RGB->LAB->RGB

MM
Posted By
Mark McIntyre
Feb 13, 2004
Views
499
Replies
28
Status
Closed
Someone brought this up in a photography website and I thought it a good question. I’ve always read that switching from RGB to LAB and then back to RGB does not alter the color space. But it does. If you fill a doc with 0 255 0, switch to LAB, and then back to RGB, there are changes in either both RGB and CMYK values or just one of them, depending on what color profile you use.

Is this correct??

How to Master Sharpening in Photoshop

Give your photos a professional finish with sharpening in Photoshop. Learn to enhance details, create contrast, and prepare your images for print, web, and social media.

CS
Carl_Stawicki
Feb 13, 2004
With the color values you mentioned (I’m assuming you meant 0R 255G 0B), the RGB levels are effected, but not the Lab’s of those colors. The Lab values of that shade of green, while in AdobeRGB, are 83L -128a 87b before and after converting. The RGB numbers end up being 21R 254G 3B. So the levels change, but not the color.

Carl.
SG
Sylvain_Gingras
Feb 13, 2004
When you change mode, I think it is best advise to do it through convert to profile, so that you can control to the way the conversion is done. But each time you convert a file there is some loss of data from rounding of numbers.

Can anyone confim?
MO
Mike_Ornellas
Feb 13, 2004
There may be rounding errors during the color space conversions. To assure little or no errors, use 16 bit files.
JS
John_Slate
Feb 13, 2004
By what version could you go to Lab in 16bit?

So the levels change, but not the color

Same color space, different numbers = different color
CS
Carl_Stawicki
Feb 13, 2004
Different combinations of levels within the same color space can have the same color. By ‘color’, I mean perceptual colors, as indicated by Lab values. Levels and colors are two different things.

Carl.
MO
Mike_Ornellas
Feb 13, 2004
John,

Photoshop CS supports 16 bit LAB.
MM
Mark McIntyre
Feb 13, 2004
Thanks.

But I”m still confused.

Does the color degrade or not?

Many authors have encouraged switching to LAB mode to sharpen the L channel, and then back to RGB or CMYK.

Is this a problem for color shifting??
BF
Bruce_Fraser
Feb 13, 2004
Yes, the conversion from RGB to Lab to RGB is subject to rounding errors.

If you convert using whole numbers between Fahrenheit and Celsius, 53° Fahrenheit and 54° Fahrenheit both convert to 12° Celsius.

8-bit/channel files only have 256 data points per channel. In a small RGB working space, those points are packed closer together than in a larger one. Lab, by definition, covers the entire range of visible color, so the data points are spaced further apart than in just about any RGB space. Hence when you convert, you get rounding errors.

Whether or not these errors are visually significant depends upon

the image
the rgb space
what you do to the image downstream

You can get a result meaninglessly close to sharpening the lab channel by either

1.) Sharpening RGB, then using Fade with the blend mode set to Luminosity, or

2.) duplicating/merging to a new RGB layer, sharpening it, then setting the sharpend layer’s blending mode to Luminosity.

Either one will avoid the rounding errors.

If you work with 16 bits per channel, the rounding error will almost certainly be insignificant, but converting to Lab is still a lost slower than either of the above alternatives, and if your file has adjustment layers, you’ll have to either flatten before conversion or suffer significant color shifts because the layers behave differently in Lab and RGB.
J
JasonSmith
Feb 13, 2004
I actually use #2 sharpening method (darn I thought I invented that).
R
Ram
Feb 13, 2004
Many authors have encouraged switching to LAB mode to sharpen the L channel, and then back to RGB or CMYK.

Yes, some do; but I’ve found it to be a dumb idea. Just loook at the color gamut in Lab and in RGB. Not hard to see that there has to be a degradation when you switch in and out of Lab.

Bruce mentions two alternative ways of doing it, but I actually prefer a different, much better route suggested in –surprise!– Bruce Fraser’s "Real World Photoshop 7". It involves getting rid of noise in the blue channel and applying different degrees of Unsharp Masking and Fading to the red and green channels separately.

Thanks, Bruce.
CS
Carl_Stawicki
Feb 13, 2004
#2 has been my choice for quite awhile now. It’s very versatile.
AS
Ann_Shelbourne
Feb 13, 2004
Even better: Paint in the USM with the History Brush using varying opacities and Luminosity mode.
MM
Mark McIntyre
Feb 13, 2004
Thanks for all the good feedback. I’ve left the same question with Dan Margulis’s color theory group as well.

Have used all three methods recommended, but good to hear what others prefer.

Gaining clarity on the LAB issue.

Thanks again.
L
LRK
Feb 13, 2004
Where is Dan Margulis’s color theory group? Can’t possibly be as interesting as Bruce Fraser though. 🙂
MM
Mark McIntyre
Feb 13, 2004
Ann, great idea.
TL
Tim_Lookingbill
Feb 13, 2004
I tried the sharpen individual red and green channel technique but always got tiny rainbow color affects (or chromatic aberations as some call it) on hard edges even in a tiff file. Jpeg was even worse.
AS
Ann_Shelbourne
Feb 13, 2004
Tim:

I have found the same thing. I don’t like the sharpen-in-color channels method — UNLESS it is in Lab’s "L" channel.
BF
Bruce_Fraser
Feb 13, 2004
FWIW, the sharpening individual channels technique was developed when blue channels were a great deal noisier than they are today—I think the example came from a Polaroid PDC 1200, if anyone remembers those. It has been redacted from Real World Adobe Photoshop CS in favor of techniques that are less of a blunt instrument, but hey, if it works for someone…
R
Ram
Feb 13, 2004
Tim,

I never work with JPEG images, ever.

… but always got tiny rainbow color affects …

Watch out for those mood swings. 😀

Seriously, I guess you mean effects, and that probably means you were being too liberal with the unsharp masking.
TL
Tim_Lookingbill
Feb 13, 2004
Hey I just figured out that this forum now substitutes our ; ) marks into those symbols automaticly. Neat affect and effect.

And, oh, how did you know I was bipolar?
R
Ram
Feb 13, 2004
his forum now substitutes our ; ) marks into those symbols automaticly

Not if you leave a space between the two characters, like you just did.
TL
Tim_Lookingbill
Feb 13, 2004
I meant to do that. 😉
R
Ram
Feb 13, 2004
I knew that. 😀
AS
Ann_Shelbourne
Feb 13, 2004
And not if you give them wiggly noses:

:~)
L
LRK
Feb 14, 2004
Thank you Mark… for the link.
PC
Pierre_Courtejoie
Feb 14, 2004
John, Mike, you could already have a 16bit LAB file in V6… and maybe before…
Yep,

I only think in the current build of the application or a least try to.

Also, if you have extra channels in an RGB file, prior to PS 8, things are grayed out so some confusion can arise as to what can and can’t be done and when.

It was just easier to speak in the context of the latest release of PS.

;o)

Master Retouching Hair

Learn how to rescue details, remove flyaways, add volume, and enhance the definition of hair in any photo. We break down every tool and technique in Photoshop to get picture-perfect hair, every time.

Related Discussion Topics

Nice and short text about related topics in discussion sections