Skip navigation
Currently Being Moderated

How to combine different saturation sliders?

Aug 8, 2012 4:22 AM

I'm not sure I understand how LR4 combines values from the global saturation slider and local adjustments. Two questions in bold at the bottom.

I made the following experiments:

1) start with a color raw image, process version 2012, untouched.

2) bring down global saturation to -100

3) add a graduated filter with saturation +100 (I added, say, sharpness -100 to see precisely to which side the filter is being applied)

I would expect half of the picture to be colored and half b&w. Instead the filter is totally ineffective (i.e. it produces absolutely no effect, you can see this rotating it). So I conjecture that the total saturation is not "additive" (i.e. -100+100) but maybe "compound" (i.e. the product of 1+p/100). if this were true, then any slider equal to -100 would bring saturation to the minimum. let's see if this holds true:

4) set global saturation to -100. any value of the filter saturation is ineffective

5) set filter saturation to -100 and global saturation to 0. now (say) in my photo the sky is blue on the left and grey on the right.

6) slowly move global saturation from -50 to +100: the grey sky changes! from a... less saturated grey to a sort of... super-saturated grey (more blilliant and slightly yellow?).

so the conjecture is false and it looks like the filter is applied first, and the global value is applied last, IGNORING what the filter possibly did (so if the filter turned blue into grey, then "saturate" the grey).

A) Is this documented anywhere?

B) In particular, if I want to desaturate everything, except for a very small area, I have to paint the complement of the area with saturation -100 which feels quite unnatural (I'd like to do it painting something only on the small area). Is there any workaround?

Thanks,
MH

 
Replies
  • Currently Being Moderated
    Aug 8, 2012 7:21 AM   in reply to mycroft.holmes

    mycroft.holmes wrote:

     

    In particular, if I want to desaturate everything, except for a very small area, I have to paint the complement of the area with saturation -100 which feels quite unnatural (I'd like to do it painting something only on the small area). Is there any workaround?

    One very simplistic variation: paint an adjustment brush over the entire picture with the desired -100 saturation.

     

    Then switch the brush to Erase mode, and you are - in effect - painting saturation, with each new brush stroke.

     

    [That's if you are determined to have this effect in the first place... ]

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 8, 2012 7:58 AM   in reply to mycroft.holmes

    Apparently you didn't like or try the suggestion I gave in the other forum.  I'd tried it, and it seemed to do what you were looking for.

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 8, 2012 9:57 AM   in reply to mycroft.holmes

    Yes, I just went back and read your question again.  And I did misunderstand what you are trying to do.  It seems to me to get that effect is going to require layers and masks, something that isn't part of the Lightroom toolbox and workflow.  That is where having Photoshop or some other pixel editing program comes into play.  One exception , of course , is the procedure Richard outlined . 

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 8, 2012 10:29 AM   in reply to JimHess

    Just as a reminder, OnOne is now giving away Perfect Layers...

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 8, 2012 10:38 AM   in reply to Keith_Reeder

    Just in case anyone is not clear, Perfect Layers like all other such utilities, is an external editor requiring a TIFF file to be made. This has very different implications than the kind of fully nondestructive adjustment, whether global or local, that happens inside LR.

     

    And that's OK, I suppose - provided everybody goes in with eyes fully open. This kind of thing is NOT a "plugin" in the same sense that one talks of calling a plugin smoothly, midway, in Photoshop. With LR, it requires a clean break from your prior editing - which does not remain changeable.

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 8, 2012 11:51 AM   in reply to mycroft.holmes

    mycroft.holmes wrote:

     

    I don't know if saturation is additive between two different local adjustments.

    Let's say I add a graduated filter with saturation -100 everywhere and paint my area with a local adjustment +100... would it work?

    From my experience the local tools can effectively be used additively or subtractive, just don't try using them "by the numbers." In other words a graduated filter with -100 Saturation may not be completely cancelled by a spot filter with +100 Saturation value, and I would not suggest using them at their "maximum settings" for that purpose.

     

    Richard's suggestion sounds like a straightforward and quick way of achieving your objective in LR, and I doubt you would be able to do it any faster using Perfect Layers, PS or any other external editor. If you’re applying this effect (selective saturation) across multiple images you can speed up your work by syncing the first -100 Saturation Brush’d image across all and then apply the eraser to each separately.

