Questions about Leica M lens performance on M11 (new information on Reid Reviews site, 3/22)

[Overgaard]

Advertisement (gone after registration)

5 minutes ago, mzbe said:

Thorsten, thanks for the detective work! I have bravely/foolishly decided to keep my M11 order and received shipping notification today, should be here in a week. Nothing beats first hand experience, I promise to test and report back on purple fringing with all my current lenses, including Summilux 21/28 and Noctilux 50.

Well, if it's not going to work out, I firmly believe we can sell our M11 cameras at premium price to someone who doesn't read this forum 🙂

 

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[harmen]

Technical question: is there any aspect of sensor design that has an impact on purple fringing?  As far as I understand axial chromatic aberration, it is the lens not focusing all the colours in the same plane.  Other than rigorously filtering out violet, what aspect of the sensor design plays a role?  Sure, more pixels mean that it’s easier to notice when looking at 100%.  I’ve not noticed a difference due to exposure setting, in the sense that underexposing the original capture makes the purple less obvious only until I raise the exposure setting in post processing at which point it looks just the same in high contrast areas.  In other words, is it entirely a ‘feature’ of the lens which is particularly strong when having the aperture wide open?

[Guest]

2 hours ago, harmen said:

Technical question: is there any aspect of sensor design that has an impact on purple fringing?  As far as I understand axial chromatic aberration, it is the lens not focusing all the colours in the same plane.  Other than rigorously filtering out violet, what aspect of the sensor design plays a role?  Sure, more pixels mean that it’s easier to notice when looking at 100%.  I’ve not noticed a difference due to exposure setting, in the sense that underexposing the original capture makes the purple less obvious only until I raise the exposure setting in post processing at which point it looks just the same in high contrast areas.  In other words, is it entirely a ‘feature’ of the lens which is particularly strong when having the aperture wide open?

Good question. I like Imatest as a somewhat reasonable source of information: https://www.imatest.com/docs/sfr_chromatic/

Please note difference between Chromatic Aberration and Purple Fringing - both defined and explained on the page. Some of the samples from M11 looked to me like Purple Fringing (which doesn't exclude CA going on at the same time ..). Ultimately the path of light traverses the lens, the top layers of the sensor stack (including e.g. filters), then hits the photo sites on the sensor. The geometry of the whole shebang matters (hence different lenses performing differently on different cameras) as well as the "causes of unintended side effects" within each component ... There's a reason why so much of modern lens design is about optimizing trade-offs, not about being able to nail down an optimal value in a single dimension.

[harmen]

That is helpful!  What it emphasizes is that if the effect is stronger when over-exposing and visible on any lens, then it’s sensor overload causing purple fringing.  When it is especially strong with wide open aperture and specific lenses, then it is chromatic aberration.   I certainly see chromatic aberration with my 28 lux.  It is not dependent on exposure.  It is dependent on aperture.  It is visible also in the center of the image in tangential but also a little in radial direction.  I guess the sensor is simply very good at picking it up.

[wizard]

vor 12 Minuten schrieb harmen:

 I guess the sensor is simply very good at picking it up.

I think that is a very valid observation. The better the sensor, the more flaws you will be able to see which so far went unnoticed. It is the same with HiFi audio equipment. The better your equipment is, the more you will notice any deficiencies in the original recording. Which can be frustrating at times.

[harmen]

I must admit to photographing a desklamp repeatedly last night.  I’ve yet to run into being frustrated by the results for an image I actually wanted to keep 😁

[Erato]

Advertisement (gone after registration)

43 minutes ago, harmen said:

I must admit to photographing a desklamp repeatedly last night.  I’ve yet to run into being frustrated by the results for an image I actually wanted to keep 😁

Please wait for the upcoming firmware update and see if the issue resolve.

[250swb]

20 hours ago, Overgaard said:

The piece by https://www.reidreviews.com is interesting and I think you can read and conclude one or the other. He test f/2.8 lenses, and for me the 21/1.4 and 28/1.4 are the real Leica jewels. 

