Contrast adjustment in RGB mode saturation increase

[larkis]

I would like to find some information that gives a non math formula explanation of why a contrast adjustment in RGB space also increases the image saturation vs doing the adjustment in LAB space or using a layer adjustment set to luminosity. Is there a website that helps visualize some of this ? I'm mostly curious of my own education.

[Mark D Segal]

There is no necessary mathematical reason for the Tone Curve or other contrast tool to change saturation with a change in contrast when adjusting photos in an RGB colour space. The Photoshop developers programmed it this way intentionally because the tonality looks more natural and pleasing adjusted in that manner. Try increasing contrast without increasing saturation, as you can do with the Luminosity blending mode, and you will see immediately the rationale. But there are ways to moderate that behaviour using layers and blending modes to suit your taste.

[larkis]

There is no necessary mathematical reason for the Tone Curve or other contrast tool to change saturation with a change in contrast when adjusting photos in an RGB colour space. The Photoshop developers programmed it this way intentionally because the tonality looks more natural and pleasing adjusted in that manner. Try increasing contrast without increasing saturation, as you can do with the Luminosity blending mode, and you will see immediately the rationale. But there are ways to moderate that behaviour using layers and blending modes to suit your taste.

Interesting, so how did the photoshop developers accomplish the saturation boost ? I see other tools behave in a similar way. Interesting that this is not a byproduct of working in the RGB space, thank you for that tidbit of information.

[Mark D Segal]

Interesting, so how did the photoshop developers accomplish the saturation boost ?

It's all in the higher mathematics of digital imaging and programming - not my field.

[digitaldog]

From an older post by Eric Chan of Adobe (san's math):


Eric Chan wrote:


Thomas and I separately explored the side effects and implications of various tone curve implementations (he in the early days of Camera Raw, I when working on color profiles in 2008 and again in 2010).  I learned a few things along the way.


One is that contrast and saturation are often correlated in the real world, which is why an increase in contrast often (though not always) works well with an increase in saturation.  Clarity is an example of a contrast control that doesn’t obey this:  it allows you to increase/decrease (local) contrast while preserving saturation.  If you’ve ever tried a strong negative Clarity adjustment on a colorful image, the result looks a bit odd.  Similarly, if you’ve taken a rather pale image and added strong positive Clarity but without punching up the Saturation/Vibrance controls, the result also tends to be artificial.  So we’ve not been entirely consistent in how we’ve dealt with saturation side effects in our contrast controls, but I’d probably give the nod to Thomas’s design (saturation side effect).


A second lesson is that the choice of color space makes a big difference to the result.  As Jeff, Andrew, and others have widely documented, we use ProPhoto as our choice of RGB primaries, which is a good thing in many cases.  For saturation side effects in curves, though, there are some issues.  In particular, due to the position of the blue ProPhoto primary, this means our current tone curve tends to have much stronger blue saturation side effects than in other hues.  In particular, if you have a typical S curve or just increase the Contrast control, darker blue tones (such as first column, third row of a standard ColorChecker, or deep blues in water reflections) tend to get overly saturated, and lighter blue tones (such as skies) tend to become overly desaturated.  There are ways to get the saturation side effects to be more perceptually uniform, and I’m investigating those for the future (would likely require a process version bump, though).


A third lesson is that our current color control set within the ACR/LR UI isn’t really good (yet) for doing 3D color edits.  This is because one cannot fully control how hue and saturation are affected as a function of lightness (or brightness, or luminance, or whatever term you want to use).  For example, when you use HSL controls to change the Orange hues, that changes them for all orange hues, light and dark.  You can’t change them separately for light vs dark hues.  It is possible, of course, to use per-channel RGB curves to bring back Photoshop-style hue twists in tone curves, but even though you don’t have full control over how those twists behave.


And finally, yes, part of film’s charm (or curse, depending on your point of view!) is the fact that it’s rather non-linear in behavior, so you do get various twists or distortions in hue & saturation as a function of exposure. 

 Eric

[Redcrown]

I remember reading a story once about the very early development of Photoshop 1.0. Back when not much was known about digital processing and everyone was learning "on the go". The story said when the "Levels" logic was first written it was very simple math. But Knoll and others argued, some saying colors just didn't "look right".

That first "Levels" was what we call Luminosity mode today. Code was then added to adjust saturation along with brightness. Some felt that didn't look right either. I can't remember which side of the argument Knoll was on. The argument was settled by leaving both methods available, and that's how Luminosity mode was invented.

So a simple answer to why saturation changes when tones change is that otherwise, it just does not look right.

