Why not 47MP , is this the limit of M system.

[Einst_Stein]

Advertisement (gone after registration)

To increase MP, taking the 47MP sensor from SL2 or Q2 seemed to me a natural choice, or even more without video. 

Not less. 

Does the 40Mp indicates it’s the limit of M system?

 

[Jeff S]

11 minutes ago, Einst_Stein said:

To increase MP, taking the 47MP sensor from SL2 or Q2 seemed to me a natural choice, or even more without video. 

Not less. 

Does the 40Mp indicates it’s the limit of M system?

 

No, the sensor is cut from the same sensor wafer for the 64Mp medium format S3, and that translates to 40.89 MP for the 35mm format. The SL2 and Q2 sensors were based on the same chip as in the Panasonic S1R as part of that partnership.  
 

Jeff

Edited August 3, 2020 by Jeff S

[Advertisement]

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

[jaapv]

No it does not indicate a limit. That depends on the sensor technology of the future. The sensor of the Q2 or SL2 are not suitable for use in the M camera, given the specific requirements of the M system and its legacy lenses. Sensors are part of an optical system and tailored to the use they are intended for. The basic technology comes from the wafer they are cut out of, the rest of the design depends on microlenses and filter array specifications.

[Einst_Stein]

6 minutes ago, jaapv said:

No it does not indicate a limit. That depends on the sensor technology in the future. The sensor of the Q2 or SL2 are not suitable for use in the M camera, given the specific requirements of the M system and its legacy lenses. Sensors are part of an optical system and tailored to the use they are intended for. The basic technology comes from the wafer they are cut out of, the rest of the design depends on microlenses and filter array specifications.

Meaning, M lens on SL2 would not perform well as expected from the 47MP sensor?

 

[Jeff S]

11 minutes ago, Einst_Stein said:

Meaning, M lens on SL2 would not perform well as expected from the 47MP sensor?

 

Not because of the MP; M lenses are optimized to work with M cameras/sensors, including cover glass, filter stack, micro lenses, etc.

Jeff

Edited August 3, 2020 by Jeff S

[Einst_Stein]

6 minutes ago, Jeff S said:

Not because of the MP; M lenses are optimized to work with M cameras, including cover glass, filter stack, micro lenses, etc.

Jeff

You mean classical M lenses? It is hard believe that M would continue whatever is was. 

It still sounds to me something to do with the M mount limitation, not the M glass.

Edited August 3, 2020 by Einst_Stein

[jaapv]

Advertisement (gone after registration)

28 minutes ago, Einst_Stein said:

Meaning, M lens on SL2 would not perform well as expected from the 47MP sensor?

 

That is a meaningless question. A lens performing well has nothing to do with the MP count of the sensor. But to transpose the situation: a sensor for an M camera is even more designed for the high incidence angles of legacy M lenses than the SL2 is, notwithstanding the fact that Leica did consider the use of M lenses on the SL cameras. It is all about shifting the design compromises.

[Jeff S]

18 minutes ago, Einst_Stein said:

You mean classical M lenses? It is hard believe that M would continue whatever is was. 

It still sounds to me something to do with the M mount limitation, not the M glass.

Huh?  See what Jaap wrote.

Jeff

[jaapv]

21 minutes ago, Einst_Stein said:

You mean classical M lenses? It is hard believe that M would continue whatever is was. 

It still sounds to me something to do with the M mount limitation, not the M glass.

It is not about the mount, after all, the adapter for the L mount has the same  M mount on the working end. It has to do with the design of legacy lenses. Leica will always maximize retro-compatibility within a system.

[Einst_Stein]

1 hour ago, Jeff S said:

Huh?  See what Jaap wrote.

Jeff

Frankly, I can’t comprehend you two yet. 

It is hard to understand why M10R choose the downgrade MP, ..., yes, I know, down grade MP does not impliy down grade IQ, but it is still suspicious.

i read your words, but I can’t extract meaningful content .

 

[Dennis]

1 hour ago, Jeff S said:

The SL2 and Q2 sensors were based on the same chip as in the Panasonic S1R as part of that partnership.  

I was reading this today. The S1R is an amazing camera. If I had to buy a camera with an EVF with more than 24MP, it would be probably the S1R.

[SrMi]

9 minutes ago, Dennis said:

I was reading this today. The S1R is an amazing camera. If I had to buy a camera with an EVF with more than 24MP, it would be probably the S1R.

M lenses work worse on S1R than on SL2. I have both and prefer SL2. That said, S1R is a mighty fine camera with several desirable features that are missing in SL2.

[Jeff S]

16 minutes ago, Einst_Stein said:

Frankly, I can’t comprehend you two yet. 

It is hard to understand why M10R choose the downgrade MP, ..., yes, I know, down grade MP does not impliy down grade IQ, but it is still suspicious.

i read your words, but I can’t extract meaningful content .

