Conversion of Color to Grayscale in Depth!

I agree. But the most important is that you are able to edit your picture in which way you like it.

Thank you for mentioning that chroma subsampling in JPEGs!

I checked my export settings in Gimp and RT. In both of them you can select very high quality settings.
Gimp manual tells there is no chroma subsampling at 4:4:4 setting. I use quality setting 100. I guess it means zero interpolation ...
Look: docs.gimp.org/2.10/en/gimp-images-out.html#save-export-image

In RT I also have Quality: 100 and Subsampling: Best Quality.

From the same picture I get the same file size in both editors with these settings.

Yes, even with "fine" selected in-camera, my embedded JPEGs come out 4:2:2 which is full-res (no sub-sampling) vertically but is 1/2 res horizontally.

see en.wikipedia.org/wiki/Chroma_subsampling

For posting on the web, I use 4:2:0 (1/2 res both horizontally and vertically} but with minimum compression.

I wasn't suggesting using the JPEG for final editing. My point was just that it may be wasteful of storage space to save two files, raw and JPEG, if your camera already has a good JPEG in the raw file. I haven't subtracted an embedded JPEG from an outright JPEG to look for differences, but Canon's embedded JPEGs seem to be about as good as the camera's HQ JPEGs.

Here's a Sigma best X3F raw conversion 4:4:4 vs. extracted from X3F 4:2:2 vs. subtraction:

Best viewed large size. Obvious difference in luminosity. Noticeable difference in overall color. Big difference in edge detail.

Comments invited.

Now I'm a little confused. When I said I extracted an embedded jpeg file from my Sony RAW file, John said he did not Subtract the embedded jpeg file from the outright JPEG of his Canon RAW and xpat showed subtracted jpeg image from his Sigma camera.

How should we convert to grayscale if we still want to do so?

I showed a third image which is the result of mathematically subtracting the one image (converted) from the other (extracted}.

When John said he did not subtract, he also meant to subtract mathematically. That is to create a new image by subtracting each pixel value in one image from the same location pixel value in another image. That is to say that, if the pixel at x,y 100,200 = 50,60,80 in one image and 45,50,65 in the other, then the result of the subtraction is 5,10,15 ... a much darker image and indeed a black image** would mean no difference at all between the two compared images.

** if a result has a negative value, that value is set to zero, in which case black may not mean "no difference".

Not sure that question is relevant while we are talking about comparing two images by subtracting the one from the other.

I'm sorry of my so called "humour".😳
I understood your comparison but could not help myself. I'll try in the future. And, yes, I'm aware of (layer) blending mode "Subtract" and I saw how the chroma subsampling artefacts became visible there.

No problem and pardon me for misunderstanding your post!

Wow. All I ever said was that it may not be necessary to shoot RAW + JPEG if the embedded JPEG in the raw is good enough for your JPEG needs. I don't have any idea what anyone is talking about anymore.

But you also said "I haven't subtracted an embedded JPEG from an outright JPEG to look for differences" - that's what I was talking about.

I think I'm basically a simpleton, I just convert to monochrome in whichever editing program I'm using nd then try different color filters and adjust the intensity to where I like them. I then adjust the curves to get the brightness and contrast to where I want I use the raw file.
Bob

I thought you showed the difference between two images, converted with different lightnesses (which dilutes the actual differences in quality by putting a faint full image in the result).

What I would be talking about is the difference between two OOC JPEGs, one explicit and one embedded, from the same capture. I don't remember having done that. I very well may have at one time and forgotten, but I have found Canon's embedded JPEGs to be good enough as far as JPEGs go, especially if you already have a raw image to work with if you want something different or better.

How can you apply filters after you have converted to monochrome? There is no differentiation due to color left to filter.

Does your program allows you to work on a color image but has a monochrome layer on top that you see? That would be a good tool.

Same capture but one JPEG was extracted from the raw file and the other JPEG was created by the raw converter on my computer.

Ah, also from the same capture but both created in-camera, i.e. a raw+JPEG shot - assuming that "explicit" means from the in-camera converter.

My example was not so, because that camera can't save raw+JPEG. Therefore my example would indeed show differences resulting from the difference between the in-camera JPEG engine and my computer's conversion software.

Understood.

I made some confusion when I started to play with words "extract" and "subtract"😱 I'm sorry.

What xpatUSA did was exactly comparing the difference between two OOC Jpegs of his Sigma camera. He extracted the embedded jpeg and subtracted it from the explicit jpeg ( or other way round, I don't know). As you can see at his example there is difference between those.

EDIT. expatUSA posted at the same time a little before me and it seems I did not understand how his Sigma camera did work.

No, sorry: one JPEG image was extracted from the OOC raw file and the other JPEG image was exported from the Sigma Photo Pro software on my computer therefore not OOC

I guess Bob is speaking of hardware called filters and used when shooting.😎

I understood. Look at my Edit in my post. We are posting at the same time now!