Contrast adjustment in RGB mode saturation increase

[Mark D Segal]

As it relates to Thomas Knoll and the ACR/LR Develop module, you would be incorrect. Thomas pretty much has free reign to do pretty much anything the ACR team really wants to do. How do you explain HDR and Pano going into ACR when it's also already in Photoshop? Why?. Because Thomas wanted it there.

No, your argument is wrong...if Thomas wanted to change the Contrast algorithm, he would do so. The fact he hasn't means the argument for changing it has failed to get traction to make it worth his (and the ACR team's) efforts.

It would require an all new Process Version and I think it's simply not worth the effort.

And if I recall correctly from the 2007 debate Thomas also dismissed the notion that there's anything more complicated about programming the tone curve with versus without some saturation co-behaviour, so this probably has nothing to do with technical problems needing a heap of budget and everything to do with what you say here.

[Mark D Segal]

I find backing off the Vibrance slider lightens blues, violets and greens in flowers. The Saturation slider show far less effect to a degree over a wider range of colors.

The effect of Vibrance adjustments isn't hard-wired into any particular colour group (but is configured to not exaggerate skin tones). It's impact is designed to vary depending on how much saturation of a colour group the adjustment it starts out with.

[Tim Lookingbill]

The effect of Vibrance adjustments isn't hard-wired into any particular colour group (but is configured to not exaggerate skin tones). It's impact is designed to vary depending on how much saturation of a colour group the adjustment it starts out with.

Mark, that isn't information that can be verified nor is it functionally useful. I don't know where you get these certainties that imply you know how a slider is engineered to deal with color since you've admitted you're not a software programmer. No where in your statement did you use one word that visually describes what you think is going on under the hood. I did. So I really don't understand what your point is in telling me how Vibrance is hardwired. I know what I see.


I only stated what I see comparing reducing Vibrance vs Saturation slider on a flower image that has the colors I mentioned. Increasing Vibrance does make mid tone to deep blues get darker and richer and does hold back on oranges as in skin tones. But the more I increase Vibrance it eventually affects warm colors as well.


Saturation slider is more balanced and even in applying saturation across a wider range of colors. I don't know nor has anyone proven how these two sliders are hard-wired.

[digitaldog]

Mark, that isn't information that can be verified nor is it functionally useful. I don't know where you get these certainties that imply you know how a slider is engineered to deal with color since you've admitted you're not a software programmer.

What he wrote however is correct in that there is skin tone protection if you will with Vibrance and the way it works is, less saturated colors are affected more than higher saturated colors. It's not a linear (if you will) boost like Saturation.

[Mark D Segal]

Mark, that isn't information that can be verified nor is it functionally useful. I don't know where you get these certainties that imply you know how a slider is engineered to deal with color since you've admitted you're not a software programmer. No where in your statement did you use one word that visually describes what you think is going on under the hood. I did. So I really don't understand what your point is in telling me how Vibrance is hardwired. I know what I see.


I only stated what I see comparing reducing Vibrance vs Saturation slider on a flower image that has the colors I mentioned. Increasing Vibrance does make mid tone to deep blues get darker and richer and does hold back on oranges as in skin tones. But the more I increase Vibrance it eventually affects warm colors as well.


Saturation slider is more balanced and even in applying saturation across a wider range of colors. I don't know nor has anyone proven how these two sliders are hard-wired.

I only report things here that I've either experienced myself or been told on competent, first hand authority. Take it or leave it. I have no interest in arguing with you.

[Bart_van_der_Wolf]

As it relates to Thomas Knoll and the ACR/LR Develop module, you would be incorrect. Thomas pretty much has free reign to do pretty much anything the ACR team really wants to do. How do you explain HDR and Pano going into ACR when it's also already in Photoshop? Why?. Because Thomas wanted it there.

Hi Jeff,

It's good to know that Thomas is pretty much in charge, but I doubt that he is not bounded by budgets and numbers of people that can be assigned to his ACR/LR team. He more than likely has to operate within such limits, and can only determine what activities he prioritizes within those bounds.

No, your argument is wrong...if Thomas wanted to change the Contrast algorithm, he would do so. The fact he hasn't means the argument for changing it has failed to get traction to make it worth his (and the ACR team's) efforts.

So it might just be that he has many things on his list that he finds more important than improving the Contrast control. Or does he still stick to his design that in the film era was probably reasonable, but by now could do with an update but he doesn't because it apparently would require a significant effort in the shape of a new Process version, Eric Chan also said as much.

