Monochrome vs. Bayer

I have a Sony A7Rii that I had converted to monochrome by removing the CFA and did some test shots in 2019 using the same subject and lens on a Sony A7Riii which has the same 42MP sensor. This illustrates the real world detail resolving difference between CFA and no CFA. The resolution improvement by removing the CFA, while barely visible, is somewhat underwhelming as is the slight improvement in sensitivity -- the main reason I did the conversion was to shoot IR.

Anyway, here are 100% crops from each camera -- the color images from the A7Riii were converted to black and white by extracting the luminance channel. There are minor differences in tonality due to the different spectral sensitivities.

9fc1ee11abf3427cb20db8322213397f.jpg

218c5978527c4136a37b6d66153195a7.jpg
its not apples to apples ,,,,, you had to use a special program like Monochrome 2 dg on the converted camera .... also thee can be " unknown unknowns " with regards to the software and electronics of a converted camera and a dedicated one
 
Does the monochrome version use microlenses?
As far as I know, the factory monochrome sensors have microlenses.
It's a choice, AFAIK
No doubt, just a question of money.
If, say, a sensor is used in UV / IR imaging equipment ...
I know of a fellow who got a monochrome astro camera to shoot UVA, but had (unexpected) bandwidth limitations due to coverglass and microlenses. Sadly, no quartz microlens option, as far as I know.
I think measuring sensor responsivity (or comparing raw numbers for the same exposure and same camera settings) may be important when comparing SNR figures.
The CFA-scraped sensors lose the microlenses in the deBayering process.
Sure.
 
I have a Sony A7Rii that I had converted to monochrome by removing the CFA and did some test shots in 2019 using the same subject and lens on a Sony A7Riii which has the same 42MP sensor. This illustrates the real world detail resolving difference between CFA and no CFA. The resolution improvement by removing the CFA, while barely visible, is somewhat underwhelming as is the slight improvement in sensitivity -- the main reason I did the conversion was to shoot IR.

Anyway, here are 100% crops from each camera -- the color images from the A7Riii were converted to black and white by extracting the luminance channel. There are minor differences in tonality due to the different spectral sensitivities.

9fc1ee11abf3427cb20db8322213397f.jpg

218c5978527c4136a37b6d66153195a7.jpg
its not apples to apples ,,,,, you had to use a special program like Monochrome 2 dg on the converted camera .... also thee can be " unknown unknowns " with regards to the software and electronics of a converted camera and a dedicated one
"Almost all digital cameras that are released today are color cameras. Several Leica and PhaseOne models are rare exceptions, but these cameras are even more of a niche product than the color cameras or digital backs made by those same companies. Nevertheless, there is demand for BW (Black and White) cameras (for reasons explained in detail below), and many photographers want BW cameras with the same lens mount as their main (color) camera, so that they can use the lenses they already own.

Since a color camera differs from a BW camera due to the existence of color filter array (CFA, Bayer mosaics) over the sensor, converting a color camera to BW is done by removing the CFA*.

After the CFA is removed, the demosaicking process that is done during RAW file processing becomes unnecessary. This, in turn, leads to higher resolution and a decrease in the number of processing artefacts (see the Bayer Moiré article), which is, essentially, why people perform the conversions of color cameras to monochrome.

However, during such conversion, the firmware of the camera doesn’t change, and the camera doesn’t know that it’s become monochrome, which leads to the following:
  1. The monochrome file recorded by the camera cannot be told apart from a color shot without a complete analysis of of the RAW data – all of the metadata corresponds to the color file.
  2. In many cameras, sensor data is processed differently for different color channels before the RAW is recorded (for example, White Balance preconditioning for Nikon: the values of the red and blue pixels are multiplied by small coefficient). In the case of a camera converted to monochrome, equally exposed pixels that originally, before conversion, belonged to different channels, receive different RAW data numbers.
The first problem leads to most RAW processing programs continuing to think that they’re processing a color file and, accordingly, preform demosaicking, which leads to resolution decreases, moiré, and aliasing. " - Monochrome 2 dg
 
...in my IR and non-IR comparisons, Erik's work (thanks, Erik!), and Jonathan Sachs's comparisons (thanks, Jonathon!), is how close the sharpness is between the demosaiced images and the ones that started out in mono.

