Leica Lens MP Resolution?

[jaapv]

Advertisement (gone after registration)

14 hours ago, ChicagoMatthew said:

What’s not to understand? Lenses have a limit to what they can precisely translate to the focal plane... higher quality glass and engineering will result in an image that is sharper, has higher contrast, better color reproduction and generally more detailed image. If that wasn’t the case I don’t think we would be spending so much money on Leica lenses. So, I am asking if Leica lenses, in all their Leica glory, can take advantage of the 61mp of this new sensor.

Maybe you should read Andy's answer again: ANY lens will benefit by a higher resolving sensor, including Leica lenses.

[ChicagoMatthew]

2 hours ago, jaapv said:

Maybe you should read Andy's answer again: ANY lens will benefit by a higher resolving sensor, including Leica lenses.

Okay, maybe I’m just not saying this clearly. I understand that any lens will benefit from a higher mp sensor. Put a kit lens on the A7r4 and the sensor will resolve as much as that lens can possible resolve, but that doesn’t mean it’s reaching the full potential of the sensor. THIS IS WHAT IM ASKING. 

When the 5DSR came out, everyone complained that the 50mp looked soft, and they did. Because the old L lenses couldn’t resolve the the fine details the sensor was capable of. 

So my question is, because the 50lux and the 35fle are film era lenses (albeit modern), will they take advantage of a 61mp sensor? 

Seems like the answer is Yes, some people above have said as much. Also, Thorston O. said they resolve up to 80mp. Which is good, because I’ll want to use them in the new SL, and take full advantage of that sensor. 

[Advertisement]

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

[jaapv]

Well, at least you have taken away that there is no cause for worry. More important is that you will never be able to get all this resolution onto paper unless you print really, really large. Which in its turn means that you can take just three photographs with your MP monster before your walls are full... And you'll have to buy an A1 printer.

[ChicagoMatthew]

54 minutes ago, jaapv said:

Well, at least you have taken away that there is no cause for worry. More important is that you will never be able to get all this resolution onto paper unless you print really, really large. Which in its turn means that you can take just three photographs with your MP monster before your walls are full... And you'll have to buy an A1 printer.

That’s true, for most people, including me, 61mp is overkill, especially if you’re printing and in fact I haven’t printed anything in years. But, there are other ways of viewing pictures and those technologies are changing as well. Computer screens, for example. Also, there is a weird satisfaction in pixel peeping. Zooming deep into a picture and still seeing lots of detail is fun. 

Also, irrational as it may be, I want my images to be the highest quality possible and this is an affordable means to that end. Do I need it? Maybe not now, but down the line who knows. 

[Chaemono]

vor einer Stunde schrieb ChicagoMatthew:

Also, there is a weird satisfaction in pixel peeping. Zooming deep into a picture and still seeing lots of detail is fun.

+1. I like to read street signs in the pictures I take, irrespective of how small they appear in the frame. The really tiny ones are the most fun. 

[jaapv]

Which is something that doesn't interest me at all. My only real objective is the printed image.

[farnz]

Advertisement (gone after registration)

4 hours ago, ChicagoMatthew said:

Computer screens, for example. Also, there is a weird satisfaction in pixel peeping. Zooming deep into a picture and still seeing lots of detail is fun. 

Then of course you're also introducing another 'resolution constraint' - the screen you're viewing the picture on, which might be a mobile phone screen or the largest Eizo.

Pete.

[Chaemono]

More MPx pictures definitely feel crispier even on a mobile phone. It’s all in our head. Just knowing that a picture is made up of 40+ million pixels feels good. I sometimes pinch to zoom when I browse the Web and try to guess the number of megapixels of each photo I encounter. 

[lct]

Sharpness is a bourgeois concept and super sharpness is...

[Chaemono]

When HCB said this, there weren’t any cellphones with Retina displays around, mind you. 

[Jean-Michel]

17 hours ago, ChicagoMatthew said:

That’s true, for most people, including me, 61mp is overkill, especially if you’re printing and in fact I haven’t printed anything in years. But, there are other ways of viewing pictures and those technologies are changing as well. Computer screens, for example. Also, there is a weird satisfaction in pixel peeping. Zooming deep into a picture and still seeing lots of detail is fun. 

Also, irrational as it may be, I want my images to be the highest quality possible and this is an affordable means to that end. Do I need it? Maybe not now, but down the line who knows. 

If you do not print, you really do not need that many megapixels. Do the arithmetic: on an iMac 5k retina it is 5000 pixels across, and, say, 3000 pixels high, that works out to all of 15 megapixels. If you get an 8k screen or television, then you might want something of the order of 50 mpx, I doubt that too many people plan to view their images on a massive 8k tv, and  those who do are unlikely to keep zooming in and out when looking at photos, not too mention the waste of mpx when viewing a vertical rather than horizontal image. 

