Dan Margulis has promoted LAB for quite a very long time. His books typically present it within them to varying degree. This one is specifically targeted to LAB. There are other versions, etc. ... but, the principles remain the same.
I used to use LAB mode quite a lot. Not so much these days, but I still employ the principles, and monitor my color picker with LAB as the secondary display.
Beside being a power (i.e. curve exponential) way to adjust the color independent of the luminosity ... it offers a way to retain a neutral balance when you start raising things, if you don't want to impart color cast, just the increase.
In order to ensure that I keep things neutral, I'd first set a "pivot point" in my curve at the mid-point. If you don't do that, your channels will impart a shift. Then I could move the channels around with more liberty, and not skew things as readily.
I got away from using LAB and mostly use RGB, coupled with blending modes these days. I may take a lap back into LAB and see how it functions with the latest iteration in PS. It's been a while since I actively used LAB. Will see how it plays again. Thanks for sharing the vid.
Raw Therapee has a large group of LAB transformations: the L, A, and B curves; sliders for lightness, contrast, and chromaticity; and curves for CC, LC, CL, and more.
Chazz5 wrote:
Raw Therapee has a large group of LAB transformations: the L, A, and B curves; sliders for lightness, contrast, and chromaticity; and curves for CC, LC, CL, and more.
Ummm, and its a free editor, (thanks) not sure what you mean by "and curves for CC, LC, CL" & more
The question is not RGB vs L*a*b. It's not either or. L*a*b is very powerful but it's also quite coarse so on images where L*a*b will benefit you you'll invariably need to make further corrections and fine tunings in RGB or CMYK with RGB being less coarse than L*a*b and CMYK less coarse than RGB but only if you need to be going to CMYK for your final output.
There is no better book on L*a*b Color than Dan's first book on L*a*b called The Canyon Conundrum. Where quite a few books on Photoshop typically have a paragraph or two - if that - on L*a*b color, Dan wrote an entire and most excellent book about the subject. I've read quite a few book on Ps in the last quarter century and the ONLY one where bells were constantly going off was that first L*a*b book. Simply the finest book on Photoshop color correction that there is. I go back and re-read it every few years. Maybe it's time again.
Peter Figen wrote:
The question is not RGB vs L*a*b. It's not either or. L*a*b is very powerful but it's also quite coarse so on images where L*a*b will benefit you you'll invariably need to make further corrections and fine tunings in RGB or CMYK with RGB being less coarse than L*a*b and CMYK less coarse than RGB but only if you need to be going to CMYK for your final output.
There is no better book on L*a*b Color than Dan's first book on L*a*b called The Canyon Conundrum. Where quite a few books on Photoshop typically have a paragraph or two - if that - on L*a*b color, Dan wrote an entire and most excellent book about the subject. I've read quite a few book on Ps in the last quarter century and the ONLY one where bells were constantly going off was that first L*a*b book. Simply the finest book on Photoshop color correction that there is. I go back and re-read it every few years. Maybe it's time again....Show more →
Yup, that's the book (linked above).
Good points about the fact that it's "power" can also be a bit coarse. I tend to think of it like the game of "Mother, may I?" with its Giant Steps and Baby steps. When you need GIANT STEPS ... LAB is worth the venture.
Dan ran a weekly editing exercise for about 3 months earlier this year via a groups.io site he has. Only about 40 or so participants ... but, the point is that Dan is still active and as passionate about color and color corrections as ever. The exercises were structured to be successively more challenging and he was promoting the use of blending tools (following different LAB renderings). Kind of his way of bringing the coarseness into further refinement, I reckon.
Tha ham-handed manner in the video above is exactly the kind of thing you want to avoid while using LAB mode. Great tool, but has to be used with a light touch.
Rajan Parrikar wrote:
Tha ham-handed manner in the video above is exactly the kind of thing you want to avoid while using LAB mode. Great tool, but has to be used with a light touch.
Well, I had to go and actually watch the damned video Rajan. The problem of course is that the narrator never explained WHY he was doing what he was doing in regards to the Curves moves and why knowing the why is so important. Ugh. Anyone who reads Margulis will instantly know the answers and recognize how a video tutorial (if you can call it that) is likely to cause as much confusion as it might help.
