Basically, I want doors that open (or close) the moment they get power. No toggle-on-and-off, no unpowered opening. They'd have higher HP and so on.
This would allow you to remotely open the doors. Why is this important?
Turrets. You could have a whole batch of turrets and keep someone on hand near a power switch so whenever a enemy gets near, the hatches slam shut in their face.
It's cheesy, but it's also a tactic that you COULD use in real life.
I'd say the game doesn't need more invitations to be cheesy/exploitative. Just because it "COULD" happen in real life doesn't make it good gameplay or necessary to include.
once again, can we look a the other applications of this though... like emergency vents for fires or hatchs to keep out man hunter packs or fend off assault by mechanoids... a hardier door that is essentially a wall piece could be super handy aside from shooting at dudes....
And this is why it won't be added to the game. Autodoors used to be remotely controlled. Tynan decided it allows too many exploits.
No, autodoors used to be instantly, magically controlled. What I'm suggesting is a semi-manual door that you have to toggle the power of(either through switch or manually toggling) to change.
It wouldn't change much. You could lead attackers one way towards turrets, then remotely close door so they have to go another way. Then close the 2nd way and open the 1st way. Tynan is not a fan of killboxes.
This could be solved with AI. Make it so you can only change paths twice before deciding to go through walls, eh?
It could also be solved by not putting it in at all :)
I'm sorry, but honestly, the wall-related AI already needs a overhaul without this.
Case in point: The Cask of Amontillado(Aka, walling in a entire trade caravan)
Hey, no one's arguing against fixing existing expoits.
The only way I could see cool would be a functional standpoint first with things like firedoors and for controlling ventilation, etc.
after that, we are talking serious AI work so it won't be majority exploiting the AI.
How would it be serious?
You could have it so the AI is willing to take X number of steps before it starts attacking walls.
You could have it assign threat to anything that inflicts injury and pursue it even if the path is cut off.
ETC
AI work is probably one of the most difficult parts in game development. Even moreso to make one that cannot be exploited. A damage pursuit flagging allows for things like tag or ping pong with the npcs or make even more complex killboxes possible.
Something simple could be open up door with turret. Guys go after turret. Close door and open up ten flanking turrets. How does one have the AI respond. Still try breaking down the door while being shredded?
Autodoors that vent temp? First thing I would do is pile wood or flammables in it. Lure a group of guys in, light a couple fires, then slam doors down, or have a subdivided hallway lined with heaters.
Essentially that is the part that makes it so difficult. Having to have an AI that acts intelligently regardless of the change in circumstances. This way it doesn't walk into otherwise obvious traps and so they know how to get back out.
The way to do that would be it would need to route regardless of doors and assume it will need to tear down doors. On top of that, have things like evading turrets. The easiest way is to have a value to break down a door or wall equal to the pathing value. Add on turrets to that and evading priority over attacking turrets if there is another route.
An AI is never really an AI. It is just programming something to react to all possibilities. Like with knowing traps so the npcs can evade them then that trap spot is useless. Making NPC's break down walls and doors is easy. But with auto open and close door exposing and ending routes to flagged threats, that is a lot of work to do without making npcs ping pong.
But you see, the thing is that in real life, you'd have 'rock and a hard place' setups like that. And in moments of panic, who says that the AI should act intelligently?
Well, in real life, there is much more complexity to the system. The ability to split up, fall back, gather intel and plan for weakspots. So in the case of my example of trying to pinball with one hatch at first with one gun, then ten opening behind, the smart AI would fall back and hit somewhere else all together.
Result would be the only way to have a killzone, is the entire perimeter is a killzone. Then your pawn's path is now the weakpoint and the killbox play is now over. An overall win since if you have fixed emplacements, and the AI is exploiting a weakness, you need your mobile pawns for defense. Now hatches aren't so bad since the AI is accounting for that and hitting accordingly. That makes it very much more like an RTS AI, but that is quite complex for a ground up build from a single person.
The only way to do it is if all fixed emplacements have their threat range as part of AI routing and the AIs better plan wall breaching. They will go towards the weakest location. That would need to include your pawn's threat as well.