And then Leica is coming with firmware update (at some point) to fix CR/purple fringing and more. It's almost impossible to conclude anything before that arrives. I think what https://www.reidreviews.com concludes on page 8 is right and a good guide.

He does do lenses testing of course, but when he's doing camera testing he pretty much sticks to the same lenses not because they are the best (although maybe because it's what most people use) but because that is how tests are done and how cameras can be compared with each other. Some lenses may coincidentally work very well with an M11, but it flatters to deceive if we don't discover if a camera falls down using an otherwise perfectly serviceable set of lenses.

[Overgaard]

11 hours ago, harmen said:

Technical question: is there any aspect of sensor design that has an impact on purple fringing?  As far as I understand axial chromatic aberration, it is the lens not focusing all the colours in the same plane.  Other than rigorously filtering out violet, what aspect of the sensor design plays a role?  Sure, more pixels mean that it’s easier to notice when looking at 100%.  I’ve not noticed a difference due to exposure setting, in the sense that underexposing the original capture makes the purple less obvious only until I raise the exposure setting in post processing at which point it looks just the same in high contrast areas.  In other words, is it entirely a ‘feature’ of the lens which is particularly strong when having the aperture wide open?

I stumbled over this 2010 discussion of M9 and Noctilux purple fringing. Same story repeats that CA/purple fringing can be both a lens chroma and microlens problem"

 

[LBJ2]

Has anyone ever tried using a Zeiss UV filter on the Noctilux 50 0,95 to see if that helps to reduce purple fringing? I think I remember reading somewhere the Zeiss UV filter in particular could be effective at limiting purple fringing. 

[IkarusJohn]

5 hours ago, 250swb said:

He does do lenses testing of course, but when he's doing camera testing he pretty much sticks to the same lenses not because they are the best (although maybe because it's what most people use) but because that is how tests are done and how cameras can be compared with each other. Some lenses may coincidentally work very well with an M11, but it flatters to deceive if we don't discover if a camera falls down using an otherwise perfectly serviceable set of lenses.

Really?  Are Sean’s lenses the ones most people use?  Leica’s great strength is not so much its cameras as its fine lenses.  Do most people spending $8,000 on an M11 use colour skopar lenses?  Sean’s choice of lenses is one of the reasons I find many of his reviews of little interest.  If he lived around the corner, I’d lend him mine …

[250swb]

17 minutes ago, IkarusJohn said:

Really?  Are Sean’s lenses the ones most people use?  Leica’s great strength is not so much its cameras as its fine lenses.  Do most people spending $8,000 on an M11 use colour skopar lenses?  Sean’s choice of lenses is one of the reasons I find many of his reviews of little interest.  If he lived around the corner, I’d lend him mine …

Well Leica wouldn't have been able to address the rainbow effect with the M9 without acknowledging the lenses that created it. It seems logical that firmware and technology should be tuned for all lenses doesn't it, or should people run out to buy another set of compatible lenses every time they buy a new body?

[lct]

To me at least lenses like Elmarit 28/2.8 asph v1 and Skopar 21/4 are sort of benchmarks to test corner rendition on M and adapted M-mount cameras. Same for Summicron 28/2 and SEM 21/3.4 which can exhibit more or less smeared corners depending upon cameras. Using more "perfect" lenses can be of little help to me from this stand point. YMMV.

Edited March 25, 2022 by lct

[pgk]

32 minutes ago, IkarusJohn said:

Leica’s great strength is not so much its cameras as its fine lenses.

An interesting comment. I would amend it to say that you should insert optically in front of fine if you are referring to M lenses that is. The optical excellence of Leica's M lenses is both an asset and an Achilles heel though as the mechanical interface and simple 6-bit coding only allow for basic corrections to be made to the resulting image files. I wonder where mea are headed with lenses and cameras if we want to get the 'ultimate' out of them. Lenses with identical angles of incidence to the sensor would seem a logical solution with a sensor optimised for them. I wonder if this is the aim of lenses produced for the SL? With regard to performance on cameras like the M11, well lenses should perform well enough but will ultimately only do so because of their optical excellence (or otherwise).

