The broken dream of the M60

[ramarren]

Advertisement (gone after registration)

3 hours ago, andrea-i said:

I think there's some confusion here Jaap, why do you think a monchrom's dng or scanned b&w film is more desirable than a desaturated color image?
From a technical stand point, desaturated color is indeed what you get with a leica monochom's dng or b&w film. ...

Not really. 

B&W films and the M Monochrom sensors have well defined spectral characteristics, they're not just RGB that's been desaturated. You use B&W filters to adjust, at capture time, the relationship between different colors and how they are represented in the DNG file. With any RGB sensor produced B&W image, you do this filtering in post by adjusting the blend of R, G, and B curves to achieve the tonal gradations and color separation important to your photograph. 

If you just desaturate an RGB sensor's capture, you get kinda blah B&W imagery. 

[Advertisement]

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

[andrea-i]

1 hour ago, ramarren said:

Not really. 

B&W films and the M Monochrom sensors have well defined spectral characteristics, they're not just RGB that's been desaturated. You use B&W filters to adjust, at capture time, the relationship between different colors and how they are represented in the DNG file. With any RGB sensor produced B&W image, you do this filtering in post by adjusting the blend of R, G, and B curves to achieve the tonal gradations and color separation important to your photograph. 

If you just desaturate an RGB sensor's capture, you get kinda blah B&W imagery. 

Yes and just like the M Monochrom has its own character, what I'm saying I'd love to have, is  the possibility to define the B&W "character" in camera, just like we do with jpegs, and output a dng with the same format as the M Monochrom.
Mine is a workflow desire of course, not a technical advantage one, otherwise just like Jaap says, best off to move the saturation slider down and be done with it. But I sure would love shooting an M-D in monochromatic mode and really embrace the limits of it just like when shooting film.
 

[LocalHero1953]

With a sensor with an overlying colour filter, you automatically get light values determined by the filter colour. So if (as is normally the case) there are twice as many green elements in a filter array than red or blue, a desaturated dng from such an image would be biased towards the brightness read by the green sensor elements. This is not necessarily 'wrong' (whatever that might be in this context), but it is definitely not the same as an image from a sensor with no colour filter array at all, where each sensel responds identically across the spectrum, not according to a particular colour.

A Monochrom would therefore be more influenced by the red and blue in a scene than the green, compared to a desaturated colour dng.

[adan]

On the tech aspects of monochrome and sensors (not the pretty but impractical M60):

Here's why you can't get a monochrome raw file (i.e. DNG) out of a color sensor. A raw file effectively already is a monochrome file - each pixel outputs simply a luminance (brightness) value. But in a section of the picture that is, for example, a red wall, the red-filtered pixels are bright, and the neighboring blue and green pixels are dark. That gives the monochrome luminance record-image a strong checkerboard pattern.

You can't just "monochrome" it - it still has to be demosaiced before conversion to monochrome. Unless you like playing checkers.

The way a color sensor "sees" the world.

Next post - the way simply "monochroming" a color DNG would look, still patterned with the checkerboard:

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 February 26, 2019 by adan

[adan]

How a "monochrome" .DNG made from that capture would look - not really what you seek.

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!

[andrea-i]

1 hour ago, adan said:

How a "monochrome" .DNG made from that capture would look - not really what you seek.

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!

Hi Adan, thanks for the insightful explanation!
But demosaicing in camera should not be impossible, the jpeg process already does that and then adds compression. Simply demosaicing and desaturating is way faster as a computing process.
For now the best workflow if someone wants to shoot pure B&W without a MM is to use jpeg, what a shame.

In any case, Leica is not likely to give us this option and more likely to keep producing a distinct line of M Monochrom, and maybe we'll never have a monochrom screenless camera. One more reason to keep dusting off our analogue Ms.

[jaapv]

Advertisement (gone after registration)

Of course the camera can demosaic the DNG - which produces a JPG file. A Bayer filtered camera CANNOT produce "pure [raw] B&W output", because there are colour filters in front of the sensor.

