How many bits of files in Monochrom MM1 and M(typ 246) ?

[a.noctilux]

Advertisement (gone after registration)

Hello,

I did search for sources of 12 bits or 14 bits said here in another thread   for Leica Monochroms.

However, I never find writing sources for this claiming "12bits/14 bits".

 

Please point me to these sources.

[Advertisement]

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

[a.noctilux]

Saw this (14bits for M246 DNG compressed or not)

[Jul]

You can use exiftool to find out on a DNG sample of each camera. It says
"Sensor Bit Depth: 14" for the M9, original monochrom might be the same.

[adan]

Try searcing "Bit depth"

https://www.imaging-resource.com/PRODS/leica-m-mono-typ-246/leica-m-mono-typ-246DAT.HTM

None specific for M10-M, but same source lists specs for M10-R (same silicon)

https://www.imaging-resource.com/PRODS/leica-m10-r/leica-m10-rDAT.HTM

Leica's owners' manuals supposedly list the bit depths of the camera in the specs, but I don't have any Monochrom manuals.

But keep in mind that few reviewers will actually have the capability to measure bit depth. They just have to take Leica's word for it.

And that the rated bit depth just means how many potential gray values the camera can distinguish and store.

Up to 4095 gray/brightness levels in 12-bit..... up to 16380 gray/brightness levels in 14-bit. No guarantee any particular picture will include the full amount.

Nor is bit-depth directly equivalent to dynamic range, since DR is further limited by a "noise floor."

The rated bit depth for a camera will always be, for example, 14 bits. Fourteen 1s and 0s for each pixel, stored in a 16-bit "computer word." - (- -101100 01111011)

But at pulled or pushed ISOs the DR will shrink - the 1s and 0s at the extremes can be garbage data (blown highlights, or random noise in shadows).

.............

Side note - you will not be able to "see" a difference" between 12- or 14-bit images....

1) on almost any computer screen, which are limited to 8-bit displays (255 gray tones, or 255 x 255 x 255 = 16.6 million colors).

2) and even then, the human eye can only distinguish about 50 shades of gray side-by-side (our vision is about 5-6-bit - just like the book says ).

But the extra data-bits allow headroom for pulling and pushing and otherwise manipulating the image without banding or posterization (creating gaps or steps in the gray range).

Here's an example - can you see what parts of this image are 5 graytones different from the background (~1/50th of 255 possibilities)?

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!

 

Edited March 4, 2022 by adan

[pop]

41 minutes ago, adan said:

can you see

Certainly. It says 'Can You read me?', and I can read it at my normal screen settings.

[a.noctilux]

Thanks for your inputs.

Yes Andy,

"Can You read me  ?" 

in the middle in four lines

 

I  know that screen is the limiting factor.

 

After searching in manuals, nothing about number of bits in Monochrom or M246.

[Jon Warwick]

Advertisement (gone after registration)

Very interesting conversation.

Perhaps it’s related, perhaps not, but I’m sure I see more tonal separation in images from the Monochroms compared to Bayer filter cameras (the latter either in color or B&W). It seems to my eyes the Monochrom images consistently appear more “3D” as a result.

Are the Monochroms (whether 12 or 14 bit) able to distinguish more grays and hence have more subtle tonal separation compared to a Bayer sensor camera?

Edit: and yes, I can read “can you read me?”, quite easily, on an iPhone and at different screen brightness settings.

Edited March 5, 2022 by Jon Warwick

[adan]

12 hours ago, Jon Warwick said:

Are the Monochroms (whether 12 or 14 bit) able to distinguish more grays and hence have more subtle tonal separation compared to a Bayer sensor camera?

There are several "moving parts" to that question, but the three keys points would be:

1) Each pixel on a Bayer sensor sees only about 1/3rd of the light - either red or green or blue parts of the spectrum. So while the actual data read off the sensor is simply a brightness value for each pixel (just like a Monochom - gray scale) it will be a "checkerboard" of different grays - an artificial pattern imposed by the color capability.

Plus the loss of light means the color image has to be exposed at a true "ISO" about 1.5 stops higher than the unfiltered Monochrom sensor, due to the loss of light to each color filter (Photography 101 - "filter factors"). That will mess with the dynamic range available, and where shadow noise kicks in.

2) That checkerboard gets removed by the de-mosaicing process when you open a Bayer image in your preferred flavor of post-processing software, which shares data between neighboring pixels, according to the software maker's algorithms, so that the final picture includes orange pixels and mauve pixels and all the other colors of the spectrum. Since "sharing data" amounts to a small blurring of brightnesses between pixels, that will also affect tonal edge transitions - although it is more obvious in regard to simple resolution; on a tiny scale the Monochrom image is sharper per pixel, since each pixel "stands alone."

(BTW the human visual system has its own role to play - it is well-known the brain can interpret contrast as sharpness and vice versa. So yes, a "sharper" image may appear to have different tonality, to some extent. As I said - many moving parts!)

3) Each software provider will have its own algorithms for doing the de-mosaicing (algorithms are intellectual property - C1 and Adobe and RawTherapee can't just copy one another's math). So that will have an influence on the final result. Additionally, there are multiple ways of converting a color image to grayscale (Channel Mixer, desaturate in the raw process, desaturate using the HSB control panel, "convert to gray scale") - each of which will have different effects on "which gray" any particular color becomes. Adobe is notorious for rendering reds too dark with "simple" desaturations.

