Burton’s suggestion of the new Margulis book on Lab color will turn your whole notion of color correction on end. It’s far too complicated to get into in a forum like this. While most books might mention Lab in a paragraph or two, this is the most comprehensive look at the subject that I know of. Buy it and you’ll be able to answer your own questions. I use the techniques in this book literally on a daily basis and my images have never looked better.
Thanks ! ! !
Brilliant!
I’m buying it.
Speaking of LAB mode, I had read in several places about the technique to sharpen or USM the L channel, supposed to give better result than sharpening the whole image in RGB mode.
Only one site (don’t remember which) warn about the image degradation that color mode change add to images.
Is it true that changing an image from RGB to LAB and then back LAB to RGB add some kind of defects, or is a totally loss less task ?
Thanking in advance for your advice.
Mario
It’s not totally lossless; there may be a few rounding errors. Not comparable to the degree of loss in repeated open/saves in JPEG (even at high quality levels), however.
Speaking of LAB mode, I had read in several places about the technique to sharpen or USM the L channel, supposed to give better result than sharpening the whole image in RGB mode.
After performing USM you can Edit>Fade Unsharp Mask>scroll mode to Luminosity and leave opacity at 100%.
This does the same thing as sharpening the L channel only, and you don’t have to convert back and forth.
The "Edit>Fade" option is one of the best options in Photoshop for lots of things.
Use EDIT>FADE for a more realistic result after using the patch tool to remove bags under the eyes.
John,
This does the same thing as sharpening the L channel only
Not exactly, but close. L in L*a*b stands for Lightness, not Luminosity. The two are different, but the end result is close to the same.
Mathias
No color mode conversion is lossless.
Converting between RGB and LAB is close to visually lossless, but can’t quite achieve that in 8 bits/channel. And there is always numerical loss (two slightly different colors become one).
Normally when one converts to Lab, it’s done as a one time round trip – RGB>Lab>RGB, or sometimes RGB>Lab>CMYK. That conversion is not going to make a visual difference, and if the reason for going into Lab is to make the image noticeably better, even if you do lose a few levels, your final image is going to be much better for it. If you were to make multiple round trips, and I’m talking 25-50, and left the dithering option checked, then, yes, you will see visible degradation, but who would ever convert that many times on a real world image. If you’re at all worried about it, it’s easy enough to test for yourself, and usually you can work in 16 bpc, which pretty much eliminates any risk.
I’ve found that sharpening the "L" channel as opposed to fading to Luminosity allows me to be more aggressive with USM, however, given workflow contstraints for commercial work, it’s not alway possible to do that and Fade to L is usually an acceptable alternative.
I’ve been having some phenomenal results with 1DsMKII images, rezzing them 200% while still in Lab and then sharpening the L channel using two passes of sharpening, one at a .7 or .8 radius and a second round with a lower amount and at .3 or .4.
Yes, there seem to be some pretty pleasant layer effetcts on the bounce from RGB to LAB.
Most of these (multiply & softlight if I recall correctly) seem to retain the differences when bounced back from LAB to RGB.
I carried out a comparison on one bounce and couldn’t spot any real differences at all
1 – source image in RGB
2 – create a bounced copy RGB to LAB & save then back into RGB & save
3 – put the bounced image on top of the source image
4 – tweak layer effect to Difference then zoomed in at 1600%
5 – confirmed on the histogram not one pixel of difference on one bounce (I didn’t try 25 though!)
I bought this book and it is changing everything. Great Recommendation!
~Rikk
Deebs, did you try an auto Levels adjustment layer on top of the difference layer?
I had a chock when ‘dither’ was enabled per default in color conversions back in Photoshop 7, I believe it was. It can crumble a clean digital file in just one conversion. Making any actions that utilizes color conversions change behaviour for instance.
I did a run through on layers (pushed for time) and wanted to see if there were any obvious differences and if these were robust through an RGB > LAB shuffle LAB > RGB bounce
The ones that caught my eye were multiply and softlight
Both have worked themselves into deebs list of methodology for doing stuff.
I didn’t get quite as far as using adjustment layers and the like.
Consider it as a first level 101 excursion into mode multiplying complexities 🙂
Thanks all for the answer to my question, really appreciated!
Mario
