Improved Bot Paths

[Lightning_Hunter]

I improved bot paths for a handful of ONS maps included in the CEONSSMapPack. Should I go ahead and upload them, or would anyone have an issue with that? I did not modify anything in these maps except the bot paths (only the bot paths were rebuilt). Here is a list of maps I improved:

ONS-Tempestade - I redid the paths completely from scratch in this map, except for the flying path nodes. The bots previously didn't even use tanks, due to there being no valid paths for the bots to drive on. I had to force the paths with almost every RoadPathNode outside to get the bots to use large vehicles. They will now dominate this map pretty quickly.

ONS-StarReach-)o(Edition-Edit8 - The bots don't fall off the edges as much now. I doubled the number of path nodes, and spent a lot of time tweaking the existing ones. The bots previously could not reach the jets in either base. This has been fixed using bot-only kickers.

ONS-ArtilleryHill-CEONSS-V2 - The path nodes have been tweaked beyond recognition. The number of path nodes is about 5x what it was before. I spent a lot of time making sure the bots use all the shortcuts. They also use jets much better, and jump in and out of vehicles less often. You will see them driving tanks and large vehicles more often than before, and rarely get stuck.

ONS-Rail-C-SP1 - The bots now use the underground tunnels. They previously wouldn't even touch the power node down there.

ONS-ColdFusion-T32 - The bots couldn't even get to a few power nodes, and would get stuck in various locations. This has been fixed.

That is all of them for now. Any interest in these versions?

[Miauz55555]

I would say just put them into Dropbox or likewise and post the link.

P.S.: Welcome to the forum Lightning_Hunter.

[N4rkoT1k]

Welcome Lightning_Hunter!

Thanks for your effort on this issue. I guess its pretty welcome to CEONSS users/admins

[Lightning_Hunter]

I was fixing up a map, and ran in to an issue. A canyon that is supposed to destroy vehicles does not. You will only die upon exiting the vehicle. There is a PhysicsVolume at the bottom of the canyon with "bdamagesvehicles" set to true, "bdestructive" set to true, and 20,000 DamagePerSec. DamageType is 'Crushed' (although I also tried 'VehicleDamageType' ). Shouldn't this kill vehicles? Why can I drive vehicles around the bottom of the canyon, only to be killed when I exit the vehicle?

[Pegasus]

Kinda tricky trying to diagnose a malfunctioning PhysicsVolume without being able to examine the map it's in or at least review its UEd props' text block, but here's a few guesses as to what might be causing the problem, based on your description:
- the PV's bPainCausing prop might be disabled
- if you've added the volume yourself, you did not rebuild geometry afterwards
- this PV is inside another customized volume, with the latter having countermanding properties and a higher priority
- the PV is non-convex or otherwise irregularly shaped and that plays havoc with the contained actors' detection

Hope that helps.

[Lightning_Hunter]

Sorry for the noob questions, but is there a way to add an instant death trigger, like in UT99? I do not wish to rebuild geometry, because that can often cause all sorts of issues. There must be an easier way than messing with a volume and rebuilding the map. I have never really done any mapping per say in ut2k4; just a lot of bot pathing (AI is my expertise).

[Pegasus]

Well, unless we're talking about actors touching a specific trigger placed in a specific location and having it cause them to die, no, not really. Since you're looking to set up an entire area (and not just a point) that will apply a certain effect to actors entering it (in this case, massive damage per sec of a specific damtype), the way to do that is through volume placing, and, if I remember correctly from even simple Volume class actors used for location naming purposes, volumes placed in UEd typically call for a geometry rebuild so that some of their attributes can be baked into the map and the engine can start taking their effects into account once ingame.
On the bright side, a geometry rebuild is a pretty standard n' straightforward editing practice, and even though there's the occasional story floating around UT msg. boards about things going horribly wrong after one (usually involving BSP holes or zones mulfunctioning), in the vast majority of mapping it's just a cause for a minute-long break, if at all. To my mind, if you're already putting in the time n' effort to help make the maps a bit better bot-wise, as long as you have your work saved before doing a rebuild (in a differently named file, just in case) and you follow up your rebuild with a ingame preview by flying around in spec cam to check everything still looks right (by clicking on the Play Map! button with the joystick icon), the chances of it going wrong should be negligible. Give a geometry rebuild a try, and if the PhysicsVolume's still not working afterwards, there'll still be other options to explore.

[Wormbo]

Actually I think volumes don't need any rebuilding, because they are generated from only a single brush and are not merged with the level's general BSP model or any other volumes. (Their model is a separate BSP tree generated only from that single brush. Same goes for simple collision models of static meshes, btw.) Things that may need rebuilding are paths though, as they may take effects like water or volume damage into account when calculating AI weights. Geometry rebuild is required after modifying non-volume brushes or when placing ZoneInfo actors or moving them between zones to associate them with the corresponding BSP leaves. In fact, most volumes can exist as non-static versions as well (Jailbreak maps make use of that for executions), even though their enter/exit logic may not work too well with immobile actors (like bored bots in jail).

One reason why a physics volume won't work in general is that it gets overridden by another physics volume of same or higher priority affecting the same area. An actors effective physics volume only changes when it left the old volume or entered a new volume with a higher PhysicsVolume -> Priority value.

[Lightning_Hunter]

Fancy seeing you here, Wormbo. We seem to run across each other in various forums. I was (and still am) a big time Worms Armageddon player (the only game I play online, really). I also asked you a question ages ago about your UT99 mod "enhanced relics". Hehe.

Interesting post. I initially thought that volumes did not need a rebuild, because I already successfully used several blocking volumes, and they worked without a rebuild. Of course, I rebuilt the bot paths (dozens of times) since they were completely redone from scratch. It's still a mystery why the physics volume won't destroy vehicles. I'll have to look closely to see if there are any other volumes touching it with a higher priority.

By the way, is there any interest in Vehicle CTF bot-improved maps around here? I've noticed that maybe 75% of all the VCTF maps have very poor bot support. The biggest issue is with bots not driving big vehicles (jumping in and out of them in most cases). Also, bots often refuse to get out of a vehicle when they near a flag, and often don't return with the flag in the vehicle. I already improved bot support in about 25 of my favorite VCTF maps. In these maps, bots will now drive tanks and other big vehicles to get flags, and actually return with the flag using a vehicle (with a few small exceptions). It's almost as if mappers of VCTF give up with pathing at the very end. In some cases, fixing the maps was as simple as forcing paths where there aren't clear white paths to the flags, and also setting a "VehiclePathName" to the flags. Bots can actually be pretty smart in VCTF. They will even attack using AssaultPaths while driving vehicles, as well as return using AssaultPaths.

[Lightning_Hunter]

Still can't get vehicles to explode... I might upload the map for someone to look at.

Is there a way to get volumes to kill bots only? I really do miss the SpecialEvent actor from UT99... It had many useful features that are missing in ut2k4...