Luminous Landscape Forum
Raw & Post Processing, Printing => Digital Image Processing => Topic started by: howardm on April 17, 2009, 08:38:22 pm
-
I've been watching a bunch of kelby & lynda courses on Lab color and I'm just kind of curious how
many intermediate/advanced PS users actually know Lab and/or spend significant time in that space doing work.
Certainly looks very flexible and powerful but can blow your foot off w/o much effort too
-
I've been watching a bunch of kelby & lynda courses on Lab color and I'm just kind of curious how
many intermediate/advanced PS users actually know Lab and/or spend significant time in that space doing work.
Certainly looks very flexible and powerful but can blow your foot off w/o much effort too
I use it on every single photograph I process. It's simple, accurate with colors, brings out color without jacking up the saturation, if you watch it, and allows you the ultimate ability to isolate and target your efforts.
My work flow is based on a simple action that converts the image to LAB, sets up three curves adjustment layers, one each for Luminance, A and B color spaces, then additional layers for levels, brightness/contrast, color balance, hue and saturation (I rarely use any but the curves layers). Sometimes I point process areas for brightness using Viveza's control points because I often just do a straight convert to TIFF in View NX without going to NX if there's not much adjusting to do (which I strive for in my exposures). My L curve is set for a fairly aggressive S curve contrast, but then knocked back to 20% opacity. Likewise the the A and B curves are preset in the action for fairly aggressive straight line adjustments then also knocked back to 20% opacity. My processing is usually just simple adjusting of the opacities, occaisionally some Viveza, once in while some Shadow and Highlights, though usually like using the Viveza U points for more contral.
The key to me is having each LAB adjustment on its own curves layers for fine control, including opacity. That's how you avoid "blowing your foot off", but truthfully LAB is no more likely to be over the top than any other processing, and messing with saturation in RGB always screws up something because it's such a blunt tool. Some of my pictures where there are already strong colors may get not much, if any of the A and B adjustments -- maybe 5% opacity for a very subltle pop, or non at all and just contrast and sharpening.
Final step is sharpening with Focus Magic, almost always with 1 pixel, now that I am using the D3X -- it doesn't need much and sometimes I skip sharpening altogether. Usually, I'll crop before converting to LAB so I'm not cropping a sharpened picture.
This is a very powerful and flexible process that's really quick and simple. For the life of me I can't figure out why more people don't use LAB. It's very gentle on the colors, very quick and easy. The only downside is you really have to work your way through Photoshop Lab Color by Dan Margulis (the Canyon Conundrum Book), but it's well worth it.
Don't hesitate to use it. You will bring much more life out of your pictures, epecially subtle color that's really there in more monochromatic or flat lighting. You can make your shots look like the scene did without over doing it. I tend to shoot a lot right after the sun has set when you get that beautiful, subtle reflected low light. LAB is really well suited for that sort of image.
YMMV, but it's free and you really won't hurt your images or your foot!
David in Phoenix
-
As a professional manipulator I've used LAB twice in my career. I know maybe one person who has used it (not uses).
You really have to take Kelby, Eismann and the like with a grain of salt.
-
As a professional manipulator I've used LAB twice in my career. I know maybe one person who has used it (not uses).
That might be me, the later (not used) <g>
You really have to take Kelby, Eismann and the like with a grain of salt.
Well both have good things to teach, certainly Eismann when it comes to retouching! But Lab seems to be the flavor of the month these days from others of the so called Photoshop guru camp. Deke's into it as well.
Dan Margulis is the guy who's put this on the map. While much of what he has to say about high bit editing and wide gamut spaces is just plain old silly, lets give him credit for being the guy who's taught Lab editing techniques from day one and is the person who one should investigate should they feel the need to use Lab corrections. Now if he'd only recognize the power of getting good data into Photoshop in the first place (either at the scan stage or more pertinent today, at Raw rendering), the need for all these Lab corrections would be reduced to nearly zero.
-
LaB manipulations are just one more tool available to you. They aren't a magic bullet but they are sometimes useful. I recommend getting Dan Margulis's book Photohop Lab Color and playing around while realizing that his "schtick" is LaB color so he is a real evangelist. I try some simple LaB manipulation (steepening the a and b curves, usually) on perhaps 20% of my images and keep it on about 1/2 of those. As with most digital image manipulations, there is more than one way to get the same result, and a lot of course depends on the result you like.
