Margulis: Photoshop LAB book

On Tue, 30 Aug 2005 12:03:30 -0500, Tom Nelson
wrote:

Is anyone else reading Dan Margulis’ new book? It’s amazing, world-changing, but I think my head is going to explode.
Tom Nelson
Tom Nelson Photography

Disclaimer: No, I don’t have any connection with Mr. Margulis or his publisher.
========================================
my ISP does not have an "x" in its name
========================================

And the title is…?

—

Hecate – The Real One

Fashion: Buying things you don’t need, with money
you don’t have, to impress people you don’t like…

"Hecate" wrote in message

On Tue, 30 Aug 2005 12:03:30 -0500, Tom Nelson
wrote:

Is anyone else reading Dan Margulis’ new book? It’s amazing, world-changing, but I think my head is going to explode.
Tom Nelson
Tom Nelson Photography

Disclaimer: No, I don’t have any connection with Mr. Margulis or his publisher.
========================================
my ISP does not have an "x" in its name
========================================

And the title is…?

—

Hecate – The Real One

Fashion: Buying things you don’t need, with money
you don’t have, to impress people you don’t like…

Actually … I’m surprised that you are not reading his Color Theory newslist. Go to

http://groups.yahoo.com/group/colortheory/

From the announcement on that list, made on Friday, August 12, 2005 at 11:20 AM:

= = = = = Begin Quote = = = = =

It is a pleasure (and a relief) to announce that the official publication date of my new book on LAB was Monday, August 8. The title is Photoshop LAB Color: The Canyon Conundrum and other adventures in the most powerful colorspace (ISBN 0-321-35678-0). It’s 384 pages in an 8×10 format. Price is $54.95. It’s the first entirely new title I’ve published in eight years.

PDFs of the Table of Contents and Chapters 2 and 9 are posted at www.ledet.com/margulis/articles.htm

The book is currently available at www.peachpit.com if you scroll through the
authors list–it has not yet entered the search database. It should be available imminently at amazon.com, and will be shown at Photoshop World in Boston.

Handled with care, LAB maneuvers are extraordinarily powerful, often achieving effects that are not even possible AFAIK in other colorspaces. LAB is not,
however, particularly user-friendly. I set out to write something that would be
at least somewhat accessible to non-advanced users, while at the same time catering to professional-level retouchers and color specialists.

Therefore, each of the first six chapters is cut in half. The first halves are quite gentle IMHO. All commands are spelled out. The basics of an LAB workflow are developed. While there is some general discussion of under what
circumstances you would want to use LAB or to avoid it, in these first halves I
simply state that certain LAB methods work better than their RGB equivalents and
leave it at that. The idea is to offer something that will give the inexperienced user striking results immediately, and a sharp improvement in color quality.

In the second halves (and in the final ten chapters of the book) the discussion gets more technical and there are many comparisons showing when LAB works
better (or worse) than other alternatives.

Some top experts have seen drafts and offered mildly positive comments. David
Biedny, the principal author of Photoshop Channel Chops and one of the great
retouching authorities, contributed a foreword in which he called the book "the most deeply advanced, inspiring, insightful, maddening, awesome, demanding,
and illuminating educational effort–in any media format–ever created for Photoshop."

Scott Kelby, who sells more books on Photoshop than anybody else, also had a
read. His comment: "This book is going to radically change how we all do color correction from this point on. Anyone not using the techniques Dan unveils
in this book will soon be a digital dinosaur. It’s that revolutionary."

More down-to-earth feedback came from a dedicated group of seven beta readers
selected from this list, a group of very diverse backgrounds and skill levels. They caught a slew of unclear areas and offered many useful suggestions. If
you like the book, you owe thanks, as I do, to Les De Moss, Andre Dumas, Bruce
Fellman, Timo Kirves, Katia Lazarova, Clarence Maslowski, and Clyde McConnell.

Also, I’d like to thank the many list members who offered images for inclusion. Particularly, I’d like to thank those who, after discussing what I was
after, were kind enough to send me a selection of *several* images. They are David
Barr, Jim Bean, Michael Benford, Hunter Clarkson, Mike Demyan, Fred Drury, Jason Hadlock, Mark Laurie, David Leaser, Mike Russell, Marty Stock, Lee Varis,
and Michael Vlietstra. I was looking for pictures that would best illustrate
the potential of LAB, and it was critical that there be a good variety to choose from. Therefore, not every one of these people is represented in the book.
However, each one of them made a significant contribution just by presenting
alternatives, and the book is better because of them.

Here’s a rundown of the contents.

1. THE CANYON CONUNDRUM
The basic LAB correction method is explained through a series of images of canyons.

2. LAB BY THE NUMBERS
How LAB is structured, what the numbers mean, how they interrelate with one another to create colors, and how they relate to the human visual system.

3. VARY THE COLOR, VARY THE RECIPE
The basic recipe of Chapter 1–an overall color enhancement–can be approximated in RGB, although the results will not be as good. When the A and B curves
run at different angles, though, the result is unique to LAB.

