LAB Color discussion -- Chapter 7 Workflow

pathfinderpathfinder Super Moderators Posts: 14,694 moderator
edited October 20, 2005 in Finishing School
Chapter 7: Summing Up: LAB and Workflow

In the first six chapters we have been introduced to the LAB color space and an initial understanding of what the numbers in LAB represent -
L -- 0( darkest) to 100 ( lightest) -- ( I have deliberately avoided saying black and white here, as the L channel does not refer to color whatsoever, so 100 L cannot be part of the description of any color. The LAB color space can describe colors that are not within the gamut of sRGB or Adobe RGB or CMYK or of a monitor or a printed page - It can even describe colors that seem impossible called imaginary colors. Like imaginary numbers, imaginary colors can be quite useful, and will be in the future.

Channel A and B are numbered from -128 to 127 ( 256 levels like other color channels in RGB or CMYK). Positive numbers (+) refer to magenta in A, and yellow in B. Negative numbers, written in parentheses rather than with a minus, sign refer to green in A, and blue in B. Values of zero are neutral and hence 0,0 is a grey color - how light or dark will depend on the value of the corresponding L channel from the very lightest possible light grey to the very darkest imaginable dark grey.

Chapter 1 introduced the symmetrical steepening of the A and B channels to drive colors farther apart when confronted with a relatively monochromatic scene like a canyon wall or a forest of green leaves. This steepening of A and B along with a nice S shaped curve in L, can give many images a lot more POP. This treatment is not usually needed on images with highly saturated colors or with broad areas of varied contrasts.

Chapter 2 continued the discussion of the numbers of the LAB color space.

Chapter 3 described the alternatives of different steepness of the A and B curves and demonstrated the usefulness of this technique in images of vegetation that need the A curve steeper than the B curve to avoid getting vegetation too yellow. A brief discussion of the correct setup values to be used in Photoshop to correspond with the book’s values. Be sure the eyedropper in NOT set to point sample, but 3x3 or 5x5. Disable the dither checkbox in the Convert to Profile dialog and use Relative Colorimetric intent. Set the Curves dialog with the darkness to the right and lightness to the left by clicking along the dark border of the curve box. Curves of the A and B space for Caucasian skin usually benefit with the B curve slightly steeper than the A curve as yellow needs to be slightly stronger than magenta, just as is true in CMYK.

Chapter 4 discusses the correcting of color casts by shifting the A or B curve away from the center point. Finding the white and black points with the Image>Adjust>Thresholds is discussed. Also covered is locking the center point of the A and B curves and moving each half separately for correcting a color cast.

Chapter 5: Sharpen the L, Blur the A and B Achieving sharper appearing images comes from increasing the luminosity variation. Image noise - color pixel noise - is best corrected by blurring the A and B with Gaussian Blur or the newer Surface Blur in PS CS 2.

Chapter 6: Entering the Forest: Myths and dangers
Moral of the fairy tale - Mothers often give good advice, even when they don’t know what they are talking about. LAB is compared to a primeval forest where monsters can lurk for the inattentive wanderer.
Myths exposed ie: NOT TRUE - Lab should be the primary workspace
True - Overuse of the L space - powerful, easy to overdo it
Not True - the best B&W conversion is the L channel as a grayscale
Not True -RGB in luminosity mode is as good a LAB
Not True - Converting to LAB “hurts” the image



And FINALLY TA DAA!!! Chapter 7: LAB and the Workflow

Truths about LAB
Generally best space for retouching
Exchanging colors is done faster and better in LAB
Certain masks and selections are easier in LAB
Several blending techniques work well in LAB and not at all in RGB or CMYK
LAB offers real advantages in working on portraits


A big questions discussed in the early part of Chapter 7, is how much time are you willing/able/allowed to spend on a single image for getting it ready for printing or display. Answers vary from 30 seconds to 30 days and it is difficult to get all photographers under the same tent ( i.e. workflow) as a result.
For those whose time is very limited a quick trip to LAB can be very effective and fast. Receive the RGB image and covert to LAB and check a few numbers in the image looking for neutral grays, whites, blacks, or known colors. Command-M to bring up the Curves dialog to allow altering the A and B curves as needed. Areas of the image that deserve to be higher contrast in the major subject areas can be identified by passing the mouse over them and seeing where they are on the L curve. The L curve can then be altered to favor the contrast in the major subject area. Sharpening the L curve can then be done and the image passed on to RGB or CMYK as needed.

