S2 Chromatic Aberration - how big an issue is it?

[PeterA]

As a 40 megapixel MFD camera - The S2 doesnt do anything better than Hasselblad - and can't do a lot of things Hasselblad does - really it is that simple for me.

As an SLR ergonomic offer - I need to see some telephoto lenses because that is what I use a 35mm SLR for - oh and the auto focus better be a lot better than Nikon as well as image stabilization technology to match.

hmm maybe they should have made an R10 - at least there was an established customer base - begging for just this camera.

 

[stephengilbert]

"[M]aybe they should have made an R10 - at least there was an established customer base - begging for just this camera."

Maybe camera makers are infected with that bragging rights thing as well: I think that while an R10 wouldn't compete with Nikon or Canon sales numbers, it would have been at least as successful as the S2 will be. But the S2 is so much "cooler," and affords superior bragging rights.... To Leica as well as to buyers.

 

[fotografz]

There is noch such thing as a perfect lens, the aberrations can be controlled to a negligible level - I don't think Leica claimed something different.
Fuji makes lenses which simply cannot be used without heavy software correction (it cannot even be used with only a few years old H-boodies!) to implement cheaper optical designs (that becomes obvious when you compare it to the optical performance to the last new Zeiss-design, the 40IF - it's superior regarding resolution and CA) - that's a big difference in comparison to Leicas approach to make designs as good as possible with the option to later control (most optical defects cannot be fully "corrected" at least not witout trade-offs) remaing aberrations within software.

I have yet to see any MF-lens (or any lens) which performs as well at wide apertures in comparison to the S-lenses, I cannot even see "proofs in the pudding" within this particular thread.
Although the particular fringing problem clearly goes beyond simple lens aberrations.

What proved to be wrong is the fact that not every DNG-capable converter can be used for DNG-files, C1 doesn't create useful files from the S2, for example. So it's an open system but not necessarily completly independent from the converter.

Doesn't matter where the choo-choo comes off the tracks ... it's the end product that matters however you get there.

For clarification of what could be read as misleading or truncated facts: the only H lenses that cannot be used with older H bodies are the ones clearly designated as D lenses ... meaning two out of 10 currently available lenses. The D lenses can be used on all H cameras from the H3D onwards ... including the H2F (which also allows use of film backs with the other 8 lenses) as well as CF digital backs with all 10 lenses .

I also disagree that the 8 non-D HC lenses cannot be used effectively without extensive software corrections .... since I use a H2F with film backs and scan on a 949 without significant purple fringing or CA.

If one feels the 40IFE is superior for the application one needs, one can simply use it with all auto aperture controls intact on any H body via the CF adapter.

Proof of the pudding? Okay, I've been reluctant to post this test series because the subject sucks ... but that is not relevant in this discussion. However, since you asked ... here's an S2 example of purple fringing the likes of which I have never experienced with any other MF lens made by anyone. I also uploaded a similar shot from the M9 and 24/1.4 which also notoriously produces fringing in similar lighting scenarios ... but is of the type that is much more easily corrected manually if needed (images A & B), and forgivable given the speed of the aperture ... stopping down virtually eliminates it.

Both are done at ISO 640 ... the S2 70mm lens was at f/3.4, the M24/1.4 was at f/1.5.

So, you are right ... the proof is in the pudding ... but to taste the S2 pudding one has to actually shoot with it in situations one regularly shoots in.

-Marc

 

[doug]

... Maybe camera makers are infected with that bragging rights thing as well...

That's what the high-end Nikon and Canon equipment is. N and C want the general public to see that "pros" (however you define them) use their equipment. The real money is in the low- and mid-level equipment that sells by the hundreds of thousands, the high-end equipment is primarily to attract attention to the brand.

 

[NotXorc]

S2 news around the web today:

1) A mini review by David Saffir (no mention of issues with CA)
http://davidsaffir.wordpress.com/2010/07/08/camera-test-the-leica-s2/

2) A few words by Stefan Daniel on the S2 system (see last paragraph)
http://blog.leica-camera.com/interview/stefan-daniel-interview-part-3/

 

[David K]

The anomaly to my premiss is that when these backs are used on a tech camera ... the optics from Schneider and Rodenstock out perform everything out there, including the Leica S2 lenses. So, I could be completely wrong about everything. Yet, I know what I see in actual practice ... so I'm confused to say the least.

Your thoughts?