4. IT’S ALL ABOUT THE CENTER POINT
Using LAB to eliminate color casts.

5. SHARPEN THE L, BLUR THE AB
LAB often, but not always, has a significant sharpening advantage over RGB, even when the RGB sharpening is done in Luminosity mode. When an image needs to
be blurred, LAB is much better than RGB/Color mode. This chapter explains not
just how but why, covering a lot of ground that is not well understood.

6. ENTERING THE FOREST: MYTHS & DANGERS.
Some avoid the use of LAB based on myths, which are debunked here. On the other hand, certain features of LAB are in fact rather dangerous if the user is
not careful.

7. SUMMING UP: LAB AND THE WORKFLOW
The first half comes to a close with a discussion of when and why to use LAB,
a question to which different users will come up with different answers. Only
those who are extremely pressed for time would want an all-LAB workflow. Everyone else needs a disciplined approach to when to use it.

8. THE IMAGINARY COLOR, THE IMPOSSIBLE RETOUCH
LAB permits us to designate (at least theoretically) colors that could not possibly exist, such as a brilliantly red black. Doing so doesn’t sound particularly intelligent, but using imaginary colors can be an extraordinarily
powerful retouching tool.

9. THE LAB ADVANTAGE IN SELECTIONS AND MASKING
The best masks usually use a single channel as a base, but few people think of using the A or B for that purpose. In fact, those channels can make selections appear out of thin air–and in one spectacular example, they make a
selection *of* thin air.

10. THE PRODUCT IS RED BUT THE CLIENT WANTS GREEN
The most effective way of making major color changes away from the art, as when the photograph portrays a product in one color but the client specifies not
just another color but gives a PMS number to match.

11. THE BEST RETOUCHING SPACE
David Biedny, who is one of the world’s most skilled retouchers, calls this chapter "nothing short of astounding."

12. COMMAND, CLICK, CONTROL
A chapter on advanced LAB curving that was one of the favorites of the beta readers.

13. THE UNIVERSAL INTERCHANGE STANDARD
We take a break from Photoshop technique to discuss LAB’s role in the exchange of documents from one colorspace to another, and also see how the difficulties of setting up a conversion of out LAB suggest solutions for other kinds of
color-matching issues, such as making CMYK matches to Pantone colors.

14. ONCE FOR COLOR, ONCE FOR CONTRAST
Four examples, one each of curves and of blends in Luminosity and Color modes. When should they be done in LAB, and when in RGB?

15. BLENDING WITH THE A AND B
The most difficult chapter of all considers blending the A and B channels into each other and/or the L for gains in contrast and color intensity.

16. A FACE IS LIKE A CANYON
We end the adventure with an easy, yet spectacular recipe for improving the believability of face shots, illlustrated step-by-step with five individuals of
various ages and ethnicities.

This book took a lot more effort than I really would have liked. Unlike Professional Photoshop, every chapter was completely new. Also, because so many of
the techniques are bleeding-edge, they haven’t been studied much, and I was learning as I went along. I am pleased with the results and think that almost
everyone will find some powerful new tools and significant improvements for their own workflow.

As previously announced, I plan a new edition of Professional Photoshop in 2006. It will certainly vary considerably from the current edition, but a lot
will depend on what the reaction to this LAB book is.

Dan Margulis

= = = = = End Quote = = = = =

Note: The most recent posting on the Colortheory newslist is

"… For those following such things, the publisher interrupted my vacation today with the news that as of this morning the LAB book was the #1 selling amazon.com title in the entire computer/internet field. …"

Tom Nelson wrote:

Is anyone else reading Dan Margulis’ new book? It’s amazing, world-changing, but I think my head is going to explode.
Tom Nelson
Tom Nelson Photography

Disclaimer: No, I don’t have any connection with Mr. Margulis or his publisher.
========================================
my ISP does not have an "x" in its name
========================================

Thanks for the hint, I’ve ordered it immediately.

Peter

Tom Nelson writes …

Is anyone else reading Dan Margulis’ new book? It’s amazing, world-changing, but I think my head is going to explode.

Before your head explodes could you post two images, one edited the "digital dinosaur" way using your pre-revelation techniques and the other edited with Dan’s new LAB techniques that shows how superior those techniques are?

Bill

Bill Hilton wrote:

Tom Nelson writes …

Is anyone else reading Dan Margulis’ new book? It’s amazing, world-changing, but I think my head is going to explode.

Before your head explodes could you post two images, one edited the "digital dinosaur" way using your pre-revelation techniques and the other edited with Dan’s new LAB techniques that shows how superior those techniques are?

Yes, Please! I haven’t even reached much dinosaur status, if any, and don’t want to learn anything that may now be obsolete.

—
John McWilliams

Remember: Opinions are like buttocks; only those which are well-formed should be shown in public.

Here ya go:
http://www.tnphoto.com/lab/

In article ,
Bill Hilton wrote:

Tom Nelson writes …

Is anyone else reading Dan Margulis’ new book? It’s amazing, world-changing, but I think my head is going to explode.