Peter
-
I use it on every single photograph I process. It's simple, accurate with colors, brings out color without jacking up the saturation, if you watch it, and allows you the ultimate ability to isolate and target your efforts.
[...]
David in Phoenix
Same here. I just find LAB way more intuitive whenever color correction is needed. In RGB I often have to struggle to achieve the same results and it takes me twice as long. I do end up coverting to RGB anyway because many filters, actions or automation are not available in LAB, but it's a small price to pay.
And, yes, Margulis' book is "the book" for LAB, I just happen to think it could have been written in half the number of pages.
-
LaB manipulations are just one more tool available to you. They aren't a magic bullet but they are sometimes useful. I recommend getting Dan Margulis's book Photohop Lab Color and playing around while realizing that his "schtick" is LaB color so he is a real evangelist. I try some simple LaB manipulation (steepening the a and b curves, usually) on perhaps 20% of my images and keep it on about 1/2 of those. As with most digital image manipulations, there is more than one way to get the same result, and a lot of course depends on the result you like.
Peter
Hi There,
Of course there is always Ian Lobb's Lobster.
Des
-
Lab is "just another useful tool", and should certainly in your list of things you know how to use and when it's useful. You can do some powerful colour effects with ab curves that you just can't easily achieve otherwise.
-
I use it on every single photograph I process. It's simple, accurate with colors, brings out color without jacking up the saturation, if you watch it, and allows you the ultimate ability to isolate and target your efforts.
When I read about Lab's colour accuracy because it represents colour (a,b ) separately from Luminosity (L), I wonder if that is true or Lab is just a math model where the concept of colour is not strictly linked to the real concept of colour.
To explain this I think of an object being shot twice, at different shutter speeds. Will the colour of the object change if we shoot it at 1/50 with respect to the colour of the object shot at let's say 1/100? (forget about any RAW clipping here)
The obvious answer is that the colour of the object will be the same, jut luminosity (exposure) will change. Right? otherwise we would admit our camera shutter speed can change colours!
Now shoot some plain colour surface twice, at two different shutter speeds. Develop the obtained RAW files with no processing at all (all parameters set to 0). Go to Photoshop with both images, and travel to Lab mode. Are colour a,b channels equal in both images?
Or simply plot a uniform colour patch in PS, look at its a and b values, overexpose it by 1EV. Did a, b remain constant?
BR
-
Now shoot some plain colour surface twice, at two different shutter speeds. Develop the obtained RAW files with no processing at all (all parameters set to 0). Go to Photoshop with both images, and travel to Lab mode. Are colour a,b channels equal in both images?
I think I understand what you are after, Guillermo. However, which raw converter will create a Lab output? If you start out with ProPhoto or whatever RGB, the result is "tainted".
Or simply plot a uniform colour patch in PS, look at its a and b values, overexpose it by 1EV. Did a, b remain constant?
Yes, they remain unchanged. However, this is not of much use: as soon as you convert it in RGB, you get the old picture.
One has to realize, that the RGB model is a primitive, temporary solution, which unfortunately became petrified, like IP and HTML. There is no way out short of a hardware revolution. Keep in mind, that the hardware is reproducing colors in RGB or CMYK format. Carry out your suggested experiment, not from raw but with filled color and switch between Lab and RGB with and w/o brightness adjustment: you get to see the same color in Lab and RGB.
-
I think I understand what you are after, Guillermo. However, which raw converter will create a Lab output? If you start out with ProPhoto or whatever RGB, the result is "tainted".
I was theorizing (is that word right?) about the assumed accurate colour modelling done in Lab mode. Didn't intend to find any practical conclusion to use, or not to use Lab, just point that it's simply a model.
Yes, they remain unchanged. However, this is not of much use: as soon as you convert it in RGB, you get the old picture.
This puzzles me. I have just set a aquare patch of colour RGB=(10,20,40). I am using a linear version of Adobe RGB so that I can follow the numbers more easily when changing exposure. Now I apply and exposure correction in Photoshop of +1EV.
The RGB number get duplicated as expected, but the corresponding a,b values in Lab change:
RGB=(10,20,40) Lab=(32,-9,15) ---> overexpose by +1EV ---> RGB=(20,40,60) Lab=(32,-12,19)
a,b changed and we only changed exposure. We simulated what we would have obtained with the camera by doubling exposure time, if a.b are only related to colour, why they should change?
