Leica S-E / S2 or something else

[DB5]

IQ /'eye-'kyoo/ noun. What my camera has that yours doesn't.

:grin:

Matt

That's funny coming from someone who owns an S. Surely you could just make do with a dslr or something?

In the case of your friend, a portrait photographer, when most competition are using a Canon or Nikon, why is it any different? He's using a 20-50+k setup when most are using a 5k setup.

When you are in an ultra competitive market, this sort of thing makes a difference.

 

[Paratom]

Sure it is. It's tangible and those things that contribute to it can be measured.

Then its easy, just tell us your measured IQ-numbers of Phase, Leica and Hassy so there wont be any further missinterpretation.

 

[Bernard]

In the case of your friend, a portrait photographer, when most competition are using a Canon or Nikon, why is it any different?

Are you sure about that? The man obviously isn't running the local grad-portrait, X-mas family photo business.

It seems to me like almost every high-end celebrity/fashion/portrait photographer I see in un-sponsored videos is running a Blad H-something. Sometimes it's a PhaseOne, sometimes a Leica S, but rarely is it 35mm. I think that the only time I ever see 35mm being used at the pointy-end of that business is in the Canon Europe newsletter. Even then, the tone is defensive.

 

[DB5]

Then its easy, just tell us your measured IQ-numbers of Phase, Leica and Hassy so there wont be any further missinterpretation.
If you are talking about resolution and DR, even how to measure DR is not so well defined. And then there are those who say maximising DR and high ISO noise comes on cost of low ISO color quality. And how you measure skin color? I dont doubt there are differences and I have no doubt Phase backs are great, but IMO IQ is a function based on many parameters and only few can be measured and are well defined.

It's not so well defined across various testing platforms, but when you are consistently testing various cameras in the same way then the data becomes relevant. Testing for all these things is a modern thing. It hasn't had decades and decades or a century of methodology like film yet and the people who are doing it are all likely doing it different ways. But these things can be measured and comparisons are easily made.

Colour depth, accuracy, purity, spectrum range, dynamic range, noise, tonality, sharpness and acuity, high ISO, low ISO, how the files work with raw developing software, the systems lens quality, how clean the signal is, the quality of the camera electronics etc etc etc all these things are contributing to IQ and they are all being developed and improved. It's totally tangible.

Maximizing DR comes at the expense of of low quality ISO in poorly developed examples. They are not going to show up in expensive medium format cameras. It's not an issue so far, but that is some of the reason why these obviously cost such crazy amounts of money.

Faithful reproduction of skin colour is measured with colour depth, accuracy, purity, spectrum range, pixel resolution, bayer related colour resolution, and also how the sensor deals with mixed lighting sources and contrast. Most of the difference between a good skin tone and a flat one, at this level, is profiling.

Phase One with C1 and Hasselblad with Phocus are so good because the've been developed cohesively. Hasselblad and Phocus has very good colour science. Sensor, Processing, Firmware, Lens and Software all being developed and integrated by the same teams makes a big difference. It's one of the reasons why Phase One and Capture One has been the standard in high end imaging for so long - it just works.

Some say it's the lens and some say it's the sensor. It's neither, exclusively. It's both together and how well they work together with a great range of variables.

 

[DB5]

Are you sure about that? The man obviously isn't running the local grad-portrait, X-mas family photo business.

It seems to me like almost every high-end celebrity/fashion/portrait photographer I see in un-sponsored videos is running a Blad H-something. Sometimes it's a PhaseOne, sometimes a Leica S, but rarely is it 35mm. I think that the only time I ever see 35mm being used at the pointy-end of that business is in the Canon Europe newsletter. Even then, the tone is defensive.

No true at all. Some of the biggest names in the business are shooting 35mm at the very pinnacle of the industry.

 

[MGrayson]

When you do not have IQ, go for EQ

I hope my sarcasm was clear. I do not believe in IQ. Nor do I believe in flat frequency response. Not that these aren't things that can be measured, just that they don't seem highly correlated with results I (and others) like. Do people go to BMW forums to point out that some other car has better cornering force or shorter braking distance and then question the sanity of the locals? We'll, probably, but it just seems pathetic.