Before your head explodes could you post two images, one edited the "digital dinosaur" way using your pre-revelation techniques and the other edited with Dan’s new LAB techniques that shows how superior those techniques are?

Bill

On Wed, 31 Aug 2005 04:36:45 GMT, "RSD99"
wrote:

Actually … I’m surprised that you are not reading his Color Theory newslist. Go to

http://groups.yahoo.com/group/colortheory/

Thanks. I tend to dislike online sources, but if it’s him it must be good, right? 🙂

From the announcement on that list, made on Friday, August 12, 2005 at 11:20 AM:

<snip>

The only thing that bothers me about that is Scott Kelby likes it, but I’m sure it’s a good book despite that. 🙂

It looks interesting. I’ll have to see if it’s been published over here yet.

Thanks for the info 🙂

—

Hecate – The Real One

Fashion: Buying things you don’t need, with money
you don’t have, to impress people you don’t like…

On Wed, 31 Aug 2005 14:22:57 -0500, Tom Nelson
wrote:

Here ya go:
http://www.tnphoto.com/lab/

Excellent.

I rarely use solely RGB anyway, well, not since Professional Photoshop by DM came out anyway. This looks like a refinement of some of what he was saying there.

—

Hecate – The Real One

Fashion: Buying things you don’t need, with money
you don’t have, to impress people you don’t like…

Margulis wrote about LAB long before his new book. Check out these articles and two chapters from the book.

http://www.ledet.com/margulis/articles.html

Margulis’ old book covers topics and techniques not found elsewhere. I do have a few problems with his writings:

– for the uninitiated, it is difficult to separate what is applicable for press printing and what is applicable for a desktop inkjet.

– the channel blending chapters skip the detail steps, making it impossible to follow for those not familiar with blending.

– his writing style.

Having said that, the real problem is that once I figure out what he is really trying to say, I find myself redoing my "completed" edits. <g>

Bill Hilton wrote:

Tom Nelson writes …

Is anyone else reading Dan Margulis’ new book? It’s amazing, world-changing, but I think my head is going to explode.

Before your head explodes could you post two images, one edited the "digital dinosaur" way using your pre-revelation techniques and the other edited with Dan’s new LAB techniques that shows how superior those techniques are?

Bill

Tom Nelson wrote:

Is anyone else reading Dan Margulis’ new book? It’s amazing, world-changing, but I think my head is going to explode.

In reality the Lab space should not be used for image editing purposes at all. A very gentle editing operation can easily throw Lab colors out of the common RGB working-spaces (and even out of the visible gamut of human vision).

Here is an animated gif showing Lab codes that are valid (in-gamut) for adobeRGB(1998):
http://www.aim-dtp.net/aim/temp/lab-in-adobergb1998.gif
It shows the (a*,b*) points at each L* plane from 1 to 100. Notice for example how very few Lab codes are valid in the dark end of the range, how irregular this mapping is, and how the majority of the valid (a*,b*) points move from segment IV to segment II as the L* increases.

I btw created this data using the AIM.XLA, it is freeware Excel Add-In for spectral and colorimetric work, available at
http://www.aim-dtp.net/aim/technology/aim_xla/index.htm

Timo Autiokari
http://www.aim-dtp.net

Timo,

‘A very gentle editing operation can easily throw Lab
colors out of the common RGB working-spaces (and
even out of the visible gamut of human vision). ‘

Dan Margulis knows this. It’s already explained in
Professional Photoshop 7 .

Best regards –Gernot Hoffmann

Timo Autiokari wrote:

Tom Nelson wrote:

Is anyone else reading Dan Margulis’ new book? It’s amazing, world-changing, but I think my head is going to explode.

In reality the Lab space should not be used for image editing purposes at all. A very gentle editing operation can easily throw Lab colors out of the common RGB working-spaces (and even out of the visible gamut of human vision).

I think you meant that LAB has a wider gamut than RGB. If editing in LAB results in out of gamut colors in RGB, is it possible to detect that’s happening to an image in Photoshop? If the out of gamut colors are invisible to human vision, they will not show on the screen in LAB space, or after conversion in the RGB space. I don’t think that they will be flagged as out of gamut warnings either.

Here is an animated gif showing Lab codes that are valid (in-gamut) for adobeRGB(1998):
http://www.aim-dtp.net/aim/temp/lab-in-adobergb1998.gif
It shows the (a*,b*) points at each L* plane from 1 to 100. Notice for example how very few Lab codes are valid in the dark end of the range, how irregular this mapping is, and how the majority of the valid (a*,b*) points move from segment IV to segment II as the L* increases.
I btw created this data using the AIM.XLA, it is freeware Excel Add-In for spectral and colorimetric work, available at
http://www.aim-dtp.net/aim/technology/aim_xla/index.htm

Timo Autiokari
http://www.aim-dtp.net

wrote:

I think you meant that LAB has a wider gamut than RGB. If editing in LAB results in out of gamut colors in RGB, is it possible to detect that’s happening to an image in Photoshop? If the out of gamut colors are invisible to human vision, they will not show on the screen in LAB space, or after conversion in the RGB space. I don’t think that they will be flagged as out of gamut warnings either.

