I'm creating something like a depth map material. I tried Camera Data approaches but I could not get what I wanted.
Render Layer Compositing is working better, but I'm try to normalize in a slightly different way. Traditional normalize spreads the values over the range 0-1, but this changes the 'scale'. I need to just shift the values to the edge of the range 0-1, as I'd like the closest point to the camera to be white. This could be done with simple addition but I need to know how much to add, i.e, the smallest depth value.
I tried using depth->minimum node but I was getting either all white or all black.