Luminance in LAB vs. RGB?

What adjustments in RGB give the same effect as the L channel in LAB?

I believe that is what is called a false premise.

I'm not very well versed in LAB, I'll admit. But I had always "assumed" that anything perfomed in one color space could be performed in another color space, although the difficulty of doing so might be increased. Is this not the case?

Try luminosity in blend mode in RGB. Ie use a curve layer with the blend mode set to luminosity.

I think that is what your trying to acheive.

Regards

Paul

Like the poster above said, use blend mode luminosity. It's close but, to the extent I understand it, not exactly the same thing. Probably close enough so that it doesn't matter.

This is probably more than you want to know, but here's some more info...

In RGB, a curve with a blend mode of "Luminosity" is close to the same as manipulating an L channel curve in LAB mode, not exactly the same, but similar. If you are using an RGB composite curve to adjust the brightness or contrast of your image, I would strongly suggest you put the curve in Luminosity blend mode as that will let your curve change brightness without changing color (within reason).

It is not exactly the same though because LAB mode maps tones differently than RGB. The L channel is calculated by using a weighted average of the R, G and B channels and then shifting the tones upwards. This gives the L channel more fidelity for adjustments in the shadow areas and less fidelity in the highlights. This can be useful because our eyes see detail in shadows much more than we see detail in highlights so it gives finer grain control in the shadows. As an example, you can more finely adjust the shadow slider in the shadow/highlights tool on the L channel than you can you the RGB composite channel.

Another difference between manipulating the L curve in LAB mode and an RGB curve is that LAB mode has extremely bright colors and extremely dark colors. In other words, a color can maintain it's color, but still be very bright or very dark in LAB mode. RGB mode cannot do that. By definition a very, very bright color must approach white in RGB since the brightest you can be is 255,255,255 and that's white. And, by definition a very dark color in RGB must approach black since 0,0,0 is black. In LAB mode, color is completely separate from luminosity so you can have extremely bright or dark colors that retain their color (100,128,0) is a very, very bright color in LAB. Those colors may or may not be printable (depending on the printer) , but they can be useful at intermediate stages during editing and some can display on screen.

Also, it's useful to sharpen the L channel instead of the RGB composite channel because the sharpening will ONLY affect luminosity and will not cause color shifts. Some people simulate this in RGB mode by sharpening the RGB composite image and then either setting the blend mode of the layer to Luminosity or using Edit/Fade to Luminosity right after the sharpening operation.

--
John
Gallery: http://jfriend.smugmug.com/portfolio

Smugmug discount coupon '3elo1xh75JSiI' to save $5

Where did you learn what you know about L*a*b mode?

Rdefen said:

Where did you learn what you know about L*a*b mode?

Click to expand...

Two main places:

Read, study and practice and discuss this book:
Photoshop LAB color by Dan Margulis

http://www.amazon.com/gp/product/0321356780/qid=1138336289/sr=8-2/ref=sr_8_xs_ap_i2_xgl14/104-2919342-3344722?n=507846&s=books&v=glance

And read this summary of the book online:
http://www.dgrin.com/showthread.php?t=18203

I've learned more from the above book than I have from the other 14 books I own on Photoshop. You not only learn about LAB mode, but you also learn a lot about channels, curves, blend modes, blend-if settings, masking, color correction, contrast enhancement, color enhancement, etc..., all things that are useful in RGB and CMYK too.

--
John
Gallery: http://jfriend.smugmug.com/portfolio

Smugmug discount coupon '3elo1xh75JSiI' to save $5

Another ploy is to use a luminosity selection (the 'claw') - alt+shft+cntl+ tilde on US keyboards - or control click on the composite rgb channel in channels palette - it selects everything on a basis of 50% gray. An inversion of the selection, of course, selects the other side.

Pretty much has been confirmed that fading in luminosity after sharpening does the same as sharpening in L.A.B.

That said, there really seems to be some color tones in LAB that would take some tricks in RGB to recreate. To my eye, there is a near pastel or leaning toward pastel tone in LAB - much like some of Maxfield Parrish's work, that can be achieved rather easily with curves in LAB.
--
Kent

http://www.pbase.com/kentc
For prior discussions on most questions:
http://porg.4t.com/KentC.html
or d/l 'archives' at:
http://www.atncentral.com

jfriend00 said:

Read, study and practice and discuss this book:
Photoshop LAB color by Dan Margulis

http://www.amazon.com/gp/product/0321356780/qid=1138336289/sr=8-2/ref=sr_8_xs_ap_i2_xgl14/104-2919342-3344722?n=507846&s=books&v=glance

And read this summary of the book online:
http://www.dgrin.com/showthread.php?t=18203

Click to expand...

If you're not sure about buying this book, try to read at least the first part of it in a bookstore, and read the threads on Digital Grin. If nothing else, you'll still have picked up on some useful techniques. But Photoshop LAB Color will give you a far deeper understanding with lots of examples. Unlike jfriend, it's my only Photoshop book, but it was easily worthwhile, and has taught me a lot more than just LAB (Blend If sliders are one of Photoshop's great, hidden features).

--
Patrick O'Leary
http://patrick.greentaperacing.us/blog/gallery/

Yeh. I've been through it once and plan to do so again. I think others would say some of what he teaches is controversial. It is interesting though.
Thanks.

You had me believing for a moment because at least you didn't say it "loads the highlights" which Scott Kelby incorrectly state in his books. But then you said "it selects everything on a basis of 50% gray." Huh? What does that mean?

What it does is load the RGB composite grayscale as a selection. The same as if you control clicked an individual color channel would load that channel as a selection.

