Skip navigation
Currently Being Moderated

What does having to make these same two hue/saturation adjustments mean?

Jan 29, 2013 2:05 AM

Tags: #lightroom #hue #yellow #saturation #yellow_hue #yellow_saturation

I keep having to make the same two yellow adjustments in Lightroom. No matter the subject matter, indoor or outdoor. And no matter the camera body. See the before and after screenshots:

 

Before: The puppy coats are too yellow (to me), even after eyedroppering the color card. The puppy whites look correct:

13.jpg

 

 

After: Yellow Hue -30 -- Yellow Saturation -20.   --The puppy coats now appear correct.to me (the puppy whites are still correct). You can also see what happens to the yellow on the color card --it looks less yellow!?  --I find myself doing this adjustment on at least 80% of all my photos. Probably the only reason I dont do it is when there is no yellow in the shot.

12.jpg

 

This is in Lightroom V4.3.  This was taken with a Nikon D4, two reflected strobes set up on either side, and with room lighting on, but none of that seems to matter.  I have several other cameras, and light source does not seem to be a factor --once I eyedropper the card everything looks good except the yellows.  The monitor/computer does not seem to matter either.  I see the same problem on both my desktop monitors and laptop, which are calibrated the same (to 6500k=white).  I have ruled out room lighting reflection because this occurs with outdoor shots, and in either case after setting white balance either with a card or an expodisc.

 

I don't know enough about this subject to say it might be my aging eyes, but I am running out of other things to consider.

 

That is why I posted this. I guess the first question is can anyone else see what I'm talking about?  If not, that would implicate my eyes.  Why do I have to see incorrect yellow on the the card in order to see correct (in this case) coat colors?    If anyone has any ideas, thanks in advance!

 
Replies 1 2 Previous Next
  • Currently Being Moderated
    Jan 29, 2013 3:20 AM   in reply to George in Seattle

    The yellow patch on the color checker looks good in the first shot, too muted in the second.

     

    I think this is something different. See the heavy blue clipping in the histogram? That's a very dark yellow somewhere, and I think that's the key.

     

    It's been my experience with all my Nikons over the years (maybe all DSLRs) that the blue channel is "weak" and goes to low clipping very easily. May have something to do with the Bayer pattern. The result is muddy, dark yellows that instantly look wrong (but it's hard to put your finger on it). One thing that shows this clearly is fresh foliage in spring, where the amount of blue light is very low.

     

    I haven't really found a good solution to this. I just try to give as much exposure as possible, and deal with it in post.

     

    But what I would do is experiment with different camera profiles. This is a case where you could also try to make your own profiles. You already have the Color Checker, so you just need to download the DNG Profile Editor.

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 29, 2013 3:31 AM   in reply to George in Seattle

    < I don't know enough about this subject to say it might be my aging eyes, but I am running out of other things to consider.

     

    As you age, your lenses do turn yellow, and by the time you are 50+ you are effectively looking through a yellow filter. I had my lenses replaced with plastic ones, and suddenly everything became more blue and less yellow.


    The gray uniforms the nurses were wearing before the op were blue after it!!

     

    Bob Frost

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 29, 2013 3:58 AM   in reply to bob frost

    Maybe so, and I'm no longer 20 myself...but I do recognize white when I see it.

     

    I think what George is seeing is real, and the blue clipping is the smoking gun.

     

    Based on what I said above, you'd think the solution was to increase blue level in the camera profile.  But once again it's not that simple. Blue is already at the correct level, it's just that it drops off the radar too quickly. That's what I mean by "weak" blue channel.

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 29, 2013 5:59 AM   in reply to twenty_one

    What camera profile are you using?  I suspect it's Adobe Standard, the default.  I recommend you change it to one of the Camera profiles - Camera Standard, Camera Neutral, or Camera Portrait.  I like Camera Neutral the best, with some increase in contrast. 

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 29, 2013 7:31 AM   in reply to b2martin_a

    I agree with twenty_one that the image appears underexposed. In both posted images with LR corrections applied the histogram appears shifted to the left about -.5 EV. Since the ColorChecker is in the picture the white patch should appear in the histogram very near 100% or fully to the right.