[IkarusJohn]

49 minutes ago, 250swb said:

Well Leica wouldn't have been able to address the rainbow effect with the M9 without acknowledging the lenses that created it. It seems logical that firmware and technology should be tuned for all lenses doesn't it, or should people run out to buy another set of compatible lenses every time they buy a new body?

Voigtlander‘s colour skopar is indicative of all lenses?  Not sure I said anywhere that you should only buy compatible lenses for each camera (though I wouldn‘t buy incompatible lenses given the choice).  For a Leica camera, I am interested in testing with Leica lenses - or at least those which are likely to be problematic.

Maybe the colour skopar is a good benchmark.  I wouldn‘t know, as I bought into Leica for the lenses.  His testing of such lenses is not very informative for me.

[IkarusJohn]

33 minutes ago, pgk said:

An interesting comment. I would amend it to say that you should insert optically in front of fine if you are referring to M lenses that is. The optical excellence of Leica's M lenses is both an asset and an Achilles heel though as the mechanical interface and simple 6-bit coding only allow for basic corrections to be made to the resulting image files. I wonder where mea are headed with lenses and cameras if we want to get the 'ultimate' out of them. Lenses with identical angles of incidence to the sensor would seem a logical solution with a sensor optimised for them. I wonder if this is the aim of lenses produced for the SL? With regard to performance on cameras like the M11, well lenses should perform well enough but will ultimately only do so because of their optical excellence (or otherwise).

Not sure I understand, Paul.

The M lenses have to provide the best image with minimal correction, as did the R lenses. In many ways, that is the challenge, and a lot of their appeal.  The Zeiss lenses (Otus?) are the same, without the compromises Leica has to struggle with to reduce size.  The M APO Summicrons are the pinnacle of this, at a price.  Conversely, the SL lenses have software corrections and little compromise to size, which would seem to result in even better performance.  Some have criticised this (was it Sean Reid who disabled the software corrections on one of his SL tests and concluded that meant the lens wasn‘t as good as it could/should have been?).

One of the reasons I keep coming back to my M cameras is because the few lenses I have (how many lenses do you need between 21mm and 75mm?) is that they perform largely consistently with my M-A, my Monochrom and M10-D.  Do they perform as well as the SL primes on my SL?  No.  Do I notice? No.  We‘re dancing on the head of a pin.  Do they perform as well on the M11?  Apparently not.  Would I gain anything by getting an M11 (and selling some other gear)?  I‘ve decided I wouldn‘t.

Speaking purely for myself, I do not chase technical perfection in my lenses or cameras.  I look at images and read tests, and I then buy what I think will be the best for me that I can afford.  I thought the 35 Summilux-M FLE looked good.  I was wrong - it didn‘t work for me.  Similarly the 75 APO Summicron, the 35 Summicron, the 90 Summicron and the 28 Summicron.  I think (hope) I have settled on lenses which do work for me - they might not be perfect, but their imperfections do not bother me.

Ultimately, it‘s content which counts.

[250swb]

25 minutes ago, IkarusJohn said:

Maybe the colour skopar is a good benchmark.  I wouldn‘t know, as I bought into Leica for the lenses.  His testing of such lenses is not very informative for me.

You ignore the point, he uses lenses with idiosyncrasies to test the camera not the lens. There are a vast wealth of M compatible lenses by Leica and others, and Leica at the moment are still trying to specify a sensor and firmware that deals with older lenses designed for film. So the idea is to choose those that have understandable problems with digital sensors to test and see how the new camera deals with them, does it make it worse, or better? Forget for the experiment whether you like or dislike a lens, it's only to gather data.