Another example, closer to L=100:
RGB=(10,20,120) Lab=(36,-18,-61) ---> overexpose by +1EV ---> RGB=(20,40,240) Lab=(50,22,-77)
Here we were closer to L=100 in the first image, the Lab model had to adapt the a,b values suggesting a different destination colour, when the only action performed was the same as changing camera exposure.
With all this I don't mean Lab is wrong, is that it's simply a colour model that is not linked to real colours in a strictly differentiated Lightness <-> Colour fashion. Lightness (L) and colour (a,b ) are still linked even if we only changed the real luminosity of the object.
BR
-
Lab is "just another useful tool", and should certainly in your list of things you know how to use and when it's useful. You can do some powerful colour effects with ab curves that you just can't easily achieve otherwise.
Absolutely, it works wonders on some slighly foggy landscapes with a lot of greens in them. I have not found any other way to get the same level of color separation.
Cheers,
Bernard
-
The whole goal of Lab is that it's perceptually linear, and treats colour separately from brightness. Lab is not perfect, but it reasonably meets these goals.
-
The whole goal of Lab is that it's perceptually linear
Not exactly, and Guillermo is right - L*a*b is NOT perceptually uniform, so there might be certain color shifts. The most common example is well known "blue turns purple" problem.
-
Hi There,
Of course there is always Ian Lobb's Lobster.
Des
I'll second that - it's very elegant and works well. I've just got V2.1 which works with CS4, should available at www.freegamma.com be any day now.
-
That might be me, the later (not used) <g>
Dan Margulis is the guy who's put this on the map. While much of what he has to say about high bit editing and wide gamut spaces is just plain old silly, lets give him credit for being the guy who's taught Lab editing techniques from day one and is the person who one should investigate should they feel the need to use Lab corrections. Now if he'd only recognize the power of getting good data into Photoshop in the first place (either at the scan stage or more pertinent today, at Raw rendering), the need for all these Lab corrections would be reduced to nearly zero.
In these forums you often note Dan's propensity to use LAB for polishing turds. However, in the introduction to his LAB book, Dan states that the misconception that LAB is for bad images is shattered by the examples in the book. He notes that most of the pictures in the book were taken by professional photographers and that deficiencies in the originals are likely the fault of the environment rather than the photographer. He concludes by stating that where LAB shines is in adding believability to images that already are reasonably good. Since Dan does not participate in this forum (he apparently prefers his own forum where he is in control), I would invite forum members who have his book to comment on his assertions.
Bill
-
Not exactly, and Guillermo is right - L*a*b is NOT perceptually uniform, so there might be certain color shifts. The most common example is well known "blue turns purple" problem.
Absolutely! Its supposed to be perceptually uniform, that was its original goal when developed mathematically from CIEXYZ. It wasn't totally successful and has, like the deltaE calculations been attempted to be "fixed" over the years.
Aside from that, Lab has been oversold. And its been oversold for a long time. Here we have Bruce Fraser discussing this way back in 1999 (in only the way Bruce could do so):
Here's a simple experiment you can try that clearly demonstrates the
shortcomings of Lab, no matter how good your perceptual rendering is.
Make a square image, and make half of the background L*48 a*4 b*-77. Make
the other half of the background L*76 a*36 b*86.
Then, place a small square on each background color, and fill it with L*65
a*-105 b*69.
You'll find that the green square on the orange background looks darker
than the green square on the blue background. If you want to make the
greens match, you'll have to edit one of them -- no pixel-by-pixel
automatic conversion will do it. The greens are colorimetrically identical,
but they look different.
This is why I feel unable to trust any automatic conversion, whether it's
done in the RIP or elsewhere, and no matter what color space you use.
Lab may be device-independent, but so are the RGB working spaces in
Photoshop 5. Given the current limitations on implementation, I prefer to
use calibrated RGB, but even if those limitations were to disappear, you'd
still have to edit the image to make the greens match if you worked in Lab.
If you want to argue the superiority of LCH as an editing interface, that's
fine. I think people should edit in the interface with which they're most
comfortable.
