Doesn't understand the logic behind ctrl-click for luminosity selection on the rgb channel

[olivierr92135410]

Hi,

I'm trying to understand the maths that are applied when I control click on a rgb channel? Apparently, it's 50% of the luminosity values (selection from 128 to 255), but when I intersect this selection by itself, (ctrl-alt-shift click), I don't get a 25% selection (0,5 x0,5 = 0,25 right?), instead I get a selection from around something like 170 to 255 (see screenshot in attachment), normally it should be 128 + 64 = 192 no?

Also when I create the luminosity masks using calculations on the grey channel in multiply mode, I get different results after repeating a few times the operation...

Thanks in advance !

Olivier

[D Fosse]

No, you misunderstand what a luminosity selection is. A pixel can be partially selected. It's not a threshold, it's not either/or.

The marching ants just mark the 50% boundary, but the selection itself is a full-range grayscale image.

If I ctrl-click RGB on this image -

- I get a selection that in reality looks like this. Dark is less selected, light is more selected:

[JJMack]

Its very easy to see this is correct:

[olivierr92135410]

Thank you for your answer,

so the marquee doesn't indicate the true luminosity selected range, but why then, when I create luminosity masks with ctrl-click and intersection, I don't get the same alpha channels that those created by the calculations method (multiply) ? (the 5 first alpha channels here under were created by  the calculations method, the following 5 by the ctrl-alt-shift click method) Also, in the alpha channels created by  the ctrl-alt-shift- click method some banding appears, altough I'm viewing them @100% zoom level and working in a 16 bits file ?

[JJMack]

I do not know what you have exactly done here or are showing here to create the steps in the alpha channels.  I do not think it would make much difference to me for a levels adjustment on those selections would most like adjust the masking effect to what I would need.  I care only about the resuling image.  I feel mask can alway be improved on.

[olivierr92135410]

I agree about improvement of mask, what I'm concerned here about, is the math applied to create those luminosity mask and most of all why the alpha channels generated by ctrl-click on a rgb channel don't get the same result as the calculations command (multiply operation and intersection of a selection are the same operations no ?)

[JJMack]

I can not answer you question I do not know the math  used to do either of if they should be the same.   I also do not care that much to want to know the answer.  All I need to know is how luminosity masking works enough to use it like for working on highlights and shadow area etc.  I do not need to program a luminosity feature for my work. I find no need to know the math or know in my math background is sufficient to understand the math required.  I have only met a few people the know and understand math like lapses transform. 

Mandelbrot may have pass by my office door a few times however, I never spoke to him.  I did not even notice if his feet touched the ground.

[OldCrelis]

Hi, I am definitely not an expert on this subject but I understood from a class I took that the cmd-click produces an 8 bit channel whereas the calculate process produces a 16 bit channel. This might explain the differences.

Ron