Skip navigation
Currently Being Moderated

Brightening shadows darkens highlights

Mar 2, 2012 8:53 AM

Tags: #artefacts #pv2012

20120216_1311_-2.jpg

Download DNG.

 

Check out the attached file and try to play with the Shadows slider. To me, it behaves somewhat unintuitively: instead of just affecting shadows, increasing the shadows also darkens the snow and vice versa — decreasing the shadows value brightens the snow.

 

Is it a bug or expected behaviour?

 

The long(er) story

I got this photo of my daughter skiing... under some very flat light and was trying to get some contrast: overall and especially in the snow.

 

First, I went what seemed the intuitive route: boosted Contrast to the max, along with setting the white and black points, then brought down highlights and lifted shadows. (That is snapshot "1-Normal" in the original). It looked fine, but I thought I'd try to get more definition in the snow.

 

So, I went on and realized that I have to stretch the highlights with -Exposure and +Whites (snapshots 2 and 3 in the original). The problem is, that under these condition, the Shadows slider starts to not behave itself, as described above.

 

Also, check out the faint large black halos and small white halos around the objects (flags and my daughter).

 

Both of these problems get worse the further you push the Exposure and White sliders aside in their respective directions.

 
Replies
  • Currently Being Moderated

    Maybe Eric will stop by, but I don't think this is a bug.

     

    This is a really, really stretched condition, and I think the new sliders do start to have some relatively unintuitive side effects when they are this far off from nominal.

     

    Cute shot, and I think you did a pretty good job of processing it.

     
    |
    Mark as:
  • Currently Being Moderated

    Hi Dorin,

     

    I have downloaded your DNG. To notice the halos you mention I had to go for 2:1 magnification. At 1:1 I did not see them at first, now with the knowledge where to look I can detect them.

     

    The shadow slider seems to do an overall compression or expansion of the whole histogram, at least once -highlights are maximized and -exposure considerably. Not so much discernible on your snapshot 1-Normal.

    I knew that shadows affects all regions, not just its own, but was not aware of the "opposite or compressing direction" before you pointed it out.

    (My test samples have not been of such a nature yet to go for extreme settings.)

     

    I guess your best expert so far could be Rob Cole, at least he is the one who has shared in most thorough detail what his develop experiences are.

    He gruntled a lot at first about *unintuitivity*, but learned to like the results.

    I looked in his current thread http://forums.adobe.com/thread/968940?tstart=0, but there he does not (yet?) give details to the behaviour of the shadow-slider alone.

    He has observed new halos as well, hopefully Eric Chan might be looking at it.

    There may be more info in this thread http://forums.adobe.com/thread/956844?tstart=30, but it is so polluted with insults that Dave Merchant mercifully locked it.

     

    If the shadow slider would behave as you and I would *intuitively* expect, we might run into another problem: if the snow would also get brightened i.o. darkened by lifting shadows, there might not be enough headroom left to counteract with the other sliders (-whites -highlights) ?

    So I see some merit in what I can observe in your example.

    Bug or Feature?

     

    Cornelia

     
    |
    Mark as:
  • Currently Being Moderated
    Mar 2, 2012 7:01 PM   in reply to Cornelia-I

    This has to do with order of adjusting sliders.  The ranges and behavior of Whites/Blacks controls depend on how the earlier controls are set (one reason why working top-down is generally recommended).  If you set Whites/Blacks earlier, this can result in interaction problems if you then go back up to the earlier sliders (e.g., Shadows). 

     
    |
    Mark as:
  • Currently Being Moderated
    Mar 3, 2012 12:23 AM   in reply to MadManChan2000

    Thanks for the answer, Eric.

     

    I need a while to digest this.... as it seems to be a big digression from current LR develop principles.

     

    So far the order of the sliders has helped you make your decisions what to do next for the image, but you always arrived at the identical result for a given set of numericals.

     

    This is now really different or did I just not understand your answer?

     

    Dorin's use case seems to be a pretty frequent one: you edit, and consider your result as sort of ok, fix snapshot1. But then you think it might benefit from further tweaks, because your visions is "to get more detail out of the snow".

    Then you change the sliders you think might affect what you intend - in the order you think.

