra wides. But the solution that Leica has to the problem goes far beyond the Sensor, they also have very thin cover glass and lens cast correction in software based on the coding of the lenses. The S-series wide angles are retrofocus designs, as they need space behind the lens for the mirror box. So the CMOSIS sensor has advantages when used with rangefinder lenses designed for film, but not with SLR lenses on the S (Typ 007).

So I feel it is a misinterpretation of the facts.

Best regards
Erik

I personally wouldn't lose sleep over the first statement but that is because I don't care much whether it is 13 or 14EV.

I am not sure about the second statement and how applicable it is to the S.

I own an A7s and 3 M lenses.  I prefer the output the FE 55mm lens (the only FE lens that I have) to the output of the M lenses on the A7s.

According to me it doesn't necessarily mean though that the Sony sensor is not adept, it just means that the Leica sensor and firmware is optimized for the M lenses and the Sony to a lesser extent.  That's all.

I appreciate Leica for doing something different from Phase/Hasselblad/Pentax.  It is a pretty bold move IMO.  I hope they will be successful, a/o also because I have a vested interest (owning M9 and S2).  Time will tell.

Joris.

[ErikKaffehr]

Hi,

Answers, see below.

Best regards
Erik

I personally wouldn't lose sleep over the first statement but that is because I don't care much whether it is 13 or 14EV.
That's OK with, it is just I want to have correct information.

I am not sure about the second statement and how applicable it is to the S.
Well, my point is that it is irrelevant for the S.

I own an A7s and 3 M lenses.  I prefer the output the FE 55mm lens (the only FE lens that I have) to the output of the M lenses on the A7s.
Hi, that can depend on many things. Leica has thin cover glass, which reduces astigmatism. Leica M also uses the 6 bit code on the lens for reducing lens cast effects. Sony doesn't do it for Leica lenses, obviously.
Zeiss has two lenses for Sony Alpha the Loxia 35/2 and a 50 mm lesn. Both are redesigns of the ZM series. This article shows the difference between the two lenses on the Sony Alpha 7r at full aperture.


According to me it doesn't necessarily mean though that the Sony sensor is not adept, it just means that the Leica sensor and firmware is optimized for the M lenses and the Sony to a lesser extent.  That's all.
I agree fully.

I appreciate Leica for doing something different from Phase/Hasselblad/Pentax.  It is a pretty bold move IMO.  I hope they will be successful, a/o also because I have a vested interest (owning M9 and S2).  Time will tell.
I have no issue with the S (Typ 007). It has live view and I feel that is important and the high ISO capability is welcome for sure. Personally, I feel that those lenses probably deserve a higher resolution sensor. Ideally, the sensor should outresolve the lens, and I guess that the S (Typ 007) needs to be stopped down to f/16 to avoid aliasing. That is what I see on my P45+ with 20-30 year old lens designs.
The sample below is a detail from a P45+ shot using a Distagon 40, upsized to 200% for better viewing. Development was in LR 5.5. I see this kind of stuff to often on the P45+.
Other raw developers may do a better job:.


Joris.

[mjrichardson]

Morning Erik

Excellent answer, congratulations, your posts always make me laugh!

Hi,

You also believe in flat earth theory? http://en.wikipedia.org/wiki/Flat_Earth

Best regards
Erik

So you don't have a problem with the sensor improvements, just that you don't feel they are relevant to the S? Well that's ok, you don't need to feel they are relevant, the Leica engineers and the people who buy it do, you'll get over it. With the ev, it's factually correct but 13.2 is closer to 13 even though the text said between 13 and 14, which is still technically factual but you prefer a manufacturer to round down rather than give a range, um, ok, that's maybe your issue rather than anyone elses.

As for the sensor, you personally feel that the S would be better served with a higher resolution sensor, well that's ok too but for reasons only Leica actually know as fact, the engineers decided that the system would deliver at 37.5, as you aren't ever going to buy one then I think your personal feelings are relevant to you but unlikely anyone else.

Your one man crusade against the S is impressive but ultimately will change nothing, what will you move on to next? The internet must be an exciting place for someone who takes issue with how data is displayed, you'll be busy!

Have a nice day.

Mat

[Theodoros]

Hi,

Can you substantiate? Why?

• Is it the red dot?
• Is it the resolution Leica 37 MP vs . Phase One P45+mat 39 MP?
• Is it the microlenses?
• Is it the smaller pixels of the Leica, 6 micron vs 6.8 micron? Could be the case!
• Or are you just looking at the price tag?

My take is that a better lens would give higher edge contrast, but would not improve rendition as it is most limited by pixel level artefacts on the P45+ image.

Best regards
Erik

Guess what Erik... All current Canon sensors have even larger pixels than the new S.... I wonder how ignorant of your basic physics and testing techniques they are.... One question  though... in your physics experiments... is it common practice to use different and have results for other? 

[peterv]

Erik is an engineer, he likes to get his facts straight, up to two digits after the comma, others like to sell cameras so maybe they bend the facts here and there to their advantage. Nothing new, really.

