I’m sorry if this has been discussed elsewhere, or if it is stupid question, but I have spent a very long time in the search function and this seems to be the solution I’m looking for.
I am trying to implement a navigation system that can be evaluated by the AI at run-time. I want to be able to take a node and evaluate it based on its position. I understand that custom nodes can be created and this is a plan for later, however I need to get my foundation in place first.
My first thought was to use a grid-graph, as I know I will be able to explore and evaluate nodes capably.
However I’m curious as to whether I will run into issues with the count of the nodes once levels reach scale, even with well targeted queries. Will having large cells in the grid result in strange pathfinding around objects? Specifically rounded edges, or objects at an angle? And if I turn granularity up to accommodate, will performance be acceptable in areas as large as 0.1km^2 or so, with humanoid-scale actors?
My second thought was to use recast, however I don’t believe I would be able to get node data as I am wanting to. Please correct me if I am wrong here, as I’ve been looking for the answer all night.
Assuming I cannot extract the node data from a navmesh, would I be able to place down empty game objects, or even a point graph, and use those nodes as my points of evaluation?
Sorry for how scattered this is. I don’t have a ton of experience with pathfinding, and I’ve been awake for far too long now.
Any feedback is greatly appreciated!