I've been adjusting saturation using the A and B channels(in curves) of Lab color lately. I've been getting great results, but I noticed something today that maybe I should have known already.
This color tweak method affects the histogram more than a little bit, so images that I had applied levels to before the Lab color tweak ended up looking a little too contrasty. Here's the tricky thing about it all though, if you do levels first and then do the Lab tweak, the histogram won't look any different when you reopen levels, it looks as though you have pulled the sliders to the correct places. However if you start the file over and do the Lab tweak first and THEN add and levels layer, you will see the histogram looks different. By doing the Lab tweak first, it seems to do some clipping of its own and thus when you open levels you don't need to pull the sliders in as far as you might have otherwise.
Hopefully this makes sense. I'd be curious if anyone noticed the same thing.
I've been a bit sloppy in developing a workflow....if I'm gonna continue using this curves method in Lab color to adjust saturation...then I guess I should be doing this step first every time, correct?
Perhaps you just need to be a little less agressive with the Lab moves and be more accurate is setting your "L" endpoints. You shouldn't really need to add Levels on top of your Lab curves. I'd work on that some more. And I would definitely not worry about your histograms, worry about your image and use the Info Palette sticky densitometers to monitor the areas you are interested in. After that make sure that you've read, re-read, and then read again Margulis' Lab book.
Personally, i do not see the need to adjust an image in LAB. You can get the same effect using Adobe Camera Raw by adjusting individual color luminance prior to opening in Photoshop.
the reason I don't set the contrast with the lightness channel in curves(in Lab mode) is that I basically use the same curves settings for the A and B channels every time to create that color/saturation pop. The result is usually a little much so I dial it back in the opacity...so I wouldn't want to lose that contrast pop.
That's probably the lazy way to do it guess...do you set your A and B channels by hand for each photo? I'm not at the point where I know exactly what effect each will have so I move both sliders in the same amount on both channels.
I really need to investigate the Margulis book....
I think what you're seeing is the fact that switching from RGB to Lab is a destructive edit in itself and if you switch back to RGB it's another destructive edit.
I think you've come to the right conclusion. You really need to treat each image as a unique problem. Very seldom do two need exactly the same correction. Dan's book will be a revelation. It seems to be roundly criticized by too many who haven't bothered to read it. There are things that you can do in Lab that are simply impossible any other way and there are things that you can do in Lab that you actually can accomplish in ACR or in complicated RGB moves, but you can almost always do the Lab moves in far less time than any other method. I use Lab for at least half of my images and probably a higher percentage of stock images that I have to "fix". Don't let anyone talk you out of one of the most powerful tools that most Photoshops users don't even know exists. Good luck.
Thanks for the advice...I've decided that I definitely need to pick up Dan's book. The few tutorials I've found online that use his ideas have all done wonders for my post work. These are results I just haven't been able to even get close to any other way.
I have read that moving from RGB to Lab and back is non destructive is that correct?
Also, just checked out your site..very inspiring work! Thanks for the help.
" I have read that moving from RGB to Lab and back is non destructive is that correct?"
No, it's not correct. Any mode change is destructive. What matters most is if you image is better off or worse off for the work you did in Lab. There are a lot of so-called experts who are so scared of losing one single pixel worth of data and to whom maintaining the original integrity of the image is so important that they miss out on the chance to drastically improve their image. The best advice is not to worry about it but that because there are some levels lost during conversions, to use 16 bit per channel files.
Often times I'll try corrections in mulitple color spaces and then compare to see what's best. Lab is not always the best solution but often times is. Part of becoming good at color and tonal correction is learning to identify quickly which tools are both the best and the most efficient.
I also use Lab mode to adjust/tweak color and contrast, usually setup as an action in my workflow. I find that after converting to Lab then creating a curves layer to adjust either the A or B channel, I will duplicate the curve layer, then change one to "Luminosity" and the other to "Color" blend modes. Then adjust the opacity of each layer to suit, usually 60-80% for the luminosity curve, and 30-60% for the color curve, merge all and duplicate the layer back to my original document, where it may be further masked or adjusted.
Reading your posts, I am still a little unsure of the exact sequence you are using. I am assuming it is Levels>Convert to LAB>Curves Adjustment of A and B>Convert back to RGB>Check Histogram. Is that correct?
If so, part of what may be causing you some problems is the manner you are doing the conversion to and from LAB. If you use Image>Mode, the return trip from LAB to RGB will always use the default Working Colorspace and Rendering Intent that you have specified in PS's Color Settings.
If you are doing the conversions with Image>Mode, you might want to try Edit>Convert to Profile instead. It will allow you more control in selecting colorspace and rendering intent during the conversion to and from.
A little more info from you on your exact working sequence and the colorspaces/color settings being used might help us pin down the problem.
I agree with jerryrock. From my own experience and reading up to date ideas on LAB (Martin Evening, Katrin Eismann and many more pro PS writers) there is no reason to use destructive editing like LAB any more. It can all be done non-destructively in CS3 and ACR.