Potentially better AI targeting even to disable firmly entrenched defences. 60 turrets at a one way in takes a lot of power. For vanilla, that means you have one heck of a power vulnerability somewhere.
Quote from: hector212121 on May 10, 2016, 09:09:07 PM
But you see, the thing is that in real life, you'd have 'rock and a hard place' setups like that. And in moments of panic, who says that the AI should act intelligently?
In other words, it's perfectly fine to build a new exploit that the AI can't cope with. I can think of someone who would probably disagree with you.
you shouldnt be naysaying design suggestions just because they can be exploited, its like saying you wont eat food because it can spoit :\
the ai needs work, everyone gets that, so it isnt a big deal to implement new ideas before that overhauls done. an if its done well, then many exploits like the ones being banged on about in this thread wont exist regardless of what a colonist controlled door will do.
quit derailing threads with chatter about exploits :\
i still think a manual blast door could be a really handy tool for the players kit, an i think we need some new interesting tools for players to play with in regards to actually playing the game. all that story stuff is fun, but i want to be actually doing fun things while i wait for the story stuff to happen
well the main exploit has often been that enemies rarely recognize trap areas and killboxes.
sapper raids were supposed to tackle this AI derp, but sapper AI can easily be exploited as well since they tend to ignore ambushers. (edit : sapper raids also tend to have fewer raiders and they often send them in waves, which makes them easier to deal with)
so i think every raid should just always have sappers (ie : sapper AI for grenadiers would only activate once they are already inside home zone of player, but act like normal raider AI prior to this. they would also have the added bonus of having tons of fellow raiders in a normal raid).
i also think bows and thrown weapons should be able pass over walls in any unroofed area (but with an aim penalty).
i also think raiders should be more able to recognize killboxes by checking number of deaths of each location. this AI death map will tell raiders which areas to avoid, like making "restricted zones" for raiders. this death map will be updated after each raid.
and since every raid is gonna bring sappers with them, it doesn't matter if players hide under a mountain or if they use power hatches or whatever. raiders would always path using the death map as a guide.
Quote from: rexx1888 on May 10, 2016, 10:48:59 PMyou shouldnt be naysaying design suggestions just because they can be exploited
The whole
point of the idea seems to be "I want a new exploit."
Quotequit derailing threads with chatter about exploits :\
The guy said flat out in the first post, it's cheesy but I want it anyway. It's not "derailing" to reply to what was said right in the OP.
The
only justification he gave is because he could imagine it happening in real life.
I guess I haven't started beating the drum on this forum yet, so mark your diaries, I'm saying this here for the first of many times:
"Realism" should always take a back seat to good gameplay, particularly in system-driven games like RimWorld.
When the ONLY reason to recommend a new system is NOT because it would be fun, NOT because it would add gameplay, NOT because it would be well balanced, but because "maybe it could happen in real life" ... that is not a good reason to suggest something.
PARTICULARLY when it's obvious at a glance
even to the person suggesting it that it is cheesy and invites exploits for chrissake!
Quote from: rexx1888 on May 10, 2016, 10:48:59 PM
you shouldnt be naysaying design suggestions just because they can be exploited, its like saying you wont eat food because it can spoit :\
No, it's like saying you shouldn't eat arsenic because it's not very healthy.
Quote from: keylocke on May 11, 2016, 12:11:13 AM
well the main exploit has often been that enemies rarely recognize trap areas and killboxes.
sapper raids were supposed to tackle this AI derp, but sapper AI can easily be exploited as well since they tend to ignore ambushers. (edit : sapper raids also tend to have fewer raiders and they often send them in waves, which makes them easier to deal with)
so i think every raid should just always have sappers (ie : sapper AI for grenadiers would only activate once they are already inside home zone of player, but act like normal raider AI prior to this. they would also have the added bonus of having tons of fellow raiders in a normal raid).
i also think bows and thrown weapons should be able pass over walls in any unroofed area (but with an aim penalty).
i also think raiders should be more able to recognize killboxes by checking number of deaths of each location. this AI death map will tell raiders which areas to avoid, like making "restricted zones" for raiders. this death map will be updated after each raid.
and since every raid is gonna bring sappers with them, it doesn't matter if players hide under a mountain or if they use power hatches or whatever. raiders would always path using the death map as a guide.
