"Some engines (most notably the latest version used by Assassin’s Creed) have even gone in a radically different direction, using the capabilities of the latest GPUs to drive rendering and effectively doing away with draw calls altogether."
I'm curious about this. Is it procedural generation? How does it work? The only thing I could find with a while of searching is:
"AnvilNext [...] is capable of generating structures in a flexible and automatic manner while following specific design rules and templates, which reduces the amount of time and manual effort required for artists and designers to create an intricate urban environment. Specific landmarks, such as the Notre Dame de Paris, are still designed by hand [...]."
He refers to a talk from the Assassin's Creed team in comments, and I discovered the person who gave the talk wrote a fairly extensive set of forum posts on it (including links to the original slides) here: https://forum.beyond3d.com/threads/gpu-driven-rendering-sigg...
Frostbite does GPU culling. This presentation is a really good explaination of their approach. It relies heavily on the ballot instruction to do parallel prefix sum in a compute shader and multidraw indirect to procedurally issue draw calls based on compute shader results.
I'm curious about this. Is it procedural generation? How does it work? The only thing I could find with a while of searching is:
"AnvilNext [...] is capable of generating structures in a flexible and automatic manner while following specific design rules and templates, which reduces the amount of time and manual effort required for artists and designers to create an intricate urban environment. Specific landmarks, such as the Notre Dame de Paris, are still designed by hand [...]."
-- https://en.wikipedia.org/wiki/Anvil_(game_engine)