58 replies to this topic

[Tombstoner]

LordKnightFandragon, on 29 January 2015 - 04:19 PM, said:

WHy is AI so hard? games many years older have made AI with all of the above and it works fine. I wont act like its easy to do, but many games before have done it, why couldnt PGI also? MW4 has AI and they can make for quite an amusing fight.

As for the AI aiming, it seems in MW4, the AI always aims for Center mass. or maybe, PGI adds a sort of TT type mechanic that will determine where the AI will aim, then as the AI is easy/medium/hard, it has a less chance to aim anywhere but the CT. No head aiming. Have it be up to the player to twist damage away as we have it now.

It can be done. for me the gunnery question is the important factor. People rail against any sort of random number generator affecting shot placement. They hate the idea of a cone of fire simulating targeting error. they want bundled direct fire weapons to alpha strike the same pixel every shot. They don't want any sort of accuracy between them and skill. its described as rolling 2d6 with an 8 or better to win... its viewed as diminishing skill. MWO is based on that design choice.

Now when you have an AI that is fully capable of head shotting you 100% of the time from max range. its game breaking.
The only way to fix it that i can see is a random number generator. The question becomes where its placed. Is it part of all weapons or is it only used on the AI. If its for everyone then the AI can target the CT or head all it wants and its no better then a human. provided you add a bit more error to compensate for the AI's ability to track with 100% accuracy. targeting issues solved.

The other solution is AI only and at that point the game become a 2d6 roll 8 or better for the win. Why is it acceptable for the AI to have a cone of fire and not the player?

[DEMAX51]

Tombstoner, on 30 January 2015 - 09:59 AM, said:

The other solution is AI only and at that point the game become a 2d6 roll 8 or better for the win. Why is it acceptable for the AI to have a cone of fire and not the player?

Because an AI has the ability to be completely 100% accurate all the time, and players don't?

[terrycloth]

AI randomness simulates player skill. You can have the degree of randomness as a variable based on how skillful a player the AI is simulating.

[0bsidion]

Asakara, on 29 January 2015 - 02:02 PM, said:

I suggest PGI starts any PvE with just AI armor and infantry, no AI mechs.


Here are 11 of my personal reasons why:

1) It would not cheapen the core game of Mech on Mech PvP by having clumsy AI Mech bots.
2) It would provide a totally different experience than the PvP one we have now.
3) PGI could reuse the turret code.
4) PGI would not need to create new code to animate mechs; just wheels, treads, and hover (and possibly rotors).
5) PGI would not need to worry about heat scale as all vehicles have enough sinks to fire all weapons.
6) Infantry (and infantry in bunkers) would give more reasons to take machine guns and flamers as well as giving a reason to put A-Pods in the game.
7) Possibly simpler pathing and combat AI especially if custom maps with triggers/scripts are used.
8) Infantry and Armor defenders are common in BT lore.
9) PGI would not have to worry about internal hits as the first point of damage which breaches a vehicle's armor knocks it out.
10) Smaller objects need less detail and less time for artists to create.
11) It could be a good stepping stone for adding combined arms into community warfare.


My $0.02.

I think you're missing some key issues with your ideas though. Even though things like soldiers, tanks, vtols, etc, might not require the level of detail as new mechs, they're still assets that have to be modeled, animated, balanced, on top of adding AI to them. It is actually a lot easier said than done.

Also, giving AI to mechs can add to QoL issues like players that for whatever reason get disconnected. The AI could step in and pilot that mech. Sure, it likely wouldn't be as good as a human but at least that's one or more mechs that are out there trying to help the team instead of being limp noodles back at the base. And then we have CW. Can't find anyone to defend a planet? No problem, an AI mech team can fill in. The potential cool part of that is the AI mech team can be filled using mechs commonly used by their respective faction, maybe even with faction specific camo and decals. And mechs are already in the game, unlike tanks and infantry, so the turn around time would be much much shorter.

[KuroNyra]

Belphegore, on 29 January 2015 - 02:09 PM, said:

I doubt we'll ever see pve in the lifespan of this game.

The Ghost drop made it.


Humans on mech against turret (IA)

[Aethon]

...but I want to be one of the tankers.

[Tombstoner]

DEMAX51, on 30 January 2015 - 10:34 AM, said:

Because an AI has the ability to be completely 100% accurate all the time, and players don't?

So your ok with playing against a Cone of fire/RNG to make it "fair", just not with using one yourself? when a COF would go a very long way in improving balance and game play over all. Particularly if it enables a drop in AI thats not a 2d6 roll.

