Rangerdown: The water should have nothing to do with it. As long as that scenero was in a dungeon (outdoor zones don't really need this node pathing), the mob should always be stuck to the assigned path, moving from node to node (train tracks I used to call them when I first started playing EQ). Since there were no nodes placed in the water in my example, the mob should not have even considered walking there. It doesn't matter that we can't tell where water is - when the nodes are manually placed, WE know where the water is, and can place nodes around water, lava, whatever, when needed.
fathernitwit: My idea was that nodes would be places manually by us in a similar fashion like what m0oni9 described : a simple # command which places a node (and perhaps also spawns a mob in that loc, so you can see where you've placed them, and do LOS checks to mobs you've already placed). I for one am 100% willing to places nodes throughout dungeons if it means we can have proper pathing and fear support in them.
The only 2 things i'm not sure of is:
1.) how expensive would this be on the CPU? If it can't be done in realtime, then perhaps some kind of script can be run after all the nodes are placed to spit out a grid file which has pre-calculated distances and LOS checks to nodes throughout the zone
2.) How would the path-finding algorithm work once the nodes are in place? I still can't figure out how to handle Example C which I listed above. Any of you guys have any ideas?
|