Please upgrade your browser for the best possible experience.

Chrome Firefox Internet Explorer
×

What should be our next map and game mode?

STAR WARS: The Old Republic > English > Galactic Starfighter
What should be our next map and game mode?

Luneward's Avatar


Luneward
08.02.2014 , 12:04 AM | #21
Quote: Originally Posted by Cerrabus View Post
How about open space. Free form flight. Joystick support. PVE mission that can be played with friends.
The problem with that is that unless you want PvE targets to be completely scripted, it's almost impossible with current tech. There's pretty much no MMO that uses any significant AI because it would be too difficult for a server to process dozens or hundreds of instances using complex AI. That's why almost all mobs in nearly all MMOs tend to be the type that either just mindlessly stand around or walk pre-designated patrol routes until they are interacted with. Something like that could only really be realistically implemented in a small team game that is hosted on a computer rather than a centrally maintained server.
Steelwind / Kaillen * The Ebon Hawk

Verain's Avatar


Verain
08.02.2014 , 01:21 AM | #22
I am reasonably certain that similarity of game experience is the driving factor, not some computational complexity.

Given that in MOST situations, mob behavior (for example) is really player behavior (specifically tank behavior) it gives the players a great deal of agency over the outcomes of their characters.

Also: You wouldn't want Raid A to be at the last boss, who has two moves that, when combined back to back, can kill an undergeared tank, and for that to happen based on Raid A's boss "realizing" it, while Raid B discovers some way (or just lucks into) some combination that convinces Raid B's boss not to stack the moves back to back. This kind of undocumented behavior would pretty much destroy world first raiding- in that example, Raid A and Raid B would both be undergeared, and some undocumented script behavior would determine the outcome. That kind of crap already happens, and it is unfortunate and unintended, but it would be EVERY boss.

Simply put, AI isn't for these types of games in pve.
"The most despicable person on the GSF forum."

Kuciwalker's Avatar


Kuciwalker
08.02.2014 , 09:32 AM | #23
Quote: Originally Posted by Luneward View Post
The problem with that is that unless you want PvE targets to be completely scripted, it's almost impossible with current tech. There's pretty much no MMO that uses any significant AI because it would be too difficult for a server to process dozens or hundreds of instances using complex AI. That's why almost all mobs in nearly all MMOs tend to be the type that either just mindlessly stand around or walk pre-designated patrol routes until they are interacted with. Something like that could only really be realistically implemented in a small team game that is hosted on a computer rather than a centrally maintained server.
omg no, I hate you people, I keep explain THIS ISN'T THAT HARD. It has been done before on hardware ten years older than what we have.

TrinityLyre's Avatar


TrinityLyre
08.02.2014 , 10:45 AM | #24
Map: Something in the flavor of the game mode I listed below.
Game mode: Cap ship battles/assault or shipyard assault/defense.
Alternate game mode: Toning down gunships and battlescouts.

Nem had some good map ideas way back, I'd like to see some of those. Moving asteroids, so that people can cry about lag even more! Wahaha!
Anastasie / Phytia The Bastion
Respected and Despised & Spouse Aggro
Check out my Galactic Starfighter Compendium and contribute!

Verain's Avatar


Verain
08.02.2014 , 12:36 PM | #25
Quote: Originally Posted by Kuciwalker View Post
omg no, I hate you people, I keep explain THIS ISN'T THAT HARD. It has been done before on hardware ten years older than what we have.

Game AI hasn't budged much since then, and the AI never used up much of the hardware of a 386.


I discovered, personally, several weaknesses of the AI in X-Wing and TIE-Fighter. It wasn't my goal to find these, nor did me knowing about them break in any meaningful fashion my enjoyment of the game, or others.

One I remember quite clearly was a rolling spiral that no enemy AI gun could hit- pretty much ever. Squadron of enemies? No big deal. This predictable and lazy spiral would throw off every enemy shot.


Now, does that sound easy to fix? Any kind of emergent behavior normally is very hard to fix. In TIE Fighter, I recall that the controls were made worse- if I held a button on the joystick that meant "make my left/right roll instead of yaw", which was a big part of this, now it ALSO meant "make my up/down throttle instead of pitch". Since I needed to roll and pitch to do the move, this made it much harder, and the spiral was now a series of small straight lines instead of all curves- and the enemies could it to some extent. I have no doubt that the control scheme was modified to make it harder to accomplish this move. The AI was never fixed- it probably wasn't possible without a redesign.