-Marc

Marc, I sometimes think you are a mind reader... and one who's better able to express the issues than I can. I've been pondering this question myself ever since I shot the Sinar ArTec. I've never seen such detail and resolution before from a back that I've been shooting for years. So it must be the lenses... right ? What else could it be since nothing else changed, e.g. RAW conversion software, etc ? Based on my own experience with Leica glass as well as what little I can understand from the experts I'm inclined to believe that the S lenses are state of the art and as good or better than anything else out there. So... "where's the Beef ?" (an American colloquialism from an old Burger King commercial for those from other parts of the world).

 

[Jack]

Proof of the pudding? Okay, I've been reluctant to post this test series because the subject sucks ... but that is not relevant in this discussion. However, since you asked ... here's an S2 example of purple fringing the likes of which I have never experienced with any other MF lens made by anyone.

Excellent "proof" Marc, thanks for posting these. This is comparable to the fringing Guy and I saw when we tested the S2 back in January, but it was so bad we simply assumed it had to be a conversion software issue; we only had C1 and LR and neither was handling it.

So while it may not be "an issue" for some shooters, it certainly is for anybody who shoots back-lit scenes with any regularity...

 

[Guy Mancuso]

I just read the 35mm at F4 up is also doing this as well. This thing just needs bloody dedicated software and that is the bottom line. Been saying this since they announced the damn thing and it seems i was right. I said my peace on this issue and i will stand by my comments whether you want to hear them or not or believe it or not because I know Leica fans do not want to hear this and are avoiding it like the plague but the digital world software has become a very very very important element in the process and it gets underestimated a great deal. If that sounds arrogant than sorry leica fans but for me that is my brutal honest assessment and what i have been stressing all along and taken a severe beating from it. I just can't push stuff under the carpet and ignore it. At this level stuff needs to work at full throttle. It maybe one sexy cam but I am not convinced what is under the hood is running on all 12 cylinders.

 

[Mozbee]

If if I'm right (correct me Guy if mistaken), you have always said that since you analyzed your move to MFDB world, when you've tried any system, when you upgraded and so on, pretty much anytime since you really started to touch any MFDB.

At the same time, I remember you saying that lenses are probably less important in MF world compares to 35mm because of the format. It feels like a company that can use to the best all the potential of what they have can kind of perform better compares to any other company where they would the king in one aspect of the chain, but weaker somewhere else.

It is also my understand from reading you for years that this aspect has been one of the main reasons that you chose PhaseOne few years ago and that your choice was quite easy once you realized that. Today, Hassy might be in a stronger position in this regards since that older time of you, now that Hassy has improved their Focus software and they integrate the whole chain so much better than before. That would make it almost a tie on that aspect alone, aside from any other considerations you may have and that would make you preferring PhaseOne like familiarity, dead C1 fan (LOL!) and easier possibility to use your back on a "technical" camera, if needs be!

Guy, am I understanding and interpreting you well?

 

[Mike M]

I've been following the saga of the S2 for awhile now because it's a genuinely interesting camera system. I'm hoping to be an Alpa owner (because of perspective control and the ability to shoot film) but the S2 still interests me a lot.

The thing that's missing from this conversation about alleged S2 aberrations is the relationship between the applied technique of the operator and the system itself. In order to adequately criticize the S2, there should be more effort at showing that problems lie specifically with the S2 and not the technique of the operator. So far, in every photo that I've encountered online with aberrations there is always what I consider to be bad technique too. Is it just a coincidence to often see bad lighting ratios, poor focus, overexposure etc on the part of the operator in S2 shots that contain aberrations. On the contrary, the S2 shots I've seen that had what I consider to be good technique don't appear to have any serious aberrations problems at all. What is the fault of Leica and what is the fault of the operator? Is it Leica's fault if a photographer decides to overexpose highlights on a pavement by as much as 2-3 stops and then complains about aberrations appearing at the line between the shadows and the highlights? Or is that just bad technique? Is it Leica's fault if aberrations occur when a photographer shoots an out of focus subject against a bald sky while overexposing the sky by several stops? Or is that the fault of the operator?

So far, I haven't seen much discussion about whether or not the operator might share some fault or may actually be at fault. I understand that technique can sometimes be a matter of opinion. But I prefer to judge professional systems by how well they perform with proper professional technique. If aberrations start popping up when excellent technique is employed, then that is definitely a sign that there could be a problem with a camera system, lens or software etc. I've owned lenses in the past that performed terrible even with excellent technique (zeiss in particular) But if the technique is bad then what can really be blamed on the camera system?