The general reasons to go into LAB are to increase variation in colors, make certain colors purer, blur away colored noise, use the Shadow/Highlight command on the L channel to open shadow detail, increase contrast in a narrow tonal range, simplify difficult retouching.

Some types of image editing are better done in a specific color space or point in workflow - i.e.: Raw Conversion, RGB, LAB, CMYK. Gross changes with curves are better done in RAW or LAB. RGB and CMYK are much poorer choices for gross corrections. Fine tuning an image is better done in CMYK or RGB as a second choice due to the ability to manipulate the Black channel. Brightening, suppressing, enhancing, or changing color are usually best done in LAB. Blurring color noise is best done in LAB in the A and B channels. The most flexible for sharpening is CMYK, but LAB and RGB can both give good results.

The Shadow/Highlight command works very well in the L channel of LAB.

As an example of the use of the Shadow/Highlight command on the L channel, I submit the following image converted from RAW in Adobe Raw Converter, hence forth called "ARC". I purposely underexposed this image in the conversion from RAW to have underexposed shadows for the S/H command. Here is the original underexposed image from RAW....
39797322-L.jpg

Note the lack of detail in the back of the chair and the rock wall and the grass at the bottom of the image.

I used the following curves in editing the image in the LAB colorspace...

39797553-L.jpg

The A and B channels were increased as shown. The L channel demonstrates the curve I wrote for the L channel and the red circles demonstrate the areas steepened and what they represent in the original image. I then sent the L channel through the Shadow/Highlight command with the following settings on the shadow commands only

39803364-L.jpg


The Amount was raised from 50% to 65% due to the difference in the apparent lightness of the L channel versus a greyscale version in RGB.
Frequently, I find heavy use of the S/H command is obviously apparent with a greying of the image at lesser amounts than 50%. But here, 65% on the L channel does not seem to have that effect. This is the final result after sharpening slightly on the L channel and moving back to RGB and saving as a jpg for loading to smuggy.

39797537-L.jpg


Certainly not an artistic masterpiece, but definitely more detail in the shadow areas of the back of the chair and the fence and the tree trunk, as well as better tonal gradation in the clouds. I think I will find frequent use of the Shadow/Highlight command on the L channel in the future on my images. For further comparison, here is the image I originally posted from the same RAW frame that I did not underexpose in ARC. This image has not been through LAB, just a straight well-done image directly from ARC

34382209-L.jpg

Marguis describes using the L, A and B channel curves on a duplicate layer as an adjustment layer - that means he creates a duplicate layer of the background original by pressing ctrl-J( not a Mac shortcut - works on the PC too) and then writes the curves on that layer. He can then blend the two layers in the Normal mode with the slider on the Layers palette to obtain the precise image he wants. I find this is helpful also with sharpening in RGB or retouching faces with the healing brush tool on an adjustment layer. He uses the example of a buffalo with the A and B curves as virtual vertical lines that create dramatic overcorrection of the colors and then correcting this with the Normal blending mode sliders.

We are now into the final part of the 7th Chapter and the easy work is behind us...
Two types of images require special care in LAB - those with different color balances in the highlights and the shadows, and those with color casts except in the highlights and shadows. Different color casts in the highlights, than the shadows frequently originate from overzealous autocorrrection in digicams that create beautiful neutral highlights ( 255,255,255 in RGB for example ) or shadows, with images with very poor color balance in the mid-tones. These cannot be simply corrected with a simple curve in the A or B channel in LAB, ( unlike RGB or CMYK where curves can affect highlights different than shadows), but requires a selection of the highlights or shadows to limit the correction to the affected areas only. This usually means a selection based on the luminosity achieved by Command - Option - 1 or (ctrl-alt-1 on the PC). That is Command Key + Option Key+ the one (1) key ( Not the L key !!) The creates a luminosity selection like this