     

    George, what LR Basic panel settings are your using, and what if any Tone panel or Local Adjustments? I also assume you are using a ColorChecker Passport created camera profile. Did you create it from this picture?

     
    |
    Mark as:
  • Andrew Rodney
    1,387 posts
    Apr 16, 2009
    Currently Being Moderated
    Jan 29, 2013 9:45 AM   in reply to George in Seattle

    George in Seattle wrote:

     

    I keep having to make the same two yellow adjustments in Lightroom.

    So why not build a preset of that yellow adjustment, then update the defaults (in Develop, click on Alt/Option key toggles Reset to "Set Default..."). Every new imported image will now have that tweak to yellow set from now on. Or just use the single preset.

     

    The initial settings you get are just a starting point you can change for all new subsequent edits. If as you say, you always have to make the same adjustment, just build that into the defaults!

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 30, 2013 8:51 AM   in reply to George in Seattle

    Post some small sRGB profile JPEGs (1200 x 800) here with the LR adjustments you think look best. This is what your clients will see, correct?

     

    The yellow adjustment issue with your ColorChecker Passport profiles indicates a problem, since you don't see this with Adobe and Camera Standard profiles. Are you rechecking the White Balance using a ColorChecker image's grayscale patches when you switch to Adobe Standard and other profiles? I use a Develop preset for each of the LR provided camera profiles containing only White Balance and Calibration (i.e. camera profile). The white balance is slightly different from the settings required for the ColorChecker Passport profile, but the same for all of the LR provided profiles (Adobe Standard, Camera Standard, etc.).

     

    Example White Balance Canon 600D:

     

    Profile                                Temp Tint

    ColorChecker Passport  5550   +23

    Adobe Standard, etc.     5600   +10

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 31, 2013 4:53 AM   in reply to trshaner

    If you have to make the same adjustments for all images, I think this means your profile needs adjusting.  If I understand correct, you are using a ColorChecker Passport profile.  It's been a while since I used the Passport software to generate profiles, but I don't believe you have the option of adjusting individual colors.  You can use the Adobe DNG Profile Editor to make the adjustments in the profile and maybe this will resolve your problem for this and all other images. 

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 31, 2013 7:54 AM   in reply to George in Seattle

    Just a quick look at this now, have to run...

     

    I brought the one you said "kind of worked" into Photoshop and boosted blue a bit in the shadows (Curves and Channel Mixer). This one's also heavily clipped in low blue. As I said earlier, I suspect this is the base problem. Look any better?

     

    WVBC12-1346.jpg

    WVBC12-1346_blue.jpg

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 31, 2013 10:17 AM   in reply to twenty_one

    I downloaded the above posted image and it looks like a simple white balance adjustment corrects the problem. The other images posted do look better than this one.

     

    Double-click on the image to see full-size

    Geroge in Seattle Yellow issue_Adjusted.jpg

     

    How does the image on the right look to you George?

     

     

    Edit: I just compared the above to twenty-one's blue shadow corrected image, and his appears to be more accurate.

     

    I'd say he's identified the problem (i.e. underxeposure), but I'm not sure what the best solution is for your usage.

     

    Message was edited by: trshaner

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 31, 2013 11:35 AM   in reply to trshaner

    Well, it's not underexposure as such. It's actually a bit more complicated.

     

    I'll have to use Photoshop to show what's happening here. This is how the green channel looks:

    green.jpg

     

    Nothing unusual there. But take a look at the blue channel:

     

    blue.jpg

     

    The histogram confirms it:

     

    histogram.png

    Not only does the blue channe contain very little information, it also drops off to full clipping much sooner than the other channels. But just shifting the white balance towards blue doesn't help - then you get the opposite problem in the highlights. This is in fact the way it turns out when you eyedrop for correct white balance.

     

    Visually, this lack of blue information translates to a heavy, unpleasant yellow without "air". I recognized George's problem here because I have struggled with it myself.

     

    One thing that may help is to make a Tone Curve adjustment to the blue channel, something like this:

     

    tone curve.png

     

    Or perhaps use the Split Toning panel. I do both, depending on circumstances. But that won't safeguard against clipping - for that you need to make certain to give adequate exposure to the shot so that you have headroom in editing.

     

    Why this tends to happen to the blue channel and not the red (which has as many sites in the Bayer pattern), I just don't know.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 1, 2013 1:18 AM   in reply to George in Seattle

    Yes, that's better, although you're killing some of the orange, perhaps unnecessarily. But in the end you'll just have to try out different things and see what works best for your camera(s).