Bruce
Let me make it clear that I'm not adamantly opposed to Lab workflows. If
they work for you, that's great, and you should continue to use them.
My concern is that Lab has been oversold, and that naive users attribute to
it an objective correctness that it does not deserve.
Even if we discount the issue of quantization errors going from device space
to Lab and vice versa, which could be solved by capturing some larger number
of bits than we commonly do now, (though probably more than 48 bits would be
required), it's important to realise that CIE colorimetry in general, and
Lab in particular, have significant limitations as tools for managing color
appearance, particularly in complex situations like photographic imagery.
CIE colorimetry is a reliable tool for predicting whether two given solid
colors will match when viewed in very precisely defined conditions. It is
not, and was never intended to be, a tool for predicting how those two
colors will actually appear to the observer. Rather, the express design goal
for CIELab was to provide a color space for the specification of color
differences. Anyone who has really compared color appearances under
controlled viewing conditions with delta-e values will tell you that it
works better in some areas of hue space than others.
When we deal with imagery, rather than matching plastics or paint swatches,
a whole host of perceptual phenomena come into play that Lab simply ignores.
Simultaneous contrast, for example, is a cluster of phenomena that cause the
same color under the same illuminant to appear differently depending on the
background color against which it is viewed. When we're working with
color-critical imagery like fashion or cosmetics, we have to address this
phenomenon if we want the image to produce the desired result -- a sale --
and Lab can't help us with that.
Lab assumes that hue and luminance can be treated separately -- it assumes
that hue can be specified by a wavelength of monochromatic light -- but
numerous experimental results indicate that this is not the case. For
example, Purdy's 1931 experiments indicate that to match the hue of 650nm
monochromatic light at a given luminance would require a 620nm light at
one-tenth of that luminance. Lab can't help us with that. (This phenomenon
is known as the Bezold-Brucke effect.)
Lab assumes that hue and chroma can be treated separately, but again,
numerous experimental results indicate that our perception of hue varies
with color purity. Mixing white light with a monochromatic light does not
produce a constant hue, but Lab assumes it does -- this is particularly
noticable in Lab modelling of blues, and is the source of the blue-purple
shift.
There are a whole slew of other perceptual effects that Lab ignores, but
that those of us who work with imagery have to grapple with every day if our
work is to produce the desired results.
So while Lab is useful for predicting the degree to which two sets of
tristimulus values will match under very precisely defined conditions that
never occur in natural images, it is not anywhere close to being an adequate
model of human color perception. It works reasonably well as a reference
space for colorimetrically defining device spaces, but as a space for image
editing, it has some important shortcomings.
One of the properties of LCH that you tout as an advantage -- that it avoids
hue shifts when changing lightness -- is actually at odds with the way our
eyes really work. Hues shift with both lightness and chroma in our
perception, but not in LCH.
None of this is to say that working in Lab or editing in LCH is inherently
bad. But given the many shortcomings of Lab, and given the limited bit depth
we generally have available, Lab is no better than, and in many cases can be
worse than, a colorimetrically-specified device space, or a colorimetrically
defined abstract space based on real or imaginary primaries.
For archival work, you will always want to preserve the original capture
data, along with the best definition you can muster of the space of the
device that did the capturing. Saving the data as Lab will inevitably
degrade it with any capture device that is currently available. For some
applications, the advantages of working in Lab, with or without an LCH
interface, will outweigh the disadvantages, but for a great many
applications, they will not. Any time you attempt to render image data on a
device, you need to perform a conversion, whether you're displaying Lab on
an RGB monitor, printing Lab to a CMYK press, displaying scanner RGB on an
RGB monitor, displaying CMYK on an RGB monitor, printing scanner RGB to a
CMYK press, etc.
Generally speaking, you'll need to do at least one conversion, from input
space to output space. If you use Lab, you need to do at least two
conversions, one from input space to Lab, one from Lab to output space. In
practice, we often end up doing two conversions anyway, because device
spaces have their own shortcomings as editing spaces since they're generally
non-linear.
The only real advantage Lab offers over tagged RGB is that you don't need to
send a profile with the image. (You do, however, need to know whether it's
D50 or D65 or some other illuminant, and you need to realise that Lab (LH)
isn't the same thing as Lab.) In some workflows, that may be a key
advantage. In many, though, it's a wash.