I salivate at additional mpx’s as much as anyone, but do find that the 18, 21, and 24 mpx that my Leicas and Canon provide are quite plenty. Some years ago, Micheal Reichmann (Luminous Landscape) made large prints from the same scene using a 10 or 12 mpx Canon G9 and by a Phase One digital back (not sure of the back’s mpx), he then asked a bunch of photographers to identify which image was made by the $600.00 or the $50,000.00 systems; none could at first, only by guessing at the different depth of field in the two image did they could discern some differences. So, since I do not plan to make 2 meter wide prints, I personally do not need more mpx’s, still, I would not say no if offered a camera with xxxxxx mpx’s. 

[pico]

18 hours ago, ChicagoMatthew said:

there is a weird satisfaction in pixel peeping. Zooming deep into a picture and still seeing lots of detail is fun.

Occasionally I test a Hasselblad with a 700mm lens on a cool clear day. The subject is a 600' high bluff two miles away. I can pick out people on the bluff under magnification. Once I showed magnified sections of the scanned negatives to a friend who was a Vietnam spook (CIA) and he said, "That's the kind of imaging we used to kill for." 

[adan]

22 hours ago, ChicagoMatthew said:

Put a kit lens on the A7r4 and the sensor will resolve as much as that lens can possible resolve, but that doesn’t mean it’s reaching the full potential of the sensor.

The point I was making is that NO lens will reach the full theoretical potential of a sensor. And no sensor will reach the full theoretical potential of a lens.

Like you, I used to think that a sensor could "outresolve" a lens or vice-versa. That the "weakest link" is the limiting factor. It is an nice, intuitive idea that, unfortunately, is not actually correct.

A little research shows that "sampling theory" and the related math says otherwise. Thus my previous simplifed equation that the total resolution you'll see in a final picture = (approximately) MTFlens x MTFsensor. Increase either, no matter how low the other, and the total increases. But is always less than the full potential of either one alone.

If you want the more complex, Fourier-transform, calculus-based equations: https://en.wikipedia.org/wiki/Nyquist–Shannon_sampling_theorem

A different take: http://www.normankoren.com/Tutorials/MTF.html

A lens samples the world at x-many lines per mm (actually, x varies across the frame and with aperture used). A sensor samples the lens image's lines per mm at Y-many pixels per mm. The two samples interact in complex ways, not just "one bigger than the other." You can get quantization errors any time you break a continuous series down into finite chunks, whether it be fractions of a mm, or pixels, or 1s and 0s.

As a practical example, suppose one "line" resolved by a lens happens to fall on the junction of two pixels. Boom, you now have that line spread over two pixels, each getting 50% of the brightness - or vanishing altogether. That can happen regardless of the actual resolution of either the lens or the sensor.

To reach the "full potential" of a sensor, you must have a lens that has infinite resolution. And to get the "full potential" of a lens, you need a sensor with an infinite number of "dimensionless" pixels. Doesn't exist in the real world.

One can avoid certain sampling problems (aliasing/moire) by using a sensor twice as "sharp" as the lens. But better than "twice as sharp" still adds continuous improvements - better tonal resolution (distinguishing grays or colors), finer "quanta" and so on - all the way up to infinitely-more sharp (which is vastly more - in fact, undefinably more - than 61 megapixels, or even 61 gigapixels, or 61 petapixels).

Now, everyone from sensor-makers camera-makers to individual photographers can choose their own "good enough." But "good enough" will always be less than the "full potential."

Edited July 22, 2019 by adan

[pico]

2 hours ago, adan said:

suppose one "line" resolved by a lens happens to fall on the junction of two pixels. Boom, you now have that line spread over two pixels, each getting 50% of the brightness - or vanishing altogether.

Too much Lewis Caroll. Sleep or a glass of wine helps.

[ChicagoMatthew]

3 hours ago, adan said:

The point I was making is that NO lens will reach the full theoretical potential of a sensor. And no sensor will reach the full theoretical potential of a lens.

Like you, I used to think that a sensor could "outresolve" a lens or vice-versa. That the "weakest link" is the limiting factor. It is an nice, intuitive idea that, unfortunately, is not actually correct.

A little research shows that "sampling theory" and the related math says otherwise. Thus my previous simplifed equation that the total resolution you'll see in a final picture = (approximately) MTFlens x MTFsensor. Increase either, no matter how low the other, and the total increases. But is always less than the full potential of either one alone.

If you want the more complex, Fourier-transform, calculus-based equations: https://en.wikipedia.org/wiki/Nyquist–Shannon_sampling_theorem

A different take: http://www.normankoren.com/Tutorials/MTF.html

A lens samples the world at x-many lines per mm (actually, x varies across the frame and with aperture used). A sensor samples the lens image's lines per mm at Y-many pixels per mm. The two samples interact in complex ways, not just "one bigger than the other." You can get quantization errors any time you break a continuous series down into finite chunks, whether it be fractions of a mm, or pixels, or 1s and 0s.