     

    But I insist that you need to keep a close eye on the blue channel. That's where the problem is.

     

    As for custom camera profiles, I haven't tried the x-rite software (never installed it) - only the DNG Profile Editor. But now that you mention it, the main reason I sort of gave that up in favor of the Adobe Standard profile, is that there was always at least one color "running off" and oversaturating to an unacceptable degree. Maybe 18 color patches simply aren't enough.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 1, 2013 3:04 AM   in reply to George in Seattle

    You're welcome

     

    I've been there, so I knew what you were talking about. But try out all the things you can think of (and report back if you stumble on something) (that goes for everyone reading this BTW), because I'm still not certain this is the ideal solution.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 1, 2013 5:01 AM   in reply to George in Seattle

    I use Camera Neutral with a tone curve modification and saturation at -5 as my default profile for my D700.  The tone curve I use is as follows:

    Input - 228, output - 233.  Input - 127, Output -127.  Input - 27, Output - 22.  It only has 3 points. 

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 1, 2013 11:39 AM   in reply to George in Seattle

    I should say that I tend to underexpose things slightly (-1/3) because of my fear of blown out white coats, and this Nikon seems to allow plenty of editing leeway in the darks --more than it does in the lights. Same goes for several other Nikon models I've had.

    Just out of curiosity: could you try (if you have time) to capture an  additional alternative frame where you deliberately overexpose, by an exposure compensation of +1/3 or 2/3 ? Does that really blow out your white dog coats? Irrecoverably?

    I find with the 3 Nikons we use that such "cautious ETTR" tends to be pretty well suited for PV2012's image-adapted controls, in most situations.

    I'd just be interested if you have this uneven channel clipping in the same degree.

     

    Cornelia

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 1, 2013 12:39 PM   in reply to Cornelia-I

    The free RawDigger application can be used to determine if you're clippiing the whites or blacks:

     

    http://www.rawdigger.com/

     

    You can't trust the LR histogram since it shows the image with the camera profile applied. Much of the white clipping you see in the LR histogram is fully recoverable with -Exposure and/or -Highlights, and Tone Curve.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 2, 2013 4:09 AM   in reply to George in Seattle

    George in Seattle wrote:

     

    Below is a zoom on a brighter part of white, and showing what I'm trying to avoid by underexposing: --the individual hairs are no longer discernable. I'm not even sure what this is because if I'm reading the histogram right, the coat is not yet blown out even though the exposure is high.  But if I had to guess, it is some sort of pixel blooming that is not recognized as clipping.  This does not look like much here, and I don't consider this amount to be a problem, but it becomes a problem in direct sunlight.  When viewed unzoomed, it makes the coat look coarser than it really is.

    31.jpg


    You still have a lot to go on here. The whites look a bit muddy to me.

     

    This is where a good monitor (correctly calibrated) really matters. I don't know what monitors you use, but if for example you take an ordinary office TN-monitor there is usually very little separation above 235. It all looks white. But with a good monitor you should be able to separate up to 252-254, if not all the way.

     

    Luminance also matters, so perhaps when you calibrate you should try a lower target value. The recommended target is around 120 cd/m², but that's for a "well-lit" environment. I have mine at 95.

     

    ---

     

    I also think it is necessary to emphasize here, so that this doesn't get sidetracked, that underexposure is not the cause of the "yellow problem". The peculiar blue response of the sensor is (and I say sensor, not profile, because it seems universal across profiles to me). But underexposure gives you a lot less data to work with in post, so that correction gets more difficult.

     

    "Expose to the right" is a well known principle in digital photography. Give as much exposure as possible without blowing out the highlights, never mind if it looks wrong on the camera LCD.

     

    Here's an in-depth explanation: http://www.luminous-landscape.com/tutorials/expose-right.shtml

    And this one by Bruce Fraser: http://wwwimages.adobe.com/www.adobe.com/content/dam/Adobe/en/products /photoshop/pdfs/linear_gamma.pdf

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 2, 2013 5:28 AM   in reply to George in Seattle

    No problem with those monitors. But 115 may be a little high for "very dim". It could also be a problem with the monitor profile itself, so recalibrate in any case.