Apologies to the OP. My only advice is to try what you can and see what you like and forget the rest.

Matt

 

[DB5]

I hope my sarcasm was clear. I do not believe in IQ. Nor do I believe in flat frequency response. Not that these aren't things that can be measured, just that they don't seem highly correlated with results I (and others) like. Do people go to BMW forums to point out that some other car has better cornering force or shorter braking distance and then question the sanity of the locals? We'll, probably, but it just seems pathetic.

Apologies to the OP. My only advice is to try what you can and see what you like and forget the rest.

Matt

Well sarcasm or not, it just seems you were implying that people who want to maximise their IQ are just doing so to say their cameras have something the others don't. I'm sure that may be the case for some but for many it just isn't the case.

Wether you choose to care about IQ or not is a personal choice - But to some it is important, and that is what they like.

How is this pathetic?

 

[Paratom]

It's not so well defined across various testing platforms, but when you are consistently testing various cameras in the same way then the data becomes relevant. Testing for all these things is a modern thing. It hasn't had decades and decades or a century of methodology like film yet and the people who are doing it are all likely doing it different ways. But these things can be measured and comparisons are easily made.

Colour depth, accuracy, purity, spectrum range, dynamic range, noise, tonality, sharpness and acuity, high ISO, low ISO, how the files work with raw developing software, the systems lens quality, how clean the signal is, the quality of the camera electronics etc etc etc all these things are contributing to IQ and they are all being developed and improved. It's totally tangible.

Maximizing DR comes at the expense of of low quality ISO in poorly developed examples. They are not going to show up in expensive medium format cameras. It's not an issue so far, but that is some of the reason why these obviously cost such crazy amounts of money.

Faithful reproduction of skin colour is measured with colour depth, accuracy, purity, spectrum range, pixel resolution, bayer related colour resolution, and also how the sensor deals with mixed lighting sources and contrast. Most of the difference between a good skin tone and a flat one, at this level, is profiling.

Phase One with C1 and Hasselblad with Phocus are so good because the've been developed cohesively. Hasselblad and Phocus has very good colour science. Sensor, Processing, Firmware, Lens and Software all being developed and integrated by the same teams makes a big difference. It's one of the reasons why Phase One and Capture One has been the standard in high end imaging for so long - it just works.

Some say it's the lens and some say it's the sensor. It's neither, exclusively. It's both together and how well they work together with a great range of variables.

I agree on most things you write about IQ. It is a short word for a whole lot of parameters and things. Thats why it is so hard to judge one cameras IQ over another cameras IQ once one has reached a certain level. Plus there is still the subjective part, like exact color is not always perceived as pleasing color, etc. etc.

Anyways, it is good that we have choices and everybody can choose his "best" solution.

 

[ErikKaffehr]

Hi Matt,

Perception is fallible and easily affected by any bias the observer may have.

In the example above, the fields 'A' and 'B' have the same colour. But, most observers would see 'A' as dark and 'B' as light.

Scientists and engineers are interested in things measurable, not because they want to rate product 'A' two points higher than product 'B', but because they need objective data. So, lens designers will calculate MTF data for lenses and they will trace ray to achieve point spread functions. Because, doing things like that turns what used to be alchemy into engineering.

Take Carl Zeiss, he used to be an engineer and he was in the business of building microscopes. Carl Zeiss figured that he could improve on the quality of his microscope if the construction would be based on science and not trial and error. So, he employed a physicist called Ernst Abbé to develop methods for microscope lens designs. Abbé developed much of the theory for correction for chromatic aberrations, but they were limited by the assortment of optical glass available, so Otto Schott, a glassmaker, joined the team. So, science laid the fundation of what now is known as Zeiss.

Zeiss needed good metrology, so they developed MTF as an objective way of measuring lens performance. But, labs don't rely on MTF only, they also analyze lenses on the optical bench.