Some might say that professional camera systems are supposed to work in harsh conditions and should hold up to the stress of bad lighting etc. That may be true for documentary shooters and some other types of photographers that don't necessarily need exercise control over the light and just need to get an image. But the light is still in charge when it comes to most professional photography. No photographers or camera systems have transcended light yet. We are still at the mercy of light, and our technique matters when it comes to capturing the light. The ability to see and differentiate between good and bad lighting is exactly what makes a professional photographer in the first place. It's the same when it comes to sharpness because no photographers have yet transcended scheimpflug and focal plane placement. We are all still at the mercy of focal planes and must properly place them in order for an image to be sharp. The ability to achieve a sharp image with properly placed focal planes, lack of motion blur, lack of diffraction etc is exactly what makes a professional photographer.

If we are going to judge the S2 as a professional system, then I think we should also judge how it works in situations where truly professional and adequate technique is applied too.

 

[Guy Mancuso]

If if I'm right (correct me Guy if mistaken), you have always said that since you analyzed your move to MFDB world, when you've tried any system, when you upgraded and so on, pretty much anytime since you really started to touch any MFDB.

At the same time, I remember you saying that lenses are probably less important in MF world compares to 35mm because of the format. It feels like a company that can use to the best all the potential of what they have can kind of perform better compares to any other company where they would the king in one aspect of the chain, but weaker somewhere else.

It is also my understand from reading you for years that this aspect has been one of the main reasons that you chose PhaseOne few years ago and that your choice was quite easy once you realized that. Today, Hassy might be in a stronger position in this regards since that older time of you, now that Hassy has improved their Focus software and they integrate the whole chain so much better than before. That would make it almost a tie on that aspect alone, aside from any other considerations you may have and that would make you preferring PhaseOne like familiarity, dead C1 fan (LOL!) and easier possibility to use your back on a "technical" camera, if needs be!

Guy, am I understanding and interpreting you well?

Let's see if I can touch on yours and Mikes comments here.

I still believe that the lenses in MF are less important because of the size of the format itself over 35mm which we all agree needs great lenses to perform the best. No argument there on 35mm. Now in the Phase or Hassy system we have a lot of older glass that we can still make use of and it seems with the higher 9 micron sensors this is a more friendly proposition to use this older glass as the 9 micron sensors still provide very good quality with the older glass. Now that the micron sensors are getting smaller we are finding that it is getting maybe more critical with lens design and case in point the P65+ and P40+ for that matter we are seeing the newer D lenses really working well with these sensors and maybe some of the older glass not doing as well as it did on a 9 micron chip. The S2 was designed for 6 micron but my bet even 4 or 5 and we won't see that until we have those sensors and I think we can make the same assumption that Phase D and Hassy newer lenses will also be able to handle the lower sensor micron sizes as well.

Now let's jump into the sensors and what most of what I have said and Jack as well is that software from the dedicated systems like Hassy, Phase, Leaf and Sinar is the software is being made at the sensor level so it is taking in all the attributes of the sensor and making those type of corrections right out of the gate. So what we see in our lenses is less aberrations right from the start because the sensor/ lens combination is being addressed early on in the software even if our lenses are not designed for lens corrections. Yes we still see them ( Mike your part) but we see this on less occasions and true some of this is operator error and we should never forget this with backlight and overexposed images we can only save the day for so long with software. But my believe is these dedicated packages in Hassy and others this is all being done in the firmware and algorithms built into the back and right out of the gate . Guess you call it total system integration they are all designed to work together. Case in point I take a Phase file stick it in ACR or LR and it does not handle it well at all, bad noise levels , bad color and more issues showing up in there software. Not the softwares fault because it does not know how to read them , just like C1 and the S2 it is dumb on how to read the S2 files and really not a great software package at this point for the S2.

Now take that info and let's apply it to the S2 and let's start with the lens are probably damn good and they are but the issue i believe is the sensor and lenses are not being worked up like dedicated software where the engineer can take it and integrate it into the software package. Now Adobe is not working at this level with Leica s2 files sure they are working to see the DNG well but they are not getting all the way down to the sensor level of corrections together in a dedicated package like Phase and others. Now Leica will say we are working with Adobe and that maybe true but I am talking very deep down to the sensor level with a lot of adjustments and corrections with firmware and algorithms built in. Have we ever seen a pure raw file. It looks like crap but as soon as lets say in Phase case we bring that into C1 it is a whole different file and a lot of corrections automatically happen. We can pick this apart in many ways but even in the case of the S2 file between C1 and LR there are vast differences in the way it looks. This clearly says to me hoe the software is interrupting the files be it bad or good there is a major difference, so to me that tells us a whole lot about software and the power of it.

