Color to Black & White: Luma, Greyscale, Desaturation, Average

[Andrea B.]

EDITOR'S NOTE:  I need to preserve this, so I'm putting it here. Comments, corrections, suggestions are all welcomed.

---

 

 

The middle strip shows Blue in varying degrees of brightness and saturation. Six different methods of conversion to Black and White are shown next to it.

 

The leftmost strip shows a kind of relative luminance conversion, sRGB luma, which is calculated on converted RGB values which are non-linear. This conversion does not do well at preserving brightness differences on the Blue strip. But in some cases, the results are better and do preserve the luminosity of colors. See below.

 

The next strip shows middle grey (128,128,128) as a Color layer over the blue strip. This produces a luminosity result very similar to the sRGB luma method. The formula has different weighting factors. (Shown below.)

 

The strip to the immediate left of the Blue strip is a greyscale conversion from Photoshop Elements. Adobe has never shared their "secret formula" for greyscale conversion. Note that greyscale conversion preserves most of the brightness and saturation differences of the Blue strip.

 

Immediately to the right of the Blue is a simple desaturation of the Blue strip. To desaturate, the max and min RGB values are averaged. This desaturation also preserves the brightness and saturation differences of the Blue strip. But desaturation does not always work well. See below.

 

The next strip shows a another simple averaging method for desaturation. The result looks good here, but you can run into troubles with this method also. See below. Please note that there are a variety of desaturation formulas.

 

And finally the rightmost strip shows a conversion based on the brightest RGB values. Obviously this method does not preserve saturation differences when brightness is the same.

 

 

 

Here are the 6 primary and secondary RGB colors:

• Top Row: Red(255,0,0) Green(0,255,0) Blue(0,0,255)
• Bottom Row: Cyan(0,255,255) Magenta(255,0,255), Yellow(255,255,0)

The 6 colors are shown at full saturation and full brightness.

 

 

 

Conversion to B&W using sRGB luma calculation.

Here the luminosity of the fully saturated, fully bright colors is preserved. The yellow block is most luminous. The blue block is least luminous. The number on each block indicates the grey value. For example, the red patch (255,0,0) has been converted to (54,54,54)

 

 

 

Conversion to B&W by layering middle grey (128,128,128) as a Color layer over the 6 blocks.

Photoshop Elements was used for the layers. The approximate formula for luminosity in PSE is shown. Again the luminosity of the fully saturated, fully bright colors is preserved. But above, there were some problems with the blue strip on the bottom half with the fully saturated but decreasing brightness patches.

 

 

 

 

Conversion by Greyscale

Here the Photoshop Elements greyscale conversion was used. The formula is unknown. It appears to be some kind of weighted calculation which preserves luminosity.

 

 

 

 

Conversion to B&W by simple desaturation.

Obviously, simple desaturation does not work so well when there is uniform brightness and saturation in the colors.

 

 

 

 

Conversion to B&W by an averaged desaturation.

Again, not so great given uniform brightness and saturation.

 

 

 

Conversion to B&W by maximum brightness, max(R,G,B).

A bit difficult to show here because each block produces 255 and (255,255,255) is pure white.

 

 

Conclusion

When converting color to Black & White, there is no best method. But it looks to me like the greyscale or luma conversions are probably a good place to start. With any method, the outcome is dependent on the intensity (brightness and saturation) of the original colors together with whatever the photographer wants to bring out in the photograph by use of filters (either physical or digital) as well as other editing tricks and techniques.

Of course, you can always let Photoshop convert to Black & White for you. 

 

 

QUESTION

What is a Monochrom* camera recording and what is it doing to the raw data?

My assumption has been that the photon wells are capturing more photons from more reflective areas and fewer photons from less reflective areas of a subject. But does the camera massage that data in any way before producing the photo? Is the resulting photo similar to greyscale or desaturated or luma conversions as shown above?

The answer is probably out there with a little bit of Internet sleuthing.

