Is 24mp the sweet spot of 36x24 format?

[hansvons]

Advertisement (gone after registration)

36 minutes ago, Stuart Richardson said:

The filmmaking case is an interesting one because the output resolutions are more or less fixed, so camera resolution is not as big a factor if it can resolve the output (I know they downsample etc). Whereas in printing there is no ceiling on resolution.

Yes, I find that highly interesting, too. It was/is so interesting, actually, that from a purely technical point of view, all of that was and remains at the heart of my work.

For perspective, the final results in 4k of the upsampled Arri files looked better than any "native" 4K camera or above produced, even when downsampled. That's not only my personal view but industry consent up to big names like Roger Deakins.

I put the native in quotes because we all know that with the Bayer pattern, a sensor’s “native” resolution only matches sensor pixel count and file resolution in monochrome files off a monochrome sensor with a 4/4 debayering pattern. In colour, a Bayer sensor resolves 1/4 R, 1/4 B, and 2/4 G. This opens up a vast field of leeway in any direction, including high-frequency filters and hot mirrors that the number 4K, 6K, or 24MP for that matter only measures the amount of the resulted pixels but never their quality. 

Enter printing. What a vast field! It can only be measured and understood with your eyes. There are much too many variables to discuss it on a numbers level. Reminds me seeing my stuff printed on proper film in a cinema back in the analogue days. Perception and emotions are everything. Love it!!

[Advertisement]

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

[helged]

17 minutes ago, hansvons said:

Yes, I find that highly interesting, too. It was/is so interesting, actually, that from a purely technical point of view, all of that was and remains at the heart of my work.

For perspective, the final results in 4k of the upsampled Arri files looked better than any "native" 4K camera or above produced, even when downsampled. That's not only my personal view but industry consent up to big names like Roger Deakins.

I put the native in quotes because we all know that with the Bayer pattern, a sensor’s “native” resolution only matches sensor pixel count and file resolution in monochrome files off a monochrome sensor with a 4/4 debayering pattern. In colour, a Bayer sensor resolves 1/4 R, 1/4 B, and 2/4 G. This opens up a vast field of leeway in any direction, including high-frequency filters and hot mirrors that the number 4K, 6K, or 24MP for that matter only measures the amount of the resulted pixels but never their quality. 

Enter printing. What a vast field! It can only be measured and understood with your eyes. There are much too many variables to discuss it on a numbers level. Reminds me seeing my stuff printed on proper film in a cinema back in the analogue days. Perception and emotions are everything. Love it!!

+1. 

[padam]

No one is disputing the advantages of higher megapixels. It is there and it can be utilised. Some of the times.
It's just that not everyone is a professional landscape photographer who needs that most of the time.
However, the disadvantages regarding usability are there. All the time. Especially on mirrorless, where video is a factor for a handful of people, line-skipped vs oversampled is far more notable (on a 1080p display without magnifying...) than higher-res for stills.

I think the measly 24MP M240 had fairly crisp files, never printed anything with that, I am sure it would fair quite well, it didn't seem to be that far off from the 47MP SL2, and it had other advantages in the field that are arguably more relevant to produce more interesting imagery than harping about resolution and print sizes, which is pretty self-evident. Of course I would have an M11, if I had a straight choice between them.
Would I pay 3-4x more to get performance I can rarely see the benefit of, and a Sony sensor I may not even like the look of? No.
Would I pay (over) double for an M10-R considering better dynamic range and ISO (that are useful all the time and of course applies to an M11 as well)?
Maybe, but I am not sure it is "the" sweet spot...

Or we can look at it from the other side. Nikon D700, still lovely, costs no money, yet, does produce files that are easier to edit than an SL2.
Not sure how "bad" they print, but color, contrast, etc. sure look good unmagnified.
But I can see that 12MP (with an AA filter) is a bit low, one of the lenses I use is a 28/1.4E, one of the sharpest DSLR optics ever made.
This probably annoys me more than size, weight, noise, etc. of course I wish it was (at least) 24MP, Nikon has many models like that.
However, all the rendering that made the D700 wonderful as it is, seems to have vanished along the way, whether that's due to Sony sensors, color filter array, who knows.

Point is: resolution is not the only thing, I wish it was just "similar but better, and yet not significantly more pricey", but it rarely is.