One thing is certain. When you work in tagged high-bit RGB, you know that
you're working with all the data your capture device could produce. When you
work in Lab, you know that you've already discarded some of that data.
Bruce
-bruce@pixelboyz.com
-
Not exactly, and Guillermo is right - L*a*b is NOT perceptually uniform, so there might be certain color shifts. The most common example is well known "blue turns purple" problem.
All perceptually uniform means that if you move a certain distance in ab it produces a similar amount of change in colour as if you move a similar distance in ab in a different colour region. There are no totally perceptually uniform colour spaces.... But lab comes close. Perceptual uniformity has nothing to do with colour shifts.
Graeme
-
In these forums you often note Dan's propensity to use LAB for polishing turds. However, in the introduction to his LAB book, Dan states that the misconception that LAB is for bad images is shattered by the examples in the book. He notes that most of the pictures in the book were taken by professional photographers and that deficiencies in the originals are likely the fault of the environment rather than the photographer. He concludes by stating that where LAB shines is in adding believability to images that already are reasonably good. Since Dan does not participate in this forum (he apparently prefers his own forum where he is in control), I would invite forum members who have his book to comment on his assertions.
First off, I've got Dan's book, its filled with turds and its kind of insulting to suggest there are professionally produced images in there. He may say the images were submitted by pros, and all of us have produced a turd here or there but this is hardly a book that provides examples of what pro's produce expect when experiencing equipment issues or brain farts. So yes, its still mostly about turd polishing IMHO no matter what he notes and you believe after reading the text and apparently ignoring the imagery.
Dan doesn't participate in this or any other forum I know of expect his so called Color Theory list. Nothing stops him from commenting expect speaking outside a highly moderated forum he controls. He can't ban anyone here from speaking about his theories in a light that doesn't generate attention to himself. And he's been invited here several times, most notably after Mark Segal did two articles on his "unique" ideas about color theory. Dan's a big boy, he can at least attempt to disseminate his views anywhere he wishes. For reasons that appear obvious to me and others, he had decided to remain behind protective walls and speak to his minions who eat up his various flat earth ideas about image processing (in this case, I'm referring to wide gamut, and high bit color theory). That he's so close minded and protective makes it far more difficult to take all his ideas seriously, even when they have merit as they do in some, note some cases, with Lab editing. But again, most of what he attempts to "fix" shouldn't be broken in the first place. Its a lot easier to produce good Raw data and render it appropriately than do, as Dan has suggested over the years, zero out all Raw converter settings and fix what now is a turd in Photoshop using Lab or any other color model. GIGO:Garbage In Garbage Out.
-
I think the techniques illustrated early in Dan's LAB book can be very useful for a lot images, especially nature and landscape. I end up making a trip to LAB for most of my color landscape images. A/B curves are still the best way to get color separation that I've found, the PS/ACR 'Vibrance' controls are just not in the same league. That said, the more advanced techniques in later portions of the book definitely fall into the category of turd-polishing IMHO.
-
And its been oversold for a long time. Here we have Bruce Fraser discussing this way back in 1999 (in only the way Bruce could do so)
Thanks a lot, Andrew, that's very interesting (as usual)
Perceptual uniformity has nothing to do with colour shifts.
Maybe I miss something... So what has to do with blue-purple, or red-orange shifts?
http://www.brucelindbloom.com/MunsellCalcH...html#BluePurple (http://www.brucelindbloom.com/MunsellCalcHelp.html#BluePurple)
A/B curves are still the best way to get color separation that I've found, the PS/ACR 'Vibrance' controls are just not in the same league.
Vibrance control is something different. You can get a/b curves color separation effect with HSL sliders, plus they're more intuitive, fast and convinient.
-
Maybe I miss something... So what has to do with blue-purple, or red-orange shifts?
http://www.brucelindbloom.com/MunsellCalcH...html#BluePurple (http://www.brucelindbloom.com/MunsellCalcHelp.html#BluePurple)
I don't think you're missing anything. Both Bruce's explain the issues well.
Vibrance control is something different. You can get a/b curves color separation effect with HSL sliders, plus they're more intuitive, fast and convinient.
And non destructive, parametric, being applied in an optimal processing order (in Adobe converters), in high bit, from the highest gamut data etc. Its interesting how old school some of the Photoshop processing techniques are for actual pro photographers capturing Raw data and using decent Raw converters. Not as sexy or controversial I guess as moves into and out of Lab.
