Chapter2 impression

[WarBlade]

With the amount of players that apparently to play at the same time, yes I think it would be too much stress on the servers to do this. If there only were some 500-1k players online at the same time then I believe that it would be technically feasible.

Imagine a trapper team blocking an entrance, such as in LW's screenshot. The melee AI would go nuts and probably be killable by a single monk with a wand.

Stress on the servers??? There are already changes to the game which have henchmen and others diving out of a hot zone as a reaction reponse to getting hit with AoE. All I'm talking about is a similar response to stop entering a defined area when two or more Rangers (or secondary rangers) are observed to be trapping like crazy.

[Green_Gloom]

Stress on the servers??? There are already changes to the game which have henchmen and others diving out of a hot zone as a reaction reponse to getting hit with AoE. All I'm talking about is a similar response to stop entering a defined area when two or more Rangers (or secondary rangers) are observed to be trapping like crazy.

For each trap you put down the AI has to find a path around. That is more complicated than the current "I've taken AoE damage too fast, move x units away" script. Each trap put down means the AI will have to calculate another path to the selected target, if you didn't then you'd be able to bodyblock the AI by spamming traps.

Then there's also the problem with melee AI and sometimes having to engage someone standing on a trap. Think of the IWAY team when the pets are dead. If all opponents are standing on traps then the IWAY team can't engage unless there's some sort of "Ignore Traps" code. The AI for this is going to take a lot of cpu cycles.

[Swiss Mercenary]

I thought that the suggestion was the AI sending in only one guy into the killzone, to trigger the traps.

[FoxBat]

I thought that the suggestion was the AI sending in only one guy into the killzone, to trigger the traps.

Something similar to that. Don't bunch up so groups are lest vulnerable to AoE and body blocking, and don't run into a trap *that is currently triggering*. If you can't path, running away for a few seconds or standing still is better while the trap unactivates. While it would be nice, I haven't suggested at all that the AI behave "suspicious" of traps it can't see like a human might.

[WarBlade]

For each trap you put down the AI has to find a path around.

Why? Players with ranged options use them. Players without ranged options stop. Why do AI-governed actors have to find an alternative path?

That is more complicated than the current "I've taken AoE damage too fast, move x units away" script. Each trap put down means the AI will have to calculate another path to the selected target,

Again, why? It could mean any number of things. It could mean an exclusion zone of radius X at point A for 60 seconds, to be overridden by successive instances of trap spamming within radius X. If the edge of the exclusion zone is reached it could mean a navigation of the perimeter to move closer to the target is activated or it could mean a stop and hold position response. It could even mean an "Activate Defy Pain and continue" response depending on the creature.

if you didn't then you'd be able to bodyblock the AI by spamming traps.

In the same way that players voluntarily opt to be blocked by trap spamming? Good! Emulation of player behaviour is exactly what I'm talking about!

Then there's also the problem with melee AI and sometimes having to engage someone standing on a trap. Think of the IWAY team when the pets are dead. If all opponents are standing on traps then the IWAY team can't engage unless there's some sort of "Ignore Traps" code. The AI for this is going to take a lot of cpu cycles.

Excessive CPU cycles . . . Why???

How much geometry are they navigating currently with all the pitfalls of the terrain, rocks, trees, moa birds and so on? Add in several temporary exclusion zones more and I can't really see how these would be anything more than a drop in the ocean.

Henchies used to get stuck on signs. Slap the same fix as a temporary exclusion on a timer at a multi-trap location and you have the basis of a trap avoidance routine already in the game. Stick flags in the AI on a per creature basis (Intelligent=TRUE, Intelligent=FALSE) and define which creatures will make use of it and so on.

[Green_Gloom]

Why? Players with ranged options use them. Players without ranged options stop. Why do AI-governed actors have to find an alternative path?

Abusability. If the melee AI always stops outside a trap then players will just use a single trapper to wall them in safely and kill the AIs with ranged attacks from 3-7 other players (depending on max players). And be a lot safer than if they had 3 monks taking care of them, vs melee only of course.

Also, writing CPU efficient pathfinding for a static world (ie. the current terrain) is a lot easier than writing it for a dynamic world, since you can place path nodes that will help the AI navigate the world. If enough traps expire to enable a shorter route while the AI is taking a long route to the target, should the AI change to the newer short route or not?