Related Question

Two objects are sitting side by side. One reflects 100% blue and absorbs all other wavelengths. The other reflects 100% yellow and absorbs all other wavelengths. If photographed with a Monochrom camera, will the two objects look the same in the photo? We perceive the yellow object as having more luminosity than the blue object. But the Monochrom camera has no way to detect that. 

 

*I am referring here to a Leica Monochrom digital rangefinder.

 

Other Conversion Methods from Color to Black & White

• Black to White gradient maps are another useful method. Experiment with the middle slider for increasing/decreasing dark/light areas or substitute dark grey for black for a different look.
• You can use the channel mixer to weight R, G and B values. Or use the channel mixer as a way to apply a filter.

 

 

BOUQUET

 

 

 

Black & White Conversion via Desaturation

Very dull !! That is, it lacks contrast.

 

 

 

Black & White Conversion via Greyscale

 

 

 

Black & White Conversion via Luma Weighting

 

 

 

Black & White Conversion via Gradient Mapping

I like this one best.

I chose black and white for the gradient.

 

 

 

The Gradient Mapping Leveled

Maybe a bit too much.

I applied the black dropper to the darkest area and the white dropper to the brightest area. 

[photoni]

51 minutes ago, Andrea B. said:

Adobe has never shared their "secret formula" for greyscale conversion

 

apart from that ... I would like to understand the difference between a debayered sensor and a full spectrum sensor

... and what kind of photographic film it is comparable to ( Ortho - Superpancromatic - Pancromatic - infrared )

 

 

P.S.

 

in LAB there is a difference in brightness,

but it is the same as the saturation subtraction with Capture One

(like all RAW processing programs the internal engine works in LAB to then be able to save the processed file in different color profiles)

 

 

 

[Andy Perrin]

Quote

 the difference between a debayered sensor and a full spectrum sensor

Very little, I think, although there may be AR coating differences, and other small differences? Not things you would probably be able to tell without instruments!

[Andrea B.]

A full spectrum conversion retains its Bayer color filter while a de-Bayered conversion does not. So it is possible that a de-Bayered conversion can capture a greater variety of wavelengths than can the full spectrum conversion because the Bayer dyes might prevent some wavelengths from reaching the sensor of the full spectrum conversion. Also, a de-Bayered conversion, IF used with the appropriate software*, will produce more detail in a photo than the full spectrum conversion because there is no demosaicing step required.

 

I don't know that anyone here knows enough about film to answer your other question. Birna might be able to.

 

* One example is Monochrome2DNG, an app made by the Raw Digger team for processing raw files from de-Bayered cameras.

LINK:  Monochrome2DNG

[Andrea B.]

Toni's flowers with leveled gradient map.

I was too heavy on the white end and lost some detail there.

[dabateman]

1 hour ago, Andrea B. said:

A full spectrum conversion retains its Bayer color filter while a de-Bayered conversion does not. So it is possible that a de-Bayered conversion can capture a greater variety of wavelengths than can the full spectrum conversion because the Bayer dyes might prevent some wavelengths from reaching the sensor of the full spectrum conversion. Also, a de-Bayered conversion, IF used with the appropriate software, will produce more detail in a photo than the full spectrum conversion because there is no demosaicing step required.

 

I don't know that anyone here knows enough about film to answer your other question. Birna might be able to.

 

 

But a de-Bayered sensor doesn't know its de-Bayered.  You now get a very magenta image as the camera is trying to compensate for the lower blue and red signals than green.  But now all are the same after de-Bayering.

This isn't a problem for a Leica monochrome camera that knows its monochrome. 

You can also program around that problem if you buy a monochrome sensor or a de-Bayered, now monochrome sensor for a Raspberry pi. However, you get this weird "screen door" effect with de-Bayered sensors.  You need to blur the image and thus loose details. The monochrome 2 DNG software has various blur options, or you can program your own with a Raspberry pi.