[digitaldog]

Brightness is a perceptual phenomena. Luminance (Luminosity) is a measure of the total radiant energy from a body. It has nothing to do with what a human perceives but rather describes the total radiant energy, such as watts/second of a source (the surface of a radiating object like a display). In Photoshop, the layer mode called luminosity is not what's occurring (I was told its something like the "Luma" which is an old TV RGB transform). If the luminance of a viewed light source is increased 10 times, viewers do not judge that the brightness has increased 10 times.


Lightness is a perceptually scaled component of color, the axis seen in Lab (Lstar) from light to dark. It IS the L in HSL.

[Bart_van_der_Wolf]

So a simple answer to why saturation changes when tones change is that otherwise, it just does not look right.

Hi,

However, some argue that the way it's implemented (sometimes) does not look right (and it's destructive), and I'd agree. That's why I prefer the ability to choose the level of saturation change myself, like implemented in a number of Topaz Lab plugins that use "Intellicolor technology". Here's a comparison.

Cheers,
Bart

[Mark D Segal]

Bart - I have nothing against plug-ins per se, but just to remind that it is also possible to quite easily control setting contrast and saturation independently in Photoshop.

Andrew - thanks for pulling up that discussion from Eric - very insightful, and confirms that for the most part these are (sensible) programming choices rather than mathematical inevitability.

[Bart_van_der_Wolf]

Bart - I have nothing against plug-ins per se, but just to remind that it is also possible to quite easily control setting contrast and saturation independently in Photoshop.

Hi Mark,

I know, but unless one applies contrast adjustments on a Luminosity blending layer, Saturation is affected with no other control than a second correction in reducing the saturation is some parts of the image and increasing it in others. The artifacts will by then be amplified and baked into the composited image. I think that Eric Chan's initial preference, before he gave in, is best from a solid processing point of view.

Cheers,
Bart

[Mark D Segal]

Hi Mark,

I know, but unless one applies contrast adjustments on a Luminosity blending layer, ...........

Cheers,
Bart

OK, but there is nothing preventing one from doing this. The Plug-in probably just makes it more convenient.

[Mark D Segal]

...... In Photoshop, the layer mode called luminosity is not what's occurring (I was told its something like the "Luma" which is an old TV RGB transform). ......

The definition of what the Luminosity blend mode does, from Fraser/Blatner in "Real World Photoshop CS2", page 357: "Luminosity is the inveree of Color. It creates a result color with the hue and saturation of the underlying color and the luminosity of the overlying color".

[Bart_van_der_Wolf]

OK, but there is nothing preventing one from doing this. The Plug-in probably just makes it more convenient.

The artifacts are probably a result of the rudimentary processing. They won't go away.

The plugin video that I linked to, was for the purpose of showing the low quality that the default (contrast+saturation) controls produce, and that it is possible to avoid issues by using better technology. Thomas Knoll may be a smart guy and a good programmer, but he is not God. Some of his compromise solutions deliver sub-par quality, sad but a fact.

Cheers,
Bart

[Tim Lookingbill]

The artifacts are probably a result of the rudimentary processing. They won't go away.

The plugin video that I linked to, was for the purpose of showing the low quality that the default (contrast+saturation) controls produce, and that it is possible to avoid issues by using better technology. Thomas Knoll may be a smart guy and a good programmer, but he is not God. Some of his compromise solutions deliver sub-par quality, sad but a fact.

Cheers,
Bart

So you thought the first demo of that video of reduced saturation and charcoal black shadows adding contrast to the woman in a dark red dress looked better than the LR version? That's higher quality technology to you? I won't be using that technology anytime in the near future.

The real flaw of that video is that they use perfectly good looking, well finished images to start with that don't need editing. I didn't see any need to use a third party plug-ins on those images. As someone with a background in making pictures look good and communicate well I have to say that's a bad way to sell a piece of software.

[Bart_van_der_Wolf]

So you thought the first demo of that video of reduced saturation and charcoal black shadows adding contrast to the woman in a dark red dress looked better than the LR version? That's higher quality technology to you? I won't be using that technology anytime in the near future.

Hi Tim,

Of the two, the second example does look more natural, although personally I'd add some saturation at my discretion and control. Given that I would not first introduce artifacts before dialing down, if at all possible, that is a vastly more robust mode of operation. I'm a bit surprised that you apparently cannot distinguish between a demonstration of differences and an example of good post-processing. Nichole Paschal (the presenter) also said that the amount of the adjustments was too much.

One could return the question, do you think the first example looked better in Lightroom? But I won't ask such a silly question.

The real flaw of that video is that they use perfectly good looking, well finished images to start with that don't need editing. I didn't see any need to use a third party plug-ins on those images. As someone with a background in making pictures look good and communicate well I have to say that's a bad way to sell a piece of software.