Of course they will be visible on screen. What would an ‘invisible color’ look like on screen anyway? Remember that you are never really looking at a LAB image, or a ProPhotoRGB image, or even an AdobeRGB image if you are using the average computer monitor. You are looking at an image that is already converted to your monitor’s color space.

—
Johan W. Elzenga johan<<at>>johanfoto.nl Editor / Photographer http://www.johanfoto.nl/

wrote:

I think you meant that LAB has a wider gamut than RGB.

Actually no, we all do know very well that Lab is larger than the common RGB working-spaces.

I meant that the Lab has a _horrendously_ large "gamut", it is skewed and even highly non-unambiguous.

I’m sure you are familiar with the nice and neat graphical illustrations of the Labs space, like it is explained and shown in the Photoshop user guide or online help (search: L*a*b model). Well, these illustrations are totally and absolutely wrong, incorrect, misleading misinformation.

Here is an animated GIF that shows how the Photoshop Lab space maps to the CIE Yxy. In this animation L* goes from 0 to 100 by stepping of 1 unit so there are 101 L* planes in the animation.
http://www.aim-dtp.net/aim/temp/lab-l-planes.gif
(it is a bit large file so please be patient)

The green "horseshoe" in this animation is called as the CIE Spectral Locus, in other words it is the gamut of the human vision. All the common RGB color-spaces are rectangles inside this "horseshoe". Of course gamuts are 3 dimensional volumes, such 2 dimensional rectangles show the maximum attainable saturation that a gamut can have, such saturation is available only at very low luminance levels, so the gamut rectangles gets smaller and smaller as the luminance increases.

Ok, back to the Lab to Yxy animation, the red dots represent the conversion of the Lab values using the full range of a* and b* codes (-127 …+128) at the given L* value. As you can see, there are major trouble at the dark end of the range where the L* is 1 to about 30. Huge amount of the codes are mapped _way_ out of the CIE locus and non-unambiguously (there are two "sets" of mapping) one covering the whole CIE locus and a horseload above and over it and the other at the bottom (totally an imaginary set). When such codes are
produced/created by an image editing operation the CMS system has an impossible task in mapping them "correctly" since from those Lab codes it is not possible to say at all what kind of color it should represent, they are just completely imaginary values with no real information. (out-of gamut RGB always has real information about the hue and luminance)

If editing in LAB results in out of gamut colors in RGB, is it possible to detect that’s happening to an image in Photoshop?

Yes, by using the gamut warning tool.

If the out of gamut colors are invisible to human vision, they will not show on the screen in LAB space, or after conversion in the RGB space.

Out-of-gamut colors are "mapped" (re-adjusted) by the color management system (either AdobeACE or MIcrosoftICM "engines") in a way or another so that they will fit to the color space they are being converted to. These two CMS engines appear to do simple channel based clipping in RGB (with Lab to RGB conversions). This kind of "mapping" introduce very difficult problems the subsequent processing of the image data.

Timo Autiokari
http://www.aim-dtp.net

In article , Timo Autiokari
wrote:

In reality the Lab space should not be used for image editing purposes at all. A very gentle editing operation can easily throw Lab colors out of the common RGB working-spaces (and even out of the visible gamut of human vision).

Dan Margulis is very forthright in talking about LAB’s ability to produce colors that are not only outside the printable, displayable or visible gamuts, they’re beyond our ability to imagine (a brilliantly yellow black, for instance). He also mentions (Chapter 8) how imaginary colors can produce real-world advantages as Photoshop attempts to reconcile their inconsistencies.

Remember that the Info palette shows readouts in two colorspaces. If you’re in LAB you can set one to "actual color" (whatever colorspace you’re in) and one to RGB or CMYK. So you can check the conversion numbers while still in LAB. Gamut Warning works in LAB, too.

Margulis makes a convincing case that LAB has substantial advantages for image correction and enhancement. It’s powerful, but that’s a good thing.

Tom Nelson
Tom Nelson Photography

Tom Nelson wrote:

Dan Margulis is very forthright

Yes, I know…

in talking about LAB’s ability to produce colors that are not only outside the printable, displayable or visible gamuts, they’re beyond our ability to imagine (a brilliantly
yellow black, for instance).

Well, it is not an _ability_ of the Lab space but a very major fault of it.

He also mentions (Chapter 8) how imaginary colors can
produce real-world advantages as Photoshop attempts to
reconcile their inconsistencies.

So… it is then the Photoshop who does the art or masterpiece, not the user of Photoshop. I do not like to work like that.

Remember that the Info palette shows readouts in two colorspaces. If you’re in LAB you can set one to "actual color" (whatever colorspace you’re in) and one to RGB or CMYK. So you can check the conversion numbers while still in LAB. Gamut Warning works in LAB, too.

Yes that is so. But we are not able to check the values of every pixel from the info palette and the gamut warning is only approximate.

Margulis makes a convincing case that LAB has substantial advantages for image correction and enhancement. It’s powerful, but that’s a good thing.