Brandon Dube, an optical designer working with OLAF labs and doing lens work for NASA among others has written an article on how lenses of different quality work with sensor of different resolutions. The article is essentially based on motion devices so it is about 8000 pixels width at Super 35 frame size that is around APS-C. It is a good read for anyone interested in the relation of sensors an lenses.

https://www.lensrentals.com/blog/2017/10/the-8k-conundrum-when-bad-lenses-mount-good-sensors/

Best regards
Erik

I hope my sarcasm was clear. I do not believe in IQ. Nor do I believe in flat frequency response. Not that these aren't things that can be measured, just that they don't seem highly correlated with results I (and others) like. Do people go to BMW forums to point out that some other car has better cornering force or shorter braking distance and then question the sanity of the locals? We'll, probably, but it just seems pathetic.

Apologies to the OP. My only advice is to try what you can and see what you like and forget the rest.

Matt

 

[DB5]

I agree on most things you write about IQ. It is a short word for a whole lot of parameters and things. Thats why it is so hard to judge one cameras IQ over another cameras IQ once one has reached a certain level. Plus there is still the subjective part, like exact color is not always perceived as pleasing color, etc. etc.

Anyways, it is good that we have choices and everybody can choose his "best" solution.

Yes subjectivity and personal preference rings the final bell. In my humble opinion any subjectivity is the place of the individual and should be left with the individual. So correct colour is the perfect destination where the individual is free to alter it in anyway they wish.

 

[ErikKaffehr]

Hi,

The main factors are full well capacity and readout noise. The full well capacity (FWC) essentially determines shoot noise, that is the amount of noise you see in highlights and mid tones. Shot noise is just a function of the number photons captured by the sensor.

DR is essentially FWC divided by readout noise. Both these quantities are measured in electron charges, e-.

So, to increase DR you can increase FWC and/or reduce readout noise. Either way is a gain for image quality. It is possible to gain some advantage in DR by using larger pixels, but it seems to matter little with modern CMOS technology, because readout noise is so low.

The example below shows that large pixel camera, like the Nikon D5 or the Sony A7s have no advantage over small pixel cameras like Sony A7rII and Nikon D850:


Increasing FWC yields lower base ISO. Somehow Nikon has achieved higher FWC so they can shoot at ISO 64. That gives Nikon a tiny advantage.

There are other factors regarding image quality, CFA design is one of them and companies have taken shunts UV and IR filtering on sensors. The Leica M8 was famous for weak IR-filtering. The reason was that they wanted a very thin cover glass, in order to reduce astigmatism with their symmetrical lens designs.

The way things work, I would say that sensor makers invest a lot of resources in improvement and those improvements make sensors better.

Making pixels smaller has some issues with lens designs having large beam angles. But, lenses built for SLR systems don't have large beam angles. So that is not a problem with the Leica S.

The most important factor is probably designs rules. The optimum pixel size varies with design rules. Tighter design rules allow for smaller pixels. Moving wiring behind the pixels also allows for smaller pixels.

Essentially, the optimum is to have so small pixels as feasible, but no smaller.

Best regards
Erik

It's not so well defined across various testing platforms, but when you are consistently testing various cameras in the same way then the data becomes relevant. Testing for all these things is a modern thing. It hasn't had decades and decades or a century of methodology like film yet and the people who are doing it are all likely doing it different ways. But these things can be measured and comparisons are easily made.

Colour depth, accuracy, purity, spectrum range, dynamic range, noise, tonality, sharpness and acuity, high ISO, low ISO, how the files work with raw developing software, the systems lens quality, how clean the signal is, the quality of the camera electronics etc etc etc all these things are contributing to IQ and they are all being developed and improved. It's totally tangible.

Maximizing DR comes at the expense of of low quality ISO in poorly developed examples. They are not going to show up in expensive medium format cameras. It's not an issue so far, but that is some of the reason why these obviously cost such crazy amounts of money.

Faithful reproduction of skin colour is measured with colour depth, accuracy, purity, spectrum range, pixel resolution, bayer related colour resolution, and also how the sensor deals with mixed lighting sources and contrast. Most of the difference between a good skin tone and a flat one, at this level, is profiling.

Phase One with C1 and Hasselblad with Phocus are so good because the've been developed cohesively. Hasselblad and Phocus has very good colour science. Sensor, Processing, Firmware, Lens and Software all being developed and integrated by the same teams makes a big difference. It's one of the reasons why Phase One and Capture One has been the standard in high end imaging for so long - it just works.

