Strange L*a*b

warszawski

i am stone, and can not understand that within a Lab image channel "a" also "b" mesure L=54 WHERE it should be L=0 ?

pathfinder

The image data are not for just the single a or b channel, but for the pixel that is represented by all three channels - A and B are 0, indicating NO COLOR - neutral - but the L value IS 54, because the pixel is a middle grey in your image. That's how LAB specifies pixels - there is NO L value on a channel, but any pixel in an image, is represented by all three channels. Any clearer now??

When I read the pixel in the grey letter "B" on my screen it reads L 54, A 0, B 0 also. I am reading the pixel data with the Digital Color Meter that ships with MAX 10.4

warszawski

L=54 is a mystic one
ok, the layer is a text "L*A*B" with a L=0, a=0, b=0 and I am "stupefied" that photoshop SHOWS a and b channels with L=54 !!!!!

Howewer, It Happens Only if a or b channel is selected in Channels palette.

For me it is an ERROR, BUG ,so why not L=44 or other value ?

L=54 is a mystic one, somebody understand it?

edgework

warszawski wrote:

ok, the layer is a text "L*A*B" with a L=0, a=0, b=0 and I am "stupefied" that photoshop SHOWS a and b channels with L=54 !!!!!

Howewer, It Happens Only if a or b channel is selected in Channels palette.

For me it is an ERROR, BUG ,so why not L=44 or other value ?

L=54 is a mystic one, somebody understand it?

You're confusing the L reading when a single color channel is selected with the L reading for the image as a whole. It's a good way to drive yourself crazy.

The actual L channel controls all the lightness/darkness values for the image. In the A and B channels, darkness and lightness values determine which end of the color scale for each channel will be emphasized. Dead neutral means no color at all. Alas, the reading in the Info palette for such a state is A=0; B=0. But if you select a single A or B channel, your lightness reading will likewise be confusing: I get L=62, which means nothing in terms of actual lightness for the image, just the value in that channel that relates to the Red/Green axis or the Yellow/Blue axis. If you switch your info readout to K (grayscale), then sample the A and B channels for a dead neutral patch, than you will see that the value is 50%, which is to be expected. The reason L=62 has little to do with lightness or darkness or much else. It's a red-herring that doesn't directly impact the operations that that you can perform in LAB, or the calculations that you have to make to know what you are doing.

Just know that anything darker than 50% in A gives you Green, in B gives you Blue. Lighter values move to Red and Yellow respectively. Taking an L reading for an A or B channel proves that we're not in Kansas anymore, but little else.

warszawski

i got it ... other "strange mappings"
ok, thank you, i got it, since there is only one question left : why neutral 50%Grey has L=54 value in Lab and not L=50 ? and then i'll be set.

here there is illustration of "my first strange mapping" i came across and understood

are they other "strange mappings" that we should be aware of?

rutt

Do you have the book? I don't have it just now, but my memory is that Dan covers this in Chapter 7 when he is explaining why to use the shadow part of shadow/highlight in LAB. Roughly the mapping from RGB into LAB is nonlinear, expanding the range on its dark end. I'm not sure where the reduction and expansion actually take place, but it's probably in the book. I'll bet Edgework knows and also can remind me of why exactly this is done.

edgework

rutt wrote:

Do you have the book? I don't have it just now, but my memory is that Dan covers this in Chapter 7 when he is explaining why to use the shadow part of shadow/highlight in LAB. Roughly the mapping from RGB into LAB is nonlinear, expanding the range on its dark end. I'm not sure where the reduction and expansion actually take place, but it's probably in the book. I'll bet Edgework knows and also can remind me of why exactly this is done.

Dang, now you're gonna make me break out the book tonight and go searching. I seem to recall that Dan was not as detailed regarding the specifics of the lightness channel other than to assert that it is NOT a valid choice for a grayscale conversion. The most he got into it was to allude to LAB's application of gamma, resulting in a lighter image than an actual grayscale version. Considering the depth of his coverage of other issues, I thought this was a bit sparce, but then, it's not really as important, other than to know that there is not a one-to-one correspondance between actual value levels and what the lightness channel shows in LAB mode.