     

    So, it seems we have identified two different issues here:

     

    1. The "dense" yellows caused by blue channel dropout from the camera sensor.

    2. Underexposure caused by incorrect white rendering/perception (leave that open for now). Again, the whites you claim to be on the verge of blowing out, look muted and muddy to me with much to go on. If you go for a "film"/retro look it'll work, but for sparkle you need to pull up those whites.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 2, 2013 8:57 AM   in reply to twenty_one

    Why is this not a camera profile problem since the blue clipping did not occur with Camera Neutral profile?

     

    I understand the Camera Neutral profile is a little flat, but this can be corrected with a tone curve. 

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 2, 2013 11:01 AM   in reply to b2martin_a

    b2martin_a wrote:

     

    Why is this not a camera profile problem

    Because the other profiles aren't any better. Here's another example from the "orchid project" I posted some examples from in another thread - and that shows the same problem with heavy yellows. First the reference shot on location:

     

    (again I need to remind everyone that inline images in the forum are not color managed. Click to show correctly).

     

    reference_01.jpg

     

    Adobe Standard:

     

    adobe_standard_01.jpg

     

    Camera Neutral, same settings:

     

    camera_neutral_01.jpg

     

    ...and here's the final corrected version:

     

    final_02.jpg

     

    This is in the camera sensor. I assume it's possible to correct it with a profile, but the 18 color patches in the Color Checker is not enough. There's no way to correct shadows and not affect middle values, thus only creating new problems.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 2, 2013 10:56 AM   in reply to George in Seattle

    George in Seattle wrote:

     

     

    This glare (for lack of a better word)  is the root of my "yellow problem" which we found was caused by underexposure since it is the only reason I lower the exposure in the first place.  Even though it is not being reported as overexposure, it doesnt look like something that can be fixed (see the 3rd screenshot from the bottom of my last reply to see what I'm talking about. It's as if the pixels are bleeding into one another to form a white blob where I should see hair contrast.

    George, in the RawDigger Histogram try checking 'Log Scale' under Y-Axis Range. That makes it much easier to determine how many pixels in each of the four channels are clipping. Below is a historgram of a Canon 5D MKII raw image of a light colored Cairn Terrier. The only clipping in this image is a few spots on the top of the dog's head.

     

    RawDigger Log Y axis.jpg

    In this image about 20,000 pixels are clipped in each Green channel, but less than 50 pixels in the Red and Blue channels. As expected there was a loss of detail in the white fur in and around the clipped areas that was not recoverable in LR.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 2, 2013 12:08 PM   in reply to twenty_one

    The reason I ask the question is because in post 16 above George in Seattle said with Camera Neutral it did not have the blue clipping issue and clicking to set white balance works better. 

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 2, 2013 12:28 PM   in reply to b2martin_a

    Yes, I realize that. This is camera-specific as well as situation-specific, so under some circumstances this will work, under others that will work. But the overall trend is unmistakeable with all the Nikons I have ever used: the blue channel drops off very rapidly in the shadows, causing those unpleasant heavy yellows.

     

    Interestingly, if you look at the "reference-shot" I posted above, using Adobe Standard, you'll notice that the darkest blue patch is actually not rendered correctly. It's too light, and too cyan (the other patches look fine). And yet this is the profile that deals with this best. As I understand, they use a lot more than 18 patches at Adobe when making profiles - so I'm wondering if Adobe Standard actually has an in-built correction for this. Pure speculation on my part.

     

    Any which way, this is a b**** to nail down.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 4, 2013 2:57 AM   in reply to George in Seattle

    You can also use the gray patches in the Color Checker as a rough guide (but can and should be adjusted on an individual image basis).

     

    The white patch should be about 245, number 4 gray should be 120, and the darkest around 50. Those are Adobe RGB numbers, which you can get at in soft-proof. They are standardized values that will often result in a slightly dull-looking image (especially for print), so take with a grain of salt.

     

    One other thing is that the dark Lightroom interface can play tricks on you, especially for highlights. They tend to seem more brilliant than they really are. For this reason I prefer to have a light background in soft-proof mode.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 4, 2013 7:02 AM   in reply to George in Seattle

    Yes, I agree about the profiles (which is just my personal experience with the specific cameras I have used of course).

     

    But every custom profile I have made has had some problem with oversaturation in some colors. It just seems to "lose grip" at that point and the colors run off, so to speak. This can be adjusted in the DNG profile editor, but then you get into deep waters in a very time-consuming exercise and I'm simply not convinced it's worth the effort. Not as long as the Adobe Standard profile behaves as well as it does, despite the minor glitches.

