Question about converting a color image to black and white

[alecspra25]

When I convert a color image to black and white I dont bother trying to get the color balance and saturation right before converting it. The reason is that I have found that by the time i settle on the black and white image that I like, the underlying color image usually looks awful. So i have always been puzzled by claims that say you need to get your colors right before converting.

To be clear, what I do is process the 32 bit color raw image to correct for chromatic aberration, distortions, and luminosity values, and sometimes tone mapping. Then I convert the image from 32 bit to 16 bit. AT that point I add a black and white adjustment layer above the color image, and tweak the color channels until I get within the ball park of what I like. Then I add a white balance adjustment and saturation layers underneath the black and white layer, then tweak those 2 layers to refine the look I want in the black and white image. Then I may add a levels and possibly additional adjustment above the black and white layer, and tweak those until I get the final look I like.

At that point if I turn off the black and white adjustment layer and look at the color image result underneath, it usually looks awful.

So I dont understand the point of trying to get the colors right before using the black and white adjustment layer. From my perspective, it sounds like a waste of time.

Does that make sense to you? Do you use a similar approach? Am I missing something here?

 

[Kevner]

Greetings alecspra25,

In response to your question, I think you will find that each photographers workflow is personalized and will differ from yours (slightly or greatly). I find that for B&W work that is intended to convey the qualities of light & color/contrast of a scene, I have better success first processing the image in color and then working own the B&W image. For images that are less representative (aesthetically pushed) I don't adjust in the image in color first. I go straight to working on the image in B&W.

Hope this is helpful,

Kk

 

[grsnovi]

I took this image of two very nearly identical film cameras.


No edits to the raw

I knew that I was going to convert to b&w.



Once I started adjusting the color sliders, so much detail appeared in the leatherette.

I never worry about getting the color right first.

I will sometimes use the camera's "baked in" b&w scene jpg option.

--
Gary
https://www.flickr.com/photos/193735606@N03/

 

[NAwlins Contrarian]

When I convert a color image to black and white I dont bother trying to get the color balance and saturation right before converting it. The reason is that I have found that by the time i settle on the black and white image that I like, the underlying color image usually looks awful. So i have always been puzzled by claims that say you need to get your colors right before converting.

To be clear, what I do is process the 32 bit color raw image

No such thing. The vast majority of digital camera raw files have either 12 or 14 bits of data per pixel. A few older compact cameras have only 10 bits; a few current medium format digital cameras have a true 16 bits, although how much of the two least significant bits is real signal instead of noise is debatable. And those aren't color; they're light (photons) making it through a red, green, or blue filter. The raw converter normally de-mosaics them to full color by referring to nearby pixels with other color filters.

to correct for chromatic aberration, distortions, and luminosity values, and sometimes tone mapping.

Sounds good.

Then I convert the image from 32 bit to 16 bit.

Again, you never had 32 bits, and you probably never had a true 16 bits, so I don't know what this is about.

AT that point I add a black and white adjustment layer above the color image, and tweak the color channels until I get within the ball park of what I like. Then I add a white balance adjustment and saturation layers underneath the black and white layer, then tweak those 2 layers to refine the look I want in the black and white image. Then I may add a levels and possibly additional adjustment above the black and white layer, and tweak those until I get the final look I like.

Sounds complicated. A lot of those controls are at least overlapping with each other.

At that point if I turn off the black and white adjustment layer and look at the color image result underneath, it usually looks awful.

Not surprising. Consider back in the days of B&W film, photographers routinely used e.g. red and yellow filters. Put one of those in front of color transparency ('slide') film and the results would typically look awful.

So I dont understand the point of trying to get the colors right before using the black and white adjustment layer. From my perspective, it sounds like a waste of time.

I think the definition of "right" has to do a lot of work here. If by "right" you or we mean 'approximating what my eyes say they saw' then it seems somewhere between a waste of time and potentially counter-productive.

But if by "right" we mean 'optimal for the B&W tones I want' then it can be useful. But that's more like thinking, 'Do I want green things like plants to be lighter or darker in my B&W? Do I want the blue sky lighter or darker?' And then you adjust the color to try to provide the best starting point for the B&W tonal relationships you want.

Does that make sense to you? Do you use a similar approach? Am I missing something here?

K.I.S.S. is a valuable principle to me, so all of that color channels plus white balance plus saturation bit sounds like a bad idea to me. Likewise for luminosity then levels etc. Personally I find curves among the most powerful and useful tools, especially when split out into separate red, green, and blue curves, and especially when applied with multiple layers and masking, and/or local adjustments.

But that's if you want to 'roll your own'. I will confess to being lazy, and often liking to use Nik Silver Efex. Currently I have a DxO Nik Collection that's two or three major versions old and has Nik Silver Efex Pro 3. So lazy means a fairly 'straight' raw conversion in DxO PhotoLab, exporting a TIFF to Silver Efex, scrolling through some of the 65 or so presets to find the one that works best, and then often using the color filter simulators (see comment about film above) to shift some of the tonal relationships.

But there's more than one way to do this, many people are more expert than I am, and tastes differ.

 

[D Cox]

I would like the converse: an AI based program to colorize B&W photos.