Result being pure white is 100% selected. Pure black 0% selected and everything else selected in proportion to it's grayscale value. The marching ants is virtually useless for displaying the selection because the selection is severely feathered because of all the gray. A trip to quick mask will show the actual selection.

That's all it does. It's very simple and very straightforward and yet constantly and consistently misstated. The only guy I've seen state it correctly is Ben Wilmore. Kelby dumbs it down like he dumbs everything down. Dear Scott: as you well know, it does not load the highlights. And also saying Matt Koslowski rocking the houseky over and over IS NOT FUNNY. It's annoying. Stop trying to do stand-up. Stand-up is possibly the world's most difficult job. That's why Dave Chappelle gets $50 million and you don't.

Rdefen said:

I think
others would say some of what he teaches is controversial. It is
interesting though.

Click to expand...

I've never heard of anything in Margulis' LAB book referred to as "controversial". What do you think people say is controversial in what he teaches?

--
John
Gallery: http://jfriend.smugmug.com/portfolio

Smugmug discount coupon '3elo1xh75JSiI' to save $5

From memory, doesn't he make a big deal about the difference between working in L and working in RGB and using blend mode L ?

Also, doesn't he basically say (for pages) that working in 16-bit mode is no improvement over 8-bit?

I think both of these statements are controversial. In particular, almost no one subscribes to the second statement about bit depth.
I'm sure there are others but I'd have to go back and review the book.

Rdefen said:

You had me believing for a moment because at least you didn't say
it "loads the highlights" which Scott Kelby incorrectly state in
his books. But then you said "it selects everything on a basis of
50% gray." Huh? What does that mean?

Click to expand...

Ok you caught me.

For the OP here is one application:

http://www.webreference.com/graphics/ps2/5.html

Rdefen said:

What it does is load the RGB composite grayscale as a selection.
The same as if you control clicked an individual color channel
would load that channel as a selection.
Result being pure white is 100% selected. Pure black 0% selected
and everything else selected in proportion to it's grayscale value.
The marching ants is virtually useless for displaying the selection
because the selection is severely feathered because of all the
gray. A trip to quick mask will show the actual selection.

Click to expand...

Thanks for the full explanation.....

Rdefen said:

That's all it does. It's very simple and very straightforward and
yet constantly and consistently misstated. The only guy I've seen
state it correctly is Ben Wilmore. Kelby dumbs it down like he
dumbs everything down. Dear Scott: as you well know, it does not
load the highlights. And also saying Matt Koslowski rocking the
houseky over and over IS NOT FUNNY. It's annoying. Stop trying to
do stand-up. Stand-up is possibly the world's most difficult job.
That's why Dave Chappelle gets $50 million and you don't.

Click to expand...

Could have left this part out....
--
Kent

http://www.pbase.com/kentc
For prior discussions on most questions:
http://porg.4t.com/KentC.html
or d/l 'archives' at:
http://www.atncentral.com

See
http://www.brucelindbloom.com/index.html?DanMargulis.html

jfriend00 said:

What do you think people say is controversial in
what he teaches?

--
John

Click to expand...

Rdefen said:

From memory, doesn't he make a big deal about the difference
between working in L and working in RGB and using blend mode L ?

Click to expand...

With good reason, as he clearly demonstrates.

Luminosity isn't the same in RGB as Lightness in LAB. Adjustments to the master RGB curve will still result in a red shift to the shadows, compared to an equivalent move to the lightness channel. LAB holds detail better as well. Particularly with skin, which is simply a series of comples gradients. Lightness moves are smoother than RGB luminosity moves, because LAB handles the gradient transitions far better. Test it and see.

Hermie said:

See
http://www.brucelindbloom.com/index.html?DanMargulis.html

Click to expand...

I don't think that the 8-bit vs. 16-bit issue has anything to do with the Photoshop LAB color book or what he teaches in it. In all the time I've been studying it, discussing it and even writing chapter summaries of it, it's never come up as a controversial topic in this book. Make your own choice for how to work, then get on with learning how to best use the techiques the book offers.

Since I personally shoot RAW and have more than enough memory in my Photoshop computer, I choose to work in 16-bit mode for as long as possible. I know the math and figure that it can't hurt me, but it could probably be shown that millions of images that are more impressive than anything I have produced were done in 8-bit mode so clearly it can't be the dominant factor. Beyond that, I don't find it productive to debate that topic.

--
John
Gallery: http://jfriend.smugmug.com/portfolio

Smugmug discount coupon '3elo1xh75JSiI' to save $5

Thanks for all of the info. The reason I was asking is on a much shallower level than where this discussion went.

I use GIMP, and as far as I know there is no luminosity blend mode. However, while playing with the surfer picture posted earlier on this board, I tried a script called "LAB Color Punch" by lasm. I was just playing around, and I decided to set the blue/yellow and green/magenta sliders to "hold" (no modification), and just saturate the L channel. I really liked the effect this had on the water and waves, and I was curious how this effect could be achieved in RGB.

Okay. I haven't used the GIMP in a while, so I don't know what the equivalent blend mode is, but you can accomplish something very similar (and perhaps better) in RGB by duplicating the image, making your curves changes while ignoring color shifts, then using a Luminance blend of the duplicate image over the original (to get the right colors back). I'm explaining this horribly because I haven't tried it yet; the technique is covered in Margulis' Chapter 14 (in Photoshop LAB Color). If you're interested, you might hit just this chapter in the bookstore.

--
Patrick O'Leary
http://patrick.greentaperacing.us/blog/gallery/