     

    Do you suggest that we should first reset the basic sliders and go anew for the different vision? Using history panel to remember the settings of snapshot1 ?

     
    |
    Mark as:
  • Currently Being Moderated
    Mar 3, 2012 1:29 AM   in reply to MadManChan2000

    Very interesting indeed.

     

    These halos are not the "shadow halos" I have refered to in other threads - I think we need a new name: how about "whites halos", since they seem to be tied to the +whites -exposure(?)

     

    Also, with this photo in PV2012, it seems nearly impossible to spread out those highlights using the normal method. You can see that Dorin had to do some intricate finessing at the top end of the tone curve to pull it off.

     

    Rob

     
    |
    Mark as:
  • Currently Being Moderated

    This photo will not exhibit the same symptoms if you enable lens corrections. That lens has really wicked vignetting...

    i.e. The funny shadow slider behavior goes away, and you can fan the highlights out using the normal method (+whites -exposure).

     

    Here's another version (neutral color profile and some color mods, same tone curve):

    - less contrasty and bright just seemed to fit the snowy conditions better to me... (plus you can see through her goggles better ).

     

    Link: http://www.robcole.com/_temp/DN_120216_104921_1311.xmp

     

    Instructions:

    --------------

    - Save xmp file as DNG sidecar, i.e. same base filename (Lightroom supports reading xmp for DNGs from sidecar, it just won't write xmp as sidecar).

    - Read metadata

    - Delete xmp file.

     

    Rob

     
    |
    Mark as:
  • Currently Being Moderated

    Oops - my bad: I must have reset it...

     

    But still, I was able to process it with far less of the funny behavior with the neutral profile - dunno what you think of it - eh? (link is above).

     

    ???

     

    But if ya like it more contrasty:

     

    * Crank contrast up to 100

    * Highlights to -100

    * Drop Vibrance to zero.

     

    Regardless, neutral profile seems to be better behaved...

     

    See related post: "How to get results in PV2012 that would otherwise be impossible".

     

    Rob

     
    |
    Mark as:
  • Currently Being Moderated
    Mar 3, 2012 6:31 AM   in reply to Cornelia-I

    Cornelia-I wrote:

     

    Thanks for the answer, Eric.

     

    I need a while to digest this.... as it seems to be a big digression from current LR develop principles.

     

    So far the order of the sliders has helped you make your decisions what to do next for the image, but you always arrived at the identical result for a given set of numericals.

     

    You still do.  The order only affects the human interaction, not the final result.

     
    |
    Mark as:
  • Currently Being Moderated
    Mar 3, 2012 7:00 AM   in reply to Lee Jay

    Yes, halos can still appear in cases of using strong settings of Highlights, Shadows, Clarity.  It is less common than before, and overall a quality improvement compared to before, but the edge-aware algorithms aren't perfect.

     

    And Lee Jay is correct:  the results depend only on the numbers themselves, not the order in which you set the numbers.  But the interaction of the controls may feel weird if you use the controls out of order.

     
    |
    Mark as:
  • Currently Being Moderated
    Mar 3, 2012 5:46 PM   in reply to MadManChan2000

    MadManChan2000 wrote:

     

    Yes, halos can still appear in cases of using strong settings of Highlights, Shadows, Clarity.  It is less common than before, and overall a quality improvement compared to before...

     

    You forgot +whites -exposure halos . And I agree - much improved and usually not much of a problem, if it all, except for the phenomenon that I sometimes refer to as shadow halos, whose symptoms are a broad very noticeable glow without particularly extreme settings. Perhaps I should be calling it shadow glow instead of shadow halos...

     

    Rob

     
    |
    Mark as:
  • Currently Being Moderated
    Mar 4, 2012 7:33 AM   in reply to Rob Cole

    No, you won't get these halos if just using Exposure and Whites.  The types of halos demonstrated here are due to the edge-adaptive algorithm used by the Highlights, Shadows, and Clarity sliders alone.

     
    |
    Mark as:
  • Currently Being Moderated
    Mar 4, 2012 4:13 PM   in reply to MadManChan2000

    Exacerbated by whites/exposure but not root cause - got it.

     
    |
    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