There is nothing for Leica to give...

 

[andrea-i]

40 minutes ago, jaapv said:

Of course the camera can demosaic the DNG - which produces a JPG file. A Bayer filtered camera CANNOT produce "pure [raw] B&W output", because there are colour filters in front of the sensor.

There is nothing for Leica to give...

 

The software engineer in me begs to differ, Jaap : )
The M Monochrom produces a specific DNG file that does not require demosaicing and instructs the PP software that it is a b&w only image.
Any of our M can produce that file, of course you loose PP flexibility just like with a jpeg, of course you'd have less resolution than with a M Monochrom, but it would make possible to shoot black and white in camera uncompressed.

[LocalHero1953]

5 minutes ago, andrea-i said:

The software engineer in me begs to differ, Jaap : )
The M Monochrom produces a specific DNG file that does not require demosaicing and instructs the PP software that it is a b&w only image.
Any of our M can produce that file, of course you loose PP flexibility just like with a jpeg, of course you'd have less resolution than with a M Monochrom, but it would make possible to shoot black and white in camera uncompressed.

The software engineer in you 😉must acknowledge the difference in hardware between the Monochrom and the colour M: the Bayer filter that screws up the response of the sensor to different colours. It yields both a distorted response to the brightness of each colour in the scene (my post) and an anomalous spatial response (Andy's post). Even if you could stop demosaicing happening, you couldn't get over these issues.

[jaapv]

As the Bayer filter is analog, software engineers tend to forget about it

[andrea-i]

37 minutes ago, LocalHero1953 said:

The software engineer in you 😉must acknowledge the difference in hardware between the Monochrom and the colour M: the Bayer filter that screws up the response of the sensor to different colours. It yields both a distorted response to the brightness of each colour in the scene (my post) and an anomalous spatial response (Andy's post). Even if you could stop demosaicing happening, you couldn't get over these issues.

Ok, but that's pretty much like saying you don't like the jpeg your camera produces, which would be fine, but a matter of personal taste.
I'm saying that from a technical standpoint, if the M Monochrom produces a DNG that is not demosaiced by the PP softwares, then it sure must be possible to save the same file from any of our digital Ms.

To sum it up, here's what each method does already (with the last method being what I'd like to have):
DNG Raw:
1) camera reads every pixel from the sensor
2) each pixel is saved to DNG, it will need demosaicing once imported back in a PP software
(I know leica's dng do have some other stuff done to the raw during saving, it's not a pure raw, but lets not take that in account, it's not crucial to my point)

Jpeg:
1) Camera reads every pixel from sensor
2) demosaic
3) post process: contrast, sharpening, etc.
4) compression
5) save to jpeg

Monochrom DNG from bayer color sensor:
1) Camera reads every pixel from sensor
2) demosaic
3) post process: contrast, sharpening, etc.
4) save every pixel to DNG in M Monochrom format uncompressed
(which in turn, tells the PP software to not demosaic this DNG)

Unless I'm missing some crucial information about how the M Monochrom DNG is saved, then this is all totally possible from a software standpoint.
 

[LocalHero1953]

OK, I accept that personal taste and willingness to live with deficiencies comes into this! 😊

Edited February 26, 2019 by LocalHero1953

[andrea-i]

9 minutes ago, LocalHero1953 said:

OK, I accept that personal taste and willingness to live with deficiencies comes into this! 😊

ahahah yes! I'd totally renounce to precious PP flexibility in order to shoot black and white, in camera, without a screen, but that's just me : )

Btw, I saw an M60 without the original enclosure and lens, for 7500 euros, I think that was a great deal if someone likes that camera but like me just wants to use it, rather than as a collector piece.

[jaapv]

1 hour ago, andrea-i said:

Ok, but that's pretty much like saying you don't like the jpeg your camera produces, which would be fine, but a matter of personal taste.
I'm saying that from a technical standpoint, if the M Monochrom produces a DNG that is not demosaiced by the PP softwares, then it sure must be possible to save the same file from any of our digital Ms.