So the Ca we are talking about in this thread it may not be so much the lens itself but the sensor lens combination and the software to interrupt and correct it right at the base level. To me it seems pretty excessive of what we are seeing over other systems and that is not Leica of what we know about them and there lens designs which are very high up there. So this has to be coming from the combination of sensor/lens and software to interrupt all that and fix it at the default level which i don't believe it is. Now I am not talking about after the fact but even at default this is way to high a issue. Not saying LR is bad far from it but not everything from the very basic to very top of the integration chain is working. Let's face it at 6k for a lens touted to have lens correction should we be seeing this, it's only logical that this should not be this excessive. Sure we will see it and same with other systems but this to me is way too much given Leica's reputation on lens design. It has to be deeper than that.


I'm on a roll here than gotta run . But our minds are still fixated on film and in those days all that made the difference was the glass . Film was all the same and the body meant nothing and the biggest key was the processing meant nothing. Today the sensor and the processing are the key differences than the film days. Now the sensor at the basic level of lets say coming out of the oven than into a cam is nothing but now the changes start at the adjustments, firmwares and algorithms that make that a viable sensor to work in photography regardless of system at this point. Than a OEM gets it and the engineers start tweaking it for there cams and to work effectively. At this point it goes into the box for shipping but let's go back the OEM's that have dedicated software can make even more software adjustments as well and improve upon it more. This I believe is the key to the total system integration take all the parts and make them work as a team effectively to knock out as much of the issues right at the default level. Than add more tools like a CA adjustment to improve upon it further. Some systems will have this some will rely on 3rd party to provide it but how much R&D is that third party going to go that deep on a cam when it supports hundreds of cams. Now I am maybe only speaking of the 10 percent that get's left on the editing floor, I don't have a number here folks but my believe is that 10 percent can make a big difference. Now maybe I am all wet here since i am not a engineer or scientist but my believe is these dedicated software packages make the difference and I really discovered this when I bought a Phase back and seen what C1 really did for it over and above what I already seen before it and I loved C1 before i even bought a back. But the combination together is what makes it sing and I am sure Hassy, Leaf and Sinar folks will say the same of there software packages in the IQ department. Maybe a kludge to work with but the IQ is great.

Anyway that is what I believe it is after 20 years of just doing digital my final conclusion is dedicated software. It makes a difference at least in my eyes.

 

[markowich]

attached are 100% crops of S2 files (in camera DNG, LR3 conversion, no PP),
taken with the 35mm f2.5 lens. the series starts at f2.5 and then the f stop is increased at half-stop values. admittedly this is an extreme example but still, these artifacts have been unseen before in the MF camp. LR3 is just hopeless for this. after paying about 40.000.-euros for the 4-lens S system i definitely deserve a dedicated RAW converter, not this LR c....
peter

 

[Guy Mancuso]

Peter the very last image I see is F4.8 with it still there. So would it be safe to assume at 5.6 this is clearing out. Now I would say extreme case here but not out of the realm of reality as I see plenty of detail in the window frame and fairly normal of a whole scene of the room itself without extra lighting of course to balance it all out.

 

[hcubell]

attached are 100% crops of S2 files (in camera DNG, LR3 conversion, no PP),
taken with the 35mm f2.5 lens. the series starts at f2.5 and then the f stop is increased at half-stop values. admittedly this is an extreme example but still, these artifacts have been unseen before in the MF camp. LR3 is just hopeless for this. after paying about 40.000.-euros for the 4-lens S system i definitely deserve a dedicated RAW converter, not this LR c....
peter

Peter, do you have a 35mm H lens for your H3DII-50 to do a similar test with and without lens corrections in Phocus? Would be interesting.
Best, Howard

 

[dfarkas]

attached are 100% crops of S2 files (in camera DNG, LR3 conversion, no PP),
taken with the 35mm f2.5 lens. the series starts at f2.5 and then the f stop is increased at half-stop values. admittedly this is an extreme example but still, these artifacts have been unseen before in the MF camp. LR3 is just hopeless for this. after paying about 40.000.-euros for the 4-lens S system i definitely deserve a dedicated RAW converter, not this LR c....
peter

Peter,

You said here you used no post processing. Was Defringe set to off? Leica recommends that this be left to All Edges by default. I think you'll see it makes a pretty big difference for the effect you are seeing here. Also, just for kicks, try setting the process method to 2003 version instead of the default 2010. I've found that the newer processing, while better overall, tends to make overexposed edges more "glowy".