Victory over an AI with a COF mechanic normalizes the skill level between human and computer. Not using one degrades the victory into a 2d6 roll... winning is more about luck then skill, where as with a balanced COF its ALL skill. Because the AI can make decisions to walk, run or stand still that affects its targeting skill, just like a player.

Its the difference between adding one cofactor to modify the COF vs. modeling the correct probability to hit for every situation or is the AI chance to hit the same at all ranges or is it range dependent. If it is range dependent the its hella easy to pick apart at range or too effective at short and vice versa.

A COF is the way to go for human and AI.

Edited by Tombstoner, 30 January 2015 - 02:17 PM.

[terrycloth]

There's no reason that COF for humans vs AIs should be linked.

The COF is a handicap to make up for the AI's perfect aim, which is unrealistic. The only reason to apply it to a human player is if you want to handicap them because their aim is too good.

And if you're doing that, you probably also want auto-aim assist for players whose aim is too bad.

Could be cool as options in private matches. I don't think it has any place in a random matchmaker.

Edited by terrycloth, 30 January 2015 - 02:54 PM.

[DEMAX51]

Tombstoner, on 30 January 2015 - 02:15 PM, said:

So your ok with playing against a Cone of fire/RNG to make it "fair", just not with using one yourself? when a COF would go a very long way in improving balance and game play over all. Particularly if it enables a drop in AI thats not a 2d6 roll.

Victory over an AI with a COF mechanic normalizes the skill level between human and computer. Not using one degrades the victory into a 2d6 roll... winning is more about luck then skill, where as with a balanced COF its ALL skill. Because the AI can make decisions to walk, run or stand still that affects its targeting skill, just like a player.

Its the difference between adding one cofactor to modify the COF vs. modeling the correct probability to hit for every situation or is the AI chance to hit the same at all ranges or is it range dependent. If it is range dependent the its hella easy to pick apart at range or too effective at short and vice versa.

A COF is the way to go for human and AI.

Yeah, no. I completely disagree, and don't see at all why it's outlandish to think an AI would use some sort of random-chance aiming mechanic. I also think it would be exceptionally stupid to add a CoF for real players.

Look - I understand. You want a CoF mechanic for players; I don't. Trying to imply that having an AI that is subject to some level of randomness would be in some way unfair doesn't make your argument any stronger.

Edited by DEMAX51, 30 January 2015 - 02:57 PM.

[Brody319]

make ghost drop turrets carry 4 ERPPCs for the energy ones, and the missile ones can carry 4 SRM 6s and 4 LRM 5s.

Ghost drops fixed.

[KuroNyra]

Brody319, on 30 January 2015 - 03:02 PM, said:

make ghost drop turrets carry 4 ERPPCs for the energy ones, and the missile ones can carry 4 SRM 6s and 4 LRM 5s.

Ghost drops fixed.

And CW killed.

[Brody319]

KuroNyra, on 30 January 2015 - 03:12 PM, said:

And CW killed.

gotta do what you gotta do.

[Asakara]

The players in their battalion of mechs vs a full combined regiment of AI vehicles and infantry would certainly make ghost drops in CW more interesting for me.

[Tombstoner]

DEMAX51, on 30 January 2015 - 02:54 PM, said:

What i want is to remove the harry potter levels of 100% magically accurate bundled weapons and overlay a system that wont harm game play or the illusion that players are good shots. A COF elevates the level of skill needed to hit the same spot with more then one weapon while moving at 150 kph. A COF if implemented correctly becomes pin point accurate if your standing still, with no actuator damage and low heat. At short range its still pin point. At medium it no longer becomes 100%. at long range yes you could miss. thats when you learn to pilot the mech. The best players will have nerves of steal while the scrubs will spray and pray.

Sorry but a COF is the best way to implement an AI. hence why its in almost every FPS with PVE that i have ever played.

[LordKnightFandragon]

Tombstoner, on 03 February 2015 - 06:35 AM, said:

What i want is to remove the harry potter levels of 100% magically accurate bundled weapons and overlay a system that wont harm game play or the illusion that players are good shots. A COF elevates the level of skill needed to hit the same spot with more then one weapon while moving at 150 kph. A COF if implemented correctly becomes pin point accurate if your standing still, with no actuator damage and low heat. At short range its still pin point. At medium it no longer becomes 100%. at long range yes you could miss. thats when you learn to pilot the mech. The best players will have nerves of steal while the scrubs will spray and pray.

