What a monochrome sensor will show you

[Flektogon]

The following is an extract from the discussion on DPReview. My reaction/reply to one forum member is as follows. I am just curious, whether I am right . What is your opinion?

ivarbg wrote:
Monochrome camera is a gimmick. Every modern camera can take pictures in BW mode or you can convert colour to BW in the postprocess. It may have limited usage in very large prints or for better photos in low light conditions.

More or less I agree with you. But to be fair, yes, a sensor without the Bayer filter (even a foveon one) will capture more details, especially tiny, coloured details. Imagine a tiny, red, for example, detail of a size of one sensor pixel. While a sensor without mosaic filter will capture it, a standard "coloured" sensor will do so only if incidentally that red detail is projected to a pixel with the red filter in front of it. Otherwise it ill be lost.


However, the human sight has a limited resolution of the coloured details anyway, so even if such a tiny red speckle will be captured, your sight will not be able to detect it anyway . But a monochrome sensor will show you it as a B&W detail, so you will see it perfectly. Actually, a monochrome sensor will show us details, which otherwise we would not be able to detect even in the original .

 

[Brian]

chimney_m9_converted by fiftyonepointsix, on Flickr

chimney_monochrom by fiftyonepointsix, on Flickr

If you cannot see a difference between these two photos, then it will not matter to you. If you do see a difference (I do) and it helps make the case, fell free to use them. I did this test after first getting my M Monochrom. I had to buy one- I called Kodakwhen the M9 was brand new and asked them to make a Monochrome version. Also called in 1992 for an IR version of the DCS200. I am 2 for 2.

Leica M9 and M Monochrom, both with the Ultron 35/1.7 at F4.

 

[Brian]

I found the thread- where the guy stated that contrast filters used for Film were just not required for a Digital camera. The reverse is true. I posted this:

With a Monochrome Digital camera- the response is Linear, and use of color contrast filters is even more important than with film. Depending on the specific spectral response, usually a deeper filter is required. Y52, O56, and R60 get the most use on my Leica. R60 got the most use on my DCS200ir- a full-spectrum monochrome camera bought 30 years ago. Using an O56 or R60- check for underexposure as the metering cells may have a different response in that region from the CMOS detector.

 

[Flektogon]

chimney_m9_converted by fiftyonepointsix, on Flickr

chimney_monochrom by fiftyonepointsix, on Flickr

If you cannot see a difference between these two photos, then it will not matter to you. If you do see a difference (I do) and it helps make the case, fell free to use them. I did this test after first getting my M Monochrom. I had to buy one- I called Kodakwhen the M9 was brand new and asked them to make a Monochrome version. Also called in 1992 for an IR version of the DCS200. I am 2 for 2.

Leica M9 and M Monochrom, both with the Ultron 35/1.7 at F4.

Brian, it's not I (it iwas ivarbg) who doesn't believe in B&W! I replied with the explanation how a sensor without mosaic filter can capture far more tiny details. But your example is not what I would use to demonstrate it! And I really don't see any difference. But take a picture of a grass lawn, for example. While a monochrome sensor will be able to capture every single stem, a "coloured" sensor very likely will deliver something close to a watercolor painting. Why? Because all tiny details, which were projected on the pixels having Red and/or Blue filter will be simply lost! And the result of demosaicing (spreading of the remaining details over the remaining pixels) will be such a fuzzy appearance.

 

[Brian]

I did a respond to the thread- if it ended up with you, I am sorry.

The example shows the artifacts introduced by the interpolation of the mosaic filter of the color image. One side has a dark outline, the other has a white line running across the edges. As stated- if that makes no difference to the user, they are mostly going to be fine with a color conversion. I used an R60 filter with the M Monochrom and a Lightroom "red" filter with the M9. The details of the Lightning Rod and the shadow cast are also better in the monochrome image. Same lens, same F-Stop, Same KAF-18500 CCD, one Bayer and the other monochrome.

I've had the M9 for 12 years and the M Monochrom for 10 years, use with Y48, Y52, O56, and R60. Both have the last generation of CCD sensors in them, I would get the Pentax before an M11. Not dumping the M Monochrom. I've also written a lot of my own DNG processing software for both.. As standard- I run my Fortran code to apply a Gamma curve converting values to 16-bits using a look up table and outputting a new DNG file. Many people shooting Leica monochrome cameras do a lot of post-processing, I just do it in Batch.

 

[Brian]

Leica - Monochrome Conversions: Using a Yellow filter with a color camera

The question came up on another forum regarding the use of color filters with color cameras when you know in advance that you will doing a conversion. Looking at the spectral response of the Red, Green, and Blue channels of the sensors used in the M8, M9, and many other cameras: the answer is...

cameraderie.org

Leica - Experiments in Leica M8 and M9 Conversions

I have been experimenting with using color contrast filters with my M8 and writing my own conversion software to work with the channels based on their altered spectral response. With a Yellow Y48 filter, the channels are mostly balanced. With an O56 filter, the Blue channel is at about 1/4th the...

cameraderie.org

Leica - Adding a "Gamma Curve" to a digital image: Thinking out Loud and Experiments

A quick Google search on how the eye perceives intensity- http://www.cs.ubc.ca/~harrison/P202/PDF/05-perception-of-brightness-4up.pdf Looks very much like how Film records intensity. Sensitometry - Wikipedia And a mathematical formula to compute an "S" chaped curve: Logistic function -...

cameraderie.org

Anyway- I wrote my own software to use color filters with the M9, custom demosaic, and convert to linear DNG. The color conversions have better highlight recovery as not all the pixels in the Bayer cell saturate at the same time as they do on the M Monochrom. In the end- I use the M Monochrom for B&W, the M9 for color, and the M8 for color and IR.