In summation: Either an M10-R or and M10-M can capture 14 bits of data per pixel (if we believe Leica's specs) - but they will each capture and store different 14-bit data. And that data will unavoidably be post-processed differently, depending on what the picture's metadata tells your software to do. "This is a Bayer color file - de-mosaic it!" or "This is a grayscale original - do NOT de-mosaic it!" Along with other "instructions" such as a specified default tone curve, black point, white point, etc.

The Monochrom may certainly distinguish more "un-messed-with" grays.

So as with most things digital, the hardware is important, but the software is equally important, if not more so.

P.S. if one shoots a Bayer-camera-created .jpg in "B&W" mode - the exact same steps have to be done. Except by the camera's own CPU, and "on the fly" enroute to the SD card. De-mosaic to "full-color pixels" to remove the checkerboard, and then desaturate the colors to grays (via who knows-what-algorithm Leica has programmed), and then "saved as" in the .jpg format.

Edited March 5, 2022 by adan

[Jon Warwick]

2 hours ago, adan said:

There are several "moving parts" to that question, but the three keys points would be:

1) Each pixel on a Bayer sensor sees only about 1/3rd of the light - either red or green or blue parts of the spectrum. So while the actual data read off the sensor is simply a brightness value for each pixel (just like a Monochom - gray scale) it will be a "checkerboard" of different grays - an artificial pattern imposed by the color capability.

Plus the loss of light means the color image has to be exposed at a true "ISO" about 1.5 stops higher than the unfiltered Monochrom sensor, due to the loss of light to each color filter (Photography 101 - "filter factors"). That will mess with the dynamic range available, and where shadow noise kicks in.

2) That checkerboard gets removed by the de-mosaicing process when you open a Bayer image in your preferred flavor of post-processing software, which shares data between neighboring pixels, according to the software maker's algorithms, so that the final picture includes orange pixels and mauve pixels and all the other colors of the spectrum. Since "sharing data" amounts to a small blurring of brightnesses between pixels, that will also affect tonal edge transitions - although it is more obvious in regard to simple resolution; on a tiny scale the Monochrom image is sharper per pixel, since each pixel "stands alone."

(BTW the human visual system has its own role to play - it is well-known the brain can interpret contrast as sharpness and vice versa. So yes, a "sharper" image may appear to have different tonality, to some extent. As I said - many moving parts!)

3) Each software provider will have its own algorithms for doing the de-mosaicing (algorithms are intellectual property - C1 and Adobe and RawTherapee can't just copy one another's math). So that will have an influence on the final result. Additionally, there are multiple ways of converting a color image to grayscale (Channel Mixer, desaturate in the raw process, desaturate using the HSB control panel, "convert to gray scale") - each of which will have different effects on "which gray" any particular color becomes. Adobe is notorious for rendering reds too dark with "simple" desaturations.

In summation: Either an M10-R or and M10-M can capture 14 bits of data per pixel (if we believe Leica's specs) - but they will each capture and store different 14-bit data. And that data will unavoidably be post-processed differently, depending on what the picture's metadata tells your software to do. "This is a Bayer color file - de-mosaic it!" or "This is a grayscale original - do NOT de-mosaic it!" Along with other "instructions" such as a specified default tone curve, black point, white point, etc.

The Monochrom may certainly distinguish more "un-messed-with" grays.

So as with most things digital, the hardware is important, but the software is equally important, if not more so.

P.S. if one shoots a Bayer-camera-created .jpg in "B&W" mode - the exact same steps have to be done. Except by the camera's own CPU, and "on the fly" enroute to the SD card. De-mosaic to "full-color pixels" to remove the checkerboard, and then desaturate the colors to grays (via who knows-what-algorithm Leica has programmed), and then "saved as" in the .jpg format.

What an exceptionally detailed and informative response. Many thanks!

[frame-it]

On 3/5/2022 at 1:26 AM, a.noctilux said:

Hello,

I did search for sources of 12 bits or 14 bits said here in another thread   for Leica Monochroms.

However, I never find writing sources for this claiming "12bits/14 bits".

 

Please point me to these sources.

download raw digger, opem a file and see the raw histogram data.

[a.noctilux]

8 minutes ago, frame-it said:

download raw digger, opem a file and see the raw histogram data.

Thanks for the info.

I don't need to know, in use the files from my two Monochroms are good enough for me.

just asking the writing sources of the claim of the post I linked.

On 2/28/2022 at 11:24 AM, xiaubauu2009 said:

I think all M except the M246 has 14bit... I just feel the 12bit is a bit too smooth in rendering to my liking. I sold the m246 pretty fast but rebought the m9m...

 

Edited March 6, 2022 by a.noctilux

[xiaubauu2009]

On 3/6/2022 at 8:11 AM, a.noctilux said:

Thanks for the info.

I don't need to know, in use the files from my two Monochroms are good enough for me.

just asking the writing sources of the claim of the post I linked.

Just use a rawdigger. I didn't know before until I tried it comparing using RAWdigger the M9M vs M246 of my own files.... 

The reason I took the test is because I just feel the files is just somehow slightly less character when push around in software, and it will have dark blotch much sooner than the CCD files. I initially put it down to CMOS vs CCD, but later found out about this 12bit business...

 

'And that the rated bit depth just means how many potential gray values the camera can distinguish and store.

Up to 4095 gray/brightness levels in 12-bit..... up to 16380 gray/brightness levels in 14-bit. No guarantee any particular picture will include the full amount.

Nor is bit-depth directly equivalent to dynamic range, since DR is further limited by a "noise floor."'

 

Just 4 times less shade of grey permissible by the m246 vs M10M/M9M. Don't understand why they reduce the bit rate. Sensor limitation? no Idea.

Edited March 7, 2022 by xiaubauu2009