Same discussion can be had regarding Leica lenses (ones that are actually Leica) with other brands (especially on M-mount).
They definitely don't represent a "sweet spot", but they also don't look quite the same, so by that point, the argument becomes difficult.

Anyone who needs higher resolution probably already has it. Others, who ask themselves may gain more benefits from other things instead.

[mauri57]

this topic is really interesting

Have you ever try to use Sigma Sd4H in superfine mode?

you get a pictures (made from 7) with a final file of 450 Mbyte...

try!

At the end the most important part of the story is the lens you are using

60 Mpx or 2000000 Mpx sensor behind a bottom of a bottle...

Edited December 11, 2024 by mauri57

[pgk]

Many years ago I remember shooting Kodak Tech Pan and processing it carefully to test lenses. To do so required a very solid tripod, careful use of a cable release, mirror lock up if available, and calm conditions. This film could yield very large prints if required and was even used by Nikon to illustrate just how good one of their macro lenses was. However it wasn't a maintstream film nor did many bother to use precise enough technique to use it. FX digital is now trying to produce files which contain information way in excess of Tech Pan. But few ever needed the precision of Tech Pan and those who did either shot carefully using it or shot larger formats. What exactly has changed 40 years on?

[jaapv]

3 hours ago, vkdev said:

Everything goes in circles. Today it's typical for computer and TV displays to be 4k. 25 years ago I used 800x600 and1024x768. Tomorrow the regular display will be 8k. and it will just happen. regardless of whether I need it or not. 

I just want to say that standards are gradually changing. chasing numbers like 100500mp is absurd, but I also don't like the idea of abandoning minimal support for the next standard.

...note that new Sony cameras have at least 33mp and not 24. I wonder what this means

That is the very definition of technology-driven marketing - Don't sell what people need but make people buy what you want to make; a variation of built-in obsolescence. 

[padam]

Advertisement (gone after registration)

We can at least applaud Leica (and Panasonic) that they don't cripple lower megapixel bodies, all is almost identical besides the sensor, up to the consumer to decide. (Well, it is no coincidence which Leica comes out earlier...) While price difference is also smaller (I'd expect an SL3-S to be up to 1500$ cheaper than an SL3), it is worth it.

In comparison, Canon, Sony etc. cripple lower-end models with a lesser EVF, screen, shutter, build, etc. things that are useful all the time.
Especially frustrating is C-Log2 on an R5II (that even an R6III won't get). While high resolution might not be useful for many (especially impractical for RAW video), that one feature really is (and all the other small things are beneficial, too).

With the M system, they might not do "entry-level" models anymore, unless it is worth it for them.

[Photoworks]

5 minutes ago, padam said:

We can at least applaud Leica (and Panasonic) that they don't cripple lower megapixel bodies, all is almost identical besides the sensor, up to the consumer to decide. (Well, it is no coincidence which Leica comes out earlier...) While price difference is also smaller (I'd expect an SL3-S to be up to 1500$ cheaper than an SL3), it is worth it.

In comparison, Canon, Sony etc. cripple lower-end models with a lesser EVF, screen, shutter, build, etc. things that are useful all the time.
Especially frustrating is C-Log2 on an R5II (that even an R6III won't get). While high resolution might not be useful for many (especially impractical for RAW video), that one feature really is (and all the other small things are beneficial, too).

With the M system, they might not do "entry-level" models anymore, unless it is worth it for them.

you got it almost right, the reason I was never happy with the Panasonic S5II is the EVF was lower resolution than the big Brother, otherwise great video camera.

[SrMi]

5 hours ago, hansvons said:

Yes. And no "however". 

That’s why I find the M system unsuitable for high MTF projects. It’s focusing mechanism and the lack of IBIS makes it prone to motion blur, aka camera shake regardless of the sensor or film stock.

However (can't help myself), if your publicised images are in the 2-4K range, the 60MP M11 is just fine as any other M. 

What is an MTF project. I am familiar with landscape photography. If you want the sharpest results, you need a tripod and manual focus. That should work well for an M camera as well. IBIS helps only at slower shutter speeds and can have negative effects in corners with wider lenses.

[Dazzajl]

14 minutes ago, SrMi said:

What is an MTF project. I am familiar with landscape photography. If you want the sharpest results, you need a tripod and manual focus. That should work well for an M camera as well. IBIS helps only at slower shutter speeds and can have negative effects in corners with wider lenses.