39899944-L.jpg

Notice the marching ants around the lighter areas of the image and not in the darker areas - this gives the selection that limits the curves to only the lighter areas and not the darker areas - and it only takes Command+Option+1 to create. Shift+Command+i will invert the selection to include the darker areas only.
Greg Gorhams B&W conversion is based on converting to LAB and going to the Grayscale image and then selecting the Luminosity channel and inverting it to affect the darker areas specifically, and then Multiply blending this layer with a new layer of solid color of your choice. Hence Shift+Command+i inverts the luminosity selection to select the darker areas of the image only.

The more challenging problem is when the mid-tones are colored but the highlights and shadows are not - you now need a selection of the mid-tones only. This selection is hidden away ( to me at least ) in the Layer Blending options. You will need a duplicate layer for this, so hit --- CTRL-J again!! Then press the adjustment layer button at the bottom of the Layers palette - the little square box with the round circle in it - and a new adjustment layer box will appear in the new layer box in the Layers palette.

39903425-M.jpg

Now we can click and hold the little tiny arrow at the upper left of the Layers Palette and select Blending Options halfway down from the drop down menu. Here is how it will look initially...

39904412-L.jpg

And then you will see this drop down menu

39904408-L.jpg

And finally this is the dialog box you will see....

39901813-M.jpg

This Blending Option dialog box describes a selection mathematically, rather than by using the selection tools. This box selects areas of the L Channel - that are affected. When the sliders are moved centrally from the outer ends, they impose limits. The setting in the next box say use the top layer unless either the top layer is very dark or the bottom layer is very light. The sliders split in half by Option clicking on them to create a gradient of the selection rather than a harsh border. The Blending Options can be selected in the upper box and include screen, multiply, difference, exclusion, overlay, hard light, soft light, etc. And there is the typical slider to adjust the blending strength as well.

40008515-M.jpg

I would recommend everyone open this dialog box and play with it - It offers a great deal of choices and they need to be experimented with to see how they work, but they do allow you to choose the central 75%, or whatever your choose, of a luminosity layer and exclude the end points that may be color neutral.

In summary, the first 7 chapters have been primarily on global maneuvers that affect the entire image without complicated selections. The easy part of the book is over. Now the fun really is about to begin. Come join the fun!!
Pathfinder - www.pathfinder.smugmug.com

Moderator of the Technique Forum and Finishing School on Dgrin

Comments

  • ruttrutt Registered Users Posts: 6,511 Major grins
    edited October 15, 2005
    Great summary, Jim! Thanks. I'm still working through the chapter, but I think having read your summary will make that easier. I'm sure I'll have some comments over the next day or so.
    If not now, when?
  • ruttrutt Registered Users Posts: 6,511 Major grins
    edited October 15, 2005
    Actually I do have a question. Those command which make selections based on lumonsity like:
    Command - Option - 1

    Are there menu equivalents? Are they just black magic? What controls what PS considers a shadow or a highlight? Frankly, I understand the blending options dialog boxes and blend-if a lot better.
    If not now, when?
  • pathfinderpathfinder Super Moderators Posts: 14,694 moderator
    edited October 15, 2005
    Page 156 in Canyon Conundrum, Margulis says, "The long way to do it is to open the Channels palette and Option-click the appropriate channel."

    This suggests to me that you could alter the luminosity mask by drawing a curve on the luminosity channel before using it as a selection with the cmd+option+1 keystrokes.
    Pathfinder - www.pathfinder.smugmug.com

    Moderator of the Technique Forum and Finishing School on Dgrin
  • DavidTODavidTO Registered Users, Retired Mod Posts: 19,160 Major grins
    edited October 15, 2005
    How does that relate to cmd-clicking on the L channel to select it? I believe that selects it as if it were a mask. Is that like selecting highlights?
    Moderator Emeritus
    Dgrin FAQ | Me | Workshops
  • pathfinderpathfinder Super Moderators Posts: 14,694 moderator
    edited October 15, 2005
    DavidTO wrote:
    How does that relate to cmd-clicking on the L channel to select it? I believe that selects it as if it were a mask. Is that like selecting highlights?
    When I use ctrl+option+1 I think I get the same thing as clicking on the L channel and saving it as a selection - sure looks the same - I have not figured out how to compare one with the other numerically yet. I'm thinkin' on it!
    Pathfinder - www.pathfinder.smugmug.com

