I’m dealing with some large worlds, talking potentially thousands by thousands of nodes wide–well beyond the scope of a single grid graph. A single large recast graph does actually do the trick for baseline pathfinding but the game itself is well suited to a grid graph logically (everything is placed on a strict grid) for modifiers/costs.
This is a large top down RTS like world in terms of pathfinding needs.
It does seem like the suggestion you’ve given in the docs and elsewhere is to use a single large recast graph. But everything in the game is placed on a strict grid and modifications to move speed etc. for agents maps well in that manner.
I know this can be accomplished with the recast graph–and have done so on smaller scales in the past–but I just want to do due diligence and make sure that multiple grid graphs would work.
Basically here’s the setup:
- Large, fully procedural, map (25+ 1000 meter square terrain tiles)
- Small grid size (2m per tile. Approximately 2500 grid cells per terrain tile)
- Individual cells on map can modify pathfinding behavior. Faster pathways, blocked paths, etc. are all being positioned cell by cell
- Approximately 200 pathfinding agents at once across the entire map is the realistic worse case scenario.
Bonus follow up question: In a world this large with nodes potentially this small, even a Recast graph takes a moment to generate–would multiple recast graphs be better (maybe 1 per 1km tile)?
From what you’re describing it looks like this may be your best bet, but I also recommend skimming through the