I am very familiar with pathfinding in general, but not with this product. I need to overcome this hurdle to get started evaluating it.
I’ve placed a 1000x1000 node grid graph on a 8 by 8 km terrain. I’d expect it to be possible to walk from most links to the other, as I don’t have many dramatic height differences in most places, nor any static obstacles. However, there’s obviously low sloping hills and such. Certainly, you’d expect that with the “max inclination setting” this would be evaluated properly.
I tried turning off collision checking without different results.
So basically, is the problem that the grid graph is too low res and that I shouldn’t even be using a grid graph in this scenario, or something else? I would prefer to use a navmesh actually, but I don’t really feel like manually creating one for the trial version… Trying to prototype the game without accumulating costs, and I need a placeholder graph.
Might add I got about 60 000 disconnected areas with default min size of 10.
Or actually, $100 is a minimal amount… Just convince me that pro generated navmesh will work in my scenario and I’ll buy immediately. Chances are I can use this product for other future projects as well.
Hi
Could you post a screenshot of your results as well as the settings you use to generate it? It is hard to help without knowing the settings.
8 by 8 km is indeed very large. If you need resolution for objects down to e.g the size of a chair, you won’t be able to get that with a grid graph. A tiled recast graph will however be able to get the resolution you need. The downside is that it is much slower to update unless you can restrict updating to using navmesh cutting ( see http://www.arongranberg.com/2013/08/navmesh-cutting/ ).
If you are really unsure about whether you should buy it or not, I could give you access to a trial version so that you can test it ( the version you are using now is the free version, not a trial version ).
Thanks for the reply, it made me more clear on how to deal with this. I have another idea: I’ll hand-roll a nav system for outdoor areas, then use your product for more complex urban areas (e.g. villages). Navigating around a couple of trees in gentle hills shouldn’t be a hard problem to solve, and indeed it might even be preferable to use steering behaviors there rather than following paths.
If it is just a couple of trees ( more specifically convex obstacles ). You should be able to use steering behaviours. But be warned, you cannot use concave obstacles. That will never work with any kind of local avoidance.
Yes, but there are workarounds. Bounding concaves with convexes, and for simple concaves manually design rules for AI on how to go in/out of concave areas. Not something you want to do for a highly complex concave but I think I can handle e.g. rocks just fine.
Obviously the sum of this is more complex and work-intensive than just slapping a general solution onto the whole terrain (which would be preferable and what I had hoped to do), but it seems at this time there’s no better option.