Hello!
This is basically a “how low do I go” discussion. Recently I’ve become more adventurous with my games development I have been heavily researching some amazing techniques. One of which be the quadtree “Level of Detail” for procedural terrains. Much of my knowledge in this field has come from Thatcher Ulrich’s white paper on “Rendering massive terrains using chunked level of detail control”
Basically my idea is to generate mesh (or mesh data) to store “out-of-core”, then based on the camera’s frustum and distance, find the correct LOD chunk required, and load the correct mesh.
I got a fair few questions regarding implementing this in Unity:
Firstly, would each chunk be a mesh asset in itself, or is that defeating the point slightly. Is the idea that the rendered mesh is updated and pushed back with more and more subdivisions. In other works will the mesh from a Unity perspective be one “whole” mesh.
Secondly I found that there was some API for low level graphics rendering in Unity using the GL class , If i was to delve in and implement this technique using the GL class would I effectively miss out on some of the high level features, like the mesh having a Transform etc?
I have got loads more questions but I think I need to wrap my head around these first. If anyone has got any knowledge in this field I would be extremely grateful for your advice and thoughts, or maybe someone has actually attempted this and has a idea on where to start.
Thanks, Caius.