-
Maybe I miss something... So what has to do with blue-purple, or red-orange shifts?
http://www.brucelindbloom.com/MunsellCalcH...html#BluePurple (http://www.brucelindbloom.com/MunsellCalcHelp.html#BluePurple)
Ok, I see what you're getting at now. In that case, Bruce is saying that if you treat Lab as perceptually uniform, when it is not, you'll get errors when you're using that space for gamut mapping.
Graeme
-
Ok, I see what you're getting at now. In that case, Bruce is saying that if you treat Lab as perceptually uniform, when it is not, you'll get errors when you're using that space for gamut mapping.
Graeme
Gamut mapping, huh?
Let's say I'm editing such a picture:
(http://members.chello.pl/m.kaluza/cielab1.jpg)
...and I want to make it lighter:
(http://members.chello.pl/m.kaluza/cielab2.jpg)
Very convinient editing space, indeed.
-
Gamut mapping, huh?
Let's say I'm editing such a picture:
(http://members.chello.pl/m.kaluza/cielab1.jpg)
...and I want to make it lighter:
(http://members.chello.pl/m.kaluza/cielab2.jpg)
Very convinient editing space, indeed.
The images that illustrate what you're saying don't work here - can you fix / repost in some way so that I can understand?
Graeme
-
The images that illustrate what you're saying don't work here - can you fix / repost in some way so that I can understand?
Graeme
Pardon, it's fixed
-
Pardon, it's fixed
Thanks. I see your point now!
Graeme
-
CIE L*a*b* was designed to measure color differences. The original Delta E metric (DE 1976), which is just the Euclidean distance, has many shortcomings which is why the CIE subsequently proposed DE 1994 and more recently DE 2000 to deal with the non-uniformities in the L*a*b* math. As Bruce Lindbloom explains on his site with examples, there are cases when moving along a straight line of constant hue angle in L*a*b* causes noticeable perceived hue shifts. This is not to say it can't be used successfully for image processing, but rather that it was not the original intent; furthermore, there have been more recent models that are better suited for image processing if "perceptual uniformity" or consistency is desired.
In some cases, image processing in a perceptually-uniform space makes sense, and other cases where it really doesn't. For example, some image processing operations assume (or work best on) linear light in which case using native camera RGB, GMCY, or some linear transformation thereof is the right thing to do.
-
Or you can open your image in RGB, add a Levels adjustment layer and slide the middle slider to like 2.00, or add a Brightness/Contrast adjustment layer and dial up brightness as desired, or even a curves layer and bump the middle -- the key is to change the blend mode of that adjustment layer to luminosity so it only affects the uh, luminosity... What's really cool is you end up with an image that is lighter like your Lab smaple, yet the red and blue still look like red and blue and not orange and lavender.
Lab has it's uses, but frankly I have never needed it for editing images.
Cheers,
-
You have to take care when using curves in luminosity mode, bright saturated colors and deep saturated colors get clipped easily. I'm generally sticking with the good ol' normal mode, and saturation mode if I want to change saturation for hi-mid-shadow tones separately. And to stay on topic, I'm not using lab apart from rare occasions I need a or b channel for masking.
-
the key is to change the blend mode of that adjustment layer to luminosity so it only affects the uh, luminosity...
But again Luminosity blend mode is just another model. In particular the well known HSB colour model (http://en.wikipedia.org/wiki/HSL_color_space) already mentioned in the thread and derived from the RGB numbers of the colour profile being used.
In Luminosity blend mode, Hue (H in the HSB model) is preserved but Saturation (S in HSB) strongly changes in a non perceptually uniform way. For example a S curve strongly desaturates the deep shadows and highlights.
As digitaldog suggested, there is not only one way to do things and there is not the perfect tool. Or we'd rather say the perfect tool is that one that helps us to reach our goal in image edition as simply as possible.
BR
-
As part of my standard workflow (in RGB color space), I use two curves layer adjustments, an S curve for contrast with the blending mode set to "Luminosity" and steepened straight line with the blending mode set to "Color" to boost color some. However, I've always wondered what are the math formulae behind these blending modes -- what specific modifications to color space coordinates are they implementing? Is this documented or explained anywhere?