Some say it's the lens and some say it's the sensor. It's neither, exclusively. It's both together and how well they work together with a great range of variables.

 

[MGrayson]

Perception is fallible and easily affected by any bias the observer may have.

Hi Erik,

I am extremely aware of this statement and the examples you cite. That is part of the reason that, while good engineering must underly our equipment, it is far from sufficient. If I were doing science, I would want and need the best calibration, sensitivity, and consistency possible. Very few photos on this site, certainly none of my own, are technically limited. We have more control over the technical aspects, and so we obsess over them. But they are far from the top of the list for good photography.

Again, I'm not saying they DON'T matter at all, just that they are a very poor measure on which to base an equipment purchase. When mjr needed to dump his Leica S system for a Phase back and a tech cam, he did it because he DID need the capabilities for a job. I am not addressing such situations. I'm talking about what will help someone engage happily and productively with photography. It is personal - I can't use a camera I don't like using, and because I'm not using it for a job, I don't have to. I could see someone feeling that they're missing something if a higher resolution, DR, whatever solution exists. I have no problem with that, as long as they admit that that's just the way they feel. There is huge resistance for us to admit our opinions. "Leica needs to do X." always means "I wish Leica would do X."

Best,

Matt

 

[DB5]

Tell Penn, Karsh, Avedon that their level of technical excellence doesn't contribute to their photography and they would have laughed you out the door. The level of obsession in their work elevates it to something above and beyond normal human endeavour. Even the most lo-fi and anti-technique of photographers have become great because of their obsession with what they do.

If you don't want to master something, that is your own choice.

 

[Shashin]

Perception is fallible and easily affected by any bias the observer may have.

Sample size can overcome individual bias in studies as well as good sampling methods.

Images are made for the human visual system. This is why, while technical criteria contribute to image quality, they don't define it. The engineers at Kodak understood this. It all comes down to the perception of results. If you take two different camera systems with one having higher numbers for specs such as DR and resolution and the result look the same, then the image quality of both systems is identical. And that is why human judgement is important.

This has alway been the problem in the discussion on diffraction. While diffraction impacts sharpness at the image plane, depth of field increases resolution behind and in front of the image plane. If two images shot at f/8 or f/16 are presented, you cannot determine which has higher image quality just by diffraction alone. A viewer may perceive the image with a greater depth of field as having greater quality. The fact that diffraction examples require two images side by side at 100% view with comments for viewers to look at certain features just indicate that diffraction for the most part does not impact image quality. And as you pointed out, humans are biased observers, which is why the bias toward sharpness affects those who care about the effects of diffraction over the properties of depth of field.

I agree the technical specification of a system are good to know, that is also susceptible to other human biases. In your plots on DR what is missing is context. If scene contrast is below the sensor DR, then the difference in DR is a moot point. People are just bad at math. But you could then argue there are condition that could exceed one sensor but not the other, making more DR preferable. But that ignores human perception. Blown out highlights do not mean that image quality is lower, it might simply be interpreted as a strong light source giving the resulting image a luminous quality. Having detail in the highlights and shadows does not always result in better quality images--HDR kind of shows the problems.

Naturally, image quality is not a comparative metric. Go into a museum and decide which images have high image quality, regardless of the content of the image, and what you will find is there is no correlation between the perception of quality with the technical specifications of the process.

 

[MrSmith]

Tell Penn, Karsh, Avedon that their level of technical excellence doesn't contribute to their photography and they would have laughed you out the door. The level of obsession in their work elevates it to something above and beyond normal human endeavour. Even the most lo-fi and anti-technique of photographers have become great because of their obsession with what they do.

If you don't want to master something, that is your own choice.

The medium should be mastered early on and become second nature no matter what you are photographing, freeing you up to create whatever imagery it is you want to make be it a A list celeb with a short temper or a cloud.

(Just my take on it)

 

[DB5]

The medium should be mastered early on and become second nature no matter what you are photographing, freeing you up to create whatever imagery it is you want to make be it a A list celeb with a short temper or a cloud.

(Just my take on it)

Photography can't be mastered early on, it's a life journey. Second nature without growth is a death. Learning never stops and the goal posts are always moving.

