Monochrom Editing / Export and "Color" Space?

[nameBrandon]

Advertisement (gone after registration)

Hey all,

Wondering how you approach maintaining the fantastic level of shadow detail we get with the monorchrom sensors when moving into an 8-bit format such as JPG, as well as standardizing on sRGB vs something like Adobe RGB 1998?

When exporting to JPG (100% quality) with sRGB, shadow detail is definitely lost and gets muddy, while Adobe RGB maintains much more detail. I've also seen posterization or blockiness if black levels go to near 0, where the conversion from the working raw/dng space (in my case in Capture One) to sRGB loses the gentle tonal transition and instead it seems as if anything close to 0 just gets mapped to 0.. 

As a result, I've started manually protecting the white and blacks by using the curve adjustment and just changing my white point to ~252 and black to 3.. but even that small shift sometimes reduces the nice dynamic range we get in raw editing. 

I'm curious if you all have run into this at all, and what you all might be doing to prevent the sRGB conversion from creating a muddy effect on shadow detail?

Attaching an example (post below as well due to file size limitations).. If you right click on each image, and open the images up in a new tab and then zoom into 100% at the top-left or the bottom-left corner of the image, then flip back between them both, you should see where some of the tone mapping difference occurs.

This is Adobe RBG (profile embedded) JPG 100%, 100% crop.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

Edited November 21, 2024 by nameBrandon

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[nameBrandon]

This is sRGB, profile embedded, JPG @ 100%, 100% crop.

 

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[nameBrandon]

accidental double post. ignore.

 

Edited November 21, 2024 by nameBrandon
accident

[pippy]

Not sure if it will help you at all - you may already be doing this - but the one big lesson I learned (fortunately very early on!) when processing-out DNG as either TIFF or JPEG was to do as much 'main' tonal-adjustment as possible at the very first stage of conversion and not wait until TIFFs / JPEGs had been obtained. Trying to achieve the same tonal-range after conversion has occurred sometimes led to my experiencing 'Blocking / Posterization / Banding' in many areas; especially in shadows and skies if there is a wide tonal-variation such as might be obtained when using Circ. Pola / Orange or Red filters on a clear cloudless day.

I do all my DNG conversion using Photoshop so I'm not sure if the same tools are available with your program.

Philip.

[nameBrandon]

1 minute ago, pippy said:

Not sure if it will help you at all - you may already be doing this - but the one big lesson I learned (fortunately very early on!) when processing-out DNG as either TIFF or JPEG was to do as much 'main' tonal-adjustment as possible at the very first stage of conversion and not wait until TIFFs / JPEGs had been obtained. Trying to achieve the same tonal-range after conversion has occurred sometimes led to my experiencing 'Blocking / Posterization / Banding' in many areas; especially in shadows and skies if there is a wide tonal-variation such as might be obtained when using Circ. Pola / Orange or Red filters on a clear cloudless day.

I do all my DNG conversion using Photoshop so I'm not sure if the same tools are available with your program.

Philip.

Thank you! That's good advice.. what I've been doing is actually decreasing contrast before I export to 16-bit TIFF and (typically) import into SilverEfx.. thought being that I've decreased the tonal range a bit so it doesn't get cut off at either end.. but might try your approach and see where that gets me!

[ramarren]

8 hours ago, nameBrandon said:

Hey all,

Wondering how you approach maintaining the fantastic level of shadow detail we get with the monorchrom sensors when moving into an 8-bit format such as JPG, as well as standardizing on sRGB vs something like Adobe RGB 1998?

When exporting to JPG (100% quality) with sRGB, shadow detail is definitely lost and gets muddy, while Adobe RGB maintains much more detail. I've also seen posterization or blockiness if black levels go to near 0, where the conversion from the working raw/dng space (in my case in Capture One) to sRGB loses the gentle tonal transition and instead it seems as if anything close to 0 just gets mapped to 0.. 

As a result, I've started manually protecting the white and blacks by using the curve adjustment and just changing my white point to ~252 and black to 3.. but even that small shift sometimes reduces the nice dynamic range we get in raw editing. 

I'm curious if you all have run into this at all, and what you all might be doing to prevent the sRGB conversion from creating a muddy effect on shadow detail?

Attaching an example (post below as well due to file size limitations).. If you right click on each image, and open the images up in a new tab and then zoom into 100% at the top-left or the bottom-left corner of the image, then flip back between them both, you should see where some of the tone mapping difference occurs.

This is Adobe RBG (profile embedded) JPG 100%, 100% crop.

<snip image>

I pulled these images into Lightroom Classic for inspection. I'm not sure exactly what you're trying to do or ask ... both of the photos you posted require EV compensation of 4+ stops to approach a normal tonal curve. They are very seriously unexposed for my way of working. Or perhaps I don't understand what you're presenting here, or what question you're asking...? 

