Using hue curve to "get the green out of yellows"

[Charles2]

This language from Ken Rockwell, where he talks about getting green out of yellows, for example, is suggestive:

Like all great art throughout the ages, the stunning color representation of the LEICA M9 polarizes all who experience it.

Some people love its unique interpretation, while others, including myself, find it hideous, and I mean hideous!
...
In Adobe Camera Raw (ACR) or some other tool, one must correct the color hues to get the greens out of the yellows and blues. One does this with the Hue sliders under the fourth green zigzag (HSL/Grayscale) tab in ACR.

In this tab, I find I like something around -18 for oranges and yellows, which makes these colors more red (less green), and +18 for blues, which pulls the greens out of the skies and makes them more blue instead of cyan (greenish).

I agree with Rockwell about Leica colors, though his saturated, oversimplified style repels me, too.

So what is the equivalent in PWP of his procedure?

One way might be to use vectors on the Color Correct hexagon that go from one color toward or away form a neighboring color rather than in toward the center or out toward the edge.

Would another way be to modify the hue curve? Here's an example on a Leica image taken by Rockwell. The DNG is available on his website, although it might need correction with a run through Adobe DNG Converter.



I prefer to work with the histogram than the curve. The diagonal vectors make the changes, while the strictly vertical vectors are intended to block color from the undesired direction. I just happened to go with HSV-H; perhaps HSL-H would be better. Anyway, is this method a more direct imitation of the ACR sliders?

(I tried to check the archives, but could not find on-topic stuff.)

[den]

Not having ACR to verify, I interpret from the quoted description as follows:
(1) "zigzag" = Broken Line;
(2) "HSL" = HSL color space model;
(3) "-18 (less green) = decreasing the G hue by 18 on a 0 to 255 scale; and
(4) "+18 for blues" = increasing the B hue by 18 on a 0 to 255 scale...

...and would purpose the following "lessGreen-moreBlue_HSL-H.cc" file shown in text form:

ColorCurves 1.0
colorspace 1
component 0
npts1 7
style1 linear
histexpand1 0
point1 0 0 0
point1 1 43 43
point1 2 85 67
point1 3 128 128
point1 4 170 188
point1 5 214 214
point1 6 255 255
npts2 2
style2 spline
histexpand2 0
point2 0 0 0
point2 1 255 255
npts3 2
style3 spline
histexpand3 0
point3 0 0 0
point3 1 255 255
end

which provides a ColorCurves HSL-H dialog in histogram and curve forms:

[Charles2]

Ken Rockwell has a screenshot of his ACR sliders. It is currently on this page
http://www.kenrockwell.com/tech/00-new-today.htm
while the presumably permanent location of the screenshot is at
http://www.kenrockwell.com/leica/images ... -08-16.jpg :



Rockwell vectors the yellows and oranges away from green. He does not change green itself, which is what DEN's curve does.

On my monitor DEN's curve does not produce the effect Rockwell makes and likes. The curve I showed produces something closer to the style he gets with those ACR tweaks. But DEN set boundaries on both sides of the change, which seems theoretically correct versus the one-sided boundaries in my example.

This discussion helps one understand how one can use the hue curves, and it appropriates language from Rockwell that I find helpful to thinking about images.

It also reinforces the view that the color curves are fundamental tools. Most if not all sliders and other manipulations in various PP programs are equivalent to certain curve changes. They might be convenient, but I think it is a scientific duty of the program vendors to relate their sliders to a common theoretical heritage like color curves. Commerce, though, trumps science.

[den]

...nothing is as simple as it seems!...

If the the slider color bars are accurate, one could Readout Tool sample:
(1) the center Reds for a locking control point;
(2) the center Oranges and the shift to the left;
(3) the center Yellows and the shift to the left;
(4) the center Greens for a locking control point;
(5) the center Blues and the shift to the right; and
(6) the center Purples for a locking control point
to determine a possible near equivalent PWP HSL-H curve.

If of interest, I can post a "*.cc" curve file in text form based on the above. The less green, more blue hue changes are not as aggressive as previously suggested.

Also I noticed in the links that NRockwell suggests "+50 Saturation" change which would suggest that it is important to make PWP equivalent adjustments in the HSL color space model... ...and ACR's Saturation tab elements need to be considered as well for a PWP equivalent.

Anyone...
Is there further interest in how to use the Readout Tool to determine Hue percents; convert to Hue 0 - 255 tones for creating a text "*.cc" curve file?

Also, a slight rounding error [or typo] in the suggested "lessGreen-moreBlue_HSL-S.cc" was posted. The Magenta locking control "point1 5 214 214" should more accurately be "point1 5 213 213".

[Charles2]

One thing this exploration reminded me is that my skills wielding the Color Range mask tool are terrible. This tool should be helpful in conjunction with modifying a hue curve and also as a different method of isolating areas that one could then run through Color Correct or simple ColorCurve-RGB changes.

[Charles2]

A coda to the discussion of HSL-H curve manipulations: The colors of an Olympus E-P1 raw file loaded into PWP were different from those in the camera JPG. In particular I wanted to imitate the JPG red on some thistles. The RGB-R curve could do it, but it also mangled other parts of the image, for example, the sky. Similarly for a saturation vector in Color Correct.

By working on the HSL-H curve, I managed to come reasonably close to the JPG, at least regarding this red stuff. I bounded the red portions of the curve at both ends and pushed in the appropriate directions within these regions.

Some people rave about Olympus colors. On the basis of a sample of one image, I suspect the camera JPG firmware does something effectively equivalent to manipulating HSL-H or HSV-H.

[cliff]

I'm still a newbie, but I would have approached the problem from a different direction. I would simply have put a gray card or a Q-13 card in the scene and used either the Color-Balance or the Color-Match Reference transform. Otherwise I would have used the rocks as a neutral gray. I know you all know this, so am I missing something too subtle for me to understand?
Cliff

[Charles2]

Looked at your product photos, Cliff. You are no newbie! The photos present the products extremely well. The images aim for an objective style (which remains a style, self-effacing as it might be). The thread began with something from Ken Rockwell, who delights in obvious exaggeration in a certain direction. He described a technical procedure to help achieve his style, and that set us off because we aspire to styles, too.

On the technical level, I presume that color balance affects every color in the image. HSL-H changes can be confined to a narrow range of hues.

[cliff]

Good answer Charles. I'm glad you replied, because I was beginning to worry that I had killed this thread.
Cliff

[den]

The illustration below may be of help in selecting hue ranges, hues, control points, and editing control points of "*.crv" files in text form. It should be noted that HSL-Hues = HSV-Hues. The color space model becomes important when further editing 'Saturations' and 'Lightness' or 'Values'. The illustration is based upon PWP's typical Color dialogues...

a_HSVcolor_hexagon_notated_400px.jpg (46.87 KiB) Viewed 6805 times