Again, it's not a tutorial about how to process an image, but a technology demonstration, which is usually made clear by exaggerating the settings and see how well it stands the abuse before producing artifacts.

If you really have a background in making pictures look good, then you should know that RGB brightness/contrast adjustments usually also change saturation (because chromaticity and luminosity are not separated), and that it doesn't look good, requiring additional adjustments. In that case, why first change the saturation by the wrong amount, having to correct it in a second adjustment. And you should then also know that first over-saturating makes it perceptually more difficult to then create the 'correct' setting than starting from a more neutral setting, even if the artifacts can be avoided.

Cheers,
Bart

[Mark D Segal]

If you really have a background in making pictures look good, then you should know that RGB brightness/contrast adjustments usually also change saturation (because chromaticity and luminosity are not separated), and that it doesn't look good, requiring additional adjustments. In that case, why first change the saturation by the wrong amount, having to correct it in a second adjustment. And you should then also know that first over-saturating makes it perceptually more difficult to then create the 'correct' setting than starting from a more neutral setting, even if the artifacts can be avoided.

Cheers,
Bart

Bart, at the risk of being repetitive, the reason why Contrast and Saturation move "in-sync" is not because they need to, but because the team of programmers, QC people and decision-makers in charge of what actually comes to users decided on the algorithms that should be satisfactory to the vast majority of their clientele. Which means that a high percentage of photographs processed in this manner are acceptable to a high percentage of the users. If they were not, and if there were a ground-swell of complaints coming to Adobe about excessive saturation resulting from increased contrast, the algorithms would have been tweaked. They aren't unresponsive dummies. If data were available, I would be surprised if more than a real minority percentage of the total number of photographs run through PS and LR in the world had their saturation dialed-back as a result of an increase in contrast. So let's keep things in perspective.

My own personal taste for LR would have been to start with a linear rendition and dial-in my own contrast and saturation, but the company has reasons, looking at the community as a whole, to do it otherwise. And really, at least at the LR or ACR stage of editing, no harm one way or another, because whether the controls are dialed up or dialed down, and no matter how many times you dial this way or that, it does NOTHING to the technical integrity of the image data because it is all metadata and any conversions include only the last instruction retained for any one of those controls. And if you don't like the rendered result, you still have the non-destructive aspect of going back to the raw file in whatever state you processed it, especially in LR where all the editing steps are retained by default unless eliminated in one way or another.

I'm not saying all this is much ado about nothing, but it may be fair to characterize the issue as "some ado about not very much".  :-)

Cheers

[Schewe]

Thomas Knoll may be a smart guy and a good programmer, but he is not God.

Not god but he is the father of this industry so a tiny bit of respect might be due.

If you don't like the way ACR/LR work, you are always welcome to use that other imaging app he created called Photoshop where's it's very, VERY easy to adjust contrast without adjusting saturation. Oh, wait, Nichole Paschal knows how to work in Photoshop too! Maybe she could create a contrast adjustment level and set it to luminance blending. Bet she could make IntelliColor™ look like Photoshop :~)

Personally, I thought her demo sucked because based on the video I saw, the LR image looked much better–although I doubt that the image needed just "a bit of adjustment" to +46 Contrast...(just a bit, really?) Oh, and I loved the way IntelliColor™ made the maroon dress look black.

Sorry nothing about that video says Thomas is wrong.

[Tim Lookingbill]

...If data were available, I would be surprised if more than a real minority percentage of the total number of photographs run through PS and LR in the world had their saturation dialed-back as a result of an increase in contrast. So let's keep things in perspective.

ACR/LR's global saturation slider is definitely not designed and programmed by dummies. I often move that slider to the left a bit just to override the contrast induced saturation boost. Luminance and most of the color hues are retained unlike the Vibrance slider.

Wish HSL panel behaved like the saturation slider when I use its sliders to reduce saturation on overly rich blue skies and shaded whites and grays.

[digitaldog]

Sorry nothing about that video says Thomas is wrong.

Or that a lot in the video is right or necessary/useful. I admit I watched half and gave up. After the bit about the display(?) of a super, super saturated tweak presumably causing banding in both awful appearing images, more in PS I knew I'd seen enough.

[Tim Lookingbill]

Just a note about that video I find a bit telling. It was posted to YouTube in 2012 and yet there are no comments after 4 years. That's very unusual for an image editing software that claims to have "advanced technology".

I'ld expect the typical Twitter and Facebook styled insipid one liners either praising or condemning it, but to not have one comment makes it appear not too many are interested. I mean there's more interest in $3 bottle of Merlot sold at Walmart on YouTube. "Oak Leaf" I highly recommend BTW.

Maybe the folks that posted the video should get some marketing bots to write some comments to liven up the place.