So @bgolus did mine or @superouman s screenshots help solve the problem or does it remain hidden? Just curious, because once that is fixed and the pathfinding is improved I think we can call the patch ready then.
Just curious. Manually moving mods changed since alpha. Took me a while to find that out. It is now in your User/Appdata/Local/Uber Entertainment on windows. You are using windows? You are sure that mod folder in that location, it is cut-pasted to move it elsewhere entirely, like your desktop, to make CERTAIN there are no mods like transparent icons? Maybe I am just being uncertain about it. If anyone else is having metal spot icons not show up, would be nice if they could come forth. Or, if you two could change settings individually 1 by 1 just testing resolutions and quality controls and see if that brings metal spot icons back. Anyone know the location of PA config settings, so one can delete it and regenerate it from default, just to see if it helps?
Just out of generic interest- are you using A* for pathfinding or or a mixture of several methods and raycasting for proximity testing/edge detection? Could it be that your raycasting interval is "missing" thin edges in obstacles thus creating the pathing artifacts? Are you using pre-generated nav-hints at all? Gawd, I'd love a blog or something on that topic :-s @sorian
Well, your wish was granted a while ago. It was something they demo'd just after kickstarter, before the game was even playable. There was a live stream on it, as well as a blog at planetaryannihilation.com and I will add the links here in an edit. Short answer, is they use flowfields. It was working shabby in the livestream and in early alpha, the reason I observed and may be incorrect was that the pathing values were too extreme. It was/is working really good in Stable build atm, but somehow adding trees affecting pathing and such has made the pathing produce unnecesary longpaths and sometimes not finding valid paths at all as if land were water outright. I would love to chitchat about this myself, its a lot of work apparently getting flowfields to work, but I am not certain a traditional pathing would be better or worse, but I am watching them getting this one working so I am impressed because this flowfields has plenty of advantages to utilize. Wholy crap, it's a small world. In a discussion about SupCom2 flowfields, someone asked about anything related to them to learn about them, and someone from Uber posted some informational links in the discussion. How interesting. http://gamedev.stackexchange.com/questions/387/how-does-flow-field-pathfinding-work
Actually the pathing stuff is being done by @Elitron. He's done some posts in the past on the topic, but the short answer is yes to all of your questions, though there's a bunch of changes to the internals since he first posted about it. edit: Flow fields are only one part of how pathing is accomplished in PA. I'll go over it from my understanding quickly, and Elitron can smack me later for messing it up. There's a nav mesh that's pretty much just the straight planet mesh unmodified. This is what units actually drive on or float above, but very little of the pathing is done at this level. The nav mesh is broken up in to sectors and islands. Islands, as you might expect, are areas that are not connected to any other area. Sectors are how each island gets broken up, these are roughly convex collections of nav mesh polygons. There's also a cost field that was stored sort of like surface texturing, but is being changed to be a full voxel representation. The cost field can also have an effect on the sector and island creation if the cost is too high to path through (ie: a solid line of buildings), as well as stored An initial a* determines the gross route by the web of sectors and their connections to other sectors. Once a route is chosen the flow fields take over. One set of these is generated for each unit size and destination in real time. The flow field is generated only for the sectors a unit is expect to travel. If for some reason they are forced to go to another sector a flow field is generated at that time. When they're moving they also try to do collision avoidance with other units, which is where they might drive outside of their initially determined path.
That link is actually how I found out about PA :-D I am reading that book right now - good stuff. I have been experimenting around with applications of flow-fields and pathing algorythms for a project of mine (not game related) and a lot of these issues with pathfinding seem oddly familiar, though in a different context. Thx for looking these up for me btw!
