How do lightness differences in bump map translate to apparent depth in terms of world units?
For example, say I want to create a cone-shaped bump on a texture, and I want the cone to have an apparent 45 degree slope from base perimeter to point. If the point at the center of the cone is 100% (white) and the base perimeter of the cone is 0% (black), what radius (in WU) would the cone need to be for that black-to-white gradient to render as an apparent 45% slope?
Alternately but equivalently, say I want a box-shaped bump on a texture, and I want that box bump to look cubic, not tall or flat. If the bump is rendered as a white square on a black background in the bump map, how big would that square need to be to appear just as wide as it is tall when rendered?
A related question: what lightness in the bump map corresponds to the apparent depth of flat, non-bumpy textures? Are all bumps raised from the flat plane (so black is flat), recessed into the flat plane (so white is flat), or both ways (so 50% grey is flat)? Or something else?