In my opinion there are no benefits from editing in the Lab space. In the past there was one procedure that was useful, the reduction of the color noise (by blurring the a* and b* channels a little) but now there are far better tools for noise reduction (that work in the RGB mode).

Timo Autiokari http://www.aim-dtp.net/

Timo Autiokari wrote:

Margulis makes a convincing case that LAB has substantial advantages for image correction and enhancement. It’s powerful, but that’s a good thing.

In my opinion there are no benefits from editing in the Lab space. In the past there was one procedure that was useful, the reduction of the color noise (by blurring the a* and b* channels a little) but now there are far better tools for noise reduction (that work in the RGB mode).

Have you actually *read* the book at all? If you have read it and THEN you make this statement, I find it interesting and would like to know where Margulis goes wrong in your opinion.

If you haven’t read it and still make this statement, it sounds a lot like the infamous ‘not invented here’ syndrome to me.

—
Johan W. Elzenga johan<<at>>johanfoto.nl Editor / Photographer http://www.johanfoto.nl/

"Timo Autiokari" wrote in message

Tom Nelson wrote:

Dan Margulis is very forthright

Yes, I know…

in talking about LAB’s ability to produce colors that are not only outside the printable, displayable or visible gamuts, they’re beyond our ability to imagine (a brilliantly
yellow black, for instance).

Well, it is not an _ability_ of the Lab space but a very major fault of it.

He also mentions (Chapter 8) how imaginary colors can
produce real-world advantages as Photoshop attempts to
reconcile their inconsistencies.

So… it is then the Photoshop who does the art or masterpiece, not the user of Photoshop. I do not like to work like that.

Remember that the Info palette shows readouts in two colorspaces. If you’re in LAB you can set one to "actual color" (whatever colorspace you’re in) and one to RGB or CMYK. So you can check the conversion numbers while still in LAB. Gamut Warning works in LAB, too.

Yes that is so. But we are not able to check the values of every pixel from the info palette and the gamut warning is only approximate.

Margulis makes a convincing case that LAB has substantial advantages for image correction and enhancement. It’s powerful, but that’s a good thing.

In my opinion there are no benefits from editing in the Lab space. In the past there was one procedure that was useful, the reduction of the color noise (by blurring the a* and b* channels a little) but now there are far better tools for noise reduction (that work in the RGB mode).

Timo Autiokari http://www.aim-dtp.net/

You posted:
"… In my opinion there are no benefits from editing in the Lab space. …."

You are entitled to your opinion.

But please keep in mind that the rest of us here … many of which *have* edited images in the L*a*b* mode … are also entitled to our opinion(s). And we think that working in the L*a*b* mode adds some capabilities that are not available in any other color "mode."

On Fri, 9 Sep 2005 22:44:30 +0200, (Johan W.
Elzenga) wrote:

Timo Autiokari wrote:

Margulis makes a convincing case that LAB has substantial advantages for image correction and enhancement. It’s powerful, but that’s a good thing.

In my opinion there are no benefits from editing in the Lab space. In the past there was one procedure that was useful, the reduction of the color noise (by blurring the a* and b* channels a little) but now there are far better tools for noise reduction (that work in the RGB mode).

Have you actually *read* the book at all? If you have read it and THEN you make this statement, I find it interesting and would like to know where Margulis goes wrong in your opinion.

If you haven’t read it and still make this statement, it sounds a lot like the infamous ‘not invented here’ syndrome to me.

Actually, it’s the "everyone is wrong but Timo" syndrome.

—

Hecate – The Real One

Fashion: Buying things you don’t need, with money
you don’t have, to impress people you don’t like…

Timo Autiokari wrote:

wrote:

[snip]

Your technical explanations are way over my head. Pragmatically, if LAB oog is the only issue, all I care is the following.

If editing in LAB results in out of gamut colors in RGB, is it possible to detect that’s happening to an image in Photoshop?

Yes, by using the gamut warning tool.

Is this the way to detect gamut warning when in LAB mode? Covert/edit in LAB, View|Gamut_Warning, View|Proof_Setup|Custom|AdobeRGB1998. This should flag LAB oog against the AdobeRGB1998 working space.

Unlike others, I don’t mind a different view, whether it is based on "facts" or opinion. In this case, perhaps a better approach is to state your view and provide a pragmatic way to demonstrate it, such as using gamut warning. If the LAB users don’t notice any gamut warnings, they can conclude that their edits are fine in LAB. If they do notice gamut warnings, they may conclude something is not right.

This way, you make your point and those who care to listen end up with a way of dealing with it.

On Sun, 11 Sep 2005 13:38:46 GMT, wrote:

if LAB oog is the only issue

out-of-gamut is not at all the only issue where Lab has poor performance.

For example, Photoshop does not provide any Lab specific editing tools. Instead the very same editing algorithms that are used for RGB data are used also over Lab data, just that those of the operations and filters that would result in a totally weird appearance are simply grayed out.

