My Duck

[mikemeister_admin]

Here is my Lab duck Greg.  I did cheat and curve it just a bit.  A little hard to resist

[ggroess]

It is always ok to cheat.  So long as the image looks better when you are done.
Our goal is always to improve the image. 

While you have hit the neutral very well something is missing from the image..

When you adjusted the curves did you adjust the reds in the A curve?  It looks like the reds and part of the greens are missing.  Remember in LAB the center of the A and B curves are neutral.  If you want to adjust a single color in LAB you have to protect the other end of the curve.

To do that you can right click on the center of the curve and select the "Pin Grid" Command.  You then can select which half of the curve you want to pin.  So if the green is on the upper right and you want to adjust the red...You would pin the Upper Right from the sub-menu.

Then you are free to adjust the reds without effecting the greens etc. 

Also be aware that adjustments to the L Curve can effect the saturation of the image.  To re-add saturation use the saturation slider in LAB.

Feel Free to play around and re-post the results...
Greg

[mikemeister_admin]

I think you nailed it Greg.  After increasing the saturation via the slider, I did some color checking with the hue clock and discovered that the small white patch on the duck's back was going noticeably towards blue on me with more saturation.  So I tried to isolate and pull that point up in the b curve away from blue to restore the white color.  As I recall, even though I had the b curve pinned all across its length, that relatively small shift did discombobulate the b curve.  I don't think I played around at all with the a curve.  (Is this even a valid approach?)  Unfortunately, lacking your eye for color, I did not notice what was missing.  I will take another stab at it tonight and post the result and have you take a look at my curve.

[ggroess]

Feel free to save the curve to your desktop and post it along with the image. 
I will be able to see what you did with the curves then.
It will be an .ACV file We have posted a few in the class board last week to illustrate specific curves.

Just click the "Save" button in CM and point to the desktop.

Greg

[mikemeister_admin]

Here's my second try Greg ... and my curves.

[ggroess]

Brian,
Much better...From your curve...Shot1 you put a lot of work into this image when a much more simple solution is possible.  Don't get me wrong...you can tackle this anyway you choose but...

Shot 1 : your curves.  You have a significant number of points on the color curves for the duck.  

Shot2  : My RGB curves.  Start by setting a shadow and highlight, open the lighting of the image by dragging the center of the RGB composite curve down a bit.  Set the neutral on the leg band.  There are multiple areas where the band is not neutral.  In fact it is harder to find a neutral than not.  The Leg band is a good place to set the neutral though because we expect it to be a neutral gray.  The Hue Clock shown in the image can be added by Alt-Clicking on a point in the image.  I placed it to verify the feathers did not have a color cast.  If they had I would have adjusted the RGB curve slightly to make them look better.  They can appear green if you miss the neutral slightly.

Shot3  : My SECOND Curve LAB: for saturation only.  Note the position of the A and B curves.  Rotation only.

The ACV curves attached are my curves for the image in RGB then in LAB.  Take a look at then and see if you can simplify your curves a bit.  When we are at this point in the class we want to keep the curves pretty simple.  you certainly are getting an understanding for how this works...but you might want to keep the curves simple and effective.

Greg

[mikemeister_admin]

Simplify . . . gotcha.  Can I ask a few questions?  (1)  Is there any advantage, or disadvantage, from setting your highlight and shadow points in Lab before heading into RGB?  (I find it a little easier to set them in Lab).  Is there any advantage to pinning an actual highlight or shadow point as opposed to just thresholding?  (3) Why did you decide to curve first in RGB rather than working exclusively in Lab?  Thanks, Brian.

[ggroess]

Simplify . . . gotcha.  Can I ask a few questions?  (1)  Is there any advantage, or disadvantage, from setting your highlight and shadow points in Lab before heading into RGB?  (I find it a little easier to set them in Lab).  Is there any advantage to pinning an actual highlight or shadow point as opposed to just thresholding?  (3) Why did you decide to curve first in RGB rather than working exclusively in Lab?   Thanks, Brian.

