Friday Facts #219 - Cliffs

[grimdanfango]

What I think should happen: The underground portion is normal on the top part of the cliff, but the other end comes out the side of the cliff (instead of the ground below it).

Actually, I think this is the best suggestion I've seen... it would make belts/pipes look like they blend in thematically with the cliffs, without needing to have a weird vertical offset (which as suggested won't happen, as it would conflict with too many systems) - they could be left in-line on both sides of the cliff. If you remove the cliff in question, you just remove that going-into-the-rockface endpiece too, and would need to lay down a new regular underground piece, or just remove both and connect above ground.

To place the cliff-entering segments, you could just place a regular underground segment, but when placed on a cliff, it swaps the graphic out for the appropriate cliff-entering variant. That way it wouldn't require any additional items.


It'd also be neat if walls would auto-blend into cliff segments if you place wall on a cliff-face, so you have a feeling that they're emerging from the cliff, rather than stopping just short of it.

[PursuedMeatl325]

Did anyone notice:

TOGoS said:
There are also some special cases about how they interact with projectiles, but for the most part, cliffs just act as walls.

I wonder what the special cases are
It may not just be cliff explosives

[Andrzejef]

Maybe, uhhh, lasercutter?
Or plasma-cutter (bundled with new kind of biters, say night-time only)

[Mr Alternative]

Cliffs in Factorio are an exciting and terrifying concept to me. Whilst I'm excited for it, as I think it's very implementable in Factorio's current state. However, what concerns me is the game A.I.; pathfinding to be specific.

Has there been any trials with biter A.I.? If so, does the A.I. make additional checks for terrain type n, and in either case, has that produced greater CPU overhead? If I were a betting man, I'd predict that swarms of biters between canyons/cliffs, agro'd by nearby pollution, yet impeded by seemingly an inability to see the cliffs in front of them, would be a frequent occurrence...

With that in mind, it'd be great to have an option to disable cliffs at world generation. I know this seems counterproductive, but it might preempt problems for the end user.

Otherwise, I'm excited by the prospect of cliffs. Probably moreso than the new shirts coming out.

[mrvn]

Cliffs in Factorio are an exciting and terrifying concept to me. Whilst I'm excited for it, as I think it's very implementable in Factorio's current state. However, what concerns me is the game A.I.; pathfinding to be specific.

Has there been any trials with biter A.I.? If so, does the A.I. make additional checks for terrain type n, and in either case, has that produced greater CPU overhead? If I were a betting man, I'd predict that swarms of biters between canyons/cliffs, agro'd by nearby pollution, yet impeded by seemingly an inability to see the cliffs in front of them, would be a frequent occurrence...

With that in mind, it'd be great to have an option to disable cliffs at world generation. I know this seems counterproductive, but it might preempt problems for the end user.

Otherwise, I'm excited by the prospect of cliffs. Probably moreso than the new shirts coming out.

Devs mentioned that cliffs are just walls. So the AI will walk around it to get to the turret behind it just like with walls. But if you have a long wall segment then the AI attacks the wall. Will it also attack cliffs? Or is there a gap in the cliffs often enough to never trigger that behaviour?

[Mr Alternative]

Devs mentioned that cliffs are just walls. So the AI will walk around it to get to the turret behind it just like with walls. But if you have a long wall segment then the AI attacks the wall. Will it also attack cliffs? Or is there a gap in the cliffs often enough to never trigger that behaviour?

You've basically hit the problem I'm foreseeing. Attacking a concrete wall is all well and good for pathfinding - it's still a straight path...and that's just for one biter. What about a swarm? Cliffs present multiple paths to a location - whats to stop biters from splintering off into smaller mobs (due to pollution/biter agro algorithm). After all, will biters be able to attack them? Why even have cliffs then, if they serve primarily to be extra large walls that the player cannot construct, but is free to destroy as well? Indeed, a compromise here, I guess, would be to allow perhaps only Behemoth-sized bugs to break cliffs...but then you must solve the problem above anyway.

Again, I love cliffs as a concept, and I've been seeing a lot of comparison to C&C, which is great too, but I must iterate that C&C (in my opinion) handled pathfinding for cliffs and similar barriers without a great deal of finesse. They hugged walls, cliffs and bridge rails on the way to their target, in one, long processional line...and that was going the 'semi-destructible' route, where you could shoot sections of the cliff (in Tiberian Sun) to bridge the higher and lower sections. Alas, AI treated these the same; one long, wall-hugging line of units.

[xng]

...

...

About AI attacking cliffs..

As I understood it, there will be special explosives to remove cliffs, which would basically mean that it is invulnerable to other attacks or those special explosives wouldn't be necessary.

[Mr Alternative]

...

...

About AI attacking cliffs..

As I understood it, there will be special explosives to remove cliffs, which would basically mean that it is invulnerable to other attacks or those special explosives wouldn't be necessary.