In my own photography with the Leica M10 Monochrome, I expose for a fully expressed tonal curve throughout the image, without clipping at the highs or falling into noise in the detail areas at the lows. I then adjust things to meet what my eye feels is satisfactory on the overall tonal scale.

G

G

Edited November 22, 2024 by ramarren

[nameBrandon]

Advertisement (gone after registration)

1 hour ago, ramarren said:

I pulled these images into Lightroom Classic for inspection. I'm not sure exactly what you're trying to do or ask ... both of the photos you posted require EV compensation of 4+ stops to approach a normal tonal curve. They are very seriously unexposed for my way of working. Or perhaps I don't understand what you're presenting here, or what question you're asking...? 

In my own photography with the Leica M10 Monochrome, I expose for a fully expressed tonal curve throughout the image, without clipping at the highs or falling into noise in the detail areas at the lows. I then adjust things to meet what my eye feels is satisfactory on the overall tonal scale.

G

G

Thanks for the response! The crops I shared are not the entire image, and they are underexposed, but this is intentional.This is one specific section of the image that is in the mainly in the darker portion of the histogram to illustrate what I am seeing... Let me try and explain it a different way..

• When you open a DNG in Capture One (or LR, for that matter), the working colorspace is a wide gamut one (ProPhoto RGB for C1, I believe).
• The export color space for web based JPG images is typically sRGB, which is far narrower than ProPhoto RGB (though not technically correct, one can think of it as if sRGB having far less colors or grey levels to work with than a wider gamut like ProPhoto RGB).
• Because of this, when editing DNG/RAW we see much smoother tonal gradation and transitions in colors (or in the case of monochrome, between grey levels)
• One of the great things about the monochrom sensor is the amount of detail we can keep in the shadows that bayer sensors would otherwise muddy or turn into noise, and many of us edit our monochrom images with this ability to retain shadow detail in mind. It is the shadow detail I'm interested here, and why the crops are underexposed to begin with.
• Once we export to JPG/PNG in an sRGB color space, this detail is often lost as we need to map tonal values from a much wider gamut into a narrower one where those tonal values don't exist.
• Depending on how extreme our shadows (or highlights) are, the mapping to sRGB can often lead to posterization, pixelation or other artifacts in areas where is a heavy density of deep blacks/shadows and bright highlights/white.

What I uploaded was a crop of an image, one with a sRGB color space / profile and the other in AdobeRGB 1998 (another wide gamut color space). If you open them both in a browser (right click, open image in new tab) or photoshop, and flip back and forth between the two images, you'll see the sRGB one is muddier than the AdobeRBG one. 

For example, if you switch between photos, you'll notice the tree in the AdobeRGB version appears to have black "holes" between the branches of pine needles were there is empty space (which is correct, IMO, and provides more contrast in the tree) whereas the same holes in the sRGB version appear to have been mapped to a lighter tone which blends in with the surrounding needles, decreasing contrast and losing detail (what I would call, "muddy").

What I'm curious about is how people typically handle this scenario and the techniques employed to retain extreme shadow and highlight detail when exporting to an sRGB color space that inherently doesn't offer as many tones to work with at those extreme brightness or darkness values.

Hope that helps!

Edited November 22, 2024 by nameBrandon

[ramarren]

Well okay: 

sRGB is a more limited color space than ProPhoto RGB (and Adobe RGB), so it is unquestionable that if you convert from either ProPhoto or Adobe to sRGB, the transformation is lossy and non-reversible. 

(In fact, ProPhoto RGB doesn't even make sense in an 8bit image at all ... there's not enough range in 8bit values to express the range of colors/tones that are the reasons for the definition of the ProPhoto RGB color space.)

The question is: When and why are you converting from these larger color spaces to sRGB? Never mind that you are by doing so tossing a huge lot of image information into the bucket ... there is simply no way to contain all of the information in an AdobeRGB color space (never mind a ProPhoto RGB color space) in an sRGB JPEG file. 

---
Context: 

JPEG (an image format defined by the Joint Photographic Experts Guild) is an 8bit, scalable compression image file format designed to be tunable for output presentation, and is a lossy format. The notion is that you define the parameters for configuration of the compression to match the size and quality of the display you intend to show the images on. 

The sRGB color space was developed in the 1990s to model accurately the color range and dynamics of a typical high resolution display of that time period. 
---

So ... A beautiful ProPhoto RGB image out of a Leica M10 Monochrom, with all it's remarkable 14-15 bit dynamic range and detailing, is going to be degraded by a significant percentage when JPEG compression is applied and it is sized for use on a typical display device (smartphone screen to desktop display), and further degraded when you apply the sRGB color space transformation to it. There's no getting around that, not at all. 