     

    I still suspect that 18 color patches (plus 6 gray) are simply not enough reference points to build a really good profile.

     

    I have one custom profile for fluorescent light, in one particular spot, but that's it.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 4, 2013 1:02 PM   in reply to twenty_one

    twenty_one wrote:

     

     

    But every custom profile I have made has had someproblem with oversaturation in somecolors. It just seems to "lose grip" at that point and the colors run off, so to speak. This can be adjusted in the DNG profile editor, but then you get into deep waters in a very time-consuming exercise and I'm simply not convinced it's worth the effort. Not as long as the Adobe Standard profile behaves as well as it does, despite the minor glitches.

     

    I have the same issues, but with my Canon DSLR bodies the CCPP profile still looks the best. The problem is that for some images the color is not always accurate and/or detail is lost in the image (i.e. flowers, dog hair, etc.). I did a little research this morning to better understand what's happening. To do this I needed two vital pieces of information:

     

    1) ColorChecker Passpot patch values, which is available here:

    http://xritephoto.com/documents/literature/en/ColorData-1p_EN.pdf

     

    2) File download of the ColorChecker with the actual values as specified above. Go to the below link, scroll down to 'ColorChecker Images" and download the ProPhoto 16 bit TIFF under the column

    Derived from GretagMacbeth
    L*a*b* D50 data
    (June 2006)

     

    http://www.babelcolor.com/main_level/ColorChecker.htm

     

    3) At this point you are good to go for comparing your CCPP images in PS, but this link also provides the values for taking the readings in LR (created by our LR forum expert Jao):

    http://lagemaat.blogspot.com/2007/09/more-precise-values.html

     

    What I discover is that none of my CCPP images came close to matching the above derived target image, regardless if I used the CCPP, Adobe Standard, or Camera Standard profiles. Next I tried using the Basic tonal controls with the CCPP image and a profile created from it. I got a very close match for the grayscale patch values using the following settings:

     

    Exposure = +.25, Contrast = -40, Highlights = -50, Shadows = 0, Whites = +30, Blacks = -40

     

    Some of the colors in the CCPP adjusted image still did not match the derived reference chart when viewed in LR or PS.

    CCPP Compare to Derived Reference with CCCP_LRAdjusted.jpg

     

    After running this test I'm starting to look at Adobe Standard profile with a less jaundiced eye.

     

    Using the above derived CC file download and CC patch values for PS and LR you can do your own testing and decide what works best.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 4, 2013 5:09 PM   in reply to trshaner

    OK, this is very interesting. I retained the Basic settings:

     

    Exposure = +.25, Contrast = -40, Highlights = -50, Shadows = 0, Whites = +30, Blacks = -40

     

    But added -25 Saturation and +25 Vibrance. The CCPP image now looks virtually identical to the reference derived image.

     

    CCPP Compare to Derived Reference with CCCP_LRAdjusted-25Sat, +25Vib.jpg

     

    This is working with my Canon images using the CCPP profile and I'm guessing it will do the same with Nikon images.

     

    Adjust the Basic panel settings as you normally would with Saturation and Vibrance both set at 0. When done adjust Vibrance to -25 and Saturation to +25. Now how does it look? I'm finding colors appear more accurate and the image has more tonal detail. You may need to readjust the -Saturation and + Vibrance amounts, but opposing settings is what seems to work. Keeping the Contrast at 0 or slightly negative seems to help as well.

     

    I'm not saying you want to do this on every image, just those with Hue & Saturation issues like those originally posted here by George.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 4, 2013 11:13 PM   in reply to trshaner

    I see you are taking a more scientific approach to this than I am

     

    But notice the darkest blue - again. It's still oversaturated, and that's bound to give you problems. And this time yellow is undersaturated. There's something funny going on with the blue channel, maybe it's sensor technology, the Bayer array is a flawed concept to begin with, I just don't know.

     

    Profiles work well output-referred, such as for monitors, since the source is reasonably well-behaved to begin with. But a scene-referred camera profile is a much more unruly beast.

     

    In the end, however, it's worth remembering that cameras differ widely by brand, model and individual copy. This is something everybody has to work out with their own equipment.

     
    |
    Mark as:
1 2 Previous Next
Actions

More Like This

  • Retrieving data ...

Bookmarked By (1)

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