To sum it up, here's what each method does already (with the last method being what I'd like to have):
DNG Raw:
1) camera reads every pixel from the sensor
2) each pixel is saved to DNG, it will need demosaicing once imported back in a PP software
(I know leica's dng do have some other stuff done to the raw during saving, it's not a pure raw, but lets not take that in account, it's not crucial to my point)

Jpeg:
1) Camera reads every pixel from sensor
2) demosaic
3) post process: contrast, sharpening, etc.
4) compression
5) save to jpeg

Monochrom DNG from bayer color sensor:
1) Camera reads every pixel from sensor
2) demosaic
3) post process: contrast, sharpening, etc.
4) save every pixel to DNG in M Monochrom format uncompressed
(which in turn, tells the PP software to not demosaic this DNG)

Unless I'm missing some crucial information about how the M Monochrom DNG is saved, then this is all totally possible from a software standpoint.
 

Where is the Bayer interpolation?

[andrea-i]

45 minutes ago, jaapv said:

Where is the Bayer interpolation?

What you refer to as bayer interpolation is indeed the demosaic process:
For each pixel take surrounding pixels and combine their values to form RGB channels.

 

[jaapv]

Which does not exist in a Monochrom DNG.  An essential difference.

[andrea-i]

9 minutes ago, jaapv said:

Which does not exist in a Monochrom DNG.  An essential difference.

Yes, there would be loss of data, almost like saving jpeg only from the camera, still better than that though because the DNG would not be compressed.
Again, for those doubting a monochromatic DNG is even possible, just look at a Leica M Monochrom DNG file.
Given the technical assumption that this CAN be done, this matter sparks the argument of whether an M Monochrom makes sense ONLY because of its increased pixel count and ISO sensibility.
This is of course subjective territory, but given that I shoot with an M8, I surely don't think pixel count or high ISO is all that important to my photography, instead I consider the M Monochrom interesting mostly because of the artistic constraints it imposes, just like shooting b&w film.
Now, we don't buy film cameras that only shoot b&w film, do we? so back to my original point, since an M-D is as close as it gets to the film experience, I think it'd be it'd be lovely to have a monochromatic DNG output : )
 

[adan]

I do get that - technically - the .DNG format is just a "wrapper" for image data.

My Vuescan software will (on request) output a "Monochrom-like" .DNG file of a B&W film scan, instead of an ordinary TIFF or JPEG. But still 16-bit data, that can be manipulated like any other raw image in Camera Raw (or presumably LightRoom) - except with color-oriented controls (calibration sliders, WB, Sat, Vib, etc.) all grayed out and disabled.

The scans are not Bayer-patterned (color comes from three RGB-filtered linear pixel arrays in the scanner) but that can be worked around. Once one has 1s and 0s, one can juggle them any way one wants.

So a 16-bit "raw-like" data file can be produced from any set of data and "wrapped" in the .DNG format. There is no programming reason that 16-bit sensor output cannot be - in camera -  demosaiced or Bayer-interpolated and converted to grayscale, while retaining 16 bits of data, and without the extra steps of WB, sharpening etc.

It's just that no one has ever bothered to do it. And it would require a third in-camera processing flow, in addition to the normal " (1) save as raw-data (color) 16-bit .DNG" or " (2) debayerize and convert to 8-bit compressed JPEG and save" flows.

Something like "(3) save as grayscale 16-bit .DNG."

The question is, how much is it worth, in terms of adding a third output "leg" to the firmware of a color camera? Size of the firmware code (and thus the PROM or whatever hardware chip stores the firmware in the camera), added menu item(s) - including perhaps "color filtering" to add the effect of a red or orange or yellow lens filter, software-engineer time, testing, extra processing time that ties up the buffer and keeps the little red processing light blinking (yeah, yeah, microseconds, but it adds up). For what percentage of users?