I've done three comparisons now, and in all three cases the improvements in sharpness were noticeable on close inspection, but much smaller than I expected them to be before I tested for the first time. I think this is a testimony to how good demosaicing has become over the years. And things would be even closer had I used Adobe superres demosaicing.

There still remain advantages to mono capture: lower noise in shutter speed limited situations, relative freedom from aliasing, better ability to use filter wheels and highly selective filters like hydrogen alpha.

Jim

--
https://blog.kasson.com
 
Last edited by a moderator:
I have a Sony A7Rii that I had converted to monochrome by removing the CFA and did some test shots in 2019 using the same subject and lens on a Sony A7Riii which has the same 42MP sensor. This illustrates the real world detail resolving difference between CFA and no CFA. The resolution improvement by removing the CFA, while barely visible, is somewhat underwhelming as is the slight improvement in sensitivity -- the main reason I did the conversion was to shoot IR.

Anyway, here are 100% crops from each camera -- the color images from the A7Riii were converted to black and white by extracting the luminance channel. There are minor differences in tonality due to the different spectral sensitivities.

9fc1ee11abf3427cb20db8322213397f.jpg

218c5978527c4136a37b6d66153195a7.jpg
Thank you!

This is massively appreciated.
 
Does the monochrome version use microlenses?
As far as I know, the factory monochrome sensors have microlenses. The CFA-scraped sensors lose the microlenses in the deBayering process.
I knew that, 🤠 but I'm sure it was beneficial to some of the newbies on this forum. LOL!!! You guys are amazing.

This level of common knowledge is repetitive. Everyone knows that the scraping of the micro lenses causes abnormalities when you work your way through the deBayering process with that new slider in LR.... Just do what I do with the Scraping Slider - move it to the left until the blankies start blinking....

--
Greg Johnson, San Antonio, Texas
https://www.flickr.com/photos/139148982@N02/albums
 
Last edited:
Yes, I used Monochrome2DNG which is necessary when processing raw files from converted color cameras. In addition to skipping the interpolation, it corrects the relative gains of the RGB channels which would otherwise produce an annoying texture. However it is possible a Leica Monochrom might produce slightly sharper images due to different firmware and software. It would be interesting to do a similar test, however one would be using different lenses on the two cameras and I don't have a Leica.

If I had to guess, assuming the firmware and software are not responsible, I would say two factors are contributing to the minimal resolution gain:

1) The interpolation algorithms used by modern raw converters are fairly intelligent and do a good job of detecting and preserving sharp edges, so some of the highly visible image detail is recovered, while in areas that lack sharp detail, interpolation errors are generally less visible anyway.

2) Very few high resolution digital images are critically sharp down to the level of individual pixels due to a combination of depth of field, focusing errors, noise, atmospheric effects, diffraction, lens aberrations, subject movement and camera movement. I eliminated as many of these factors as I could in the test (heavy tripod, very careful focusing, good lens, stationary subject at a great distance, low ISO) in the hope of seeing more of a resolution gain -- under less optimal conditions the resolution gain would likely be even less.
 
Yes, I used Monochrome2DNG which is necessary when processing raw files from converted color cameras.
Sufficient, but not necessary. You can extract the image from the raw file and operate on that.
In addition to skipping the interpolation, it corrects the relative gains of the RGB channels which would otherwise produce an annoying texture.
I do that correction in Matlab.
However it is possible a Leica Monochrom might produce slightly sharper images due to different firmware and software.
With a monochrome camera, you can look at the raw data and see how sharp it is. Use MTF Mapper or Imatest for quantitative results.
It would be interesting to do a similar test, however one would be using different lenses on the two cameras and I don't have a Leica.