 

[MrSmith]

Second nature without growth is a death

I totally agree with you there, I guess what I'm trying to say is technique should be second nature freeing you up to see and create.

Visual awareness and 'seeing' not half a stop extra of DR or an extra 10lp of resolution is where I want to see gains.

 

[DB5]

I totally agree with you there, I guess what I'm trying to say is technique should be second nature freeing you up to see and create.

Visual awareness and 'seeing' not half a stop extra of DR or an extra 10lp of resolution is where I want to see gains.

I agree with that too. But the rate at which we want technology to improve is unfortunately slower than how fast it does actually develop and it's still less than what we had before in Large Format film. I will still take whatever gains there are which help me to get what I want from my camera to communicate what I want.

I personally think that photographers have a quest to uphold how great photography can be in this age of insta-overload. There's so much crap out there nowadays and the art of photography is slipping into mediocrity. Having a great image isn't enough to me. There was a day that Large and Ultra Large Format was what photographers did. You look at it and it makes your eyes pop and your heart stop. There are very few people flying that flag anymore.

 

[ErikKaffehr]

Hi,

I would not agree on that, at least not fully.

A lot of research has gone into image quality and it is pretty well understood. Take the issue of diffraction.

Very clearly, once you are past optimum aperture and that may be f/4 for a highly corrected system or f/16 for a poorly corrected one, diffraction comes into play. So you can stop down to improve DoF but you loose sharpness in the plane of focus. In digital photography, sharpening also comes into play. You can sharpen more if you have a clean image.

Regarding sharpness, Ed Granger at Kodak developed SQF that is regarded to correlate well with perception of sharpness. In the real world, it is not that easy.

Let us be a bit realistic...

• If we make small images using wide angle lenses a cell phone will achieve very good results.
• If we make large prints with the main subject in focus and don't care about the rest, SQF is a good measure.
• If we want to achieve selective focus, with background out of focus we want good performance at large apertures. SQF will be a good measure, but out of focus rendition may play a major role. Cell phones have small sensors with short focal lengths, so they will not work for this.
• If we want focus from foreground to background, diffraction will be the limiting factor. The only cure is the ability of using tilts for tilting the focal plane.

In general, having low noise levels is always beneficial as it allows for more processing. Except at high ISO or in extreme darks, the only factor determining noise is the full well capacity and how well we utilize it. We need to expose to the right to utilize FWC. If we have a good FWC, like most modern CMOS sensor have, we can be a bit more conservative regarding exposure.

I am quite serious about this. Any modern camera you can buy today is capable of delivering seriously good image quality. Luminous Landscape has an interview with Ctein, whom Kodak regarded the greatest master printer. Ctein now uses APS-C or may be even 4/3 for all his work. It is absolutely good enough for A2 (16"x23") prints and as good as 6x7 film ever was. I don't have seen Ctein's prints from 4/3, but I am pretty sure they would put anything I achieved from my 37x49mm P45+ to shame. Why? Because he knows how to print!

What I would say is:

• Any decent camera today can produce decent pictures. If not so, the photographer is at fault.
• To have the best image quality, you would look for the sensor having the least noise and the highest full well capacity.
• For best tonality, you would look for an efficient CFA paired with a low noise sensor with high full well capacity.
• To extract maximum detail and avoid artefacts you would look for a sensor that matches the resolution of your lens.Excellent lenses need excellent sensors to make them justice.

But, all this fails if the photographer doesn't do an optimal work.

I don't say that you need to be an expert in optics, sensor technology or image processing. But, if you want to make the best of your equipment, a basic understanding of the processes behind will certainly help!

Best regards
Erik

Sample size can overcome individual bias in studies as well as good sampling methods.

Images are made for the human visual system. This is why, while technical criteria contribute to image quality, they don't define it. The engineers at Kodak understood this. It all comes down to the perception of results. If you take two different camera systems with one having higher numbers for specs such as DR and resolution and the result look the same, then the image quality of both systems is identical. And that is why human judgement is important.