Why is manual focus sharper? 

[Einst_Stein]

6 hours ago, Stuart Richardson said:

This is the myth that will never die. Perhaps it is just how you worded it, but motion blur has nothing to do with sensor resolution other than increasing sensor resolution will show more of the blur that was also there for the low resolution camera. If you take an identical photo on a 24mp and a 60mp sensor, and you can see blur at 100% on the 60mp but not on the 24mp, if you were to sample the 60mp file down to 24mp you would no longer see the motion blur. Motion blur has to do with magnification, not sensor resolution.

To be honest, I am actually surprised how much disinformation is in this thread. I know I don't know everything, but after 25 years in the photo world, almost twenty since I trained under a master printer and sixteen since I started printing exhibitions for museums and galleries, there are two things I feel pretty confident saying:

1. You can make a good print from nearly any camera ever made. That is to say that at the end of the day, the photo itself is the most important thing.

2. If you want to make large prints, they look better with more resolution, unless you are going for a lofi look.

There are two main ways to achieve this:

1. Increase the resolution of the camera and lenses.

2. Lower the magnification ratio (i.e. use a significantly larger piece of film or sensor).

With increasing resolution comes better tonality and microcontrast, less moire and all else being equal, more processing headroom (because digital processing is reliant on data, and there is a lot more of it in higher resolution cameras).

Sensor resolution is just one part of the imaging chain, but it is a fundamental one. Just how you can't outrun sloppy technique, you can't get any more resolution than your sensor provides. AI interpolation tends to look unnatural and worse than simple bicubic interpolation because it is uneven and inconsistent in its application. Details look strange and there are a lot of soft to hard, smooth to rough transitions that put the images into the uncanny valley. There is no free lunch. You can't create information that was never recorded. Well, you can, but it will be a machine generated imitation of the world, not a recording of it. The imaging chain starts at subject motion and stability of the photographic platform (i.e. tripod, IS or hand stability), and runs to focus quality, lens quality, aperture, shutter speed, camera vibration and eventually winds up at sensor resolution.

Is 24mp enough for most people? Yes. Is it enough for everyone? No. Is it where camera resolution will or should stop? That seems highly unlikely to me given the last few thousand years of human technological development.

With the same sensor technology, increasing MP per area means degrade the sensor sensitivity, including low light performance and dynamic range. That is probably the main reason Leica offers SL2S. That is why more people prefer SL2S over SL2. 

Low light performance might be overcome with tripod  if shooting static subjects. I think your comments might come from your landscape photography. But it is not a solution for no static subjects.

Dynamic range degradation is hopeless unless your subject does not need is soak to suffer that.

Of course people can choose to live with that ”enough” dynamic range, or “enough” low light sensitivity, or “enough” resolution. Then shut up the discussions. 

 

Low light performance might be overcome with tripod, which I think suits your landscape photography. But this is not a solution for non static shots. 

Dynamic range 


I cannot believe you did not notice this. 

[Einst_Stein]

6 hours ago, vkdev said:

Everything goes in circles. Today it's typical for computer and TV displays to be 4k. 25 years ago I used 800x600 and1024x768. Tomorrow the regular display will be 8k. and it will just happen. regardless of whether I need it or not. 

I just want to say that standards are gradually changing. chasing numbers like 100500mp is absurd, but I also don't like the idea of abandoning minimal support for the next standard.

...note that new Sony cameras have at least 33mp and not 24. I wonder what this means

In automobile, there used to be a spec claiming how few seconds it can ramp up speed from zero to 100miles. That is like MP today in camera, or 8K TV. 

Since the automobile find miles per gallon is much more meaningful, the focused spec shifted away from seconds to 100miles.

[SrMi]

35 minutes ago, Einst_Stein said:

With the same sensor technology, increasing MP per area means degrade the sensor sensitivity, including low light performance and dynamic range. That is probably the main reason Leica offers SL2S. That is why more people prefer SL2S over SL2. 

No. With the same sensor technology, increasing MP does not degrade sensor sensitivity. SL2 has a different sensor technology than SL2-S. Therefore, SL2-S has better DR. SL3 has sensor technology similar to SL2-S, and the DR compared to SL2-S is similar.

Nikon's and Sony's offerings are other examples of cameras using different resolutions but the same sensor technology. The DR is similar. 