If I had to guess, assuming the firmware and software are not responsible, I would say two factors are contributing to the minimal resolution gain:

1) The interpolation algorithms used by modern raw converters are fairly intelligent and do a good job of detecting and preserving sharp edges,
Amazingly so.
so some of the highly visible image detail is recovered, while in areas that lack sharp detail, interpolation errors are generally less visible anyway.
Unless they are operating on aliased data.
2) Very few high resolution digital images are critically sharp down to the level of individual pixels
And that's a good thing. That would mean beaucoup aliasing.
due to a combination of depth of field, focusing errors, noise, atmospheric effects, diffraction, lens aberrations, subject movement and camera movement. I eliminated as many of these factors as I could in the test (heavy tripod, very careful focusing, good lens, stationary subject at a great distance, low ISO) in the hope of seeing more of a resolution gain -- under less optimal conditions the resolution gain would likely be even less.
Thanks, Jonathon.
 
Does the monochrome version use microlenses?
As far as I know, the factory monochrome sensors have microlenses.
It's a choice, AFAIK
No doubt, just a question of money.
Maybe not just a question of money, but a question of judgement, application, and money (if the end user aren't prepared to pay premium for custom design, the choice is often to use stock microlenses).
 
Yes, I used Monochrome2DNG which is necessary when processing raw files from converted color cameras. In addition to skipping the interpolation, it corrects the relative gains of the RGB channels which would otherwise produce an annoying texture....

<snip>
I did not know about Monochrome2DNG (!), and I've been playing with an IR converted back for almost a year. It did take me a while to figure out a way to remove the demosaic lattice noise in the image.

I ran an image through Monochrome2DNG and compared it to my current process:



Left Monochrome2DNG  - right Demosaic, channel balanced in RawTherapy @ 300%
Left Monochrome2DNG - right Demosaic, channel balanced in RawTherapy @ 300%



My Monochrome2DNG has better acutance, but still shows the filter pattern. The overall sharpness is about the same, which is a testament to the maturity of the demosaic algorithms. I find my current demosaic process has a very "film" look to it (I shot a fair bit of IR film back in the day).

I remember reading an article by Jim that mentioned the deeper IR signal was typically the same on the 3 channels once you remove the visible wavelengths, but that is not my experience with this sensor, and it appears that Monochrome2DNG is not balancing the channels.

I'd like to improve my Monochrom2DNG process (... what I'm missing?).

(P65 back with the hot mirror replaced by a 780nm cutoff filter on a Hasselblad 2000)
 
at some point these claims lead to one or all of the following conclusions ....

the senior project engineers / color scientists / test technicians at Leica Camera Wetzlar Germany :

1) did not know what they were doing when they released the mono Leica

2) released it knowing there was no advantage over a color version just to get money

3) hoodwinked some of the worlds best photographers who have been shooting for decades

4) managed to keep the farce going for a decade that the models have been out
None of your conclusions are supported by either Erik’s or my claims.
Neither you or Erik nor any of your claims were mentioned. I don't see the post to which you replied as being about you. I understood the context of the remarks to be a reaction (not a set of conclusions) to some of the comments made by a variety of people in this thread which began the monochrome discussion carried over into this thread.
You have constructed a straw man.
Who constructed a straw man?
 
at some point these claims lead to one or all of the following conclusions ....

the senior project engineers / color scientists / test technicians at Leica Camera Wetzlar Germany :

1) did not know what they were doing when they released the mono Leica

2) released it knowing there was no advantage over a color version just to get money

3) hoodwinked some of the worlds best photographers who have been shooting for decades

4) managed to keep the farce going for a decade that the models have been out
None of your conclusions are supported by either Erik’s or my claims.
Neither you or Erik nor any of your claims were mentioned.
Then whose claims are being questioned?
I don't see the post to which you replied as being about you
The assertion in the post I replied to was that all the various claims led to the same conclusions. I was being specific about two people.
I understood the context of the remarks to be a reaction (not a set of conclusions) to some of the comments made by a variety of people in this thread which began the monochrome discussion carried over into this thread.
You have constructed a straw man.
Who constructed a straw man?
pentax mono.
 