[pgk]

36 minutes ago, IkarusJohn said:

Not sure I understand, Paul.

We are as you say 'dancing on the head of a pin'. Today's lenses, optical or optical combined combined with software correction, are extraordinarily good. But many are still demanding better (apparently we do need better lenses to produce even better photographs). As I see it we are going to end up with specialist lenses for specialist applications if we are not careful much like the graphic lenses to be used at specific apertures and specific reproduction ratios, but we want them to work in variable situations. And we want cameras to be ever better.

This all comes at a price and that price is unsurprisingly precision technique during usage. M lenses are small jewels capable of extraordinarily good results. But they are unlikely to remain the pinnacle of imaging 'quality' if that is, we persist in chasing the dream of technical specifications.

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[Tailwagger]

9 hours ago, harmen said:

...I certainly see chromatic aberration with my 28 lux.  It is not dependent on exposure.  It is dependent on aperture.  It is visible also in the center of the image in tangential but also a little in radial direction.  I guess the sensor is simply very good at picking it up.

As I mentioned elsewhere, I have this thing about fringing. I f-ing hate it. It rips at my soul, gets me all wound up and results in my entering the red mist (a term from racing where something pisses you off to the point where you're intent on crashing into anything just to be done with it).  And of course when I do manage to go asymptotic, thinking clearly isnt on the table.  Hence some of my rather strong posts on the problem. 

But I've got a gig this weekend that is going to require I shoot with the 28 lux outdoors in what I expect to be strong light. So I decided to run an experiment shooting a scenario that was sure to induce excess purple across a series of apertures to better characterize things and build a mental picture of what I can and can't get away with.  I shot a scene with lots of bare branches, into bright sun at F16, F8, F4, F2 and F1.4.  In every instance there was evidence of fringing.  And as you'd expect, the amount increases as the aperture widens. But having now calmed myself, I also decided to finally get off my ass and do a profile of lens and camera to see if that had any positive effects. That said, I'm still a cheap, lazy bastard so I still haven't summoned either the money for a K meter nor the energy to use anything other than AWB.  

My impression following this quick test is that there are likely at least two things going on. First, yeah, more pixels, more fringe. To be expected. But after some basic processing, I also suspect that things are amplified by the AWB issues and the lack of an official profile.  As has been suggested elsewhere, the M11 seems to favor a purple/magenta cast, clearly evident below.  After doing some basic rebalancing via the dropper and applying the profile, I've concluded that while the problem by no means disappears, it becomes far more manageable. 

Worst case scenario at F1.4. All that was done set the camera profile to adobe standard, press auto, adjust white balance, apply the profile I generated.  By F4, there's almost no need to apply any specific defringing. Obviously a single test doesn't prove much, but I thought it was worthwhile path to share as a potential strategy to lessen the effect.  

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

Edited March 25, 2022 by Tailwagger

[IkarusJohn]

12 minutes ago, pgk said:

We are as you say 'dancing on the head of a pin'. Today's lenses, optical or optical combined combined with software correction, are extraordinarily good. But many are still demanding better (apparently we do need better lenses to produce even better photographs). As I see it we are going to end up with specialist lenses for specialist applications if we are not careful much like the graphic lenses to be used at specific apertures and specific reproduction ratios, but we want them to work in variable situations. And we want cameras to be ever better.

This all comes at a price and that price is unsurprisingly precision technique during usage. M lenses are small jewels capable of extraordinarily good results. But they are unlikely to remain the pinnacle of imaging 'quality' if that is, we persist in chasing the dream of technical specifications.

All of which reinforces my decision to stick with what I have, and increases my worry that Leica will not produce a camera which appeals to me in the future - the APS-C owners are furious about the lack of a CL2; I have an SL, but prefer the M system; but I don‘t want 60MP, increased pressure on technique, cropping and pixel binning.

I love the fact that Leica are always trying to do better, but I would love to be able to buy the best 24MP, M mount camera that can be done in, oh, 5 years?