    Moderator of the Technique Forum and Finishing School on Dgrin
  • ruttrutt Registered Users Posts: 6,511 Major grins
    edited October 15, 2005
    pathfinder wrote:
    Page 156 in Canyon Conundrum, Margulis says, "The long way to do it is to open the Channels palette and Option-click the appropriate channel."

    This suggests to me that you could alter the luminosity mask by drawing a curve on the luminosity channel before using it as a selection with the cmd+option+1 keystrokes.

    I still find this very mysterious. What determines the threshold for the mask?
    If not now, when?
  • DavidTODavidTO Registered Users, Retired Mod Posts: 19,160 Major grins
    edited October 15, 2005
    rutt wrote:
    I still find this very mysterious. What determines the threshold for the mask?


    I think it's handled like any other mask. Black is opaque, white transparent, gray is right in between. But that's just me thinking...not really knowing.
    Moderator Emeritus
    Dgrin FAQ | Me | Workshops
  • pathfinderpathfinder Super Moderators Posts: 14,694 moderator
    edited October 15, 2005
    DavidTO wrote:
    I think it's handled like any other mask. Black is opaque, white transparent, gray is right in between. But that's just me thinking...not really knowing.

    I think thats is, David. That is why I said that you might be able to do a curve on the L before selecting it as a luminosity mask. But I have not tried that so ne_nau.gif
    Pathfinder - www.pathfinder.smugmug.com

    Moderator of the Technique Forum and Finishing School on Dgrin
  • ginger_55ginger_55 Registered Users Posts: 8,416 Major grins
    edited October 16, 2005
    May I interject a question since you have that blending thing up there. How do you split the sliders?

    I have tried many times. Once it happened by "magic", I never could get it to happen again. I have other uses for that especially some stuff in books, they tell you to split it, but they don't say how.

    You said by the options key, or something. What do you mean? Please, how precisely do you split the slider?

    (excuse me, smile)

    I read a very good and thorough review of this book and LAB in Rangefinder magazine. It also mentioned the type of people who might like learning it. The magazine is not in here. I will type it out some time.

    ginger
    After all is said and done, it is the sweet tea.
  • DavidTODavidTO Registered Users, Retired Mod Posts: 19,160 Major grins
    edited October 16, 2005
    ginger_55 wrote:
    You said by the options key, or something. What do you mean? Please, how precisely do you split the slider?

    ginger


    PF was talking macspeak. Option = Alt. So in windows you should be able to alt-click to split the sliders.
    Moderator Emeritus
    Dgrin FAQ | Me | Workshops
  • pathfinderpathfinder Super Moderators Posts: 14,694 moderator
    edited October 16, 2005
    DavidTO wrote:
    PF was talking macspeak. Option = Alt. So in windows you should be able to alt-click to split the sliders.


    What David said!! On a PC, hold down the ALT key and click and hold down the left mouse button to slide the sliders apart. On my MAC the sliders are kinda finicky, but it works if you are patient. :):
    Pathfinder - www.pathfinder.smugmug.com

    Moderator of the Technique Forum and Finishing School on Dgrin
  • ginger_55ginger_55 Registered Users Posts: 8,416 Major grins
    edited October 16, 2005
    Thanks, I wrote down the info. Will try it as soon as the opportunity presents.

    That has been one of life's little mysteries. Amazing what you can learn and where, smile.

    thanks again,

    ginger
    After all is said and done, it is the sweet tea.
  • ruttrutt Registered Users Posts: 6,511 Major grins
    edited October 16, 2005
    One thing PF didn't cover was Dan's description of digital workflow in the "old" days, circa 1990. There really was no reason to include this in the summary, it really doesn't mean much to someone who is trying to leran today's workflow. But I found it fascinating.