 

[Brian]

Fortran Output.

 

[Brian]

Brian, it's not I (it iwas ivarbg) who doesn't believe in B&W! I replied with the explanation how a sensor without mosaic filter can capture far more tiny details. But your example is not what I would use to demonstrate it! And I really don't see any difference. But take a picture of a grass lawn, for example. While a monochrome sensor will be able to capture every single stem, a "coloured" sensor very likely will deliver something close to a watercolor painting. Why? Because all tiny details, which were projected on the pixels having Red and/or Blue filter will be simply lost! And the result of demosaicing (spreading of the remaining details over the remaining pixels) will be such a fuzzy appearance.

The RGB Dye used in the Mosaic Filters have a lot of overlap. They are not hard cut filters like the older Technicolor Process where true separation was used. The interpolation schemes used for color cameras are more sophisticated. All that stated- I still prefer the monochrome camera.

What I've tried to do is compare the best I could get from the M9 with the M Monochrom, so when someone asked what the real advantage is- could show them an A versus B comparison.

 

[Brian]

BLUF of why the experiment was worth doing-
"The question came up on another forum regarding the use of color filters with color cameras when you know in advance that you will doing a conversion. Looking at the spectral response of the Red, Green, and Blue channels of the sensors used in the M8, M9, and many other cameras: the answer is Y48. The Medium Yellow cuts out about 1/2 of the spectrum, cuts out "traditional color-separation Blue", cuts out about 1/2 of "color separation Green". The Medium Yellow filter is also known as a "cloud filter", it darkens the blue in the sky and brings out clouds. So why not just do all this in software and kill the Blue channel? The color dye used in Mosaic filters of camera sensors are not true separation filters, but are sensitive way beyond the traditional cutoffs. The Blue channel response extends almost out to Red. By using the Yellow filter, you get all of the Blue response that extends past 480nm. With the Leica M8, that portion is almost as efficient as the Red channel. You get a much more balanced image to start with.

Most software will attempt to demosaic and process the image not knowing that you used a yellow filter. I guess you can fight it, means post-processing."

and I will not post hundreds of lines of Fortran code here, but it is in the link above. I ended up doing histogram equalization and custom demosaic. Even with the Yellow filter, the Blue channel has a lot of sensitivity left in green. So- you still have information to work with.

 

[Dino26323]

The following is an extract from the discussion on DPReview. My reaction/reply to one forum member is as follows. I am just curious, whether I am right . What is your opinion?


More or less I agree with you. But to be fair, yes, a sensor without the Bayer filter (even a foveon one) will capture more details, especially tiny, coloured details. Imagine a tiny, red, for example, detail of a size of one sensor pixel. While a sensor without mosaic filter will capture it, a standard "coloured" sensor will do so only if incidentally that red detail is projected to a pixel with the red filter in front of it. Otherwise it ill be lost.


However, the human sight has a limited resolution of the coloured details anyway, so even if such a tiny red speckle will be captured, your sight will not be able to detect it anyway . But a monochrome sensor will show you it as a B&W detail, so you will see it perfectly. Actually, a monochrome sensor will show us details, which otherwise we would not be able to detect even in the original .

Yes I too have been following some posts on the thread on the other forum too.
There are a lot of different opinions and all of them are not wrong exactly. I can see some truth in all the different opinions. We all have different opinions on black and white or about black and white so there is not going to be an absolute answer to the question of whether the monochrom sensors are relevant necessary or worth it.
All of the results that I have seen are very interesting even though none of them say they used a filter. There are quite a lot of Leica test shots that I have seen going back many yrs. .Like any technique in post I suggest that the monochrom sensor , with no filter attached to the lens, will work with some shots and not so much others. Certainly there appears to be a look sort of going on because the whole photo is directly just displaying the luminosity of tones and is blind to colour separation or at least is almost blind to favouring any colour channel. This in itself appears to be successful on many of the test shots BUT granted we are only being shown the shots that worked best from such test shots.
I personally am intrigued by the results. There was a lot of talk about it , that is, the sensor monochrom being in the wrong pentax camera. I do not see it that way. Because it is an ApSc camera then there are a ton of options like street photography and Landscape photography whereby two bodies are taken with me. It works out pretty good I think being ApSc and I would take two cameras with me. I would buy filters but also I would try and use no filters as much as I can because there is a lot that can be done in post just as one person kept trying to tell everyone.
One person seemed annoyed with all the greys being shown and wanted more contrast into near black and white. I have found I like the look of the tones from the test shots whereby there are lot of different grey tones. EVERYONE is going to be different in regards opinions on this camera. I believe Ricoh will be glad they offered this option and may end up offering it on many other Ricoh Pentax cameras. Whether you think it is right is irrelevant if Ricoh sell more K3iii as well as K3iii monochroms because of its release to the market.

Regards
Dino26323