Or, examine for example the color Lab=0,120,0 or the color Lab = 0,0,-120. Both naturally are absolutely black since L=0, in the reality that is.

Is this the way to detect gamut warning when in LAB mode? Covert/edit in LAB,
View|Gamut_Warning, View|Proof_Setup|Custom|AdobeRGB1998. This should flag LAB oog against the AdobeRGB1998 working space.

Yes it is.

better approach is to state your view and provide a pragmatic way to demonstrate it

I already posted one, here is a much better version of it: http://www.aim-dtp.net/aim/technology/aim_xla/lab2rgb-ani.gi f

If the LAB users don’t notice any gamut warnings, they
can conclude that their edits are fine in LAB. If they do notice gamut warnings, they may conclude something is not right.

No. For example Lab space editing _very easily_ changes hues incorrectly, e.g. in-gamut orange can turn into in-gamut yellow or in-gamut red, or blue can turn into in-gamut cyan or in-gamut violet, etc.

In all cases, if one really wants to work in Lab, one has to go there from the 16-bit/c mode. Try this:

1) Open any 8-bit/c RGB image
2) Duplicate it
3) Convert the duplicate to Lab
4) Convert it then back to the RGB space of the original
5) Then drag it as a layer on top of the original
6) Put the layer into difference mode
7) Flatten the image
7) See the Histogram, take per channel views

there are error up to 20 level units. Notice that this is difference view, the errors are not there in the deep dark end in the actual images.

Timo Autiokari

In article , Timo Autiokari
wrote:

On Sun, 11 Sep 2005 13:38:46 GMT, wrote:

if LAB oog is the only issue

out-of-gamut is not at all the only issue where Lab has poor performance.

For example, Photoshop does not provide any Lab specific editing tools. Instead the very same editing algorithms that are used for RGB data are used also over Lab data, just that those of the operations and filters that would result in a totally weird appearance are simply grayed out.

Photoshop doesn’t provide separate RGB-specific or CMYK-specific editing tools but I don’t see that that’s a problem…

Or, examine for example the color Lab=0,120,0 or the color Lab = 0,0,-120. Both naturally are absolutely black since L=0, in the reality that is.

Right, Don’s book contains an illustration of LAB 5, -50, 0 against 5,0,8 (a brilliant green black against a brilliant neutral black). The green is clearly visible. The primary advantage of such shenanigans (a minor one, IMO) is that you can force a color into a burned-out highlight.

better approach is to state your view and provide a pragmatic way to demonstrate it

I already posted one, here is a much better version of it: http://www.aim-dtp.net/aim/technology/aim_xla/lab2rgb-ani.gi f

Your link demonstates that LAB is a huge colorspace. Like any wide-gamut colorspace it must be used wisely. But it can be used.

In all cases, if one really wants to work in Lab, one has to go there from the 16-bit/c mode. Try this:

1) Open any 8-bit/c RGB image
2) Duplicate it
3) Convert the duplicate to Lab
4) Convert it then back to the RGB space of the original
5) Then drag it as a layer on top of the original
6) Put the layer into difference mode
7) Flatten the image
7) See the Histogram, take per channel views

there are error up to 20 level units. Notice that this is difference view, the errors are not there in the deep dark end in the actual images.

I got a pure black image (i.e., no difference between the layers).

Tom Nelson
Tom Nelson Photography

Timo Autiokari wrote:

On Sun, 11 Sep 2005 13:38:46 GMT, wrote:

if LAB oog is the only issue

out-of-gamut is not at all the only issue where Lab has poor performance.

But that seems to you the primary reason you gave (your words below) for not using LAB.

"In reality the Lab space should not be used for image editing purposes at all. A very gentle editing operation can easily throw Lab colors out of the common RGB working-spaces (and even out of the visible gamut of human vision)."

For example, Photoshop does not provide any Lab specific editing tools. Instead the very same editing algorithms that are used for RGB data are used also over Lab data, just that those of the operations and filters that would result in a totally weird appearance are simply grayed out.

That may be a shortcoming on Photoshop’s part, but not an objective reason to claim that "the Lab space should not be used for image editing purposes
at all".

[snip]

Is this the way to detect gamut warning when in LAB mode? Covert/edit in LAB,
View|Gamut_Warning, View|Proof_Setup|Custom|AdobeRGB1998. This should flag LAB oog against the AdobeRGB1998 working space.

Yes it is.

Thanks for confirming.

[snip]

If the LAB users don’t notice any gamut warnings, they
can conclude that their edits are fine in LAB. If they do notice gamut warnings, they may conclude something is not right.

No. For example Lab space editing _very easily_ changes hues incorrectly, e.g. in-gamut orange can turn into in-gamut yellow or in-gamut red, or blue can turn into in-gamut cyan or in-gamut violet, etc.

But such changes are visible on the monitor and in print. If the LAB users can notice these and find them problematic, they will tune their edits to remove them. Failing that, they will stop using LAB. If they don’t notice the differences, who cares?