Last edited:
On the topic of what Leica is willing to produce and sell, within a vast constellation of limited editions was this $50,000 M-10P kit.
...Six sets were gifts to the royal family, one set was retained by the ThaiBev museum and one presented to Leica's museum. All of the remaining sets were donated to twenty-two different charitable foundations and sold at auction to raise funds.
That’s laudable.

The post was a response to the claim that unless a model offered superior image quality, Leica wouldn’t offer it for sale. The was a counter-proof.
This is a straw man argument as no such broad generic claim was made to prompt a response or counter-proof.

https://www.dpreview.com/forums/post/67258958
 
Last edited:
at some point these claims lead to one or all of the following conclusions ....

the senior project engineers / color scientists / test technicians at Leica Camera Wetzlar Germany :

1) did not know what they were doing when they released the mono Leica

2) released it knowing there was no advantage over a color version just to get money

3) hoodwinked some of the worlds best photographers who have been shooting for decades

4) managed to keep the farce going for a decade that the models have been out
None of your conclusions are supported by either Erik’s or my claims.
Neither you or Erik nor any of your claims were mentioned.
Then whose claims are being questioned?
It's not specified and perhaps intentionally, so as to avoid this kind of back and forth.
I don't see the post to which you replied as being about you
The assertion in the post I replied to was that all the various claims
Where does it say "all" the various claims? It's not specific as to which claims are being referred to and again perhaps for the same reason as above or perhaps it's just as a broad general reaction to the tone of some previous posts.
led to the same conclusions.
It actually says "to one or all of the following conclusions".
I was being specific about two people.
The poster to whom you replied wasn't.
I understood the context of the remarks to be a reaction (not a set of conclusions) to some of the comments made by a variety of people in this thread which began the monochrome discussion carried over into this thread.
You have constructed a straw man.
Who constructed a straw man?
pentax mono.
I don't think so. But people are always entitled to form their own opinions.
 
Last edited:
at some point these claims lead to one or all of the following conclusions ....

the senior project engineers / color scientists / test technicians at Leica Camera Wetzlar Germany :

1) did not know what they were doing when they released the mono Leica

2) released it knowing there was no advantage over a color version just to get money

3) hoodwinked some of the worlds best photographers who have been shooting for decades

4) managed to keep the farce going for a decade that the models have been out
None of your conclusions are supported by either Erik’s or my claims.
Neither you or Erik nor any of your claims were mentioned.
Then whose claims are being questioned?
It's not specified and perhaps intentionally, so as to avoid this kind of back and forth.
I don't see the post to which you replied as being about you
The assertion in the post I replied to was that all the various claims
Where does it say "all" the various claims?
It says "these claims". If there were claims excluded from the discussion, it should have been specific.
led to the same conclusions.
It actually says to one or all of the following conclusions.
I was being specific about two people.
The poster to whom you replied wasn't.
He didn't bound the scope of his assertion.
I understood the context of the remarks to be a reaction (not a set of conclusions) to some of the comments made by a variety of people in this thread which began the monochrome discussion carried over into this thread.
You have constructed a straw man.
Who constructed a straw man?
pentax mono.
I don't think so.
What claims by whom support one or more of the conclusions above in light blue?
But people are always entitled to form their own opinions.
 
On the topic of what Leica is willing to produce and sell, within a vast constellation of limited editions was this $50,000 M-10P kit.
...Six sets were gifts to the royal family, one set was retained by the ThaiBev museum and one presented to Leica's museum. All of the remaining sets were donated to twenty-two different charitable foundations and sold at auction to raise funds.
That’s laudable.

The post was a response to the claim that unless a model offered superior image quality, Leica wouldn’t offer it for sale. The was a counter-proof.
This is a straw man argument as no such broad generic claim was made to prompt a response or counter-proof.

https://www.dpreview.com/forums/post/67258958
Thanks for the input! Super-great that you have the time to go back through old threads searching for opportunities to comment on perceived logical issues.
 