As a photographer, I’m grateful for the scientific input of the engineers on this board, since this helps me to put the marketing speak in perspective.

I’m also grateful for the input of people like David, Doug and Steve. They have acces to -interesting- background information that may be helpful when making a buying decision.

In second quarter 2015, when the new S arrives, we can make tests against the 50 MP Sony 33x44 mm sensor * and see what all this means for our photography.

* By then, their might also be a 50 MP 24x36 mm sensor.

[Theodoros]

Never the less, one can't avoid noticing, that Erik uses examples with comparison of Sony's 24mp Cmos 24x36 sensor and Kodak's old 39mp CCD MF sensor ...and then addresses his (questionable IMO) findings as if Leica's new sensor will behave exactly the same as the old CCD MF sensor!!!  All this of course is despite the fact that Leica's new sensor has much more many commons with... Sony's sensor (both of exactly the same pixel size, both Cmos, both with micro lenses)!!!!

[Carl Glover]

Hi,

I would say that Rolleiflex Hy6 Mod2 may be an attractive alternative or the Alpa FPS, depending on needs, of course.

http://www.dhw-fototechnik.de/en/rolleiflex-slr/rolleiflex-hy6-mod2.html
http://www.alpa.ch/products/lenses/macro-tilt-swing/apla-lens-adapter-rolleiflex-600x-hy6-sinar-hy6-leaf-afi.html

I already have a hy6, my second one in fact.

I would use the S2 as a replacement for my Canon 5DIIs (except when it comes to the two outstanding T/S lenses) but using my favourite Rollei glass.

[bjanes]

There is one big issue though. The lenses. No one that I know uses a digital camera without a lens. The Leica S lens line is Superb from the 24mm to the 180mm. They really get every bit of detail out of the sensor. Plus you can use Contax 645, Hasselblad H, Hasselblad V, Mamiya 645 and Pentax 6x7 lenses.

Nor does one use a lens without a sensor. As this post by Jim Kasson and the ensuing discussion by Jim and Bart van der Wolf point out, given reasonably good lens performance, one gets the most bang for the buck by using a higher resolution sensor rather than increasing lens performance. Take a look at Jim's quiver plots. The Leica lenses are undoubtedly outstanding, but they can not get every bit of detail out of the sensor since MTFs multiply and the MTF of the lens is less than unity. The cost sensitive photographer would likely get more for his money by choosing a sensor with higher resolution than is offered by the Leica and using less expensive but still excellent lenses with this sensor. The Pentax MFDB would give more bang for the buck, but there is not that much difference between 37.5 and 50 MP.

Bill

[mjrichardson]

I can appreciate I may be alone in thinking this but to me sharpness at minute levels is a important but by no means make or break when it comes to buying in to a system. I now have m4/3rds, Canon APSC, Nikon full frame and Leica S, the look of the file coming from the Leica is just incredibly deep and sumptuous for want of a better word, surely the image as a whole is made up of many elements, contrast, colour etc? Why is the focus simply on how many pixels and their size are required to get a sharper image?

Mat

[Bart_van_der_Wolf]

The cost sensitive photographer would likely get more for his money by choosing a sensor with higher resolution than is offered by the Leica and using less expensive but still excellent lenses with this sensor. The Pentax MFDB would give more bang for the buck, but there is not that much difference between 37.5 and 50 MP.

That's correct. We're talking about a sampling density difference between 5.3 micron (645Z), and 6.0 micron (S, typ 006 and 007). That's not a huge difference, but would probably be enough to reduce the risk of aliasing (Erik's point) a little bit, also marginally helped by per pixel diffraction. The physical sensor array sizes are close enough to make no difference regarding subject magnification factor for a given field of view.

The original question why Leica chose to keep at the same resolution may be as trivial as having no readily available sensor for the relatively limited quantities they would require. The rest is guesswork.

The Leica lenses would benefit from a higher sensor resolution, but there is probably no viable sensor available off the shelf. The rest is left to marketing.

Cheers,
Bart

[peterv]

And there you have it. Six pages of forum discussion in a nutshell.

Thanks Bart, for stating the issues clear and simple.

[bcooter]

I can appreciate I may be alone in thinking this but to me sharpness at minute levels is a important but by no means make or break when it comes to buying in to a system........snip......the look of the file coming from the Leica is just incredibly deep and sumptuous for want of a better word, surely the image as a whole is made up of many elements, contrast, colour etc?

Mat

I agree with the look trumps all, but I think we're in the minority on this.

My question to everyone but Mat is  . . . who cares?  

I bought a Leica because I wanted it and never needed anyone to validate my purchase by agreeing with me.  

I didn't buy it to prove it's a better camera, test it against tree trunks or produce charts.   I don't sell or compete with cameras.  

I buy them to make a living and hopefully enjoy my day and in all honesty am glad so many people don't rush the Leica doors.  


IMO

BC