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 8, 2012 11:19 PM   in reply to mycroft.holmes

    Local saturation works like global vibrance, except locally - it was a misnomer.

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 9, 2012 8:44 AM   in reply to mycroft.holmes

    I think the OP touched on the difference between global and local controls:

     

    It appears the Graduated Filter and Brush saturation control settings (and others?) are applied to the image prior to the global controls. That would make them non-linear, since even the Basic panel default settings ('0') apply toning to the image. This may be why local saturation appears to work more like global vibrancy, but don’t know.

     

    I checked the local control sliders using the Graduated filter and the only settings that appear to be linear (i.e. local = global) are the Highlights and Shadow sliders. You can use these two controls with opposite and equal settings to fully cancel their effect on the image, but not the others.

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 9, 2012 5:54 PM   in reply to trshaner

    We might use one adjustment for tone locally. Another to tweak hue. But those are separate matters, tackled independently. Either in the same pin or in different pins, provided they are combining, rather than fighting, with each other - and with the global adjustments - that is IMO much easier to manage and also to predict.

     

    Common example: using a sufficiently large and soft-feathered brush we can near-enough make a smooth gradient adjustment for a sky. Then (as we could not with the grad tool) we can erase from this same brushed mask the mountain, person or whatever else sticks up into that sky - achieve a single selection, IOW.

     

    This seems to me far better than using a simple grad to darken the whole upper part of the picture, then a brush which tries to restore the tone of all the bits which should ideally not have been darkened by the grad. Two different kinds of adjustment are never, realistically, going to cancel each other exactly; we are better off using another approach which will not create that particular problem.

     

    As I see it the OP wanted to use one adjustment to take out all the colour, and a different adjustment to put it back locally. This made difficulties, or opened questions of interaction, which using a single selective adjustment instead simply... sidesteps.

     

    Message was edited by: richardplondon

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 9, 2012 12:57 PM   in reply to trshaner

    trshaner wrote:

     

    I think the OP touched on the difference between global and local controls:

    It appears the Graduated Filter and Brush saturation control settings (and others?) are applied to the image prior to the global controls.

     

    My guess: Lightroom consolidates globals with locals then incorporates. Different algorithms for different sliders, but all have some kind of roll-off and/or limit...

     

    If you consider local saturation to be vibrance instead of saturation, that would explain it, right?

     

    To quote Eric Chan in paraphrased fashion:

     

    "Local saturation is more like vibrance than saturation".

     

    It's possible that it's 100% exactly like vibrance and it's just a simple misnomer, but I don't know enough to say. But in my experience, he's right: it's definitely more like vibrance than saturation.

     

    Rob

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 10, 2012 5:53 AM   in reply to Rob Cole

    Rob Cole wrote:

    My guess: Lightroom consolidates globals with locals then incorporates. Different algorithms for different sliders, but all have some kind of roll-off and/or limit...

    The "simplest implementation" would sum (i.e. consolidate) the local and global sliders before application, but as the OP has discovered this is not the case. From the quick test that I performed the only sliders that appear to work this way are the Highlights and Shadows.

     

    You can test this using a raw image file that has moderate to heavily clipped highlights. Create a virtual copy, apply a graduated filter with +2 Exposure to the whole image, and then set global exposure to -2. Now turn on the highlight clipping indicator and compare clipping between the copy and original, which should show a loss of dynamic range. Now try the opposite settings using -2 Exposure graduated filter and +2 global Exposure global. Both the images and histograms should be nearly identical, which indicates to me that the local controls are applied before the global controls.

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 10, 2012 2:59 PM   in reply to trshaner

    Woah! - applying locals before globals seems bass freakun' ackwards to me, but I confess I can't come up with an alternate explanation. I mean, there is no question that any consolidation that is done is far from "simplest implementation" (sum), yet consolidation does not explain things like:

     

    If you drop exposure far enough with the local that the blacks are clipped too much, there is *no way* to recover them using global exposure - they are lost forever. (in like fashion to your test with the highlights).

     

    Wow - I'm floored. I'm gonna have to re-think the whole business of balancing the globals with the locals. I'm sure this accounts for some exasperation in previous editing sessions, and/or suboptimal results.

     

    I'll go back and read the rest of this thread now - I had previously only skimmed enough to conclude that y'all didn't know what you were talking about .