[snip]

I learn and use new tools and techniques to achieve improvements in my image edits. If I can observe the improvements, I don’t care much about how the tools and techniques work behind the scene. If I can’t see any improvements, I’ll use something else. Bottomline: it is the subjective evaluation that counts, not the objective technical babbles.

On Tue, 13 Sep 2005 12:19:55 GMT, wrote:

I learn and use new tools and techniques to achieve improvements in my image edits. If I can observe the improvements, I don’t care much about how the tools and techniques work behind the scene. If I can’t see any improvements, I’ll use something else. Bottomline: it is the subjective evaluation that counts, not the objective technical babbles.

Exactly. The only relevant questions are "Does it work?" and "Does improve on what I already do?"

—

Hecate – The Real One

Fashion: Buying things you don’t need, with money
you don’t have, to impress people you don’t like…

"Hecate" wrote in message

On Tue, 13 Sep 2005 12:19:55 GMT, wrote:

I learn and use new tools and techniques to achieve improvements in my image edits. If I can observe the improvements, I don’t care much about how the tools and techniques work behind the scene. If I can’t see any improvements, I’ll use something else. Bottomline: it is the subjective evaluation that counts, not the objective technical babbles.

Exactly. The only relevant questions are "Does it work?" and "Does improve on what I already do?"

The problem with that though, is that what may work under certain conditions may fail under (slightly) different conditions. As an analogy, it may be perfectly riskless to stick-shift from 4th gear to 1st when standing still, it would cause disaster if you did that when driving 60 miles per hour or more.
A slightly more fundamental understanding of what goes on "under-the-hood" may prevent unpleasant surprises (which alway present themselves at the worst moment).

The same applies to editing in Lab mode, with proper precaution (and some prudent rendering intent towards output) trouble can be avoided/reduced. Just being aware of the risks may already help some.

Bart

Bart van der Wolf wrote:

"Hecate" wrote in message

On Tue, 13 Sep 2005 12:19:55 GMT, wrote:

I learn and use new tools and techniques to achieve improvements in my image edits. If I can observe the improvements, I don’t care much about how the tools and techniques work behind the scene. If I can’t see any improvements, I’ll use something else. Bottomline: it is the subjective evaluation that counts, not the objective technical babbles.

Exactly. The only relevant questions are "Does it work?" and "Does improve on what I already do?"

The problem with that though, is that what may work under certain conditions may fail under (slightly) different conditions. As an analogy, it may be perfectly riskless to stick-shift from 4th gear to 1st when standing still, it would cause disaster if you did that when driving 60 miles per hour or more.

I fully agree. This is demonstrated in most of the PS books and tutorials. The example images used by the authors seem to be hand picked to suit a particular tool or technique. The authors typically will fail to point that out. It is a rude awakening to a reader when other images do not work as well.

A slightly more fundamental understanding of what goes on "under-the-hood" may prevent unpleasant surprises (which alway present themselves at the worst moment).

The same applies to editing in Lab mode, with proper precaution (and some prudent rendering intent towards output) trouble can be avoided/reduced. Just being aware of the risks may already help some.

Again, I fully agree. But Adobe does not go out its way to explain "what goes on "under-the-hood"" with each PS tool, nor do most of the PS book authors. We are left at the mercy of generous "experts" to share their wisdom, which may or may not be valid. Even if all these info are available, it will be rather daunting to absorb them all.

Whatever tool or technique I use, I always keep an eye out on the basic infos provided by PS, such as the histograms, the color values, and the gamut warnings, etc. When these seem to be out of the ordinary, it becomes a warning sign. (Hence my suggestion to check oog when using LAB, assuming that oog is that big a problem.) Very often, I find myself trying out different tools to see which is most suited for a situation. Not the best approach, but that’s all I can do.

Since the end results of edits are visual images, the most valuable tool is our visual evaluation of what are on the screen and in prints. Unfortunately, critical evaluation is easier said than done. Even with the best calibrated equipments, there are many problems in the evaluation process, especially for those who are not trained (including myself), such as:

– extreme highlight/shadow and netual are easy to evaluate by the numbers, but colors are much more difficult
– the evaluation is subjective
– our eyes can lie to us
– we change our mind after a day
etc.

There are plenty of books on PS tools, but what we really need is an education on critical visual evaluation of screen and print images, followed by an official explanation of what is going on with each tool.

On Tue, 13 Sep 2005 12:19:55 GMT, wrote:

But that seems to you the primary reason you gave (your words below) for not using LAB.

Yes. The (horrible) Lab gamut is _one_ of the _main_ reason why not using Lab. But it has many other problems also.

But such changes are visible on the monitor and in print.

No. Such changes/errors sure are _displayed_ by the monitor. But the user _very easily_ adapts to the poor quality, he/she can not see/notice all errors.

Here is a demonstration about adaptation of the vision to gradually decreasing quality, please go to
http://www.aim-dtp.net/aim/evaluation/gamma_error/processing _space.htm the page is about another matter so scroll down to the end of the page, there is the test image set and some option button.

When you arrive there the option button:
"jpeg copy of the original"
is selected.

1) Now, click the option button:
2) "Linear editing"
3) and look at (inspect) the image about 5 seconds or more.