(1)  Is there any advantage, or disadvantage, from setting your highlight and shadow points in Lab before heading into RGB?  (I find it a little easier to set them in Lab)
no real advantage or disadvantage. I think the LAB space is easier to navigate but there are images like this one where RGB has a distinct advantage.  Lab can be a sledge hammer when it comes to colors. The L channel is more forgiving than making the composite curve adjustments in RGB. I would recommend you get used to doing both thresholding and pinning in the color spaces then you do not have to switch color spaces and apply changes every time.  As we get deeper into this process you might not want to commit changes to frequently.

2) Is there any advantage to pinning an actual highlight or shadow point as opposed to just thresholding?
A qualified depends...I frequently try both as many times the pinned value I have set is an educated guess. I threshold that point and see if there is a way to get more details.  Look at the ducks breast feathers...If I move the highlight too much they blow out...they are very important but which feather is the brightest??  I pick a point and then drag it around the image a bit to check myself. REMEMBER the highlight and shadow are supposed to be set on the significant areas you want full details in.  They are frequently not the darkest or brightest point of the frame.  If the clouds are blown out but not important to the image do we really care?? If the shadow under the car is totally black is that more important than the details in the weave of a black hat the person in front of the car is wearing??

(3) Why did you decide to curve first in RGB rather than working exclusively in Lab?

I didn't decide at the start.  I actually started in LAB and fought the image for the color of the feathers.  Having looked at your curves before I started I knew you fought for the color as well.  I started in LAB to see if you had just over adjusted it to death and found myself adding control points all over the curve.  At that point I decided to quit on LAB.  I knew I could come back for saturation as needed.  I also have learned, and you will too; that RGB is a much more subtle color space for adjustments.  Just as CMYK is even more subtle than RGB. I also knew that by setting a highlight and shadow; I was going to lose saturation in the image so I was coming back to LAB for the finish.  This decision freed me up to fix the problem without worrying too much about saturation.  BTW I can correct the image totally in LAB.  It has been done many times but I wanted to illustrate the diminishing returns you can get with too much adjustment.

Great questions...
Greg

[mikemeister_admin]

Thanks for great answers.  It helps to know that this is even challenging for the masters.  OK . . . here is my third effort,  I took it into RGB to set a shadow, highlight and neutral.  I then lifted the composite RGB curve to add lightness and contrast.  I tried playing around with the individual curves to see if I could get the hue clock on the back feathers to register closer to true black, but could not without disturbing other elements of the image, so I gave up on that.  Then over to Lab to rotate the a & b curves for increased saturation.  In comparing the end result to my second image I thought at first that I had somehow introduced a blue cast, though it wasn't showing up on the hue clocks.  After looking at it more, I convinced myself that that I had actually introduced a bit of a green cast in the second image and that this third image was actually the more color accurate of the two and just needed some more contrast.  So . . . back into RGB to give the duck a slight S-curve on the composite curve.  But the hue clocks on the back feathers still point to "G."  Obviously I am still not at the simple stage.

[ggroess]

I like it the image has a richness the first two are lacking....the Breast feathers appear to be a bit too hot but that is a simple adjustment. If you need to go into LAB you can use a "Lizard Tail" to try to bring the highlight details back. Lab could have added the contrast back without messing up the color as well. Lizard Tail shown in Shot1 note the change only in the Breast feathers. 

http://www.curvemeister.com/wiki/index.php?title=Special_Curves:_Lizard_Tail

BTW in RGB Color and Luminosity are linked.  If you adjust the individual colors you effect the rest of the image.  Use the color Worm to help you "see" the range of the image you are effecting.

http://www.curvemeister.com/wiki/index.php?title=Color_Worms

We will learn a technique called "By The Numbers" where you correct the image using RGB to achieve / set multiple neutrals based on brightness.  That is coming up in the next week or so...

Greg