And that is the core problem with AIs- players often don't understand which games get beat up hard and which don't, and in a single player game when you figure out a thing that the AI can't do or won't do, that's correct play. If, say, raid bosses were designed this way, you'd have a really rough time of it. If PvP objectives were designed this way, you REALLY would.





Here's an example:

You eagerly launch the new game mode, which involves each side having capital ships, each of which has objectives on it to destroy. They are guarded by the enemy team and also by some NPC turrets.

The design is plain- the turrets are supposed to be overwhelming for one player, who should deal some damage and die, but if you bring three ships you should clear a turret and be able to get out of range before anyone takes hull damage.

In this world, if you figure out a way not to get hit by the turret- say, railguns outrange them, or they have to actually predict your future position to fire such that you figure out something that they can't hit- then suddenly you are some unhittable god, but a new player just gets shredded. Then the devs are like "ok, we fixed it so that railguns can't be queued, and we made it so that you can't be missed by the turret". Then your play and experience don't mean less, they mean nothing, and a ship type is gone from the mode.

Do you see how they didn't have a KNOB in that example? They couldn't just say "make it better". If you make something with realistic assumptions- such as the X-Wing and TIE Fighter games, where a laser could pass between your solar panels and above your cockpit- and then have the pace of the game entirely perceivable by humans, then humans will always just dominate the scripted and predictive AIs. You need another level of software (which doesn't exist in easy to implement, easy to maintain, or computationally cheap manners) to actually be able to learn.


What type of software would have been able to target me firing my spiral? The motion for the spiral was pretty simple- I'd take my stick left while holding the "roll" button, then move the stick up and down. Right worked just as well. This meant that I was constantly rolling, and constantly varying my pitch. The code obviously predicted where I was going to be based on my current position and vector, and it also "knew" that I could accelerate or whatever. What it couldn't understand was the reasonably basic pattern that I would continue to increase my pitch to a max, then decrease it. As such, it would land very very few shots. A human would figure it out instantly.


And again- I want to emphasize- this isn't something they can patch. Not cheaply. AI requires ludicrous hours of test compared to everything else, and the AI has to key based on something- so that if you gave it the ability to understand that roll, it might think the roll was happening at other times. You don't get to say "when the player is doing X" -> go look at that code. It's an emergent behavior. Fixing it would be really hard.


So you'd have an uneven play experience, with some players thinking something is hard and others thinking something is easy. You'd lose games because your AI was outsmarted, and win because it picked randomly correctly. You'd face a myriad of exploits that wouldn't exist in the game universe, you'd grind against the rough edges of an AI built to be challenging but beatable, a very very hard and arbitrary line that varies based on player.

So no, I don't think you'd like any of the AI things.




I think you COULD do stuff like:

> Four turrets in a ball, rotating.
> Turrets don't shoot through each other.
> Turrets can individually acquire targets.
> Turrets can rotate infinitely fast.
> Turret shots will always hit unless evasion stops it.
> Ball of turrets follows a set path.

Now you, the friendly player, know that this turret ball will be a small speedbump on your enemy- he'll have to stand and snipe them, or he'll have to fly in and burst them, but if he just flies past he'll get harried by their shots and worked down.


But this isn't what most players mean when they say AI. The game could deliver you something akin to a raid experience, with some concessions, but it can't be like Wing Commander or whatever without having all the problems that all those games had that would REALLY be a big deal in a multiplayer pvp environment.
"The most despicable person on the GSF forum."

Verain's Avatar


Verain
08.02.2014 , 01:19 PM | #26
Quote: Originally Posted by TrinityLyre View Post
Nem had some good map ideas way back, I'd like to see some of those. Moving asteroids, so that people can cry about lag even more! Wahaha!
The asteroid movement could be independently calculated locally (lagless). Moving asteroids wouldn't be possible to lag into.

The problem seems to be that the engine doesn't understand changing terrain, or something?
"The most despicable person on the GSF forum."

Kuciwalker's Avatar


Kuciwalker
08.02.2014 , 02:33 PM | #27
Verain, I'm aware of all of those things. However, for "PvE" style objectives in PvP, where you have to destroy an enemy capship or whatever, as long as the defenses are static things like turrets it's not that hard to make them not very exploitable. (Especially if the enemy team is likely to be around to disrupt any exploit you try to use!)