What I do is: 

- When I finish whatever rendering work I want to do on the DNG files, I export the rendered masters to ZIP-compressed, full-resolution, 16-bit, ProPhotoRGB color space TIFF files for printing and archiving, and future editing. 

- When I post photos on line, I export to JPEG/sRGB files at a reasonable (if not original) resolution @2400 pixels on the longest edge. That's good enough for most display systems to present my work satisfactorily. 

Is the output a perfect representation of the original master? Hardly. But it's good enough ... 

I spent time doing a detailed section by section comparison of your sRGB and AdobeRGB image fragments: On my 5K Apple Studio Display at 400% image display, yes I can see the differences ... But they are literally insignificant and inconsequential unless the only task I'm doing is to determine what's different between them at a pixel level, and at any normal resolution displaying the entire image, there is almost literally no difference between them. 

G

[nameBrandon]

23 minutes ago, ramarren said:

 

The question is: When and why are you converting from these larger color spaces to sRGB? Never mind that you are by doing so tossing a huge lot of image information into the bucket ... there is simply no way to contain all of the information in an AdobeRGB color space (never mind a ProPhoto RGB color space) in an sRGB JPEG file. 

 

Thanks for the reply! The conversion is a function of preparing a JPG for web / social media. Agreed that we are losing a ton of information, but it's unavoidable if one wishes to standardize to sRGB.

My question was primarily around the techniques people employ to limit the impact of that conversion to their images. Sounds like you accept the loss and move on, which is perfectly fine. I was curious if others perhaps embedded profiles that had a wider gamut like Adobe RGB and relied on browser interpretation to provide the appropriate rendering.. or perhaps people may have looked into extremes of their images and then proofed the output in target space in photoshop and adjust accordingly by perhaps raising the black and white points, etc..

The real concern I have (which is not unfounded, see below) is the additional compression that some image hosts / provider then re-apply to the already converted JPGs, which can cause even more severe artifacting, particularly if those extreme areas are not well managed in the first place upon initial color space conversion.. 

This is an extreme example, and is a combination of a browser issue as well as extreme dark areas, but this is why I started looking into it in the first place..  Look at the blocks on the right hand side. These are obviously not present in my original, nor in my exported JPG. They are a function of how the image host is further compressing the jpg I uploaded as well as the particular browser (in this case). But I did test this, and they are avoided if I raise the black point of the image by +10 or so, and follow the same processing chain. 

Appreciate the response(s), but it doesn't seem like there is anything novel here that anyone is doing to manage this, or that I've missed any standard techniques that people might take when working with monochrom files in a narrow color space. 

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

Edited November 22, 2024 by nameBrandon

[ramarren]

Unfortunately, there's no practical solution that works for all cases. Outputting to 8Bit Adobe RGB profiled data doesn't work properly unless you know that everyone in your audience is going to be using browsers that read the image profiles and apply them properly ... and of course, if the images are going through some intermediate server/distribution source that is also massaging the data, all bets are off. sRGB means that even unprofiled displays have a good chance of displaying images within a decent range of proper color balance and gradation.. 

At least using JPEG/sRGB as a basis is minimally compatible with nearly everything out there and if you do a decent job of tweaking your photos to minimize artifacts as best you can ... well, that's as good as you're likely to get. This is why I don't put a lot of effort into trying to devise "better" solutions, and I rely upon the ("knowledgable") viewing audience to understand that what they see on a display is only a halfway decent representation of the photograph, nothing more. (The "not knowledgable" viewing audience likely can't see the differences, nor care about them...)

In my opinion, an exhibition print is the final word, not what shows up on a computer display. The latter is just for convenience in distribution, and cost.

G

[ph0toni]

@nameBrandon There are some issues to consider
1- almost all the images reproduced on this site do not have a color profile (only some accounts can do it... I don't understand!)
I am convinced that your images saved on the server have been rewritten in sRGB.
2- when I saved the images on my Macbook Pro they were without a color profile
3- if I open the images in photoshop and assign the same profile they are identical, if I assign different profiles sRGB seems better.
4- You have to consider that Capture One displays on the monitor in 8 bit, but processes the display of a RAW at 12 or 14 or 16 bit
5- It makes no sense to talk about gamut or color depth for BW images, they have three identical channels, the differences between sRGB and AdobeRGB are only visible in RGB images on a professional monitor.

.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[jaapv]

Actually ACR (Photoshop workflow) by default opens the image in Greyscale, not RGB (although it is easily changed) If you do not actually need RGB, like for toning or some plug-ins etc., Greyscale is an excellent choice to edit in, only "dumbing down" when you need to at finally saving your image. The previews that you see when editing in Lightroom are created on the fly from Prophoto.