It would require an all new Process Version and I think it's simply not worth the effort.

Not worth the effort. Is that in the light of other improvements that have more impact for you or Adobe, or because you think it's not an issue anyway. Whichever it is, that's fine with me, I'm just wondering. No problem with solidarity to Thomas either, but really wouldn't you prefer to be able and avoid the issues that come with the current implementation of the Contrast control? Or to phrase it differently, would you avoid using Contrast if it gets fixed and doesn't offer a "Use Legacy" toggle?

Anyway, I've attached a ZIP archive with 3 TIFF files. They are gradient versions of 3 of the ColorChecker patches (Light Skin, Sky Blue, and Foliage), each basically with constant Hue and Saturation, but with linear Brightness going from 0 to 100%. Applying a Contrast curve to them will reveal if and how much the original constant Saturation is altered when sampled with the colorpicker in HSB mode.

Cheers,
Bart

[fdisilvestro]

This thread reminds me of Plato's Allegory of the cave

[Mark D Segal]

This thread reminds me of Plato's Allegory of the cave

I'm not so sure, because the people who you probably think are in a cave really aren't. They've known and thought about this issue since a long time back. I think there is just a simple disagreement about two questions: (1) is the issue important, and (2) would a substantial lessening of the linkage between contrast and saturation create a higher quality and more efficient starting point for image editing? There is no consensus here on either question and I doubt that one will emerge from this discussion. So the bottom lines are simply: (a) we wait for Adobe to decide whether they will amend this linkage going forward (they know the issue too and they know how to program either implementation) and (b) in the meanwhile, those who prefer other kinds of renderings should use the applications that provide them.

[Schewe]

It's good to know that Thomas is is pretty much in charge, but I doubt that he is not bounded by budgets and numbers of people that can be assigned to his ACR/LR team. He more than likely has to operate within such limits, and can only determine what activities he prioritizes within those bounds.

Yeah, ya know, I'm not gonna get into the specifics of budgets and personnel. But I'm pretty sure if Thomas wants to do something, Thomas can do it. Adobe kinda defers to the guy that wrote Photoshop ya know? The fact he's not changed Contrast kinda leads me to believe he doesn't think it's needed at this point.

[Tim Lookingbill]

What he wrote however is correct in that there is skin tone protection if you will with Vibrance and the way it works is, less saturated colors are affected more than higher saturated colors. It's not a linear (if you will) boost like Saturation.

That's a much better and simpler to understand explanation, Andrew, thanks.

So the Saturation slider acts on color (more) linearly vs the more non-linear with Vibrance. I already knew about Vibrance's skin protection years back when it was first discussed. However, from editing over 1000 Raw files over the years it's clear to me different images add and subtract from the intensity of non-linearity depending on color content variances. On some images the Saturation slider will reduce luminance slightly in highlights of fully saturated pinkish orange sunset clouds but I'm never sure if this luminance hit caused by the gamut limit of my sRGB-ish display.

[Lundberg02]

I am a simple man, as OReilly is fond of saying. What IS saturation? Never found a good explanation.

[Bart_van_der_Wolf]

I am a simple man, as OReilly is fond of saying. What IS saturation? Never found a good explanation.

Hi,

It depends on context.

In this thread's context, it's known as Colorfulness or purity of color, see here for a more precise description ("Saturation is the colorfulness of a color relative to its own brightness").

So when the relative ratio between the dominant colorchannels changes, i.e. dominant R, G, or, B increases or the others decrease, Saturation changes. Contrast on the other hand is when a given color reflects or emits more of the same color, e.g. when shaded from illumination or more illumination is added.

Then there is also the psychophysical aspect, where human vision play a lot of tricks on what we see, but that is beyond the scope of this thread, but it does play a role. For handling that, Color Appearance Models (e.g. CIECAM02) are designed, and they too (have to) treat Chroma and Luminance separately.

Cheers,
Bart

P.S. And for those who mistakenly think that a better tamed implementation of a Contrast control would lead to unsaturated and dull colors, I've attached a reprocessed (with C1 Pro version 9) version of an older image of mine.

[digitaldog]

I am a simple man, as OReilly is fond of saying. What IS saturation? Never found a good explanation.

Colorfulness:


https://en.wikipedia.org/wiki/Colorfulness



http://rit-mcsl.org/fairchild/PDFs/AppearanceLec.pdf

[Mark D Segal]

Thanks Andrew - the PDF is very good.