For the record, when I'm working in L*a*b, I always duplicate the file and convert that copy to L*a*b, do what needs to be done, flatten it and then drag it back on top of the RGB where the whole thing can be blended, masked or whatever to your heart's content. What you often find when you're working in L*a*b is that you tend to overcook the image so having that extra bit of control on the back end is often what's needed to keep you from having to re-do the whole shebang. Only on occasion will I saved the layered L*a*b file. Actually only once in recent memory when I wasn't sure how the whole thing was going to convert to CMYK and thought I might have to revisit the whole correction.
For the record, I did post those examples a couple of years ago and using L*a*b was the only way to change the color of the night sky in the JPL images I was using, where the entire nebula just didn't flat out disappear, and the challenge was getting a deep, saturated, intense, blue to convert to CMYK without the whole thing turning to purple mush.
Rajan Parrikar wrote:
Tha ham-handed manner in the video above is exactly the kind of thing you want to avoid while using LAB mode. Great tool, but has to be used with a light touch.
Never really heard heavy handed as ham handed. I guess different parts of the world use different terminology, ex America car hood/trunk, England car bonnet/boot etc. etc.
Peter Figen wrote:
Well, I had to go and actually watch the damned video Rajan. The problem of course is that the narrator never explained WHY he was doing what he was doing in regards to the Curves moves and why knowing the why is so important. Ugh. Anyone who reads Margulis will instantly know the answers and recognize how a video tutorial (if you can call it that) is likely to cause as much confusion as it might help.
For the record, when I'm working in L*a*b, I always duplicate the file and convert that copy to L*a*b, do what needs to be done, flatten it and then drag it back on top of the RGB where the whole thing can be blended, masked or whatever to your heart's content. What you often find when you're working in L*a*b is that you tend to overcook the image so having that extra bit of control on the back end is often what's needed to keep you from having to re-do the whole shebang. Only on occasion will I saved the layered L*a*b file. Actually only once in recent memory when I wasn't sure how the whole thing was going to convert to CMYK and thought I might have to revisit the whole correction.
For the record, I did post those examples a couple of years ago and using L*a*b was the only way to change the color of the night sky in the JPL images I was using, where the entire nebula just didn't flat out disappear, and the challenge was getting a deep, saturated, intense, blue to convert to CMYK without the whole thing turning to purple mush. ...Show more →
I think the why in the case of the video was for color adjustments in A & B, vs using using Hue/Saturation He showed how flat the color was in RGB adjusting Hue/Saturation. vs LAB towards the end of video.
Then he made 2 curves in L 1 for contrast on Mtn, and 1 for brightness of trees and then adjusted each using the opacity slider, to control any heavy handed work with those adjustments
I do like your idea of keeping the original file RGB, and copy into LAB so you have leverage to adjust as needed in the end.
Glad to have stimulated another read of Dans book. I have a lot to learn about LAB for other ways of using it for editing.
IndyFab wrote:
Never really heard heavy handed as ham handed. I guess different parts of the world use different terminology, ex America car hood/trunk, England car bonnet/boot etc. etc.
IndyFab wrote:
I think the why in the case of the video was for color adjustments in A & B, vs using using Hue/Saturation He showed how flat the color was in RGB adjusting Hue/Saturation. vs LAB towards the end of video.
Then he made 2 curves in L 1 for contrast on Mtn, and 1 for brightness of trees and then adjusted each using the opacity slider, to control any heavy handed work with those adjustments
I do like your idea of keeping the original file RGB, and copy into LAB so you have leverage to adjust as needed in the end.
Glad to have stimulated another read of Dans book. I have a lot to learn about LAB for other ways of using it for editing....Show more →
There was still no talk of WHY there's more separation of color in L*a*b when steepening the "a" and "b" curves, only that there is a difference. That's what I'm talking about. And, the reason, of course, is that by steepening the "a" and "b" curves, you're essentially moving all the colors further apart from one another as they're becoming more saturated, so there's not as much mashing together and loss of tonal separation.
And again, nothing at all was mentioned of the importance of rotating those "a" and "b" curves around the neutral center point in order to not impart a huge color cast on the image, even though that's just what he did. That is one of the most important things to know - y'know - so you know when to play by the rules and when to not. And, along the same line, it'd be nice to mention just what the "a" and "b" channels actually ARE - again, y'know - yellow vs. blue and then magenta vs. green - and Dan has an entire section explaining just that with examples and the why. Those are the things you need to understand in order to move beyond simple rotational movements around the center axis of the curve. Unfortunately, without that foundational information all you get is semi useful information without any deeper understanding. Maybe that's enough for some.