Edited December 11, 2024 by SrMi

[jaapv]

2 hours ago, padam said:

We can at least applaud Leica (and Panasonic) that they don't cripple lower megapixel bodies, all is almost identical besides the sensor, up to the consumer to decide. (Well, it is no coincidence which Leica comes out earlier...) While price difference is also smaller (I'd expect an SL3-S to be up to 1500$ cheaper than an SL3), it is worth it.

In comparison, Canon, Sony etc. cripple lower-end models with a lesser EVF, screen, shutter, build, etc. things that are useful all the time.
Especially frustrating is C-Log2 on an R5II (that even an R6III won't get). While high resolution might not be useful for many (especially impractical for RAW video), that one feature really is (and all the other small things are beneficial, too).

With the M system, they might not do "entry-level" models anymore, unless it is worth it for them.

Of course Leica has to keep up-more or less - with the times, but basically their products are aimed at optimizing photography, not keeping up with the Joneses technologically. 

[SrMi]

57 minutes ago, Dazzajl said:

Why is manual focus sharper? 

Because AF is not always reliable, and MF (magnified and with focus peaking) is more trustworthy. AF is quicker, but that is not an issue for landscape photography.

[Einst_Stein]

5 minutes ago, SrMi said:

No. With the same sensor technology, increasing MP does not degrade sensor sensitivity. SL2 has a different sensor technology than SL2-S. Therefore, SL2-S has better DR. SL3 has sensor technology similar to SL2-S, and the DR compared to SL2-S is similar.

Nikon's and Sony's offerings are other examples of cameras using different resolutions but the same sensor technology. The DR is similar. 

Are you serious? Or this is a joke? You are impossible. 
I will leave you alone.

[SrMi]

6 minutes ago, Einst_Stein said:

Are you serious? Or this is a joke? You are impossible. 
I will leave you alone.

I am serious, and so are many on this and other forums who believe that knowledge trumps belief. Have you ever compared an image of a lower and higher-resolution camera? I have both SL2-S and SL3  and have compared them.

P2P measured the DR of SL2-S and SL3, which are very similar. DPR studio scenes compare low and high-resolution cameras with the same technology, and the DR (noise) is very similar.

What is your basis for believing that the DR of lower-resolution sensors is better than the DR of higher-resolution sensors?

[padam]

40 minutes ago, jaapv said:

Of course Leica has to keep up-more or less - with the times, but basically their products are aimed at optimizing photography, not keeping up with the Joneses technologically. 

While that's true for the M, the SL system goes in the same basket as Panasonic (photo And video), and a lot of users expect the same from it. In fact, the SL2 has Log for video, while an S1R does not. So, for the most part, they provide similar features as well, it is the hardware that seems to be limited, at least for the time being. The problem with an SL3-S (if there is one) is that it will be an even smaller update than the SL3.

Meanwhile, others are keen to bring out quite frequent and notable updates. For instance, while an M has a blackout free viewfinder, stacked sensor cameras also offer this in a usable manner in their own way and it is now reaching midrange level cameras (alongside better EVFs as well, which used to be a highlight of Leica when the system started, and it is a bit disappointing that they haven't moved on further from that level), might as well try it eventually.

Edited December 11, 2024 by padam

[Advertisement]

Hallo,

Look here: Leica L-Mount

[Dazzajl]

22 minutes ago, SrMi said:

Because AF is not always reliable, and MF (magnified and with focus peaking) is more trustworthy. AF is quicker, but that is not an issue for landscape photography.

Okay, I see what you’re saying. MF is not sharper, you find it more reliable. 
 

Although with the camera locked down on a tripod, a point set with single AF should be more reliable than MF, which has room for human error. 

[BernardC]

1 hour ago, padam said:

In fact, the SL2 has Log for video, while an S1R does not.

Is that a typo? Leica has L-log, Panasonic has V-log.

1 hour ago, padam said:

(alongside better EVFs as well, which used to be a highlight of Leica when the system started, and it is a bit disappointing that they haven't moved on further from that level)

They still have the best EVF image, partly thanks to the fact that they use all-glass/high-cost EVF optics. Some brands have EVFs with more megapixels, I've tried them in stores and wasn't impressed. It's a textbook case of specifications vs. real-world. You need to trust your eyes; if the image looks de-saturated and low-contrast, it is. Don't listen to the salesman shouting "it's 8K!" in your ears.