You may want ot check out Dan MArgulis’ book on working in Lab mode.
Personally I never colour correct in CMYK if I can help it. I always do all my work in RGB then convert to the CMYK profile needed (always keeping a master file in RGB) If the need arises, I will do minor tweaks in CMYK.
I find correcting in CMYK is much harder than in RGB
(BTW haven’t read Dan’s Book yet, but I’ve heard it’s got interesting stuff)
Correcting in Lab is definitely not for sissies. Lab offers corrections that are not possible in any other color space, but Lab corrections are generally coarser than those made in RGB or CMYK. Lab often has the opportunity to pull things from an image that no other method can, but you will almost always have to fine tune it afterward in either RGB or CMYK.
Whereas most books on Photoshop might have a single paragraph or possibly an entire chapter on the Lab color model, Margulis has devoted an entire book to it, and for me, it is an invaluabel resource that I constantly go back to, as there is far too much to absorb in even several readings.
What Margulis will emphasize is that no single color space is perfect for all corrections and proficiency in one will only help the others. He also goes to great length to help the reader clearly identify those images which will clearly benefit from Lab corrections and which are best left to more traditional spaces.
Peter Benedetti,
Color correcting in Lab is hardly amateur stuff. Just the opposite. Apparently, many people find it hard to think in Lab color mode, but it’s not hard at all. In Lab, images have a "red-green" axis (a) and a "yellow-blue" axis (b).
A color cannot have both redness and greeness. It is either-or (or neutral).
A color cannot have yellowness and blueness. It is either-or (or neutral).
But a red color or a green color can have a yellow or a blue attribute. A yellow color or blue color can have a red or a green attribute.
It helps to look at a stylized model of a slice through the Lab color space to understand that colors are mutually exclusive red-green or mutually exclusive yellow-blue.
It’s very useful to be able to vary the brightness of an image (Luminosity) without affecting color at all. Or the opposite. You can’t do that in RGB or CMYK. And it’s possible to control exactly how you want particular color ranges to behave, while leaving everything else alone.
One of the problems, though, in Lab is that small adjustments can result in huge changes that can quickly get way out of control. But once you understand the model, you can effect just about any color change via Lab. Changes that are impossible in CMYK.
While I can think easily in Lab, like most others, I do most of my editing in RGB, relying on the color I see on my calibrated screen (with other visual references on screen and off-screen).
I can’t "think" at all in RGB. It makes no sense to me whatever, to try and predict the result of changing any one value as to the real visual result. Give me a trio of RGB numbers and I cannot "see" the color in my head. I just rely on what I see on screen and let the RGB values become what they may. But, I can "see" the right color when given CMYK or Lab numbers.
"Thinking" in CMYK is very easy, as is editing in that space. I know, for instance what I need to make a sky, or grass, or gray look correct on one press vs another, etc.. I hardly ever need to make such changes, however, as I rely on ICC profiles to get the color right going from RGB to CMYK.
A BIG problem, editing in CMYK, is that such files are generally 8 bit, and any color compression/expansion is going to result in data loss and posterizing.
I stay in 16 bit as long as possible especially if converting in and out of Lab and RGB.
Margulis has invaluable advice about both CMYK and Lab editing. But other things he has to say about bit depth and image quality, the printing process itself and digital files are the confused rantings of a madman.
Wow! Peter & Reed said a mouthful!
I used to edit in Lab. After I found ProPhoto RGB, I started doing most of my editing with it. After reading Peter’s & Reed’s posts, I’m embarrassed to say I have forgotten the benefits of editing in Lab. Time to consider adding that back to my workflow. Especially for stubborn images.
Lab did come quite easily to me. Try a few dozen images when you have some leisure time and see for yourself if you like it!
Leisure time. LOL! 🙂
First of all thank you for all your help,
I just wanted to clarify one thing…., when I am speaking of color correction I am refering to corrections for print which for those of you who are familiar with prepress know that when we are talking about ink on paper and color correcting with that in mind is what I am doing. I am not sure if any of you are talking about correcting images for other purposes, such as more vibrant web colors or something along those lines….
Logic tells me as a "printer" that since my final output file must be CMYK then that is what I should be correcting in, because ultimately thats what the final file gets seperated as.
I understand LAB and RGB have a larger color gamut.
but whats actually achievable when it comes to ink
on paper is a result of CMYK inks…..
So I guess now my final question would be…
as a "printer" (prepress person) – Shouldn’t
I be working from the start in CMYK since
simulations of LAB or RGB are in many cases not achievable on paper without the addition of special spot inks,
not to mention it seems pointless to color correct
in either RGB or LAB since my final file will
always be CMYK?
It’s not necessarily about the gamut, but about what you can do within a particular color model in terms of color or tonal correction. There are types moves that, while may be close to achievable in RGB, are so fast and so effective to do in Lab that it almost feels like cheating. As I said before, it’s not the be all and end all, but one of many powerful tools at your disposal for a better image. I’ve been pretty much totally immersed in Photoshop since the beginning of ’95 and I had more eye-opening, someone turned the light on experiences reading Dan’s Lab book than all the other books I’ve consumed put together. All I can tell you is that you have to go into it with an open mind and a real desire to add this most competitive tool to your personal arsenal. What Dan does is expain something fairly complicated in a way that makes it immediately accessible to people of varying skill and comprehension levels. That in itself is no mean feat.