David

 

[David Grover / Phase One]

Fuji makes lenses which simply cannot be used without heavy software correction (it cannot even be used with only a few years old H-boodies!) to implement cheaper optical designs (that becomes ovbious when you compare it to the optical performance to the last new Zeiss-design, the 40IF - it's superior regarding resolution and CA) - that's a big difference in comparison to Leicas approach to make designs as good as possible with the option to later control (most optical defects cannot be fully "corrected" at least not witout trade-offs) remaing aberrations within software.

Nonsense as we have spoken about before.

The HC lenses were developed before any kind of lens correction via software existed.

The HCD lenses (28 and 35-90) were designed with extra thought on what could be achieved with the additional consideration of lens corrections via software.

Therefore with the HCD28 lens we were able to build something smaller and lighter than before, with excellent edge to edge sharpness and zero distortion. Try doing that completely optically for a 28mm lens and achieving the same results at the same cost.

Ultimately the end result whether it involves corrections or not is the important thing to the photographer.

Any lens will be improved with digital corrections, as we have shown by including V lens corrections as well within Phocus.

Please stop spreading slander and misinformation, as usual.

 

[Jack]

Some facts:

1) It is physically impossible to design a "perfect" lens,

2) As digital resolution increases (pixels get smaller) lens anomalies are easier to find

3) When designing lenses, correcting one anomaly generally increases another. (For example, as you adjust your design to reduce illumination falloff, you'll increase linear distortion.)

3a) Thus lens design becomes a series of tradeoffs for lens designers; ultimately they do what they can to optimize the overall performance, though different companies often bias towards different sets of tradeoffs

4) #1, 2 and 3 become harder still when a fixed lensmount/flange-focal distance requires more extreme retrofocus or telephoto optical designs and/or as lens free aperture size increases

5) Digital corrections can work in conjunction with a well optimized lens to get one closer -- nay, VERY close -- to optical "perfection"

Some corolarry's:

A) #4 above is two of the main reasons why digital-specific tech camera lenses are freer of anomalies than fixed-flange lenses

B) Many fixed-flange MF lenses are "good enough" when used with today's highest resolution digital backs, the problem arena typically lies with extreme wide angle (heavy retrofocus) and optically fast lens designs

C) Any current high-end digital "system" should be using BOTH optimal lens design AND digital corrections for their proprietary lenses that need it -- moreover, and any company doing so should be applauded.

 

[markowich]

Peter,

You said here you used no post processing. Was Defringe set to off? Leica recommends that this be left to All Edges by default. I think you'll see it makes a pretty big difference for the effect you are seeing here. Also, just for kicks, try setting the process method to 2003 version instead of the default 2010. I've found that the newer processing, while better overall, tends to make overexposed edges more "glowy".

David

david,
i tried all that and LR3 cannot cope with it satisfactorily, at f2.5 to f4.0.
defringe in LR is a joke. beyond f4.0 things get sorted using cyan/red at -15 or so.
peter

 

[brianc1959]

Nonsense as we have spoken about before.

The HC lenses were developed before any kind of lens correction via software existed.

10+ years ago you could download Panorama Tools for free and do a superb job of correcting all manner of lens issues (regular distortion, gullwing distortion, lateral color, etc.). I started doing this in ~1999, and it was a well-established procedure even then.

Have there really been substantial improvements in software correction over Panorama Tools of the mid-1990's? (which is still available as far as I know, and is still free).

 

[markowich]

Peter, do you have a 35mm H lens for your H3DII-50 to do a similar test with and without lens corrections in Phocus? Would be interesting.
Best, Howard

howard,
i do have an HC 35mm lens but unfortunately it is in my UK home and now i am in vienna. anyway, it would be an unfair test -sort of- because the HC lens starts at f3.5, where the S 35mm fringing already becomes moderate and kind of controlable, even in LR. i did try with the HCD 28mm and it is much better in controlling CA when using phocus. i do not blame leica's lens design, they developed (relatively) high speed lenses but we pay the price for their decision not to invest in software.
peter