I agree with the statement that its not Lab vs RGB. It isjust another tool from which to choose.
Up front it is important to note that in my opinion you want to be in 16 bit mode if you are going to jump to Lab or the color components (a and b channels) can be posterized (the video RGB was 16 bit mode).
Its also not hard to create out of gamut colors in Lab Space which may or may not be of concern (or a feature) as when jumping back to RGB space the Color Management will render it back into the RGB gamut.
Tool of choice has a lot to do with experience and which one gives them them best result in their own opinion.
A curves adjustment Layer can exactly duplicate many other Adjustment Layers yet it usually takes a bit more experience as the adjustments can be sensitive. Not so different for some work in Lab Space as well.
Ones productivity with a given tool is pretty important to me as well if I can get the same results. For me, my go to tool is Adobe Camera Raw filter. For me it is quite powerful and quite intuitive on how to get results.
I note that the video goes from 16 bit RGB to 16 bit lab, several curve Adjustment Layers, mask added to each, then adjusting opacity (to bring it back to reality?), and then back to ACR for fine tuning.
I took the same starting image and just applied a few sliders in the basic panel of ACR (no lab involved) and got a pretty decent result. That did not include any adjustment brush or any other tuning.
For me that was easy peasy for even a basic result done in 20 seconds and no masking. I include the image of the ACR basic panel. That was even without trying to put much energy into it such as using the vast variety of other tools in ACR.
I do use Lab yet my goto is ACR filter as a first stop. Just my opinion
It has been over decade since I read Dan Margulis' classic. But I remember clearly his suggestion to not use (or exercise great caution) the LAB mode on images that are colour saturated to begin with.
Peter Figen wrote:
There was still no talk of WHY there's more separation of color in L*a*b when steepening the "a" and "b" curves, only that there is a difference. That's what I'm talking about. And, the reason, of course, is that by steepening the "a" and "b" curves, you're essentially moving all the colors further apart from one another as they're becoming more saturated, so there's not as much mashing together and loss of tonal separation.
And again, nothing at all was mentioned of the importance of rotating those "a" and "b" curves around the neutral center point in order to not impart a huge color cast on the image, even though that's just what he did. That is one of the most important things to know - y'know - so you know when to play by the rules and when to not. And, along the same line, it'd be nice to mention just what the "a" and "b" channels actually ARE - again, y'know - yellow vs. blue and then magenta vs. green - and Dan has an entire section explaining just that with examples and the why. Those are the things you need to understand in order to move beyond simple rotational movements around the center axis of the curve. Unfortunately, without that foundational information all you get is semi useful information without any deeper understanding. Maybe that's enough for some.
Thanks for the explanation Peter
2 take away's
1) "the importance of rotating those "a" and "b" curves around the neutral center point in order to not impart a huge color cast on the image"
2) "nice to mention just what the "a" and "b" channels actually ARE - again, y'know - yellow vs. blue and then magenta vs. green"
Gimp quite recently implemented 16 bits/channel and it allows more and repeated operations without banding and other artifacts of lost precision. I waited a looooong time for deep pixels. Perceptual transformations go back at least to 1948 (before computers). I need to see if Gimp can now handle L*a*b* in an effective way.
IndyFab wrote:
I think the why in the case of the video was for color adjustments in A & B, vs using using Hue/Saturation He showed how flat the color was in RGB adjusting Hue/Saturation. vs LAB towards the end of video.
Then he made 2 curves in L 1 for contrast on Mtn, and 1 for brightness of trees and then adjusted each using the opacity slider, to control any heavy handed work with those adjustments
I do like your idea of keeping the original file RGB, and copy into LAB so you have leverage to adjust as needed in the end.
Glad to have stimulated another read of Dans book. I have a lot to learn about LAB for other ways of using it for editing....Show more →
If you really want to learn about using LAB, I'd highly recommend Dan's book (linked above) ... subtitled as Peter mentioned, that most folks recognize it as "Canyon Conundrum". I saw that there's a V.2, not sure about Dan's latest book, if it has elements of a V.3 or not. But, once you "get it" ... you "got it".
Then, after that ... you might reach out to Dan himself via his groups.io. If you join his group, you'll be able to read through the exercises that are posted with all manner of comment by both members of the group and Dan's critique / comment as well. I've not talked (groups.io / email) with Dan for a couple months now, so I'm not sure what his "next move" is ... but he's still in the game, even if he's slowin' down a bit.
