Bahamut mission trouble

[Misery]

Ok, so,  in my current game, I FINALLY found one of the Bahamut missions.  Only found one so far. 

But I havent completed it, and not because it's difficult... but because it simply takes an absurd amount of time, as well as probably way more ammo than my entire team has total.  Or so I'm guessing;  I havent had the patience to complete this one yet.  It may as well be 3 or 4 normal missions in one long string.  And each individual turn takes an unusual amount of time, even if enemy bots are not moving.  That seems to happen sometimes in any mission.  Still, I am thinking that there's probably just not enough ammo to finish it, the enemies simply take too many hits.

Why is this?   Because the entire mission.... every single time.... spawns about 2 gazillion Thunderbots, in huge, dense clusters (in addition to occaisional patches of Silence or Ninja types).  I dont know that I've encountered THAT many robots in any other mission.  In other missions this'd be lessened by the fact that they'd do their "shoot at cover" thing, but there's no cover in this mission type at all.  Most weapons cant even hit them, not without the Exo taking fire (Assault and Science cannot fight them whatsoever, since they always take more than one hit, and those two cannot fire from far enough away) particularly with Sentries glitching out all over the place, but those get shot quickly anyway).  They cant be fought in groups that enormous, so I have to slowly pull one or two at a time, and have the Brawler (with Dissolver fired around a corner) or Sniper deal with them after dragging them to specific positions.

If there was enough ammo, it'd be beatable by simply using that tactic (which is how I often deal with those guys if they dont explode themselves), but it'd take ages.

It's the only mission type that's done anything like this, really.  Everything else has worked quite well.

[x4000]

Point taken on the missions being too large, I will shrink them.

What do you think about making it so there are fewer bots in general, but they all have something like 3x as much health?  Say, half as many bots with 3x or 4x as much health?  That way the combat with single bots is tougher, but there are not so many bots.

The slowdown in missions comes from too many bots trying to pathfind to you from too great a distance, and isn't something I've been able to figure out how to solve.  It's just frankly a lot of data to chew through, and they can't precalculate it because each bot's movement, and yours, affects the decisions of the later ones.  So it's a linear process that can't go in a separate thread because it would still be something you'd have to wait on anyhow, and putting it in another thread would actually be slower because of having to transfer a bunch of data.

In other words, it's an intractable problem unless the bot density is simply culled.  We had to do the same thing with the ships in AI War at one point, although the per-ship pathfinding there is trivial compared to here (by design).  It's simply a different problem in AI War, and the complex AI can indeed be shunted to a different thread that isn't blocking.  Not so here.

The lower difficulties already use fewer bots, and this would just change the upper ones to use the same number of bots, but with higher health, rather than using more bots with the same health.  Probably this change would mirror the Easy number of bots, and so Easy and Casual would see no change at all from my proposed changes.

Thoughts?

[nas1m]

Point taken on the missions being too large, I will shrink them.

What do you think about making it so there are fewer bots in general, but they all have something like 3x as much health?  Say, half as many bots with 3x or 4x as much health?  That way the combat with single bots is tougher, but there are not so many bots.

The slowdown in missions comes from too many bots trying to pathfind to you from too great a distance, and isn't something I've been able to figure out how to solve.  It's just frankly a lot of data to chew through, and they can't precalculate it because each bot's movement, and yours, affects the decisions of the later ones.  So it's a linear process that can't go in a separate thread because it would still be something you'd have to wait on anyhow, and putting it in another thread would actually be slower because of having to transfer a bunch of data.

In other words, it's an intractable problem unless the bot density is simply culled.  We had to do the same thing with the ships in AI War at one point, although the per-ship pathfinding there is trivial compared to here (by design).  It's simply a different problem in AI War, and the complex AI can indeed be shunted to a different thread that isn't blocking.  Not so here.

The lower difficulties already use fewer bots, and this would just change the upper ones to use the same number of bots, but with higher health, rather than using more bots with the same health.  Probably this change would mirror the Easy number of bots, and so Easy and Casual would see no change at all from my proposed changes.

Thoughts?

Sounds good to me!

[Misery]

Sounds like it's worth a try to me.  Particularly as it also sounds like this thing could be a potential issue for those with weaker PCs, or at least that's what I tend to assume.

That being said, most of the missions havent been too much of a problem with this, but the issue has still been noticable to at least some degree in all of them;  there's a distinct speed increase at the end of a mission when I'm moving around and grabbing items before leaving.

There is the issue of balance though; certain bots with enough health, and moreso depending on what they're paired with, could make a situation where the player is inevitably pushed back and beaten by the force of sheer numbers. Numbers as in stats, not the bot army size.  This can already occur to some degree in the current version I've found, after running into the CommandBot's health-boost bug multiple times, though it's likely less of an issue on lower difficulties.  Even weaker bots become a dramatic threat when there's more than a few of them in that state.  Granted, I cant imagine that the health boost would be THAT much, but it illustrates that point well enough.   Possibly.

That'd be the main issue with it, I think.   I cant imagine this being easy to balance out.  Other than that, I dont see any problems with the idea.  It would give the higher difficulties a different feel if it works out, that's always a good thing.

[Teal_Blue]

Hi,
             I haven't gotten out of normal yet, so i am still struggling with making non-optimal mistakes, but i thought i would throw out 2 suggestions and ask if they are feasible?

1. Have groups of bots pathfind in groups, by that i mean if we have 3 groups, thunderbots, then silence, and then ninjas, have the first 10-15 of thunderbots pathfind first, the rest of the thunderbots, as well as the other 2 groups remain active, but do not yet approach. In this way it is perhaps like a que where only the first 10-15, or 15-25? Whatever number you feel is appropriate and doesn't lag the machine, nor throw too many bots at the player all at once. Does this sound workable?

Also,

2. Give one of the exos, maybe assault, or the science exo as misery says they seem the weakest on range? and give them one weapon, perhaps only a weapon available in the expert mode, that has no limit on its ammo. So the missions become doable. Not necessarily easier, but doable.

Does this sound like anything useable?

My intention here is not to throw out unworkable ideas, or things that take the game off course, but to hopefully contribute something that can be useable and helpful. 

Thanks for listening,
-Teal

 

[x4000]

Good points, Misery -- hard to balance out to some extent, but we do have time.

Teal -- pathfinding in groups really only works well in RTS-like games with very granular positions.  We do that sort of thing in AI War from time to time, but it doesn't work with a very coarse positional system like a roguelike.  In terms of unlimited ammo, that's something that we assiduously want to avoid, as scarcity of ammo is pretty important to the game design.

[Teal_Blue]

Okie Doke, i can understand now why this is such a thorny problem. Thank you for explaining it though.

-Teal

 

[keith.lamothe]

Misery, one thing you may find helpful with the new Overload ability (replacing Self Destruct in 0.914) is that:
- It's short range, but by nature of radial AOE it can "hit around a corner"
- If your exo is tanky enough it can do a _lot_ of overloads, kind of functioning as a very short-range high-ammo weapon.  This probably corresponds exactly with those exos that you have trouble giving enough range to attack without being killed first

Naturally this comes at a tradeoff in terms of less available stealth and trap-skill, but so it goes.