3D Pathfinding Theories
#3

Quote:
Originally Posted by Gamer_Z
View Post
Just create a navigation mesh of SA, It probably already exists somewhere in the game but it would be visualised like this:


You can look in the unreal engine source how that nav mesh is made and how navigating through it is done.

Your method could work for creating those meshes, probably the only way to do it without access to the SA engine.
Also, how would you navigate only roads?
Yeah this generated mesh would be more for walking around. Your GPS data would still be best for vehicles. For walking on sidewalks though, we could use an algorithm that supports weighted nodes and weight the nodes that are near the pedestrian nodes defined in gta3.img. With that, the generated paths would tend to walk mostly on sidewalks, but if the destination was somewhere else the path would go off the sidewalk. Like say the target was across the street, the path would go along the sidewalk until crossing the street was optimal then it would cross the street.

EDIT: So anyways, I've been running my method in PAWN on my already shitty computer. And damn is it slow... It has been running for hours on just a 62.5x62.5 area (so 125x125 points, each with at least one node and a lot with multiple level nodes). The getting the nodes (on all levels) part is quick (2 seconds everytime), scanning for connections is what is slowing it down so much. Obviously it'd be a lot quicker in a plugin and on a better PC (seriously, this needs more than just a 1.55ghz single core CPU, burning 100% constantly at the moment).
Reply


Messages In This Thread
3D Pathfinding Theories - by Crayder - 31.03.2016, 04:32
Re: 3D Pathfinding Theories - by BurnZ - 31.03.2016, 13:31
Re: 3D Pathfinding Theories - by Crayder - 31.03.2016, 15:00
Re: 3D Pathfinding Theories - by Crayder - 04.04.2016, 16:58

Forum Jump:


Users browsing this thread: 1 Guest(s)