undefined | Better HN

0 pointssbergot2y ago0 comments

We can only speculate one the difficulty of the fixes. I am not an expert but I believe culling and LoD are two big things for engines. Not being able to use existing implementations is really unfortunate. I guess the "tricks" are well-known but I don't think the implementation is easy.

0 comments

1 comments · 1 top-level

Ciantic2y ago

I'm not a GPU programmer, but I became interested to dig out what is the state of art of culling, and found out Arseny Kapoulkine [1] has implemented those in MIT licensed C++ code base:

Cluster cone culling, frustum culling, Automatic occlusion culling, triangle culling, Meshlet occlusion culling, Optimizing culling. On paper sounds impressive, and maybe Colossal Order could learn from those.

https://github.com/zeux/niagara

j / k navigate · click thread line to collapse