I think there’s merit to the idea, but I’m not finding a way to do it without creating a large collection of such profiles, and spending more time figuring out which one to use…
I shoot (mostly) underexposed, letting my camera’s highlight-weighted matrix metering do it’s thing. Doing this, I almost never blow highlights; when I do, it’s usually light sources or clouds and the librtprocess highlight recovery routine works quite well to fill in the blotches wtih some detail. But as a result, most of my exposures require some lifting of shadows, and it varies from scene-to-scene. I recently shot a wedding for a friend of my daughter, 600 images from three cameras, and I had to spend a few hours going through each batch of shots from given scenes. I’ve become quite adept with my wonky filmic’s four parameters, getting the right amount of toe, slope, and rolloff.
What I did to my software that’s really helped with this is to make one mode of my batch processing an “apply the current processing”. So, I open one of the images of a collection, process it to taste, then delete the other JPEGs of the group and kick off the batch command. It collects the toolchain for the currently open image, and the command line tool applies the new processing to make the missing JPEGs.
For the wedding photos, I went back and reviewed the processing, but I couldn’t find any patterns that would warrant collection into a named profile. I think that’s the consequence of any sort of ETTR-oriented capture: I can divine no good heuristics for characterizing the processing required…