[Gnu3dkit-dev] Goal: No imposed limit on the number of objects in a scene

[Brent Gulanowski]

General Strategy: Partition the scene into sub-scenes and load only 
those which are potentially visible.
Sub-goals:
- Asset Management: re-use assets (textures and meshes) as much as 
possible; identify explicit sub-scenes where they are used (and 
thereby, which sub-scenes where they are *not* used). Identify re-use 
of assets.
- Paging: load only potentially visible portions of the scene
- Potentially Visible Sets and "Super-sets": we can know what is 
visible from a single point, but can we know what is visible from 
points within X units of that point? The size of X is dependent on the 
speed of the viewer's movement, and the scene complexity.
- Detail Suppression: Partition a scene based on detail level. Leaving 
out details can reduce scene complexity. The question: "When can you 
get away with less detail?" The tentative answer: "When the user is 
moving too fast or is busy with other things." These are often the 
same, and are determined by how quickly the camera is tracking/rotating.

--
Brent Gulanowski                                address@hidden

Ce n'est pas une signature.     -Magritte