If you want to dynamically change the selected path then you add a lot of extra AI-pathing calculations. You have to check all traps for all AIs present every time a trap is placed or expires. Then you have to find the "best" route to the target from the current position for every AI. And that path may become blocked a second later by a new trap, which means all the AIs have to recalculate their path.

Giving them a formation to use until they engage the selected target would be a more usable solution.

[WarBlade]

Abusability. If the melee AI always stops outside a trap then players will just use a single trapper to wall them in safely and kill the AIs with ranged attacks from 3-7 other players (depending on max players). And be a lot safer than if they had 3 monks taking care of them, vs melee only of course.

Bad answer. There's no sense in make a sitting duck situation as part of a fix and historically these developers look to be fairly effective at avoiding problems and quick to fix new bugs.

Also, writing CPU efficient pathfinding for a static world (ie. the current terrain) is a lot easier than writing it for a dynamic world, since you can place path nodes that will help the AI navigate the world. If enough traps expire to enable a shorter route while the AI is taking a long route to the target, should the AI change to the newer short route or not?

Every monster in the game presents an obstacle. They move around a lot and when encountered they make you stop. I've considered the idea of a trap exclusion zone as a temporary 'monster' moreso than something akin to a cliff wall.

There are some problems with creatures getting stuck

If you want to dynamically change the selected path then you add a lot of extra AI-pathing calculations. You have to check all traps for all AIs present every time a trap is placed or expires.

No you don't. New traps can override old traps, so the expiry of old traps is not useful information for calculating exclusion zones. If 8 rangers trap like crazy without moving position you have 8 markers to check at any given time. Define a radius for override distance and you can knock that back to as little as 1 exclusion zone to keep track of.

The game already has something similar happening with spirits overriding spirits.

Then you have to find the "best" route to the target from the current position for every AI. And that path may become blocked a second later by a new trap, which means all the AIs have to recalculate their path.

That's a point that could still use work in this game IMO. More with respect to actor collisions, but once properly solved you'll be left with the solution for henchmen colliding with an invisible barrier caused by a trap recognition function.

If Stefan follows Somewhat Zen directly North and bangs into a Tengu, he stops. But what if the game marked the off-set angle of that Tengu at 345° and offer up a 90° direction change bypass option taking the closest angle (to the target Somewhat Zen) of the two possible resulting directions? Maintain the new heading for 1 second before returning to the original target point and I see the skeleton of a fix beginning to emerge.

Giving them a formation to use until they engage the selected target would be a more usable solution.

Err . . . I'm not so sure about that. Certainly maintaining a set distance between units in a line astern formation would be easier, but anything else would be about the same complexity IMO.

The formations currently used in the game are further evidence that trap exclusion zones are quite achievable. Run around with the Henchmen for a bit and watch where they stand in relation to your character. Hit "X" and watch the formation collapse and reform as your character spins to face the opposite direction. Each unit maintains a set distance and position around a mobile point...

Just as they can be set to maintain a specified distance from a hot zone of traps.

[Green_Gloom]

No you don't. New traps can override old traps, so the expiry of old traps is not useful information for calculating exclusion zones. If 8 rangers trap like crazy without moving position you have 8 markers to check at any given time. Define a radius for override distance and you can knock that back to as little as 1 exclusion zone to keep track of.

Can you guarantee that every team with trappers will have their trappers stay in the same spot to place traps? If not then you have to check for every trap when it is set and when it expires.

If traps can be exploited as an invisible wall then you can be certain that people will use them as such.

[WarBlade]

Can you guarantee that every team with trappers will have their trappers stay in the same spot to place traps? If not then you have to check for every trap when it is set and when it expires.

No I've already pointed this out: New overrides old on a tight radius basis, much tighter than the actual exclusion zone. That way if several traps are positioned within a tight area you only have to keep track of the last one dropped.

[Green_Gloom]

No I've already pointed this out: New overrides old on a tight radius basis, much tighter than the actual exclusion zone. That way if several traps are positioned within a tight area you only have to keep track of the last one dropped.

Consider the following scenario:

If the two top traps are renewed at their excact locations while the lower 3 are replaced by the New Traps. This will leave a gap big enough to run through, but if you don't update the pathing AI when the old traps expire then the AIs will believe that they are still blocked.