     

    trshaner - thanks for clubbing me a bit more upon my thick arrogant skull...

     

    Eric Chan - How about a white paper or blog post about this?

     

    Rob

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 10, 2012 4:12 PM   in reply to Rob Cole

    It kind of surprised me as well and does seem backwards from how you would expect the local controls to be applied.

     

    The local controls are gonna bite you eventually if you're not aware of this behavior. Fortunately most people use them for purposes that shouldn't affect the dynamic range, such as applying +EV local exposure to underexposed areas, and -EV on overexposed areas.

     

    I am sure the LR team had a reason for implementing the local controls this way......or should I say "I hope they have a reason?"

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 10, 2012 4:42 PM   in reply to trshaner

    MH, if you want to desaturate an entire image and paint bnack in stauration with either the Adjust ment Brush or the Grad filter. First in the HSL Panel set all Saturation sliders to -100 and then use whichever of the two aforementioned tools to paint back in saturation/colour where you want with a setting of +100 Saturation.

    Is this what you want..........

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 11, 2012 3:54 AM   in reply to mycroft.holmes

    Perhaps some of our surprise as to the sequence in which global and local adjustments are happening, goes away once we describe the local tools in functional terms as an override, instead of as a dependent adjustment?

     

    If our conception of what is happening, is: the output from one operation passes on to a further operation which somehow magically retrieves colour out of grey, or detail out of clipping - then our explanations as to how such a thing can happen, are logically awkward ones.

     

    If on the other hand our conception is simpler: the global tools have set the processing such that there is grey or clipping, but the local adjustment has overridden the processing of those parts of the picture to some other outcome - there is then, surely, nothing surprising to explain away... and we have permission once more, to regard the actions of the tools as intuitive!

     

    If the local adjustment has overridden the saturation to -100, it makes no difference what the global settings are, we will not see any colour. This does not mean the local adjustment has happened FIRST and so denied the global adjustments any hue information to work with [we can see that this is not the case because HSL still discriminates on a hue basis, changing whitebalance alters relative tones, and so on]. On the contrary - the local adjustments have "won" by having the LAST WORD on saturation. The same is true when it is the  global adjustments which say the image is desaturated or clipped; the local adjustments have the authority to overrule that.

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 11, 2012 8:03 AM   in reply to richardplondon

    This is all speculative since none of us forum members responding are Adobe employees. What is certain is that the local and global controls are NOT simply summed (i.e. consolidated), which has the potential to reduce dynamic range in the processed image. This also makes using the local controls more difficult and "less predictable," as the OP originally outlined. Regardless of the actual implementation, the local tools work well when used for the intended purpose of making slight to moderate corrections to selective areas of the image.

     

    The OP also pointed out that the local control settings are additive (i.e. summed) when using multiple graduated filters and/or brush adjustments (i.e. -2EV Brush cancels a +2EV Graduated Filter). So we do have a way of applying opposing corrections to the image that are very predictable and linear.

     

    Special thanks to the OP mycroft.holmes for starting an interesting discussion and providing good feedback.

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 11, 2012 3:06 PM   in reply to trshaner

    trshaner wrote:

     

    The OP also pointed out that the local control settings are additive (i.e. summed) when using multiple graduated filters and/or brush adjustments (i.e. -2EV Brush cancels a +2EV Graduated Filter). So we do have a way of applying opposing corrections to the image that are very predictable and linear.

     

    The locals may always be added to each other before having their effect, but it's worth noting, when combined with globals:

     

    The combined effect of some becomes very much non-linear at higher values, and some sliders take on a whole different meaning in different ranges (e.g. local sharpness has 3 distinct ranges that do different things). I haven't scrutinized all yet, just a heads up...

     

    Rob

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 11, 2012 7:14 PM   in reply to Rob Cole

    Thanks Rob, I was afraid this might be the case. If a local setting pushes any highlight or shadow pixels into clipping the control may appear non-linear or have a different effect. This is because the global controls may not be able to recover the clipped pixels. This is true for even the local sharpness slider, especially for dark images against a very bright background and vice-versa.

     

    Please keep us posted with your findings.

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 12, 2012 4:06 PM   in reply to trshaner

    Yep.