 

[Andrea B.]

Yes, the de-Bayered camera's internal software/firmware continues to demosaic the data and that produces the described weird effects in the JPG stored within the raw file as displayed on the LCD. But if the raw file is processed in an app which does not demosaic -- such as the Monochrome2DNG you mentioned -- the screen artifact does not happen. 

 

Side Note: I thought that the Mono2DNG app reconstructed the data in order to suppress the screen rather than blurring the screen? 

(I'm not sure that "reconstruct" is the correct wording.)

 

 

I added a reference to Monochrome2DNG above where I mentioned "in appropriate software".

 

********

 

The original topic here is about converting an RGB image to Black & White by using various weighted RGB formulas (greyscales) or RGB averaging (desaturations).

It is not about Monochrom cameras whether original or de-Bayered.

[photoni]

@Andrea B. When it comes to conversion it is mandatory to start from a RAW and use a soft that "translates" from the raw LAB data such as AdobeCameraRaw or Lightroom or Capture One or Monochrome2DNG etc etc ...

Each conversion with photoshop or others is "lame" because it uses already processed data.

[Andrea B.]

Yes, we know this. In practice I only process raw files.

 

But it is difficult to create raw files for illustration of B&W conversion techniques! For example, my blue strip above. How does one create an ARW, RAF, CR or NEF with a blue strip of different saturations and brightnesses. Not easy.

[Kai]

With all these considerations about the technical possibilities, I ask myself:
What purpose should an image (color or b+w) serve?

- Is it a matter of measuring the diffuse reflectance (for semi-scientific evaluations)?

- Is it about making any differences/structures optimally visible (e.g. in forensics or criminalistics)?

- Is it an aesthetically pleasing, artistic image?

- Is it a grayscale image that reproduces the visual impression of brightness as accurately as possible? (There can be no equivalent in the UV and IR.)
 

That's why I don't have "the" conversion process, but adapt my conversion on b+w to the respective purpose.

[Alaun]

Just to mention it,  when you shoot IR, e.g.  with 800nm, the cameras behave almost like b&w.  You can take the WB from there and take pictures in UV.  I used that with my Panasonic cameras.

 

[Andy Perrin]

19 hours ago, Kai said:

With all these considerations about the technical possibilities, I ask myself:
What purpose should an image (color or b+w) serve?

- Is it a matter of measuring the diffuse reflectance (for semi-scientific evaluations)?

- Is it about making any differences/structures optimally visible (e.g. in forensics or criminalistics)?

- Is it an aesthetically pleasing, artistic image?

- Is it a grayscale image that reproduces the visual impression of brightness as accurately as possible? (There can be no equivalent in the UV and IR.)
 

That's why I don't have "the" conversion process, but adapt my conversion on b+w to the respective purpose.

Yes, this is a very good comment. To know what the "best" b+w image is, you have to specify the purpose first.

[photoni]

On 5/13/2022 at 9:24 PM, Kai said:

With all these considerations about the technical possibilities, I ask myself:
What purpose should an image (color or b+w) serve?

- Is it a matter of measuring the diffuse reflectance (for semi-scientific evaluations)?

- Is it about making any differences/structures optimally visible (e.g. in forensics or criminalistics)?

- Is it an aesthetically pleasing, artistic image?

- Is it a grayscale image that reproduces the visual impression of brightness as accurately as possible? (There can be no equivalent in the UV and IR.)
 

That's why I don't have "the" conversion process, but adapt my conversion on b+w to the respective purpose.

 

 

it is a response that resembles that of a politician before the elections.
if you tell me about white balance there is a standard way with PTFE, aluminum, steel, etc ... that gives you a standard
if you use the standard when developing the RAW file you will have a "the" parameter
If you use "the" parameter and remove the color information (all development programs use Lab) you will only have the L brightness channel, this is also "a" BW standard.
It is not aesthetic or specific taste for a particular use.

 

I believe :)))

Antonio