Pseudo-displacement, real displacement, methods...
There a number of ways (beside modeling) to get 3D on a surface. I've used various ones, and I'm starting to get confused by their relationships. Maybe someone can clarify? (Still using 2.79b.)
My starting point: Cycles, glossy plane, not subdivided. Using an image texture for depth.
- Only connect image texture to Displacement input of Material output. Render looks 3D; obviously geometry unchanged.
- Only connect image texture to Height input of a Bump node. Bump Normal output to shader Normal input. Render looks 3D, no geometry change. In fact, it looks identical to #1, at least to me. Maybe calculated the same?
- Subdivide the plane, A LOT. Probably add Subsurf too. Add a Displacement modifier and insert the image texture into it. Render looks 3D because geometry is 3D (pretty rough).
- Subdivide the plane, just a little. Enable Experimental and add adaptive subsurf. Only connect image texture to Displacement input of Material output. Render looks 3D. Most optimum and efficient choice?
Adaptive subsurf and Displacement modifier seem to be mutually exclusive.