Sorry but a COF is the best way to implement an AI. hence why its in almost every FPS with PVE that i have ever played.

CoFs are cool and work well except when taken to WoT levels of random, then they just become aggravating, and while WoT, skill still matters, it definitely nullifies it to a degree. A exaggerated CoF like WoT gives even the scrub a chance to beat a skilled player......We need one more like Planetside 2....mild enough to persuade against long, endless bursts, but small enough to let you know its there and deviate some shots.

But at this point, I see a massive overhaul to the Heat scale, making worthwhile heat penalties and reducing the amount of stuff we can realistically shoot all at once, to be a much better solution.

Edited by LordKnightFandragon, 03 February 2015 - 07:03 AM.

[Johnny Z]

Tombstoner, on 30 January 2015 - 09:59 AM, said:

It can be done. for me the gunnery question is the important factor. People rail against any sort of random number generator affecting shot placement. They hate the idea of a cone of fire simulating targeting error. they want bundled direct fire weapons to alpha strike the same pixel every shot. They don't want any sort of accuracy between them and skill. its described as rolling 2d6 with an 8 or better to win... its viewed as diminishing skill. MWO is based on that design choice.

Now when you have an AI that is fully capable of head shotting you 100% of the time from max range. its game breaking.
The only way to fix it that i can see is a random number generator. The question becomes where its placed. Is it part of all weapons or is it only used on the AI. If its for everyone then the AI can target the CT or head all it wants and its no better then a human. provided you add a bit more error to compensate for the AI's ability to track with 100% accuracy. targeting issues solved.

The other solution is AI only and at that point the game become a 2d6 roll 8 or better for the win. Why is it acceptable for the AI to have a cone of fire and not the player?

Cone of fire is correct because bots dont "aim" they get tied to items like the mech skeleton etc. This is my understanding anyway.

[Screech]

Initial PvE should be limited to the 4 man drops. Trying to do it on a larger scale to start would be a mistake.

[Tombstoner]

LordKnightFandragon, on 03 February 2015 - 07:02 AM, said:

CoFs are cool and work well except when taken to WoT levels of random, then they just become aggravating, and while WoT, skill still matters, it definitely nullifies it to a degree. A exaggerated CoF like WoT gives even the scrub a chance to beat a skilled player......We need one more like Planetside 2....mild enough to persuade against long, endless bursts, but small enough to let you know its there and deviate some shots.

But at this point, I see a massive overhaul to the Heat scale, making worthwhile heat penalties and reducing the amount of stuff we can realistically shoot all at once, to be a much better solution.

Exactly a badly done COF is detrimental to any game. To be fair the COF in WOT is meant to reflect the guns real performance as measured buy WOT sources, typically historical government documents. Players make bad choices taking a large caliber short barrel with a large COF against a long barreled high velocity medium sized round... and QQ when they miss.

i was thinking something a bit smaller then planet sides but yes you got the idea. a dynamic COF that changes depending weapon type and movement.

I do agree that at this point any changes to targeting cant happen. And ghost heat needs to be tied to a global cool down and the damage potential of the combined shot.

Johnny Z, on 03 February 2015 - 07:27 AM, said:

Cone of fire is correct because bots dont "aim" they get tied to items like the mech skeleton etc. This is my understanding anyway.

That is also mine. An AI needs a system that breaks its perfect accuracy in a way that put players on an = footing as the AI and not a 2d6 win loose helping hand simulator.

Screech, on 03 February 2015 - 08:18 AM, said:

Initial PvE should be limited to the 4 man drops. Trying to do it on a larger scale to start would be a mistake.

I agree. that way you can get 4 friends or 2 real and 2 AI to fill in the blanks.

I see PVE as one of the best money makers for PGI.... Why mission packs. a set of 10 missions for solo play with a decent performance based reward to be earned in the end.

[LordKnightFandragon]

Screech, on 03 February 2015 - 08:18 AM, said:

Initial PvE should be limited to the 4 man drops. Trying to do it on a larger scale to start would be a mistake.

PVE Coop should only be 4 man drops or if you try the clan side, 5 man.

What should kinda happen is like Mechwarrior 4 mercenaries, they run the same campaign, but basically if you run through the IS version, you can only pilot IS mechs and you will face the campaign based on the IS perspective.

Then, if you run it through on the Clan side, its mostly the same basic objectives and stuff, but you are locked into clan mechs. MAybe it could be switched to where the IS Are on defense, while if you play clan, you have attacking based objectives.....

BUt you get just your little 4 or 5 man coop unit...would be fun.