The brains I want on the camera are for things that actually involve taking the shot. Give me intelligent capture of images for stacking. That entails two things:
1) HDR exposure. Point the camera at something, select HDR. It takes the exposure and examines the frame for any pixels near the extremes of the sensor. If there are any pixels near the top it reshoots exactly the same shot but with a shorter exposure time. Repeat until there are no really bright pixels. On the other end, if there are any really dim pixels reshoot with a slower exposure, repeat as needed.
2) Focus stacking. Manual focus, pick a point. Pick another point. The camera shoots a sequence of exposures moving the focus between the two points.
I recently made a photobooth from an old Canon DSLR and a rPi running gphoto: a script takes the images out of the camera and posts them on a server, and people can see them in almost real time.
It's really great, but it would be even better if it was all done in camera.