devlog.txt

Cellular Automata development log

Implement 1d and 2d variations simple enough.

3d another matter, exponential increase in size.

Original plan to use voxels and render as cubes a la minecraft. However
dumb 'brute force' approach proved impractical > 1000. Considered ray tracing
for vis/occlusion culling, seemed overkill. Examined minecraft technique,
seems it's an algorithm that unifies nearby similar geometries and renders
as a solid mesh. This will not work for my non-uniform, almost random, cellular
behaviour. I'm essentially trying to solve the toughest "noise" case.

Compromise
----------
Implemented a 3d point cloud which worked but wasn't visually striking enough.
At this point, realise delving any further would be 80% writing a 3d engine and
thus losing sight of the original objective. So opt for pseudo-3d (2d blocks
on a 3d terrain) and go for real 3d with birds demo as the dataset is much smaller.



References

1. Flocks, herds and schools: A distributed behavioral model (http://dl.acm.org/citation.cfm?id=37406)
2. Nature Of Code: Autonomous Agents (http://natureofcode.com/book/chapter-6-autonomous-agents/)