[Lundberg02]

So saturation is like porn. We know it when we see it, but we can't explain it.

[digitaldog]

So saturation is like porn. We know it when we see it, but we can't explain it.

I'd be happy to explain porn to you 

[Lundberg02]

Ten minutes of sophistry and hand waving, as W L Root used to say.
So if contrast also adjusts saturation, then some sort of reciprocity theorem tells me adjusting saturation also adjusts contrast.

[Tim Lookingbill]

P.S. And for those who mistakenly think that a better tamed implementation of a Contrast control would lead to unsaturated and dull colors, I've attached a reprocessed (with C1 Pro version 9) version of an older image of mine.

That's the first I've seen of that image so I don't know what to compare it with to support your point. I don't see dull colors in that (C1 Pro version9) Beach Slide image, but then I don't know what it looked like before you added contrast that would affect saturation more or less to a degree.

The issue with adding contrast is that it's subjective on how dark the user wants the shadows. So that slide image could start out sort of unnaturally flat looking more or less and then the user decides the shadows need to look overly dark thus increasing saturation.

We don't know how that image started out so it would be more informative to show a progression of individual edits representing increased contrast to get an idea how well C1 Pro's controls saturation applying more contrast.

Nice looking image BTW, Bart.

[Denis de Gannes]

Ten minutes of sophistry and hand waving, as W L Root used to say.
So if contrast also adjusts saturation, then some sort of reciprocity theorem tells me adjusting saturation also adjusts contrast.

I guess so and dependent on which software application you are using to render raw files, Your camera manufacturer's software, ACR / Lightroom, DxO optics pro, Capture One Pro, and any other raw processing application the resulting tone and colour will be different.

[Bart_van_der_Wolf]

That's the first I've seen of that image so I don't know what to compare it with to support your point. I don't see dull colors in that (C1 Pro version9) Beach Slide image, but then I don't know what it looked like before you added contrast that would affect saturation more or less to a degree.

Hi Tim,

The main purpose of adding that image was to caution some people against getting the wrong idea about suppressing the Saturation boost. It does not make images look dull. And over-saturated Contrast enhanced images also do not have to look over-saturated, because there are tools to somewhat correct that. However (!), since the Contrast induced Saturation changes are highly non-linear, it is almost impossible to fully correct it (should one wish to).

The issue with adding contrast is that it's subjective on how dark the user wants the shadows. So that slide image could start out sort of unnaturally flat looking more or less and then the user decides the shadows need to look overly dark thus increasing saturation.

Yes, the amount of contrast we bestow on an image is rather subjective, and also depends on the scene contrast, and whether we used proper technique to capture the image to begin with. People e.g. tend to underestimate the benefits of using a properly dimensioned lens hood, it improves overall Contrast and Saturation. A foggy scene may not improve if we attempt to fill the histogram, and adding non-linear Contrast will also affect the highlight roll-off, which is why I already start with a linear tone curve (to avoid adding a Contrast roll-off to the already added roll-off of a more film curve response).

We don't know how that image started out so it would be more informative to show a progression of individual edits representing increased contrast to get an idea how well C1 Pro's controls saturation applying more contrast.

Nice looking image BTW, Bart.

Thanks. A problem with showing a progression of editing steps, besides the subjective choices (some of which are subtle), is that it does not convey the actual user experience of not having to second guess if the color we currently see is somewhat resembling the captured color (after the camera profile added its flavor of rendering).

Also, not everybody has 'perfect' color vision or memory, and there are psycho-physical factors involved that I mentioned earlier (automatic adaptation to the current image colors, even if they are wrong, e.g. White-Balance). So it's mentally easier to start with 'correct' Saturation and add to flavor, than starting with the 'wrong' Saturation and remove to flavor (which is almost impossible to do perfectly due to non-linear Saturation changes). Also every time Contrast is changed with the traditional approach, Saturation should be checked again. And many people settle for good enough is good enough, and what's good enough for one, isn't necessarily the same for another.

I currently can open, but cannot edit, some(!) TIFFs in Capture One version 9 (despite enabling that in the preferences), so I cannot show a side by side comparison (TIFF input should avoid Raw converter / profile differences) with e.g. ACR/LR. I'll have to figure that out first. Not that I normally use Capture One for my postprocessing, but it can occasionally be useful.

Cheers,
Bart