That is to say, how many Leica users would want to permanently throw away the color/chroma data, but not get any of the benefits of a true Monochrom-sensor image (higher rez, ISO and DR)? Because if it's <1%, how much time and money does Leica want to devote to it?

More or less, Leica has said: "If you are really devoted full-time to B&W, buy a Monochrom and get the extra image quality. If you want a color camera that occasionally produces B&W pictures - do your own (much more flexible and user-controllable) converting in post-processing. Or shoot JPEG."

Edited February 26, 2019 by adan

[Advertisement]

Hallo,

Look here: Leica M

[andrea-i]

1 hour ago, adan said:

I do get that - technically - the .DNG format is just a "wrapper" for image data.

My Vuescan software will (on request) output a "Monochrom-like" .DNG file of a B&W film scan, instead of an ordinary TIFF or JPEG. But still 16-bit data, that can be manipulated like any other raw image in Camera Raw (or presumably LightRoom) - except with color-oriented controls (calibration sliders, WB, Sat, Vib, etc.) all grayed out and disabled.

The scans are not Bayer-patterned (color comes from three RGB-filtered linear pixel arrays in the scanner) but that can be worked around. Once one has 1s and 0s, one can juggle them any way one wants.

So a 16-bit "raw-like" data file can be produced from any set of data and "wrapped" in the .DNG format. There is no programming reason that 16-bit sensor output cannot be - in camera -  demosaiced or Bayer-interpolated and converted to grayscale, while retaining 16 bits of data, and without the extra steps of WB, sharpening etc.

It's just that no one has ever bothered to do it. And it would require a third in-camera processing flow, in addition to the normal " (1) save as raw-data (color) 16-bit .DNG" or " (2) debayerize and convert to 8-bit compressed JPEG and save" flows.

Something like "(3) save as grayscale 16-bit .DNG."

The question is, how much is it worth, in terms of adding a third output "leg" to the firmware of a color camera? Size of the firmware code (and thus the PROM or whatever hardware chip stores the firmware in the camera), added menu item(s) - including perhaps "color filtering" to add the effect of a red or orange or yellow lens filter, software-engineer time, testing, extra processing time that ties up the buffer and keeps the little red processing light blinking (yeah, yeah, microseconds, but it adds up). For what percentage of users?

That is to say, how many Leica users would want to permanently throw away the color/chroma data, but not get any of the benefits of a true Monochrom-sensor image (higher rez, ISO and DR)? Because if it's <1%, how much time and money does Leica want to devote to it?

More or less, Leica has said: "If you are really devoted full-time to B&W, buy a Monochrom and get the extra image quality. If you want a color camera that occasionally produces B&W pictures - do your own (much more flexible and user-controllable) converting in post-processing. Or shoot JPEG."

I know, not a chance : ) 
Not to mention the fact that they have an actual interest in keeping the digital monochrome experience exclusive to their line of MM bodies.

All this talk made me think of my ideal PP software, which would be very different from lightroom and is probably material for another thread ; )
 

[jaapv]

4 hours ago, andrea-i said:

Yes, there would be loss of data, almost like saving jpeg only from the camera, still better than that though because the DNG would not be compressed.
Again, for those doubting a monochromatic DNG is even possible, just look at a Leica M Monochrom DNG file.
Given the technical assumption that this CAN be done, this matter sparks the argument of whether an M Monochrom makes sense ONLY because of its increased pixel count and ISO sensibility.
This is of course subjective territory, but given that I shoot with an M8, I surely don't think pixel count or high ISO is all that important to my photography, instead I consider the M Monochrom interesting mostly because of the artistic constraints it imposes, just like shooting b&w film.
Now, we don't buy film cameras that only shoot b&w film, do we? so back to my original point, since an M-D is as close as it gets to the film experience, I think it'd be it'd be lovely to have a monochromatic DNG output : )
 

I'll quote Erwin Puts: The Monochrom sensor is inherently better for monochrome photography because it lacks the optical aberrations by the Bayer filter and the deterioration by the processing pipeline. A pixel on the sensor is a pixel in print. FWIIW.