    At the time a computer program to perform digital photo manipulation required a >$100k machine which would still have been 100s of times slower than even the slowest acceptable machine today. An application of curves would take long enough to go out for lunch. The largest most expensive disk would hold just a few images. RAM memory was measured in single digits of megabytes at most.

    The scanner (also a major capital investmnet) operator was the best paid employee. His goal was to make every scan perfect enough that no manipulation was required. Dan implies that he was a highly paid manager/consultant who would be called in to help the scanner operators improve their batting average.

    The reason for Dan's telling this story, besides the fact that he loves to write about this sort of thing, is to contrast it with digital image acquisition today. Today, nearly every image will receive some PP manipulation and the most successful acquisition process is not one that requires this to happen less, but one that makes it easier. For example, the old scanner operator tried to acheive perfect contrast with good white and dark points. Today, it's easy to adjust contrast afterwards, and an acquisiton with perfect white and dark points can make it much harder to color balance or expose hidden details in the highlights and/or shadows. So when using the raw converter (our equivalent of the drum scanner) it's best to be conservative. Make the grossest of corrections, such as WB and don't try to get the contrast or color balance perfect. That is done better and easier later on.
    If not now, when?
  • ruttrutt Registered Users Posts: 6,511 Major grins
    edited October 16, 2005
    Going to far in layers as part of the one minute workflow
    Pathfinder wrote:
    Marguis describes using the L, A and B channel curves on a duplicate layer as an adjustment layer - that means here creates a duplicate layer of the background original by pressing ctrl-J( not a Mac shortcut - works on the PC too) and then writes the curves on that layer. He can then blend the two layers in the Normal mode with the slider on the Layers palette to obtain the precise image he wants. I find this is helpful also with sharpening in RGB or retouching faces with the healing brush tool on an adjustment layer. He uses the example of a buffalo with the A and B curves as virtual vertical lines that create dramatic overcorrection of the colors and then correcting this with the Normal blending mode sliders.

    This is a very useful technique, one that Dan many PS experts, including Dan, use often. A part of the picture I didn't have before reading this particular secion, is that it can be used to make it easier to get good results FAST. Dan's context for this particular example is what he calls the "one minute workflow". By working in adjustment layers, he can take a quick swipe at each adjustment erroring on the side of overadjusting. His example has both shadow/highlight and LAB curve steepening in separate layers. Then he can turn down the opacity of each adjustment as final quick fine tuning step.

    In sum, I like the emphasis in this chapter on the time budget for image adjustment. This is the voice of the true professional. He has it in his ability to improve almost any image and to keep going. Where is the point of diminishing returns? The one minute workflow is within nearly every time budget except one which calls for instant on the spot printing or displaying proof sheets.
    If not now, when?
  • ruttrutt Registered Users Posts: 6,511 Major grins
    edited October 17, 2005
    I love the insight that the shadow part of shadow/highlight works better in LAB and the highlight part works better in RGB (and I love knowing why!)

    Here is an old favorite that I never really finished. First my recent version, using what I've learned so far from the LAB book, especially the shadow/hightlight tip.

    40393051-L.jpg

    For reference, here is the original:

    36868919-L.jpg

    And here is my best effort from 3 weeks ago:

    37166400-L.jpg

    (Well, my best effort with no selection, layer mask, etc.)

    OK, so now I:
    1. Straigntened and cropped (including a little clonign lower left to allow a slightly wider crop on that side to get in more of the man's head.
    2. Used shadow/highlight twice; once in RGB to recover the highlights, like the in woman front's face, hair, and blouse;
    3. and once in LAB to recover the shadow details, the stuff inside the door.
    4. After this I A+B steepened by 15% on each endpoint without touching the L curve. I didn't save the result at this point, but it was already better than anything I had before.
    5. Next I moved (through RGB) into CMYK, made a duplicate layer, and
    6. steepened its K curve into very steep S (something David suggested.)
    7. Then I added a layer mask and painted out the stuff that got "plugged" by this move, in particular inside the door (done with a rectangular selection and the paint bucket) and a few other places, faces and shadows on the clothes. This was really really easy because the paint strokes dodn't show against the light background, only against the jeans. Just using a soft brush and full opacity was enough to unplug what needed unplugging and I didn't have to be very accurate at all. For example, you can't see the rectangular selection I used to unplug the interior, even though I extended it well down onto the sidewalk (to below the woman's feet.)
    8. After flattening, I sharpened the K chanel,
    9. moved back to LAB and sharpened the L channel on a different layer.
    10. Moved back to RGB and duplicated both USM channels,
    11. made a darken and lighten version of each and adjusted opacity.