4) Then click the option button:
5) "Gamma-space 1.2 editing"
6) and look at (inspect) the image about 5 seconds or more, inspect _carefully_ the differences/changes that you discern at the time the image changes.

Continue the steps 4, 5 and 6 with the rest of the 1.4, 1.6, 1.8, 2.0,
2.2 cases. Each time carefully inspect the changes that you see in the
image after each change of the image.

Now, finally you have the "Gamma-space 2.2 editing" version shown and you have not detected much of changes, usually people describe this along the lines "the noise seem to change a little at some of the steps".

Now then… click the "Linear editing" option button and …Lo and Behold.

Bottomline: it is the subjective evaluation that counts, not the objective technical babbles.

The real world bottomline is that in case one is not skillful enough to turn the images into masterpieces in a well behaving RGB color-space then one _certainly_ will not be able to do that in the very badly behaving Lab space, instead the resulst will be much worse.

And the corollary from that:

In case one is able to create masterpieces in a well behaving RGB space then there is absolutely no need to use the horribly behaving Lab space.

Timo Autiokari

Tom Nelson wrote:

Your link demonstates that LAB is a huge colorspace.

Huge does not describe the situation properly.

This animation shows Lab L* planes on CIE Yxy chart.
http://www.aim-dtp.net/aim/temp//lab-l-planes.gif
The green horseshoe is the Spectral Locus, chromaticities that are out of that Spectral Locus are totally incorrect, un-usable. As can be seen the Lab is just horrible under L* < 30.

Like any wide-gamut colorspace it must be used wisely.

Common RGB spaces are all inside the Spectral Locus, in that sense they have no wasted codes nor imaginary color codes like the Lab space has.

But it can be used.

Yes, like a Trabant can be used as a car.

I got a pure black image (i.e., no difference between the layers).

The instruction was not to evaluate the difference image visually but to go to the Histogram (or Levels) dialog and to see the situation on the individual channels. In a non-linear color-space such as adobeRGB it is quite difficult to see the light that RGB codes at about level 20 output since they are severly dark, and this was not the subject matter, you are not supposed to see the color of these _difference codes_. Maybe the meaning of the difference layer mode is not clear to you, each channel value of each pixel location is calculated between the layers 1 and 2 as: R=abs(R1-R2), G=abs(G1-G2), B=abs(B1-B2). So neither of the colors that have some difference are not necessarily near black like their difference is. An error of 20 level is quite severe for such colors that are visible.

Timo Autiokari

On Wed, 14 Sep 2005 13:25:56 +0200, "Bart van der Wolf" wrote:

Exactly. The only relevant questions are "Does it work?" and "Does improve on what I already do?"

The problem with that though, is that what may work under certain conditions may fail under (slightly) different conditions. As an analogy, it may be perfectly riskless to stick-shift from 4th gear to 1st when standing still, it would cause disaster if you did that when driving 60 miles per hour or more.
A slightly more fundamental understanding of what goes on "under-the-hood" may prevent unpleasant surprises (which alway present themselves at the worst moment).

You misunderstand. I wasn’t saying don’t bother to investigate and understand what is going on – to take your analogy – I want to know how to drive correctly, but I don’t need to know precisely how a specific gear box works to do so – just the basic principles involved. 🙂

The same applies to editing in Lab mode, with proper precaution (and some prudent rendering intent towards output) trouble can be avoided/reduced. Just being aware of the risks may already help some.

I agree completely. I think that you should be aware of the risks in whatever you do – it helps you to know, for example, when to break the rules and when not to 🙂

—

Hecate – The Real One

Fashion: Buying things you don’t need, with money
you don’t have, to impress people you don’t like…

"Tom Nelson" wrote in message

Is anyone else reading Dan Margulis’ new book? It’s amazing, world-changing, but I think my head is going to explode.

There you go, Tom, it’s a dynamite book.

Seriously though, don’t expect to get it all in one pass. Read each chapter once through, and count on reviewing them later. Dan’s writing is entertaining, but very dense with new ideas and material.

Some of the Lab ideas, particularly those in the first two chapters on increasing color and contrast, as well as some material on skin tones in Lab, are covered in the beginner’s Curvemeister class. The class is free and may be taken using the Curvemeister demo:
http://www.curvemeister.com/support/class/
—

Mike Russell
www.curvemeister.com

On Fri, 16 Sep 2005 22:53:41 GMT, "Mike Russell" wrote:

"Tom Nelson" wrote in message

Is anyone else reading Dan Margulis’ new book? It’s amazing, world-changing, but I think my head is going to explode.

There you go, Tom, it’s a dynamite book.

Seriously though, don’t expect to get it all in one pass. Read each chapter once through, and count on reviewing them later. Dan’s writing is entertaining, but very dense with new ideas and material.

Very true – I read Professional Photoshop at least 3 times. The book is out here on the 30th and I’ve reserved my copy 🙂

—

Hecate – The Real One

Fashion: Buying things you don’t need, with money
you don’t have, to impress people you don’t like…