HSB Average from Selection + HSB Adjustment

Question

Little bit of an unusual one for you here; but here it is anyway... Is there any way to get the average HSB value of a randomly selected area without applying the average filter, using script?  The selection is already created before the script is run, so all the script needs to do is get the selected area's average HSB value.  Additionally, is it possible to then adjust that selected area's HSB values to a predetermined value?  I want it to keep the detail, but say, darken an area if it's brighter than the preset or desaturate it if the color is too strong.  Any ideas on how to do these/if it's possible to do these?  I'm pretty sure the second one's doable, but not so sure about the first, so any help is appreciated.
dgolberg

Answer 1

Well, I'm in RGB/8 color mode, but I'm using the HSB (via the color picker window) to determine the difference in brightness of the resulting color. 
Anyway, below is an example of what I'm trying to do (I couldn't use the actual project due to NDA, but it's the same methodology).  The first image is the original photo with a color palette for various stains in the lower right corner.  On the palette, I took the original stain, selected it and averaged it out to get a base color to work with.  I then cut out the object (in this case the cabinet) to apply the "stain" to. 
The method I had been using was to then create a fill layer below the cabinet layer using one of the stain averages, desaturate the cabinet layer and set it to soft light and duplicate it.  After seeing your mention of luminosity and linear light, I tried those, but got even darker results.  Instead, I found 1 soft light layer and 1 luminosity layer over that gave a slightly better result than 2x softlight layers, but they still come out the same brightness.  The 2nd and 3rd picture are examples of my results after merging the color layer and 2 overlay layers together.  The 2nd picture is the top stain, the 3rd is the next one down from that.  Now, they're not TOO far off in this one; but sometimes the original we work with has a darker stain to it, so the result comes out too dark compared to the palette, even though it looks good otherwise. 
Obviously, some stains are just too dark to do anything with, but the ones we're having the most trouble with we can still get to look right, albeit through a very lengthy and manual process of curves, saturation, and color balance; and often the manual method comes up with inconsistent results.  Basically, what I'm trying to do is automate the process to not only speed things up; but to get more consistent results than we currently are getting from the manual method (often times others have to do the process, and their proof colors may be off or they simply have trouble adjusting the colors correctly).  So far, this method is getting our colors much more consistent, but the brightness is still a bit off due to the intensity of the original; which is why I keep referring to HSB, as I can see the Brightness of the texture this way.
All I need to do now is create a script that can get the color of a random location in the image (basically the layer that was cut out from the background then altered and merged back into a single layer like images 2 and 3 above), and compare the averaged brightness from the stain palette to the averaged brightness from the merged layers.  I know how to get the color of a specific location, but not of a random location that is ever changing (the object(s) we're working with are usually in different places in each photo), which is what I'm stuck on now.  I would assume that, if it even is possible, it would require either a selection of the area or a transparent layer where it can pick the color from the areas that have pixel color and ignore the transparent sections.
Does that help any?