    Do I have to tell you that this turned out not to be a 1 minute workflow. Chalk it up to a learning experience. But I like this particular image enough to try to make it the best it can be.
    If not now, when?
  • DavidTODavidTO Registered Users, Retired Mod Posts: 19,160 Major grins
    edited October 17, 2005
    rutt wrote:
    Here is an old favorite that I never really finished....

    Best version yet, Rutt. Can't see anything I would change.

    You can stop now. ;)
    Moderator Emeritus
    Dgrin FAQ | Me | Workshops
  • dandilldandill Registered Users Posts: 102 Major grins
    edited October 20, 2005
    rutt wrote:

    • After flattening, I sharpened the K chanel,
    • moved back to LAB and sharpened the L channel on a different layer
    • Moved back to RGB and duplicated both USM channels,
    • made a darken and lighten version of each and adjusted opacity.
    Thanks very much for this detail, which I find very helpful. Can you say more about the steps above? That is, how did you get the two (sharpened?) USM channels (from CMYK and from LAB) into the the final RGB document? I ask because I thought when you go to a new color space, the channels get flattened?


    Thanks
    Dan
    Dan Dill

    "It is a magical time. I am reluctant to leave. Yet the shooting becomes more difficult, the path back grows black as it is without this last light. I don't do it anymore unless my husband is with me, as I am still afraid of the dark, smile.

    This was truly last light, my legs were tired, my husband could no longer read and was anxious to leave, but the magic and I, we lingered........"
    Ginger Jones
  • ruttrutt Registered Users Posts: 6,511 Major grins
    edited October 20, 2005
    dandill wrote:
    Thanks very much for this detail, which I find very helpful. Can you say more about the steps above? That is, how did you get the two (sharpened?) USM channels (from CMYK and from LAB) into the the final RGB document? I ask because I thought when you go to a new color space, the channels get flattened?


    Thanks
    Dan

    It asks whether to flatten and tells you that bad things might happen if you don't. In this case, the goal is to end up in RGB with a layer each for dark and light USM halos and to use lighten and darken as blending modes, so it's pretty clear that the blending is going to end up doing what I wanted.

    Take a look here, for a more detailed description of this kind of light/dark sharpening: http://www.dgrin.com/showthread.php?t=9739
    If not now, when?
  • spocklingspockling Registered Users Posts: 369 Major grins
    edited October 20, 2005
    rutt wrote:
    [*]moved back to LAB and sharpened the L channel on a different layer. [*]Moved back to RGB and duplicated both USM channels,
    [*]made a darken and lighten version of each and adjusted opacity.
    [/list]

    Do I have to tell you that this turned out not to be a 1 minute workflow. Chalk it up to a learning experience. But I like this particular image enough to try to make it the best it can be.


    Just to add here, I have been following along a bit, and based off this thread and this discussion on sharpening, I have made an action that acomplishes this. It can be downloaded from here.

    Just trying to help.
  • dandilldandill Registered Users Posts: 102 Major grins
    edited October 20, 2005
    rutt wrote:
    It asks whether to flatten and tells you that bad things might happen if you don't. ...

    Take a look here, for a more detailed description of this kind of light/dark sharpening: http://www.dgrin.com/showthread.php?t=9739
    That's what I needed. Thanks again.
    Dan Dill

    "It is a magical time. I am reluctant to leave. Yet the shooting becomes more difficult, the path back grows black as it is without this last light. I don't do it anymore unless my husband is with me, as I am still afraid of the dark, smile.

    This was truly last light, my legs were tired, my husband could no longer read and was anxious to leave, but the magic and I, we lingered........"
    Ginger Jones
Sign In or Register to comment.