     

    In addition to the local vs. global ordering issue of clipped pixels, most algorithms have limits, and what happens when local adjustments are pushing those limits depends on the adjustment.

     

    For example:

     

    local highlights:

     

    (assuming global highlights = 0)

     

    -50 & -50: effect same as -100

     

    but

     

    -50 & -50 & -50: effect is stronger than -100, but not as strong as -150.

     

    why?

     

    because you're reaching the end of what -highlights can do, and instead of just stopping working completely (and suddenly), the effects of additional locals begins to roll off (taper off). Similarly for shadows. The threshold for highlights/shadows is 100 - anything greater than that and you get diminishing returns.

     

    FWIW - this is why the globals for those are limited to 100 - so there is enough headroom for the locals to still have some effect, even if not as much as you might expect.

     

    Also big issues due to ordering when negative local values have different meaning than positive values.

     

    One of the more critical, especially if converting 2003 photos with local negative sharpening (radical change to negative (local) sharpening introduced in PV2010):

    total local sharpening amount >0 adds to global (dunno what happens toward max limit).

    -1 to -50 (partial to total sharpening is masked)

    -50 to -100 (blur)

     

    Another biggie - local clarity. Since negative local clarity is being applied before global positive clarity (as opposed to being consolidated), and negative clarity works radically different than "less positive" clarity, there is a huge difference between setting global clarity high and applying negative clarity (locally) to overly clarified regions, versus applying global clarity lower, and supplementing under clarified regions using local clarity with positive values. This is one that has bitten me a bunch, but thanks to the issue raised by this thread, it won't any more...

     

    I really wish Adobe would document stuff like this so we would know about it from the begininning instead of discovering it later...

     

    Rob

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 13, 2012 4:29 AM   in reply to Rob Cole

    Whew, that's enough to choke a horse. Anyone care to write the book, 'The Complete Idiot's Guide To Using Lightroom Local Controls?'

     

    With all of these issues it begs the question as to why the Adobe LR Team gave local controls precedence over the globals. I hate to sound negative as I've been a happy LR user since version 1.0, but these types of issues coupled with the LR4 PV2012 learning curve are getting a bit old. At least now we better understand the limitations of the local controls. Thanks again to Rob and everyone else for your due diligence in providing assistance with work arounds.

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 13, 2012 6:47 PM   in reply to mycroft.holmes

    mycroft.holmes wrote:

     

    the issues that you describe are somehow not surprising

     

    Perhaps not so surprising once you know, but I bet most people would be surprised by some things. For example, I would bet most people who set global clarity at 25 then paint with local clarity = -10 are expecting a net clarity of 15, as opposed to negatively clarifying the region by 10, then positively clarifying the negatively clarified region by 25. It's natural, in Lightroom, to think that locals would be consolidated with globals, just like locals are consolidated with other locals. I mean, it's obvious in a program with layers or stages, like Photosop or NX2, that the output of one op is the input to the next, but that's not how Lightroom seems to work. Although locals are applied before globals, in most cases the net result is cumulative / associative, thus the surprise when it does not work that way.

     

    There are good things about how it's been done:

     

    For example, the way negative sharpening works is really cool once you know it:

     

    Negative local sharpening essentially reduces sharpening by a percentage, which is better than a negative offset, since you don't have to redo all your locals whenever you change the global, if you want the net to be zero, for example.

     

    I just wish it were clearly documented, and clearly indicated in the UI (e.g. different coloration for the different ranges). Here's the feature request for it:

     

    http://feedback.photoshop.com/photoshop_family/topics/lightroom_acr_in dicate_the_different_behaviors_of_local_sharpening

     

    And I wish clarity worked the same way:

     

    negative local clarity reduces global clarity by a percent, then if pushed further actually negatively clarifies.

     

    In fact, here is the feature request for it, please go vote for it if you approve:

     

    http://feedback.photoshop.com/photoshop_family/topics/lightroom_acr_ha ve_negative_local_clarity_work_like_negative_local_sharpening

     

    Rob

     
    |
    Mark as:

More Like This

  • Retrieving data ...

Bookmarked By (0)

Answers + Points = Status

  • 10 points awarded for Correct Answers
  • 5 points awarded for Helpful Answers
  • 10,000+ points
  • 1,001-10,000 points
  • 501-1,000 points
  • 5-500 points