Years ago I developed a mask selection process for use with satellite data that had some similarities with the mask area selection process used in dt. In that process (which was only 8 bit rgb) a small selected rectangle would directly ‘find’ the range of colors for that area and then the area would be modified using plus/minus buttons along with the cursor, to permit the area to grow as needed.
The system in dt is of course much more sophisticated but the user needs to initially determine which band, or combination of bands would be the most useful. That is not always obvious.
I wonder whether something similar would make sense in dt?
The current system is really great … but just a bit ‘academic’. Or maybe there is some good reason for keeping it this way!
I am not suggesting that the current system be replaced but maybe, since masks are used in all functions, it would be sensible to have some degree of “auto-select” as a part of the process.
There are auto select pipette’s for the parametric channels have you tried them??
You can use the boost to expand the mask… also use the C and M keys …C will show you the image broken down for the chosen channel and M will show the mask from the current slider postions…
I’m only adding this because I didn’t understand what Todd wrote above, even though I’m familiar with the tool and use it quite often. Maybe it’s only because I haven’t had my morning coffee yet 
The boost factor slider allows the range of values targeted by the parametric mask sliders to be extended. It may be used in scene referred editing, where luminance values may extend beyond 100%, to target highlights. This slider is only available for channels where it is meaningful.
When I use GIMP it has select by color and fuzzy select tools which are a form of parametric masks. One feature I like about selections in GIMP is the ability to grow or shrink a selection. When using the auto select tools in GIMP the selection can be improved usually by growing the shrinking the selection by just 1 or 2 pixels. I wonder if a similar option in DT for selections would be practical to implement and if others could see the benefit in this?
Maybe it is just me … but I do not find this slider to work in a useful way. To extend the basic mask I find it better/easier to fiddle-around with tags.
The idea of the booster sounds great … it just does not appear helpful to me.
(could be to do with the git version that I am running??)
Further testing and it appears that the ‘boost-slider’ is not correctly functioning with the current git.
I will make an appropriate report.
It may only work on Luma channels… I thought it was an expander originally but I think its for something like when you use Jz. That parameter has a mid point of about .62 not 18% or 0.5 and its based on the absolute candela scale I think so to target highlights you have to use the boost.
You can see this by looking at an image. Try a parametric mask on Jz…use a an image with fairly bright elements use the left picker which will show the value… if you get a value greater than 1.0 then you need to use the boost to add EV so that you can target those spaces with your mask…otherwise 1.0 becomes like a threshold and everything above it will be lumped together…
It does not change the mask; rather, it changes the range off the selectable values, so you can select values above 100%.
In darktable, you can use parametric masks, drawn masks, and a combination (you draw a mask and then restrict it based on parameters). Drawn masks can be resized.
Additionally, you can refine the masks: feather them, change the contrast, take detail level into account.
Well, I used it on ‘exposure’ which is pretty basic … furthermore I used on the cz channel which should be just fine.
Above is before
Above is after
If the boost slider is not supposed to work under these conditions then maybe it should not be shown.
Again … maybe I simply do not understand how this should work
I have been duly enlightened by the design team and closed my issue.
I did comment as follows for the information of others:
“OK then, I totally misunderstood the use of the booster-factor. What I now understand is that the selection process is a “hit and miss” process. If I try one setting of the boost and that does not provide a correct response, the I change the boost and start all over again.
I do get it now. My misunderstanding. This functionality is totally unlike how I possibly imagined that it could be designed.
Sorry that I wasted everybody’s time.”
By the way … if anybody can actually get this function to work (in any way) on a current git version, I would like to know.