This has alway been the problem in the discussion on diffraction. While diffraction impacts sharpness at the image plane, depth of field increases resolution behind and in front of the image plane. If two images shot at f/8 or f/16 are presented, you cannot determine which has higher image quality just by diffraction alone. A viewer may perceive the image with a greater depth of field as having greater quality. The fact that diffraction examples require two images side by side at 100% view with comments for viewers to look at certain features just indicate that diffraction for the most part does not impact image quality. And as you pointed out, humans are biased observers, which is why the bias toward sharpness affects those who care about the effects of diffraction over the properties of depth of field.

I agree the technical specification of a system are good to know, that is also susceptible to other human biases. In your plots on DR what is missing is context. If scene contrast is below the sensor DR, then the difference in DR is a moot point. People are just bad at math. But you could then argue there are condition that could exceed one sensor but not the other, making more DR preferable. But that ignores human perception. Blown out highlights do not mean that image quality is lower, it might simply be interpreted as a strong light source giving the resulting image a luminous quality. Having detail in the highlights and shadows does not always result in better quality images--HDR kind of shows the problems.

Naturally, image quality is not a comparative metric. Go into a museum and decide which images have high image quality, regardless of the content of the image, and what you will find is there is no correlation between the perception of quality with the technical specifications of the process.

 

[DB5]

Hi,

I would not agree on that, at least not fully.

A lot of research has gone into image quality and it is pretty well understood. Take the issue of diffraction.

Very clearly, once you are past optimum aperture and that may be f/4 for a highly corrected system or f/16 for a poorly corrected one, diffraction comes into play. So you can stop down to improve DoF but you loose sharpness in the plane of focus. In digital photography, sharpening also comes into play. You can sharpen more if you have a clean image.

Regarding sharpness, Ed Granger at Kodak developed SQF that is regarded to correlate well with perception of sharpness. In the real world, it is not that easy.

Let us be a bit realistic...

• If we make small images using wide angle lenses a cell phone will achieve very good results.
• If we make large prints with the main subject in focus and don't care about the rest, SQF is a good measure.
• If we want to achieve selective focus, with background out of focus we want good performance at large apertures. SQF will be a good measure, but out of focus rendition may play a major role. Cell phones have small sensors with short focal lengths, so they will not work for this.
• If we want focus from foreground to background, diffraction will be the limiting factor. The only cure is the ability of using tilts for tilting the focal plane.

In general, having low noise levels is always beneficial as it allows for more processing. Except at high ISO or in extreme darks, the only factor determining noise is the full well capacity and how well we utilize it. We need to expose to the right to utilize FWC. If we have a good FWC, like most modern CMOS sensor have, we can be a bit more conservative regarding exposure.

I am quite serious about this. Any modern camera you can buy today is capable of delivering seriously good image quality. Luminous Landscape has an interview with Ctein, whom Kodak regarded the greatest master printer. Ctein now uses APS-C or may be even 4/3 for all his work. It is absolutely good enough for A2 (16"x23") prints and as good as 6x7 film ever was. I don't have seen Ctein's prints from 4/3, but I am pretty sure they would put anything I achieved from my 37x49mm P45+ to shame. Why? Because he knows how to print!

What I would say is:

• Any decent camera today can produce decent pictures. If not so, the photographer is at fault.
• To have the best image quality, you would look for the sensor having the least noise and the highest full well capacity.
• For best tonality, you would look for an efficient CFA paired with a low noise sensor with high full well capacity.
• To extract maximum detail and avoid artefacts you would look for a sensor that matches the resolution of your lens.Excellent lenses need excellent sensors to make them justice.

But, all this fails if the photographer doesn't do an optimal work.

I don't say that you need to be an expert in optics, sensor technology or image processing. But, if you want to make the best of your equipment, a basic understanding of the processes behind will certainly help!

Best regards
Erik

There is no way that an APS-C camera will create the same image quality as Medium Format Digital. No way on Earth. Linear sharpness can be equal but that is all.

Without reading quotes or interviews or fully understanding the context from which this has supposedly been read or taken there is no way APS can be compared with Medium Format film of digital. No way on earth. The only thing it is comparable is linear sharpness, which is the least interesting thing about larger formats. Post can certainly get them closer in terms of the tonality loss but there is no way it will match a Medium Format digital or film print that's been equally printed by a competent printer. Good enough, is another very subjective thing.