I see a lot of folks convert to the LAB color space (and back again) and you most often say you're doing it so you can edit luminance and color informaton separately.

My question is, why not use the HSL color space instead? It seems more intuitive. Photoshop has the Hue, Saturation and Luminosity (Luminance and Luminosity are very similar) blend modes, so you can extract HSL channels and do your HSL modifications interactively within an RGB image--no need to switch modes.

I understand all about how the LAB color gamut covers the whole range of human color perception and then some, but this capability isn't of much use unless you have a display device to match.

My inquiring mind wants to know.

Bart

..... It seems more intuitive....It may seem that way, but if you've ever tried working in that mode you pretty soon discover that it is very abstract.

As you said Luminance and Luminosity are about the same thing, so let's cancel thmn out. That leaves us with the colour information, and how this information is presented.

The LAB mode uses Cartesian coordinates where (simplifying) one dimension is A (Red to Green) and the other is B (Blue to Yellow).
The HSL uses polar coordinates, where H measures the angle (0 = Red, 120 = Green, 240 = Blue and 360 gets back to Red again) and S measures the distance from the Grey centre.

While polar coordinates can be good for some things, we are much better at interpreting Cartesian (x,y) stuff.
Try looking at the H channel of an image, it's very difficult to visualize the "real" colour. Saturation is usually a bit easier but it tends to get out of hand (undefined actually) in the white/grey/black areas.

Having said that, you could compare it to FFT techniques where the "working" image is completely incomprehensible but for some special circumstances it is exactly the right tool to use.

So, yes, there maybe some good specific use for HSL somewhere - our friend Stroker has put in some good work there - but for now, the "killer app" hasn't appeared.

Rô

Hi Ro,

Actually, I usually use Color, Saturation, and/or Luminance. For every Lab tutorial or article I've seen, one would merely need Color (R,G,B) and Luminance (greyscale). The only time I've ever used Hue was for specific types of selections (eg., if you want to select the red part of a piece of clothing.)

I use the saturation "channel" for control of saturation in a manner that the HSL adjustment layer really isn't able to do.

Regarding polar versus rectangular coordinates, both color spaces project onto difficult-to-envision 3-d volumes. HSL projects onto a symmetrical double-cone--I think when one looks at a picture of that shape, it's fairly intuitive. The cylindrical coordinate system of HSL makes sense for that sort of a shape. Lab projects onto a weird asteroid-shaped object that can never be intuitively captured in any coordinate system. The a and b channels are simply variables that can be projected onto a rectangular chart (you could plot HSL coordinates into a cube if you wanted), but I don't see how that provides anything other than a false sense of intuition.

But all the technical stuff aside, I guess I was just arguing that you can do the sort of stuff people do in Lab mode using simply color and luminance layers in R,G,B without switching to Lab, but you don't have the downside of loosing your adjustment layers when you change between modes. I suppose one could just stay in Lab mode all the time--then do CSL splitting within Lab mode.

Bart

I feel your pain, Bart. People touting seperation as a major selling point of Lab even though it can be done in RGB mode with various tools. And it doesn't even take much to do. I've already done one tutorial along these lines and I'm planning on doing more.

(One that really gets on my nerves is the subject of over-saturation in RGB. "Don't do this because you will over-saturate. Instead, use this long-winded hack." Ug!)

For me, Lab is excellant for two major reasons:
1. based on human perception
2. Cartesian

Notice that seperation isn't on that list.

I have been having a lot of success recently pulling HSL out onto layers.
Hue it's very difficult to visualize the real colour. Well this may be so but, to me. It seems far easier than trying to visualise colour by looking at the ‘a’ and ‘b’ channels

Now maybe if I could pull the a and b channels out onto layers and put the colour back in then I might find LAB useful. Trouble is, I can’t find a way to do that in LAB.

Ken

Now maybe if I could pull the a and b channels out onto layers and put the colour back in then I might find LAB useful. Trouble is, I can’t find a way to do that in LAB.
I'm kind of headed in this direction in the FM thread.