Whereas, yes, I completely recognize that a pure-PvE not-on-rails version would not have the same consistency of experience of something like a raid. I just don't care. I want the PvE anyway.

Ramalina's Avatar


Ramalina
08.03.2014 , 02:13 PM | #28
What it sounds to me like a lot of people are craving in GSF are:

Story: What exactly are we doing here? GSF is quite weak here. Class story or X-wings series style story is clearly beyond the GSF budget, but with the whole Kuat Drive yards mini-expac there would have been a natural fit for answering the question, "why are we doing this."

Rational Mission Objectives: I think this is what a lot of the, "more PvE please," requests stem from. What exactly is the strategic and tactical objective of the battle?

You don't shoot down enemy pilots just for the sake of shooting them down. You shoot them down because you want to do something and they are an obstacle to doing it. Same for PvE elements. Blowing stuff up for the sake of blowing it up gets old fast, even if the secondary explosions are really pretty.

You want something like: Shoot down fighter screen>Destroy particle shield generators>Destroy airdock door control (think drydock, but for spacecraft instead of oceangoing craft, welding underwater ain't easy, neither is welding in hard vacuum) > GOAL stop construction of enemy capital ships at this facility by wrecking the shipbuilding equipment inside the airdock.

I realize that Admiral Aygo is really keen on having us fly in circles around unmanned satellites, but I think he must be on some really, uh, interesting medications to believe that this is a useful expenditure of military resources, compared to say, defending the Republic shipbuilding facilities on the Kuat orbital rings.

Stronger incentives for team play: The thing about the ground PvP vs GSF and PvE vs GSF is that the simple scripted objectives in GSF aren't that great at helping people learn or desire team play. In group PvE the mechanics FORCE a certain level of team play for all non-trivial content. In PvP this is not true, but some of the objectives have very easy and obvious ways to benefit from team play, and as players get better they can build on that if they want to (though it doesn't happen all that often in PUGs). GSF has more space for teamwork to benefit players than in the ground game I think, but does a poorer job of getting them thinking about that in some ways.

If the above are really the desired outcomes, then doing things like capital ship battles, escorts, hardened target bombing, etc. is more feasible because you're not trying for a war simulation, but just trying to provide certain attributes of gameplay experience. Something that is more within grasp of the engine's limitations.
"A padawan's master sets their Jedi trial, Rajivari set mine."
- Zhe Lian, Sage.

Twitch

Itkovian's Avatar


Itkovian
08.06.2014 , 07:18 AM | #29
Quote: Originally Posted by Nemarus View Post
I think some type of hardened objective assault/defense (like capital ship) is what would garner most popular support and hype.

And it shouldn't be too hard to implement in the current GSF engine, as long as the objective doesn't move.

Basically, just make a big capital ship, put it in interesting terrain that blocks long range LOS to some elements (like a huge repair facility), and slap a very big number of very durable targets/turrets on it--each with different defensive qualities that reward different weapon choices, ship classes, and components.
This. Most definitely.

One particular advantage to such a mode is that it would make bombers actually act as bombers. Load up protorps, go bomb the enemy! Likewise, Strike fighters would play a greater role.

In fact I believe there was some datamined info that hinted at a game mode like that. Hopefully that mode wasn't scrapped, but merely put on the back burner.

Regardless, I think the single most important update to GSF would be a new game mode to add actual variety to the gameplay.

Verain's Avatar


Verain
08.19.2014 , 09:27 AM | #30
Quote: Originally Posted by TrinityLyre View Post
Map: Something in the flavor of the game mode I listed below.
Game mode: Cap ship battles/assault or shipyard assault/defense.
Alternate game mode: Toning down gunships and battlescouts.
I think one of the things is, if they would give us different trays, each tray could have different rules. So you'd have your hangar for game mode A, your hangar for game modes B and C, etc. This could allow the game to launch a game that is "scouts only", using the tray you can only put scouts on, for instance (and if you didn't, well, you wouldn't get that queue pop!)

I mean, FPS games have a LOT of variation that we don't see in GSF, but much of it would apply. You have the idea where one character is supercharged and the enemy team gets a bonus for killing him, you have the idea where bases are established and meaningful (and several variations on that), you have capture the flag (and variations on that), etc. We don't see any of that in the current game, but there's space for all of it.
"The most despicable person on the GSF forum."