I like the idea of a deathmap, especially as a component to a larger AI overhaul. Another problem is raiders sending absolutely silly people at you that add "fuzz" to the point value used for raids. The smarter the loadouts are about distributing points, the less raids where a dozen people with excellent shivs and superior shields try to dig through a wall.
Better loot for the player is easily matched by smarter AI and better loadouts for attackers. I'd be much more intimidated by three assault rifles held by competent marksmen than the current "brawler with a sniper rifle" randomness. And it's much easier to balance the game around consistency in threat level at point thresholds.
summed up: possible, but needs a lot of AI work to make it unexploitable. as that AI work would greatly benefit the game anyway, why the hell not? although the game should probably give courses in military tactics, then..
The thing about a deathmap is that after a while, the map would say the same thing about everywhere...because everywhere would be the same.
Perhaps a "security map" where enemy pawns who ESCAPE(!) remember where security is that they've LOS'd and judge based on security "strength" and turret count--so a single, say, Reinforced Steel(from a mod) turret would be worth more than two normal steel turrets because it has ten times the health, meaning it'll probably deal 2.5 times more total damage(assuming that the normal ones get targeted evenly).
QuoteThe thing about a deathmap is that after a while, the map would say the same thing about everywhere...because everywhere would be the same.
deathmap would just update itself after each raid. for example : if a lot of raiders die in X location, then they would avoid pathing through that location.
raiders would always take the path with the
least amount of deaths, so even if the entire map is full of killboxes, raiders would always take the safest path. (ie : path with least amount of deaths)
Quote from: hector212121 on May 10, 2016, 07:02:55 PM
How would it be serious?
You don't understand how difficult programming is. You keep using high level concept, but computers operate on very low level terms. Computers are complete idiots and have to be given instructions for every possible case.
Easy to make AI would most likely be easy to exploit.
As a simple programmer, I can easily say that the work the AI needs is beyond any "simple" code. You would need to spend time on something like this, and it adds more fixes than it does solutions, so no, your idea isn't a good idea. I'm sorry.
And for the record, computers run on true and false as previously stated. This means a true and false needs to be given for every situation an AI runs into. EVERY situation. It's not simple.
hmm.. iirc, pathing in rimworld works like this :
if you order a pawn to go across the marshland from one point or another, each floor tile have different levels of traversal value. ie : walking on dirt or mud is much slower than walking on cemented pavement.
so the navigation system of pawns already take into account the traversal value of each different type of floor tiles to reach it's target destination by taking the fastest and most efficient path.
a deathmap, essentially just adds a layer of information over the value of these floor tiles after every raid. ie : a floor tile could have X traversal value + deathmap value.
so the pawns will just take both the traversal value and the deathmap value in choosing the most optimal path, the same way that each pawn already takes each floor tile's traversal value by default.
Quote from: hector212121 on May 11, 2016, 09:20:53 PM
The thing about a deathmap is that after a while, the map would say the same thing about everywhere...because everywhere would be the same.
ah, but if you balance the lethality value of one tile against any other tiles, while total deaths in different locations would go up, the "death quotient" relative to other areas would still be reflecting the actual lethality of the location. so while, after playing for (ingame) decades, any spot on the map would have collected hundreds of deaths, your defensive positions would have collected a few thousand - and the death map would still be accurate.
on another note, I suggest that the death map would reset to its previous state if you kill all attacking raiders, but if some of them flee, the death map for that faction would be permanently updated.
Quote from: blub01 on May 12, 2016, 08:55:56 AM
on another note, I suggest that the death map would reset to its previous state if you kill all attacking raiders, but if some of them flee, the death map for that faction would be permanently updated.
oh yiiiss.. leave no witness alive.. hahaha. ;D
Quote from: keylocke on May 12, 2016, 09:37:35 AM
Quote from: blub01 on May 12, 2016, 08:55:56 AM
on another note, I suggest that the death map would reset to its previous state if you kill all attacking raiders, but if some of them flee, the death map for that faction would be permanently updated.
oh yiiiss.. leave no witness alive.. hahaha. ;D
exactly.