....Sort of. I think you've missed what I was saying. If the AI cannot destroy cliffs, if it is too difficult to go around them (this is purely pathfinding - how pollution/agro with regard to finding paths to pollution sources is unknown to me), you must make a decision: should you make the biter AI able to destroy them too, like they do with player-constructed barriers? Can every biter destroy them, or just the bigger ones? Or maybe smaller biters can, if en massed? If yes to any of the previous three questions - how big of a hole would the biters make? just enough for one, or will it be programmed to make a sizeable hole for a swarm to fit?

If no (they can't destroy them), then how will the biters pick the best path when there are a lot of them, and a lot of potential paths through cliffs/canyons are presented? This is not an unlikely situation, especially if a biter base is a significant distance from a pollution source and, say, 'map elevation' were set to high at map generation.

Then you have to consider the gameplay ramifications of each decision. If you make it so only the player can destroy cliffs, you can run into a number of issues (as described in previous posts) that other, similar games have also had to deal with. If both can destroy cliffs, they become glorified concrete walls, which the player can't destroy until explosives are researched/produced, whilst the biters have that ability from the start.

Then there are corners - what if the AI paths itself into a corner? Well, you might get the 'stuck swarm effect' I described in my second post, if biters can't destroy cliffs. In fact, they'd keep building in that corner, like a swarm...

[mrvn]

...

...

About AI attacking cliffs..

As I understood it, there will be special explosives to remove cliffs, which would basically mean that it is invulnerable to other attacks or those special explosives wouldn't be necessary.

....Sort of. I think you've missed what I was saying. If the AI cannot destroy cliffs, if it is too difficult to go around them (this is purely pathfinding - how pollution/agro with regard to finding paths to pollution sources is unknown to me), you must make a decision: should you make the biter AI able to destroy them too, like they do with player-constructed barriers? Can every biter destroy them, or just the bigger ones? Or maybe smaller biters can, if en massed? If yes to any of the previous three questions - how big of a hole would the biters make? just enough for one, or will it be programmed to make a sizeable hole for a swarm to fit?

If no (they can't destroy them), then how will the biters pick the best path when there are a lot of them, and a lot of potential paths through cliffs/canyons are presented? This is not an unlikely situation, especially if a biter base is a significant distance from a pollution source and, say, 'map elevation' were set to high at map generation.

Then you have to consider the gameplay ramifications of each decision. If you make it so only the player can destroy cliffs, you can run into a number of issues (as described in previous posts) that other, similar games have also had to deal with. If both can destroy cliffs, they become glorified concrete walls, which the player can't destroy until explosives are researched/produced, whilst the biters have that ability from the start.

Then there are corners - what if the AI paths itself into a corner? Well, you might get the 'stuck swarm effect' I described in my second post, if biters can't destroy cliffs. In fact, they'd keep building in that corner, like a swarm...

I think aliens shouldn't be able to destroy cliffs. Because if they could why didn't they do so in the last million years before out arrival?

All the rest I think is fine. If the aliens can't find a way down the cliff then they can't attack. As long as they do't start using 100% cpu trying to find a way over and over that is fine. The path finding does allow for huge detours around water and cliffs should have ramps way more often. So I doubt that there would be a problem as long as the path finding handles cliffs as indestructible obstacles.

[xng]

I think aliens shouldn't be able to destroy cliffs. Because if they could why didn't they do so in the last million years before out arrival?

All the rest I think is fine. If the aliens can't find a way down the cliff then they can't attack. As long as they do't start using 100% cpu trying to find a way over and over that is fine. The path finding does allow for huge detours around water and cliffs should have ramps way more often. So I doubt that there would be a problem as long as the path finding handles cliffs as indestructible obstacles.

I agree, I think the biters are supposed to be "dumb" and not always be able to figure out a way to get inside the base if there are natural hinderance. I would still welcome the introduction of some more intelligent antagonist in the future, like an AI rogue robot or something, and also some biter variants that maybe can climb cliffs. Spitters could still be a danger even if they can't find a way up/down.

Cliffs are my absolute favorite new feature, I also wish we didn't have landfill to begin with but could only build bridges. (As default setting, with landfill existing in creative mode and easy mode)

[Mr Alternative]

...

...

About AI attacking cliffs..

As I understood it, there will be special explosives to remove cliffs, which would basically mean that it is invulnerable to other attacks or those special explosives wouldn't be necessary.

....Sort of. I think you've missed what I was saying. If the AI cannot destroy cliffs, if it is too difficult to go around them (this is purely pathfinding - how pollution/agro with regard to finding paths to pollution sources is unknown to me), you must make a decision: should you make the biter AI able to destroy them too, like they do with player-constructed barriers? Can every biter destroy them, or just the bigger ones? Or maybe smaller biters can, if en massed? If yes to any of the previous three questions - how big of a hole would the biters make? just enough for one, or will it be programmed to make a sizeable hole for a swarm to fit?

If no (they can't destroy them), then how will the biters pick the best path when there are a lot of them, and a lot of potential paths through cliffs/canyons are presented? This is not an unlikely situation, especially if a biter base is a significant distance from a pollution source and, say, 'map elevation' were set to high at map generation.