As a practical example, suppose one "line" resolved by a lens happens to fall on the junction of two pixels. Boom, you now have that line spread over two pixels, each getting 50% of the brightness - or vanishing altogether. That can happen regardless of the actual resolution of either the lens or the sensor.

To reach the "full potential" of a sensor, you must have a lens that has infinite resolution. And to get the "full potential" of a lens, you need a sensor with an infinite number of "dimensionless" pixels. Doesn't exist in the real world.

One can avoid certain sampling problems (aliasing/moire) by using a sensor twice as "sharp" as the lens. But better than "twice as sharp" still adds continuous improvements - better tonal resolution (distinguishing grays or colors), finer "quanta" and so on - all the way up to infinitely-more sharp (which is vastly more - in fact, undefinably more - than 61 megapixels, or even 61 gigapixels, or 61 petapixels).

Now, everyone from sensor-makers camera-makers to individual photographers can choose their own "good enough." But "good enough" will always be less than the "full potential."

So you’re saying if you have a low quality lens and a high quality sensor, the strong sensor props up the weak lens.... and if you had a low quality sensor and a great lens... the lens would prop up the sensor... the strong element brings up the total quality. Which makes sense.

But if you have a super new top of the line sensor and you pair it with a film era vintage lens, by the logic above you would reach a total quality level, but you’d be using all the lens had to offer, thus out-resolving the lens...right?! 

 

[farnz]

I don't think that Andy's saying that - it's a case of the lowest common denominator is the weakest point in the image 'chain'.   A low resolution lens will always produce a low resolution picture irrespective of how good the sensor is and vice versa.

Pete.

[jaapv]

5 hours ago, ChicagoMatthew said:

So you’re saying if you have a low quality lens and a high quality sensor, the strong sensor props up the weak lens.... and if you had a low quality sensor and a great lens... the lens would prop up the sensor... the strong element brings up the total quality. Which makes sense.

But if you have a super new top of the line sensor and you pair it with a film era vintage lens, by the logic above you would reach a total quality level, but you’d be using all the lens had to offer, thus out-resolving the lens...right?! 

  

I am more than  a bit puzzled about the negative perception of "lens era vintage lenses"  Some of the very best lenses ever built are from the "film era" and are higher resolving and better corrected than their present-day counterparts. Just think of the Leica APO-Telyts. For instance the 280/4.0. That lens is diffraction-limited, one of the very few lenses that is.
Nor is resolution  in the plane of focus the only quality criterium for a lens. It is about the state of correction of ALL aberrations. Both in the sharp and unsharp areas. ESPECiALLY  in the unsharp areas.

This smells of the misconception that lenses are "built for digital" nowadays. Which is complete baloney.

[lct]

Several if not all lenses are designed for digital use nowadays. Focus shift for instance is a flaw i had no perception in the film days. Lenses like the 35/1.4 FLE would not exist w/o digital i suspect. Also we had no italian flag or other color shift problems with film. Modern lenses, especially wides, are designed to avoid those issues on rangefinders. Comparing lenses like CV 21/4 and 21/3.5 is quite obvious from this viewpoint.  

[adan]

2 hours ago, farnz said:

I don't think that Andy's saying that - it's a case of the lowest common denominator is the weakest point in the image 'chain'.   A low resolution lens will always produce a low resolution picture irrespective of how good the sensor is and vice versa.

Pete.

Not exactly. What I am saying is that in the real world, all sensors and lenses are "weak links" - they all have MTFs (or resolution) of less than "perfect" (100%/1.00/∞ lppm).

Suppose you have a sensor capable of 99% MTF, and a lens capable of 75% MTF. You use them together and the final system MTF is not 75% - it is 0.75 x 0.99, or 74.25%. A very, very good sensor will still "bring down" the apparent performance of a lens. And, of course, since multiplication is transitive ( 2 x 4 = 4 x 2), a very, very good lens will still bring down the performance of a sensor, at least from its "full potential."

An analogy would be - blaming the man who fails to make the final score in a game for the loss, and calling him "the weakest point". No - it was the collective failure of the whole team (the total system performance) to score more goals during the whole game that produced the loss.

The good news is that a sensor of 80% and a lens of 50% will produce system MTF of 40%, while a sensor of 99% and a lens of 50% will produce a system MTF of 49.5%. Not the full potential of the sensor - but the sensor still improves the total performance with any lens. And the converse - the better* the lens (even if it doesn't "outresolve" the sensor), the better the final total result.

____________________

* for a given value of "better."

 

[Advertisement]

Hallo,

Look here: Leica M

[Chaemono]

vor 7 Stunden schrieb ChicagoMatthew:

So you’re saying if you have a low quality lens and a high quality sensor, the strong sensor props up the weak lens....

What Andy said and to paraphrase Otto from another thread, if the resolution of a lens is very poor, the improvement in overall resolution by pairing it with a high MPx sensor will likely be infinitesimally small. It’ll still be there, but one won’t be able to see it. 😂