How many megapixels in the next M?

[adli]

Advertisement (gone after registration)

I would expect that would depend on what and where they shoot. I tend to lurk around dark places so for me high ISO is preferred over DR, but I would expect a lot of Leica users would prefer higher DR. We can wish for both

 

A good question would be to know if Leica users prefer more DR or more isos in general. I prefer more DR personally but i'm not sure the majority would vote like me.

[lct]

Sorry i wanted to say more DR or more MP... 

[jaapv]

Still, if a photographer, new to Leica, would would want to buy a camera in a fit of megapixel mania, the M would be the worst possible choice. The camera is too much designed to a concept, not to general specifications.

[darylgo]

I am not sure that DR and ISO are totally separate, higher DR implies improved shadows as does higher ISO. Nonetheless improved DR gets my vote.

[Paul J]

If you want to start a signature war, you state in yours "Perception. Not perfection". Nob the less you claim that you need 60-80 MP without being able to tell how your image will be better perceived with more pixels. Amusing.

 

My sig is a statement on the discussions where people are saying that they need this and they need that, unable to differentiate their needs from their desires. I am of course aware that in many types of photography, you will need different focal lengths, but in traditional photojournalism, you will get a long long way with the classic reporage combination of 35 and 90.

Sorry, I don't understand what you are saying there. Perception not perfection relates to photography, the act of taking pictures, and my point of view. Pixel count has nothing to do with that but is a technical requirement for what I do, for what my clients need and expect. What ever you find amusing about that is of little consequence to me but it does speak measures...

[adli]

Your clients are all producing billboard size commercials?

[Exodies]

Advertisement (gone after registration)

Given the choice between DR, ISO, and pixel count - more DR please. Perhaps a logarithmic rather than linear response from the sensor.

[AlanG]

The question is not about how many megapixels do you want. Is more about what do you think Leica will do. 

So if they went from 18 to 24 then the next step would be either 30 or 32. To make a significant change in how large the image could be reproduced, you'd have to double the resolution by going to 96MP. And that would assume that the lenses and camera technique would be able to consistently get that. On a large print viewed from a distance equal to its diagonal or closer, depth of field might become very thin to the point where much of the image doesn't look so sharp anyway. Viewed from father away the resolution would become less important. So smaller increases in MP are not really going to make much difference in most images. Of course the sensors will most likely improve in other ways.

[AlanG]

Given the choice between DR, ISO, and pixel count - more DR please. Perhaps a logarithmic rather than linear response from the sensor.

I think we generally look at exposure as a logarithmic response but it can be expressed either way.  A 13 stop dynamic range can be expressed as an 8,000 to one brightness range or alternatively as log 3.9. (Every step of .3 is one stop) So a 15 stop dynamic range would be 32,000 to 1 (linear) or log 4.5

 

Regardless of the dynamic range, the brightest full stop of exposure will use half the tonal range of the sensor. Each subsequent stop equally divides the remaining information similarly until you get down to where an additional stop of shadow detail won't actually exist because it will have too few tones to separate into an image and will mostly just be noise. So it is a logarithmic response.  A linear response would divide up the number of available steps evenly among stops.  E.g. if the are 15 stops (32,000 discrete steps,) each stop would have about 2,133 shades of grey per color channel. But I don't see how that is possible. Instead the first stop uses 16,000 discrete steps. 

 

This is what makes it so difficult to increase dynamic range. A sensor with 20 stops of range (log 6) would need to be able to capture a scene brightness range of 1 million to one.  

 

At this point cameras increase their sensitivity by doing sophisticated image processing to amplify and boost the ISO of the sensor. And they are boosting the dynamic range this way too. But at some point the native capture will be cleaner with a greater range and that will allow for greater exposure adjustments in post.

[pop]

....Regardless of the dynamic range, the brightest full stop of exposure will use half the tonal range of the sensor. Each subsequent stop equally divides the remaining information similarly...

That's the description of a linear response. In a logarithmic response, each doubling of the light intensity would increase the output by a constant amount.

[AlanG]

That's the description of a linear response. In a logarithmic response, each doubling of the light intensity would increase the output by a constant amount.

OK if you look at it with the concept of film density on a 45 degree reproduction scale, every .3 of density (density is a logarithmic value) requires twice as much light to be recorded. So the first .3 density takes 2 units of light, the second .3 density requires 4 units of light, the third .3 units of density requires 8 units of light. The units of density are equal steps. E.g. you could put two pieces of .3 density in a sandwich to achieve .6 density. 

 

How are you going to change that?  Is the first stop going to require two units of light and the second stop also going to require two units of light instead of 4?  What will differentiate them? Every stop of output already requires doubling the amount of the light.

[pop]

....How are you going to change that? ....

 

I'm not going to change anything. Exodies suggested introducing sensors with a logarithmic response and I pointed out that your description was not of a logarithmic but of a linear sensor.

[mjh]

Creating a logarithmic sensor would amount to reducing each pixel’s sensitivity in proportion to its accumulated charge. That would probably be doable somehow but it won’t be easy.

[Soden]

I'm more interested in a sensor with white balance and ISO per bayer cluster, rather than one per sensor.

[Joshua Lowe]

I'm a serial cropper so more resolution appeals to me but I'd sooner have more DR, less noise, better low light performance, etc.

[jmahto]

Maybe we can have two different sensors to address wide DR. One optimized for day light range and the other for night time light. Just like our eyes have different sensors for very dark scenes. Like rods and cones.

[darylgo]

Maybe we can have two different sensors to address wide DR. One optimized for day light range and the other for night time light. Just like our eyes have different sensors for very dark scenes. Like rods and cones.

Fuji did this, the Super CCD:

 

https://en.wikipedia.org/wiki/Super_CCD

 

Leica has this currently, the M240 cone version for color, the M246 rod version for B&W.  

[colonel]

I would say that a decent jump from 24mp is 1bp

 

http://phys.org/news/2015-09-breakthrough-photography.html

[adan]

Technically, that should be 1Gp (Gigapixel) - Giga being the neo-Greek prefix for a billion (as Mega is for a million). Of course, a billion means either 10⁹ or 10¹², depending on culture/language, but Giga always refers to 10⁹.

 

But a gigathanks for the link, in any case.

[Advertisement]

Hallo,

Look here: Leica M

[Jeff S]

 IIRC, the S is still CCD, as is my D, which in my experience means taking far greater care with highlights in the context of less dynamic range to play with. It was just such frustrations that pushed me to buy a Z.  

 

Nope, the S 007 is CMOS.  Time for you to reassess.  

 

Jeff