Then you have to consider the gameplay ramifications of each decision. If you make it so only the player can destroy cliffs, you can run into a number of issues (as described in previous posts) that other, similar games have also had to deal with. If both can destroy cliffs, they become glorified concrete walls, which the player can't destroy until explosives are researched/produced, whilst the biters have that ability from the start.

Then there are corners - what if the AI paths itself into a corner? Well, you might get the 'stuck swarm effect' I described in my second post, if biters can't destroy cliffs. In fact, they'd keep building in that corner, like a swarm...

I think aliens shouldn't be able to destroy cliffs. Because if they could why didn't they do so in the last million years before out arrival?

All the rest I think is fine. If the aliens can't find a way down the cliff then they can't attack. As long as they do't start using 100% cpu trying to find a way over and over that is fine. The path finding does allow for huge detours around water and cliffs should have ramps way more often. So I doubt that there would be a problem as long as the path finding handles cliffs as indestructible obstacles.

Indeed, the majority of points I bring up are 'doable', but that really doesn't answer my initial questions, so I shall phrase them for clarity:

How does the current state of the biter A.I. react when cliffs are introduced into the game? Does it produce any greater overhead (in a small, 'lab' situation, say) than current situations?

This is not an exhaustive list of problems cliffs might require for pathfinding, so I'm curious as to whether the devs have tried it in-game and whether it has presented issues similar to past games of the genre. In addition to my concerns about pathfinding, I am concerned about how late-games will be affected by the addition of cliffs. Factorio is a CPU intensive game, and personally I don't see more cycles being dedicated to calculatiing an off-screen swarm of biters path, whom now must now navigate a variable maze (I say 'maze' because I don't know how dense map generation settings may end up being, so I'm assuming that dense concentrations are possible (certainly with perlin noise)) being that beneficial to smooth frame-rates in the late game.

By bets are on the AI being very derpy.

[Rseding91]

Indeed, the majority of points I bring up are 'doable', but that really doesn't answer my initial questions, so I shall phrase them for clarity:

How does the current state of the biter A.I. react when cliffs are introduced into the game? Does it produce any greater overhead (in a small, 'lab' situation, say) than current situations?

This is not an exhaustive list of problems cliffs might require for pathfinding, so I'm curious as to whether the devs have tried it in-game and whether it has presented issues similar to past games of the genre. In addition to my concerns about pathfinding, I am concerned about how late-games will be affected by the addition of cliffs. Factorio is a CPU intensive game, and personally I don't see more cycles being dedicated to calculatiing an off-screen swarm of biters path, whom now must now navigate a variable maze (I say 'maze' because I don't know how dense map generation settings may end up being, so I'm assuming that dense concentrations are possible (certainly with perlin noise)) being that beneficial to smooth frame-rates in the late game.

By bets are on the AI being very derpy.

Currently they'll treat cliffs like anything else they can't go through and go around them. It does add potential overhead to path finding since it's 1 more thing they have to go around. Other than that if a cliff doesn't exist on the path they want to take it adds no additional overhead.

[mrvn]

How does the current state of the biter A.I. react when cliffs are introduced into the game? Does it produce any greater overhead (in a small, 'lab' situation, say) than current situations?

This is not an exhaustive list of problems cliffs might require for pathfinding, so I'm curious as to whether the devs have tried it in-game and whether it has presented issues similar to past games of the genre. In addition to my concerns about pathfinding, I am concerned about how late-games will be affected by the addition of cliffs. Factorio is a CPU intensive game, and personally I don't see more cycles being dedicated to calculatiing an off-screen swarm of biters path, whom now must now navigate a variable maze (I say 'maze' because I don't know how dense map generation settings may end up being, so I'm assuming that dense concentrations are possible (certainly with perlin noise)) being that beneficial to smooth frame-rates in the late game.

By bets are on the AI being very derpy.

No bet. The AI is derpy.

But my bet is that it will find a way around the cliffs and attack a nearby turret if it exist. If nothing is near the cliff then all bets are off. Aliens attack trees too for no reason, why should cliffs be different?

[Andrzejef]

It's not for no reason, they are after the tasty grubs living there under the bark

[QGamer]

But my bet is that it will find a way around the cliffs and attack a nearby turret if it exist. If nothing is near the cliff then all bets are off. Aliens attack trees too for no reason, why should cliffs be different?

Trees are made of wood. Cliffs are made of rock, which must be much harder to eat through.

[Andrzejef]

Maybe they have some special enzyme in their saliva? I mean, you never know until one chomps on your leg

[Dron__]

I offer several options for lowering the pipes

1.png (199.9 KiB) Viewed 8642 times

you can also dig a tunnel for pipes

1

drip a tunnel.jpg (318.3 KiB) Viewed 8629 times

[mrvn]

Since cliffs are just walls you have to put an underground pipe there. So it goes into the ground before the cliff and comes out of the ground after the cliff.

[hitzu]

It's not optimal, but it works.

[5thHorseman]

It's not optimal, but it works.

Not only does it work, but it's how it works in the game.