9 Replies Latest reply on Nov 10, 2014 10:45 AM by G.Hoffmann

Why does eye dropper Lab show a different value than the Lightness channel?

Can someone help me understand how the L in Lab color in the eye dropper tool maps to the actual Lab Lightness channel?  Both are on a scale of 0-100, but they clearly show different values.  Here's a video that show what I mean (note that if I completely desaturate the color, I get the same results, it isn't a color thing).

• 1. Re: Why does eye dropper Lab show a different value than the Lightness channel?

The lightness channel is affected by the color. If you notice the lightness channel will appear lighter than a grayscale image of the same scene. When you're measuring the color it is taking into account the color. When you measure the lightness channel, you are measuring gray. I know it seems odd, but that's how L*A*B* works.

• 2. Re: Why does eye dropper Lab show a different value than the Lightness channel?

8 bit/channel L* is stored as 0-255, even though the CIE range is 0-100.

And L in HSL really doesn't relate to L* in CIE L*a*b*.

• 3. Re: Why does eye dropper Lab show a different value than the Lightness channel?

Even if I use a completely neutral gray color, I see the exact same issue.

• 4. Re: Why does eye dropper Lab show a different value than the Lightness channel?

This is the fundamental difference between RGB and Lab color models. Better illustrated than explained:

• 5. Re: Why does eye dropper Lab show a different value than the Lightness channel?

I cannot read much useful in the OP's video, but I think, the OP is not talking about

HLS- or HSL-Lightness (not available in PS CS6) or HSB-Brightness, but only about

CIELab Lightness in two different modes:

Figure 1:

Figure 2:

The area was filled with a gray L=50 a=0 b=0.

In Figure 1 all channels are selected. The values are correct.

In Figure 2 only the Lightness channel is selected. The value L=54 is wrong.

What might be the reason?

Measurements were always taken inside the area. The RGB-values are the same in both modes.

Please don't verify this by measuring in these Forum-images. These are PNGs in RGB.

Please verify by repeating the trial.

Best regards --Gernot Hoffmann

• 6. Re: Why does eye dropper Lab show a different value than the Lightness channel?

Yes.  I shouldn't have posted that first video with color, seems to be confusing people.  If I work in Lab mode, using the Lab readings on the eye dropper and use only grey paint; I still see a difference between the L value reported by the eye dropper on the background and the value of the Lightness channel.  They are clearly different, I'm just wondering how the mapping between these works.  a and b are different as well, as the eye dropper shows values of -128 to +127 for the eye dropper but 0-100 for the a and b channels.

Here's a redo on the video showing that grey also shows this difference:

• 7. Re: Why does eye dropper Lab show a different value than the Lightness channel?

Thanks for the feedback. The effect is the same, but the deviation in your test is even larger.

The deviation depends on the settings for gray: either Dot gain or Gamma. L = 54 for G = 2.2.

For Dot gain =10%we get L = 70 instead of L = 50. L=54 is valid fro G=2.2.

I don't know the purpose of this internal calculation. It's probably not a simple error.

Perhaps it's an attempt to show the L-channel as an editable image with respect to

the color settings, e.g. for the application of sharpening or contrast manipulations.

Best regards --Gernot Hoffmann

• 8. Re: Why does eye dropper Lab show a different value than the Lightness channel?

Bingo, that's it!  The dot gain impacts the channel separation in RGB as well.

So now the question is, what is the meaning of it?  Is the color picker showing Lab for a display space and the channel separation for a pseudo printed space?  I had always thought that Lab was just Lab (especially since its gamut is massively bigger than any printer or screen can display).

• 9. Re: Why does eye dropper Lab show a different value than the Lightness channel?

I had always thought that Lab was just Lab

That's true. The transformation between Lab and any calibrated RGB space, like sRGB or Adobe RGB

is unique in both directions (for all in-gamut colors).

A gray image (R=G=B) in Lab by three channels L, a=0, b=0, is handled correctly by Photoshop, but the

single channel L of Lab is handled as a Grayscale (one channel).

A gray image as Grayscale can have any gray space like Dot gain = 10% or Gamma = 2.2: the appearance

on the monitor will be the same in Photoshop. Color Management is even valid for gray images.

This is not so for the L single channel.

The encoding by different Dot gains or Gammas has the purpose, that the so generated file can be sent

directly to an image setter or plate setter without any modification.

What has this to do with the L channel of Lab? Nobody (let me use this formulation) would ever like to

send the L-channel directly to a printer for K-only printing. But we (at least I) would like the L-channnel

as an invariant lightness representative.

Therefore I'm considering now the observed behaviour as a bug, unless I knew better arguments.

Best regards --Gernot Hoffmann