Feature #2197
Updated by skyjake almost 8 years ago
The old experimental bias lighting system has two components:
* Direct light from point light sources. Tracks configured light sources and determines whether individual vertices are lit by them. Does simple raycasting to determine occlusions (shadows).
* A volumetric grid is used for distributing soft ambient light from lit sectors to the immediate surroundings.
The problem with these is that the implementation is entirely CPU-based and operates only on vertices. This is Both of these are incompatible with the future direction of maximum GPU utilization and per-pixel lighting. Therefore the old experimental implementation should be removed, and a better GPU-based lighting system should be implemented for Doomsday 2.
* Direct light from point light sources. Tracks configured light sources and determines whether individual vertices are lit by them. Does simple raycasting to determine occlusions (shadows).
* A volumetric grid is used for distributing soft ambient light from lit sectors to the immediate surroundings.
The problem with these is that the implementation is entirely CPU-based and operates only on vertices. This is Both of these are incompatible with the future direction of maximum GPU utilization and per-pixel lighting. Therefore the old experimental implementation should be removed, and a better GPU-based lighting system should be implemented for Doomsday 2.