 

It didn’t downgrade MP; it was increased from 24MP in the M10 to 40.89 MP using the same sensor architecture as in the M10 Monochrom and the 64MP S3. Like many companies, Leica spreads costs by using the same wafer to cut sensors for multiple products.  They then bond cover arrays, filter stacks, etc to the sensor that maximizes performance for each respective product.  The M10R sensor architecture improved in many more ways than just MP, as Stefan Daniel explains, including 10% higher photon capture, more active vs inactive (supporting electronics) sensor area, and more. 
 

The SL2, Q2, and S1R sensor architecture and technology is different, developed at a different point in time, but was likewise shared to create economies of scale.  
 

How these various product lines evolve in sensor technology, and MP count, as well as many other characteristics remains to be seen. Leica will decide not just what’s feasible, but what will sell at a desirable price and margin.

Jeff

[SrMi]

27 minutes ago, Einst_Stein said:

Frankly, I can’t comprehend you two yet. 

It is hard to understand why M10R choose the downgrade MP, ..., yes, I know, down grade MP does not impliy down grade IQ, but it is still suspicious.

i read your words, but I can’t extract meaningful content .

 

Using the same sensor wafer that M10M, M10-R and S3 have, you cannot fit more than 40MP on the 35mm sensor format.

According to Leica, the more commonplace (and cheaper) sensor wafer used in SL2 and Q2 would not perform as well as the sensor used in M10-R. 

[Jeff S]

3 minutes ago, SrMi said:

M lenses work worse on S1R than on SL2. 

In part due to one less filter layer for the SL2.  This is an example how the same sensor architecture doesn’t necessarily translate to the same performance. The same Sony sensors, for instance, have been used with a Fuji X-Trans based camera as well as other cameras incorporating traditional Bayer arrays.  Not simple stuff.

Jeff

 

[rramesh]

The M is fundamentally a rangefinder. Photo taking manual. You see through an optical window and you take a picture. All focusing is within the limits of the human vision. As long as the sensor can provide a similar look it would be deemed sufficient. Barnack though 1MP was sufficient for a postcard sized snap. We have 40MP today. More than sufficient for a wall.

Focusing with an an EVF is different. Photo taking is electronic. You view the image electronically, zoom as close as you want with available technology, focus (manual or let the camera do it) and then take the picture. With technology advancement, it benefits from higher and higher MP (EVF and sensor) and the ability to zoom to the smallest area to confirm focus. Much like a satellite or microscope photo. 

I think as a photographer I prefer the former.

 

Edited August 3, 2020 by rramesh

[adan]

1 hour ago, Jeff S said:

No, the sensor is cut from the same sensor wafer for the 64Mp medium format S3, and that translates to 40.89 MP for the 35mm format.

I have a real problem with this particular "slang" description of the concept. I'm all for idiom in the right place, but this is just far too misleading: it does not sound "hip," it just sounds "uninformed."

One cannot pick up an S3 sensor (or a wafer of S3 sensors) and a pair of shears - or a band-saw - and "cut" an S3 sensor down to make an M10-R sensor.

Sensor CCD/CMOS chips are more than just the waffle-pattern of identical pixels. They have a frame - a border - of a specific design (approximately 24mm x 36mm, or 30mm x 45mm or whatever) that includes reference pixels, and I/O points, and other "house-keeping" necessities, that cannot just be sliced and diced into functioning chips of different sizes.

Sensors are built up from identical individual pixels (of known characteristics such as size and efficiency and light response, and perhaps even the same Bayer filter and microlens specs), in the same way that one can build a shack or a mansion out of the same-spec bricks.

Although even that is a poor metaphor - a sensor is actually assembled in a CAD program as boxes on a computer screen, and only then output as a finished complete "picture" or blueprint (actually, several layered "blueprints") of the entire sensor, which is burned/etched/engraved/fabricated into the silicon wafer** by photolithography, lasers, or other means at the manufacturing stage. As multiple copies of the identical chip (border and all), which then are in fact cut apart at some point.

So long as that is understood, it is quite obvious that Leica and their sensor-maker found it easy to "recycle" the existing pixel architecture of the S3 in a whole new chip, produced on its own wafers. Just as Kodak repeatedly used the identical 6.8micron pixels in units of different sizes (on different wafers) for: Kodak's medium-format CCD sensors, the digital-module-R, the M8, the M9 and the Monochrom v1. Covering about 10 years of production.

And I do agree with Jeff and Jaap that the slightly-larger (and thus less numerous) S3 pixels are a better choice for an M-sensor, with all its special needs for legacy wide-angle lenses - special microlenses on each pixel, and minimized structural pixel "leakage" in particular.

__________________

** silicon wafer: a disk of pure silicon of 200-300mm diameter, sliced thin from a silicon crystal ingot just as a salami is sliced. But useless until a particular architecture is fabricated onto it.