A quick note on the "a" and "b" channels. The "b" channel is yellow vs. blue, but the "a" channel is magenta vs. green, not red vs. green, something that Dan takes part of a chapter explaining.
Hi Reed,
In no way am I doubting you or calling you out on it, but I’m curious–what are some of the things that come from Dan Margulis’s soapbox that you question?
alan
Peter Benedetti
"Logic tells me as a "printer" that since my final output file must be CMYK then that is what I should be correcting in, because ultimately thats what the final file gets seperated as.
"So I guess now my final question would be…
as a "printer" (prepress person) – Shouldn’t
I be working from the start in CMYK"
Actually, no.
Most of the time you should edit in the space that gives you the most control over your images. Since most of them are probably photographic, that’s RGB (or Lab). Then let your output profiles translate the best you can get the image to look into output color space. Your profiles know your presses better than you do.
Sure you can tweak a CMYK image on the few occasions when you need to make some minor change.
By converting into CMYK before you have optimized the original gamut of the image, then try to make color adjustments, you have thrown away image quality before you even start. Especially subtle detail that makes the difference between average color and great color.
If you’re creating graphics in Illustrator or InDesign (or Quark XPress) working in CMYK is probably the best bet. (I simply can’t work in RGB for that kind of thing).
Peter Figen,
I think Dan has gone off on one of his useless tangents about the "green-magenta axis" in Lab space. Why doesn’t he argue for yellow-purple instead of yellow-blue? Or lime-magenta instead of green-red.
The color "red" in the red-green a Lab axis is not meant to be a particular hue. It means that colors in that direction are "red-like" hues. Magenta is a "reddish" hue, as are fire engine red, brick red, and purple red, all on that end of the axis. At the other end are "greenish" hues. That’s all. He just doesn’t get it.
About once a year he pops off in some similar way. He once proclaimed that the release of Photoshop 5 had placed the entire graphics arts industry in jeopardy, when it came out with its improved color management capabilities as default settings vs the older, less capable Photoshop 4 color settings.
Alan,
He’s hopelessly confused about the way an offset press uses oil (ink) and water to put an image on paper.
He’s contends that increased line screen halftone frequency cannot render higher image detail or quality. He bases that opinion on use of increased line screen images under printing conditions meant for coarse screen.
He insists that there is no image quality advantage to 16 bit images vs 8 bit images. He compares the two under conditions that convert 16 bits to 8, then back to 16, then compares that to 8 bit images, "proving" that 16 bits offers no additional image information.
There is no place in his world for disagreement (with him) or discussion.
Other than that, he’s damn good at CMYK and Lab editing.
CIE LAB was designed as a model of human visual response, not for a particular user’s editing pleasure.
Dan, what’s the first version of Photoshop you ever used? In a serious way, 2.0, because earlier versions didn’t really support CMYK.
What’s your favorite Photoshop tool?
Curves. Do other tools exist?
You’ve been a prepress guru for years– what’s your hardware setup? Like many others who have been buying new Macintoshes every two or three years since the beginning of time and can’t abide the thought of throwing any of them away, my network looks like a spaghetti factory. My desk has two 21" monitors on it so that I can multitask. One is connected to an 833mhz Silver G4 with 1.5 gig of RAM, the other to a machine that was very hot three years ago. But I still recall the days when a 32-mb RAM upgrade was $1,000.
I know you’ve done a lot in Photoshop– any good Photoshop stories or anecdotes? Only about calibrationists.
What’s your favorite thing about Photoshop?
The power it gives to users to break away from past practices.
What’s your least favorite thing about Photoshop?
It’s monopoly position and the arrogance that creates in its development team.
Do you have a web site?
Of sorts, at www.ledet.com/margulis
Who are the Photoshop experts that you look up to?
David Biedny and Bert Monroy.
Last question– what is Photoshop 10 going to be like?
It will require 3 terabytes hard disk space and 30 gb RAM. Other than that, it’s hard to say.
Check out my tip: Faces and the Luminosity Blend
If a 833 g4 is good enough for Dan, it’s good enough for me.
Reed,
first off thanks for the input on everything…,
Question,…
For the sake of this discussion I was not refering
to art created in illustrator or In Design, I am refering to photoshop "images" only…."Continuous Tones"….
I know you suggest making your edits in LAB or RGB before converting to CMYK, but that is assuming I am supplied with the image as RGB or LAB correct?
What would you suggest in a scenerio
where I am supplied a CMYK photoshop image.
Is it then point-less to convert to lab just to do a correction?
As you have explained your image once converted to
CMYK has "suffered" Gamut wise so to speak…. So at that point would you still suggest converting to LAB or RGB to do your correction, then back to CMYK once again or simply make your correction in CMYK,
OR does it make a difference either way?