"It can all be done non-destructively in CS3 and ACR. "
Again, these are historically the weakest arguements for not using Lab. What I see more often is people not proficient enough iin Lab to reallly understand why it's so great. Every adjustment, no matter where you do it is going to be destructive once the image is flattened. If you can't see the "destructiion" in any output, then does it really matter? I'll repeat myself again. Yes, there are some moves that you can duplicate in RGB, but almost always you can get there much faster in Lab and often in ways that are subletly better. If you've got a couple hundred images to bang through on a tight deadline, you generally want the most efficient route possible.
"What I see more often is people not proficient enough iin Lab to reallly understand why it's so great. "
The same argument can be used for Adobe Camera Raw.
"Yes, there are some moves that you can duplicate in RGB, but almost always you can get there much faster in Lab and often in ways that are subletly better. If you've got a couple hundred images to bang through on a tight deadline, you generally want the most efficient route possible."
This statement is your weakest link. Adobe Camera Raw is very capable of batch processing.
Jerry, I'm not talking about batch processing raw files, I'm talking about applying specific corrections to individual images to provide optimum enhancement for whatever intended use of the image. I'm also talking about the ability to asses an image to determine the best route forward. My statement is not a weak link at all. Maybe you don't yet quite grasp what's really possible when working in Lab, and just how fast and effective those moves can be. Compared to the relatively crude sliders in ACR/LR, the Curves in Lab are simpler, more effective, have more deliberate control and, once you start to understand what they can do, are plain more intuitive. I rarely make batch adjustments other than the color temp of a raw processing, but that is almost always done in C1, which, for me, is far more intuitive, faster and provides better quality raw conversions than ACR.
I have no idea what your involvement in photography or digital imaging is, and googling your name comes up with nada. Where can I see your work? Here in Los Angeles, I work for demanding high end clients all day (the latest Metro billboards are up all over town now) on deadlines that would make most people faint. You HAVE to develop workflows that allow you to get the most out of the image in the least amount of time. Time really is money, and using Lab color is one of the most powerful tools I have at my disposal. Have your read The Canyon Connundrum even once let alone several times? Out of all the books on digital imaging I've read, and I've got quite a library going back to RWPS 3 in 1995, Margulis book is simply (well maybe not simply) the best book in digital imaging I've ever seen. Dan has a way of explaining some very complex ideas in a way that they become easily accessable. There were more "light bulb" moments that happened reading that book that in all the other books combined. The lessoned learned therein turned my into a MUCH better imager almost overnight, and I was very very good prior to that. I'm not just an armchair photographer here. I do this every day for real money and have no hidden agenda.
I don't really think I deserved that verbal assault. There are many ways of processing an image depending on your intent. I have never claimed that there is one processing workflow that works for everyone. My statement that ACR processing can accomplish the same output as converting from RBG to LAB and back to RGB is valid. My opinion is that LAB is not the way to go for optimum workflow.
I applaud you for being an accomplished billboard photographer and book collector.
I have been a photographer for over 30 years, both film and digital. I used to process all of my film and prints in my color darkroom. Most of my photographic work has been landscape, fine art prints, portraits, commercial (product) photography with the occasional wedding. My work has been published in Elle Decor magazine and I have had several gallery showings of my photographic prints. I have recently completed a degree in Graphic Design with formal training in the entire Adobe creative Suite as well as Quark XPress to round out my knowledge.
Jerry some people in life see things only one way. Some are very successfull ,some are not. They will venomously argue their point till you agree. It is best to just let them be.
I'll admit up-front that I'm NOT a pro but I am curious about one thing. Can you really reproduce LAB's ability to break colors apart in ACR?
For example, you have different shades of beige and you want to break them apart to show more color contrast. I don't see how you do that with ACR, even with use of the HSL tab. Sure you can turn yellows green and darken blues, etc., but I don't find those sliders very effective on subtle colors nor on trying to break apart colors that reside in the same region of the color wheel.
If there is a way, I would like to understand what it is.
Interesting thread, looks like there are some definite lines drawn, so to speak. I thought I might chime in too. I tend to use LAB quite a bit on certain types of pictures. LAB does have some distinct advantages as opposed to the other colors spaces for some things. I use it when making selections based on color, some channel blending techniques, often when adjusting the shadows with the shadow/highlight command, sometimes with USM in certain situations, often when limiting an adjustment to color with the blend-if sliders, dodging and burning, etc. I don't see any degradation with multiple conversions between RGB and LAB. I sometimes use straight line adjustment to the a and b channels in curves to separate color, but often right curves in the a and b channels as well. I usually save setting endpoints and contrast curves for RGB or CMYK. There are just some things you can do in LAB that you can't do elsewhere.
One thing you have to watch out for are gamut problmes if your normal RGB space is sRGB. I usually use aRGB and it's not so much an issue, still you have to watch the info pallet. Both current books by Margulis are very informative and beneficial to read and understand.
Thanks for the advice....didn't mean to start a skirmish! ha.
It seems like some people don't know exactly what I'm talking about or how I'm employing Curves in LAB mode to tweak color. This tutorial sums it up nicely, it's based on techniques by Dan Margulis(I have his book on order...)
Peter, I see what you mean about multiple color spaces. Lab works great for a lot of things but CKMY is easier for me right now when editing headshots(getting skin tones right).