I suspect that the matter discussed here has to do with bit depth rather than  the RGB variant as a Monochrom file contains no colour data so there is no colour gamut. All three colour channels are identical, unless deliberately skewed during postprocessing (AKA toning) 

So my advice is: convert to any RGB but keep your bit depth at 16 as long as possible.

[jaapv]

34 minutes ago, ph0toni said:

(only some accounts can do it... I don't understand!)

In Photoshop one can convert to (and embed) a profile and bit depth at will. The site does not rewrite the profile, but not all browsers can handle for instance Adobe RGB and many computer screens will only render sRGB. 

[ph0toni]

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

.

.

.

.

@jaapv as you can see I posted an image in BW (one channel) and an AdobeRGB, if you save both photos they are rewritten without color profile and both with three RGB channels

Edited November 23, 2024 by ph0toni

[ramarren]

Hmm.

Adobe Lightroom Classic reads any file (raw, PNG, JPEG, TIFF, PSD) and promotes it internally (in memory and in the LR previews) to 16-bit per channel, ProPhoto RGB for editing. Doesn't matter whether the file is monochrome or not; 16-bit RGB and ProPhoto RGB cover the maximum image editing space possible with current imaging systems.

Obviously, with a monochrome source file, all three channels are identical. If you go into the Tone Curve tool, with a Leica M10 Monochrom raw file, the editing mode is set to grayscale (all three channels simultaneously) and the Monochrom's embedded tone curve is displayed by default. You can, if you so desire, click on  each of the three channels and adjust the color of the file, creating some fun special effects, and export that image file in any file type you want with any color profile you want, in either 8- or 16-bit depth (depending upon the limitations of the file type; JPEGs are always 8-bit). It's worth noting that bit-depth and color profile is applied ONLY on export ... you are ALWAYS editing in 16-bit per channel, ProPhoto RGB, and when you export a file, LR Classic does its best to make any combination of file format and color profile look as identical to what you see on the editing screen as possible. 

With that in mind, I did a test to see how close the results looked: I took one of my recent M10 Monochrom files with a broad lighting range from bright highlights very close to the white clipping point to deep, dark tones very close to the black point. I exported it as 16-bit TIFF, ProPhoto RGB and Adobe RGB as well as 8-bit JPEG (compression @80) Adobe RGB and sRGB. I examined the exported files, side by side, in Affinity Photo (so as to decouple as much as possible from any Adobe/LR-PS sneakiness) at 100% and 400%. 

The export files do show differences in the histogram curves ... however, the differences in the output images whatever differences might be there are, even at 400% display, so insignificant as to be considered non-existent. 

If you are interested to see the files, let me know and I'll figure a way to send you the .zip file containing all four of them. The image is this one, posted yesterday: 

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

The Wall Around The Olive Grove - Santa Clara 2024

Leica M10 Monochrom + Voigtländer Heliar Classic 50mm f/1.5 VM; ISO 160 @ f/2.8 @ 1/350 sec

G

[ph0toni]

@ramarren I meant that my images have lost the color profile
my monochrome BW is saved on the site as an sRGB
my AdobeRGB file is saved on the site as an sRGB
While your BW image, Photoshop recognizes it as grayscale without color profile
.
This image LINK , Photoshop recognizes Adobe RGB
while this image LINK , Photoshop does not recognize the color profile.
.
very strange

[ph0toni]

2 hours ago, ramarren said:

It's worth noting that bit-depth and color profile is applied ONLY on export ... you are ALWAYS editing in 16-bit per channel, ProPhoto RGB, and when you export a file, LR Classic does its best to make any combination of file format and color profile look as identical to what you see on the editing screen as possible.

Yes, of course
I have been using Photoshop (version 2.5) since 1993, in 2005-2007 I tried Aperture and Lightroom... then I deleted them.

As I wrote in the first post on this site LINK , I am also interested in ultraviolet photography (to emulate wet collodion) and infrared photography.
Camera Raw works well for IR ... but it doesn't work at all with UV, so for 6 years I have been using Capture One to develop my Raw, then I use PH for everything else.

C1 works like LR, but as I said it sees UV files much better and is very convenient with the tethering function.

[jaapv]

I need to correct myself: The newer editions of ACR will not open in Greyscale any more, it must be any of the 16bit RGBs; it needs to be converted in Photoshop itself to Working Gray (which is 16 bits) . But in that case you must upload to the forum as the smallest  .PNG  ( and reduce to 1920 PX) to stay under the sponsored 5 MB.

[Advertisement]

Hallo,

Look here: Leica M

[ph0toni]

I didn't talk about ACR, I said that the files on this site open in Photoshop in a strange, inconsistent way.
Ramarren's .jpg file opens in grayscale, mine doesn't.

I

[jaapv]

ACR is the raw converter for Photoshop...  I suspect that Ramarran is using an older version.

t has nothing to do with this site, but by the way the files are labelled and configured.