Dan doesn’t like people who calibrate their color or their presses. So I guess that makes him the guru of random color.
I’m still waiting for a Dan Margulis/Bruce Fraser tub-o-pudding death match.
And Timo Autiokari can take on the winner, with colored strobe lights a’flashin’ and a’reflectin’ on their Barco monitors.
C.C. will officiate and M.O. can provide "color" commentary.
XD
Peter,
I used to convert to LAB in order to bump up contrast without affecting the color as happens with in rgb or cmyk.
Of course then you have be careful that (especially if the image was supplied and you don’t know the separation setup used) the image is converted back to 4c with the same GCR. I don’t know why and others have had different experience but in many images I found that going to rgb then cmyk from lab produced a better separation.
All that aside it was recently pointed out to me that I can do the exact same thing (add contrast without affecting color) in cmyk if I put the mode to "Luminosity". It makes such sense once you know it. You are only adding contrast to the luminance, which is pretty much what the L channel is anyway.
alan
Peter Benedetti,
If you are supplied an image in CMYK, you really have no choice but to use what you’ve been given. Of course, then you stay in CMYK.
And you’ve just described one of the worst problems in printing – use of CMYK files as though the file has meaning for any condition other than the specific one for which it was made. A CMYK file should be created from the "source file" (RGB usually) as one of the last steps in preparation of a job going to a specific printing press, known to be printing to a specific set of conditions (halftone line screen, paper type, ink set, sheet fed vs web, etc., etc., etc). Some of those conditions may stay constant from job to job on a particular press in a given printing house, but not only is every press unique in its printing characteristics, but many presses are used to print under a bewildering variety of conditions during the course of a week, or even a single day.
That’s where profiles come in and allow proper translation of the "source" images to the actual printing conditions. If you are supplied with CMYK files, how did the person who converted those files to CMYK know about the final printing conditions? If he or she did know, and handled the conversion right, you really don’t need to be doing any corrections. But so many designers and even printers have no idea that one press’s CMYK files have no business being used on another press and they pass around CMYK files like CMYK is a "universal format." It is anything but.
I don’t know which is worse, for a pre-press department to receive CMYK files made under "mystery" conditions or to receive RGB files with no embedded profile. Unfortunately, unless a printing house has complete control of the preparation of a job (and assuming it knows what it’s doing), more often than not, it’s going to be receiving CMYK files and untagged RGB files from outside agencies for output to plate. And when that happens, color management and quality color is poor.
Dan Margulis has shown that it’s not necessary to even have a color monitor (to say nothing of a calibrated color monitor) to do color correction. It’s all done by numbers.
Actually, he’s right, it IS possible to do color correction using a monochrome monitor (even a green screen!) and color correct "by the numbers."
And the color that results is about the quality and consistency that used to come out of shops during the 70s and 80s.
Dan’s a dinosaur. But he pretty much held color printing together with his bare hands in the dark ages before we had the kind of control we now have.
If he ran a carpentry shop, he wouldn’t let his employees use a tape measure, because "real carpenters" don’t need to measure before they cut.
If Dan was a musician, I’d guess he’d want to play everything exactly as it is written on an officially certified score.
On the other hand there’s Bruce, and his participation in wild ass cool soundscape experimentation <
http://www.oranj.com/improv.html> wherein the rules are learned, then thrown away, because he understands that subjectivity and creativity matter.
Word on the street is that Timo toots a mean kazoo.
😉 XD
Okay, I get the score (to use your musical metaphor) now.
I can’t dissagree completely with Dan Margulis. There are times when it is important to know the numbers because the Monitor won’t tell you everything, i.e.
If you have a proof done on a paper that has a cast to it (and there are thousands upon thousands of offset papers) that there isn’t a profile for one needs to have an idea what taking some cyan out here or magenta there.
And then there is the GCR thing. You just can’t see that on a monitor. There are some things to learn from Dinosaurs (beside setting up an giant Asteroid ballistic system).
alan
Alan,
You apparently like to fly by the seat of your pants. Yeah, I know, we all do it. But it’s crap.
Dan knows squat about GCR.
And manual attempts to do anything about GCR, black generation, etc just result in awful press conditions. Like I said above, results in the kind of color that used to get printed in the 70s and 80s.
Here’s something to ponder on. HP is releasing 2 new high end ink jet printers to compete with the likes of Epson 7800/9800. They will be 6 color and 12 color, respectively, each available 24 or 44 inches wide. They will have BUILT IN X-Rite spectrophotometers and software to allow AUTOMATIC profiling for any paper or inkset to be used. Just press a button, wait 20 minutes and you have your profile. The equivalent of a 1500 swatch profile run using today’s stand alone X-Rite equipment and X-Rite/Macbeth software.
The point? It’s absolutely necessary to be able to profile the output conditions in order to convert the "input" color files so that the result is accurate printing. That kind of automation has got to make it through to the printing industry in general at the printing press/paper/ink level.
