release the hounds!

[MalReynolds]

Please bring back T2 hounds. Yeah, the highly annoying ones that danced around in a room, mostly out of sight, until you enter, when they yappingly converge. Those ones. The ones that would then melt back to the walls when you went back to the corridor; and would only follow you if they had damaged you sufficiently and smelled blood. Man's best friend.

587e8615c7f34281911b2748c84563.jpg (2.62 KiB) Viewed 2314 times

Make them rare if you want; but I miss them.

They can't be any more annoying than pack after pack of cold hatchlings.

[Crim, The Red Thunder]

Or at least bring back the pack mentality ai they used.

+1 for this.

[MalReynolds]

You're right; I think we should add the hounds-hang-in-rooms dynamic for some critters, but I think packing behavior is even more important.

Creatures with pack_in=true would do better at surrounding the player. Assuming they are melee, if a move which lessens distance to the player is unavailable, they would choose any unoccupied spot adjacent to the player and try to path there, even if they have to go a ways around. That would include hounds, probably melee orcs, things like that.

I think this could be part and parcel of adding better AI about when a monster is forming an obstacle to packing in. It's a pretty simple situation to check.

(1) Monster X is adjacent to @. (2) There is a monster Y which is adjacent to X, but which has no path to @. (3) X has pack_in=true. (4) X has a square adjacent to it and also adjacent to @ which is unoccupied. (5) X moves to this square. There might be some additional fiddling needed, but I think 1-2-3-4-5 would take care of 95% of all available situations.

Ideally, things like hounds and melee orcs should be trying to surround @ to cut his mobility down to zero.

Packers have one goal: take an extra hit to pack the player, increasing attack exposure and cutting down escape.

This would require much more active management by @ to create bottleneck situations and managing escape routes, and would avoid than this ridiculous situation, which makes a mockery of basic AI:

Code: Select all

#.......#
#...@...#
####h####
####h####
####h####
####h####

[MalReynolds]

Actually, the algorithm above would produce this:

Code: Select all

#..h.h..#
#..h@h..#
####h####
####h####
####h####
####h####

The goal is to truly surround--to get rid of that top escape spot. With perfect packing, you need a monster to move to the top spot even though there are no adjacent pathless monsters, if, say, there are a sufficient number of pathless monsters in LoS. The situation right above would be harder to rectify; but I'm sure there's a tweak to do it.

Try this as an addition to 1-2-3-4-5:
(A) Enough pathless monsters in LoS; (B) monster X has a square both adjacent to X and to @ which is empty; (C) X moves to that spot.

This would lead to some flow around @ as packing in happens; but would stop the flow if there aren't enough adjacent spots to make moving to an empty spot worthwhile (if there aren't enough packers to completely surround). Ideally you would want the monster farthest away from the pathless monsters to be the one flowing; that would usually be the case if the flow is being set up around a relatively passive player, but could lead to oddities if a player kills a monster close to the pathless monsters, opening up a spot which then might be filled by a non-optimal move.


Just a thought; an interesting problem.

[Jeoshua]

I've noticed that a lot. Creatures lining up for the slaughter, that is. And with so many good options to attack all of them at once, combat quickly becomes a game of "find the bottleneck". That's par for the course in a lot of roguelikes, but that doesn't mean it SHOULD be like that.

One situation I've noticed a lot is something like this:

Code: Select all

  @  
###x #
###xo#
###  #

There is clearly enough room for the "o" monster to get through, but they just sit there waiting for their turn to die!