To arrive at monochrome images from a color sensor requires recording light intensity from various scenes thru color filters with different transmission characteristics, combining and interpolating the resulting recorded values into RGB pixels, and then balancing and reinterpreting those RGB values into monochromatic intensity values for a grayscale image...

...It's certainly possible to produce very similar results to direct monochrome capture and image processing using a color sensor. But given the number of variables and inherent differences along the color path; I would expect to also find some differences as well. What would the range of potential differences be, if any?
There are edge cases that favor one or the other. One that comes to mind in favor of monochrome is a low light situation where shutter speed is constrained and the temperature of the blackbody-like illuminant is below 3000K. However a monochrome sensor most likely overcounts photons at the edges of the visible range, so SNR results may be higher than what the Human Visual System would see, a milder version of what you get with cameras without a UV/IR filter. To each their poison.

Jack
Thanks once again for your thoughtful reply. I have linked your full reply and then edited much of it for the sake of space. These posts can get very long at times over the course of a discussion!

The noticeable difference I've seen in monochrome images is the improvement in apparent sharpness due to eliminating interpolation estimates (and errors) being added to each pixel during demosaicing in reconstructing (as best a given application is able) the missing values which were not recorded during capture. There may also be some other very subtle but perceptible differences in noise, tonal gradations, etc. under some circumstances I suppose — but I can't say I've noticed or gone looking for them.

The first digital capture systems I used, back in the Digital Stone Age, utilized monochrome sensors (Leaf DCB II Live and Leaf Volare) which produced color using sequential captures thru RGB filters. Those 4 and 6 megapixel dinosaurs are the only devices I've used on a regular basis without a Bayer color filter array covering the sensor. If, however, you're not looking to produce color; then I can understand why you may prefer using a monochrome sensor with a dedicated camera that supports it.

I've not had any interest in creating side by side comparisons using the same cameras or backs, with and without a color filter array, to compare monochrome results. There are however comparisons out there for those interested in searching them out. The various Phase One Achromatic backs made over the years, all of which have color counterparts, would be one place to start looking. Leica models have been mentioned as well. There have also been digital cinema cameras produced in versions with monochrome sensors.
 
Last edited:
On the topic of what Leica is willing to produce and sell, within a vast constellation of limited editions was this $50,000 M-10P kit.
...Six sets were gifts to the royal family, one set was retained by the ThaiBev museum and one presented to Leica's museum. All of the remaining sets were donated to twenty-two different charitable foundations and sold at auction to raise funds.
That’s laudable.

The post was a response to the claim that unless a model offered superior image quality, Leica wouldn’t offer it for sale. The was a counter-proof.
This is a straw man argument as no such broad generic claim was made to prompt a response or counter-proof.

https://www.dpreview.com/forums/post/67258958
Thanks for the input! Super-great that you have the time to go back through old threads searching for opportunities to comment on perceived logical issues.
The original post I linked and to which you replied is just three days old. Still within the window for guests and fish as the old adage suggests. But you can accept or reject it as you like — as with fish and houseguests.
 
Last edited:
On the topic of what Leica is willing to produce and sell, within a vast constellation of limited editions was this $50,000 M-10P kit.
...Six sets were gifts to the royal family, one set was retained by the ThaiBev museum and one presented to Leica's museum. All of the remaining sets were donated to twenty-two different charitable foundations and sold at auction to raise funds.
That’s laudable.

The post was a response to the claim that unless a model offered superior image quality, Leica wouldn’t offer it for sale. The was a counter-proof.
This is a straw man argument as no such broad generic claim was made to prompt a response or counter-proof.

https://www.dpreview.com/forums/post/67258958
Thanks for the input! Super-great that you have the time to go back through old threads searching for opportunities to comment on perceived logical issues.
The original post I linked and to which you replied is just three days old. Still within the window for guests and fish as the old adage suggests. But you can accept or reject it as you like — as with fish and houseguests.
Everyone should have a hobby they enjoy.
 

Keyboard shortcuts

Back
Top