Wafers cannot produce pictures (or anything else) - chip designs cannot produce pictures, until they are fabricated onto a wafer. Therefore I am fortunate that my cameras do not contain "wafers" - they do contain completely fabricated image sensors.

Image link to wikimedia

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!

[Jeff S]

9 minutes ago, adan said:

I have a real problem with this particular "slang" description of the concept. I'm all for idiom in the right place, but this is just far too misleading: it does not sound "hip," it just sounds "uninformed."

One cannot pick up an S3 sensor (or a wafer of S3 sensors) and a pair of shears - or a band-saw - and "cut" an S3 sensor down to make an M10-R sensor.

Sensor CCD/CMOS chips are more than just the waffle-pattern of identical pixels. They have a frame - a border - of a specific design (approximately 24mm x 36mm, or 30mm x 45mm or whatever) that includes reference pixels, and I/O points, and other "house-keeping" necessities, that cannot just be sliced and diced into functioning chips of different sizes.

Sensors are built up from identical individual pixels (of known characteristics such as size and efficiency and light response, and perhaps even the same Bayer filter and microlens specs), in the same way that one can build a shack or a mansion out of the same-spec bricks.

Although even that is a poor metaphor - a sensor is actually assembled in a CAD program as boxes on a computer screen, and only then output as a finished complete "picture" or blueprint (actually, several layered "blueprints") of the entire sensor, which is burned/etched/engraved/fabricated into the silicon wafer** by photolithography, lasers, or other means at the manufacturing stage. As multiple copies of the identical chip (border and all), which then are in fact cut apart at some point.

So long as that is understood, it is quite obvious that Leica and their sensor-maker found it easy to "recycle" the existing pixel architecture of the S3 in a whole new chip, produced on its own wafers. Just as Kodak repeatedly used the identical 6.8micron pixels in units of different sizes (on different wafers) for: Kodak's medium-format CCD sensors, the digital-module-R, the M8, the M9 and the Monochrom v1. Covering about 10 years of production.

And I do agree with Jeff and Jaap that the slightly-larger (and thus less numerous) S3 pixels are a better choice for an M-sensor, with all its special needs for legacy wide-angle lenses - special microlenses on each pixel, and minimized structural pixel "leakage" in particular.

__________________

** silicon wafer: a disk of pure silicon of 200-300mm diameter, sliced thin from a silicon crystal ingot just as a salami is sliced. But useless until a particular architecture is fabricated onto it.

Wafers cannot produce pictures (or anything else) - chip designs cannot produce pictures, until they are fabricated onto a wafer. Therefore I am fortunate that my cameras do not contain "wafers" - they do contain completely fabricated image sensors.

Image link to wikimedia

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!

We went through this already in the following discussion, including posts by you (post 45) and by Nicci (posts 22,38, 73), which directly address the issue and related terminology.

Jeff 

[Einst_Stein]

Now I start to get it. But I would not put the point on the same wafer, even if it really does make the economy sense.

If I understand it right, what you are saying is that with this "new" technology (41MP vs. 47MP), Leica decided to cut back the MPs in exchange of per sensor performance to get a better overall performance (IQ).   I can buy that. This is what I observed comparing the M240 and Sony A7RII (24MP vs. 40+MP, I sold the Sony).

Whether it has to be on the same wafer is irrelevant, and I am not sure it makes engineering sense (to cut cost). I would skip this long and technology intense argument for now. 

Edited August 3, 2020 by Einst_Stein

[Advertisement]

Hallo,

Look here: Leica M

[Tailwagger]

8 hours ago, Einst_Stein said:

Frankly, I can’t comprehend you two yet. 

It is hard to understand why M10R choose the downgrade MP, ..., yes, I know, down grade MP does not impliy down grade IQ, but it is still suspicious.

i read your words, but I can’t extract meaningful content .

 

To understand the problem, look up telecentric and retrofocal lens design.  Just look at the pictures, not need to read anything. Follow that up with a high school physics lecture on the subject of diffraction.  Many wider M lenses, particularly the early designs, are retrofocal. That makes them a problem in the corners due to diffraction. 

Why? In any engineering business things happen for two completely different reasons... engineering and  business.  Perhaps, (business) due to cost, market constraints, cross technology xfers, long standing relationship or even a licensing restriction by Panasonic as to use of the sensor for an ILC is restricted to L-mount.  Or (engineering)  perhaps it just didnt fit. Or its power demands were too high for the smaller M battery. Or  41Mpx is at the current limit of the M microlensing technology.

Most likely?  The Pano sensor is FF,  there is no medium format version of it.  Where does Leica find an MF sensor for the S cameras. Sony sensors are 3x4 not 2x3, so as an off the shelf part, they are out. Therefore if MF is to continue, Leica is forced to do a custom design.  A custom design purely for S cameras is likely to be cost prohibitive. Solution? Kill two birds with one stone,  provide the S with a lower cost sensor by sharing its development and production costs with the M which will sell a 100 fold more cameras than the S.