17 réponses à ce sujet

[ClarkHewis]

Procedural generation has great merit to an exploration/shooting franchise but can it be applied better. For the RPG/FPS genres maybe instead of innumerable planets to explore it could be applied more storyline centred by using the engine to create innumerable enemy types and variation. The engine type is already being applied to wildlife npcs and thats through the same variation systems we've seen in other games. Preposing discussion on a more combat orientated model...

 

Application: -

Modelling templates, combined with armour compatiblity, designed for form movement

Tech levels effecting projectiles used by weapons

Player classes or types used as archetypes in enemy force composition, as in applied to skills or tactics used

Sense models to create stealth and detection alternates

Saving feature to distribute to community so preferred code has a wider chance to be experienced by more players

 

Requirements for storyline: -

Galaxy spanning travel

 

This method allows use of the current multiplayer without dedicating more resources to information storage for the company yet allows a large number of enemies, with a community interaction that selects new enemy races through popularity. These races could then be used as DLC.

Critique welcome.

[Sartoz]

                                                                                       <<<<<<<<<<(0)>>>>>>>>>>

 

Seems more arcade like than a Mass Effect game =  aRPG.

 

 

[TurianSpectre]

Procedural generation has great merit to an exploration/shooting franchise but can it be applied better. For the RPG/FPS genres maybe instead of innumerable planets to explore it could be applied more storyline centred by using the engine to create innumerable enemy types and variation. The engine type is already being applied to wildlife npcs and thats through the same variation systems we've seen in other games. Preposing discussion on a more combat orientated model...

 

Application: -

Modelling templates, combined with armour compatiblity, designed for form movement

Tech levels effecting projectiles used by weapons

Player classes or types used as archetypes in enemy force composition, as in applied to skills or tactics used

Sense models to create stealth and detection alternates

Saving feature to distribute to community so preferred code has a wider chance to be experienced by more players

 

Requirements for storyline: -

Galaxy spanning travel

 

This method allows use of the current multiplayer without dedicating more resources to information storage for the company yet allows a large number of enemies, with a community interaction that selects new enemy races through popularity. These races could then be used as DLC.

Critique welcome.

I thought this was what they had already implemented? gonna have to elaborate more for me lol

[SKAR]

Procedural generation has great merit to an exploration/shooting franchise but can it be applied better. For the RPG/FPS genres maybe instead of innumerable planets to explore it could be applied more storyline centred by using the engine to create innumerable enemy types and variation. The engine type is already being applied to wildlife npcs and thats through the same variation systems we've seen in other games. Preposing discussion on a more combat orientated model...

Application: -
Modelling templates, combined with armour compatiblity, designed for form movement
Tech levels effecting projectiles used by weapons
Player classes or types used as archetypes in enemy force composition, as in applied to skills or tactics used
Sense models to create stealth and detection alternates
Saving feature to distribute to community so preferred code has a wider chance to be experienced by more players

Requirements for storyline: -
Galaxy spanning travel

This method allows use of the current multiplayer without dedicating more resources to information storage for the company yet allows a large number of enemies, with a community interaction that selects new enemy races through popularity. These races could then be used as DLC.
Critique welcome.

Interesting read. I'm indifferent.

[Monk]

Procedural generation has great merit to an exploration/shooting franchise but can it be applied better. For the RPG/FPS genres maybe instead of innumerable planets to explore it could be applied more storyline centred by using the engine to create innumerable enemy types and variation. The engine type is already being applied to wildlife npcs and thats through the same variation systems we've seen in other games. Preposing discussion on a more combat orientated model...

 

Spoiler

Application: -

Modelling templates, combined with armour compatiblity, designed for form movement

Tech levels effecting projectiles used by weapons

Player classes or types used as archetypes in enemy force composition, as in applied to skills or tactics used

Sense models to create stealth and detection alternates

Saving feature to distribute to community so preferred code has a wider chance to be experienced by more players

 

Requirements for storyline: -

Galaxy spanning travel

 

This method allows use of the current multiplayer without dedicating more resources to information storage for the company yet allows a large number of enemies, with a community interaction that selects new enemy races through popularity. These races could then be used as DLC.

Critique welcome.

 

This line doesn't make much sense. A mass, unnamed mob doesn't a story line make. While it would free-up resources used to create misc. mobs*, this does nothing for the storycraft itself. The story is where the art is and needs careful crafting, not some a dumping of "lego blocks".

 

* And as noted in a post above, i'd be surprised if it's not already utilized for doing just this.

[UpUpAway]

One application, I thought, would be that in a single playthrough, only a certain percentage of those "innumerable" side-quest planets would generate and on the next playthrough the specific planets that would generate could be somewhat randomly different.  The side quest battles (enemy spawns and such) associated with the number of planets could adapt themselves to the changed terrain from playthrough to playthrough... making each playthrough somewhat unique even though they involve the same basic side quests.

 

For example, lets use the ME3 N7 missions... it wouldn't really affect the story for most of them if the layout of the battlefield generated at little differently from playthrough to playthrough as long as it still contained the basic elements for the battle part of mission.  E.g.  The specific layout of Benning is not really  important to the story as long as there is a staging area of for the scared couple to be found and a way to be cleared to get them safely to the shuttle and the battlefield has somewhere on it some dog tags to be found.  I think it would add some replay value if these set pieces and the enemies within them would generate a little differently from playthrough to playthrough.

[goishen]

It could happen.   If you take a look at what'll happen in space and theories suggesting space battles, then they'll almost have to have procedural content creation.

[Sanunes]

I don't think I would enjoy procedural generation for it could make the game a lot more uneven gameplay wise for one time you might be doing a mission and can ignore everything and then another one it is non-stop combat that might be more difficult then intended because of spawn locations.

 

The other problem I see with procedural content on a developer side of things is the frustration with save file bloat, for they are going to have to store the variables somewhere.  Now Bethedsa is the champion of bloat in save files for every item you interact becomes part of your save file even after you discard it.  Of course one way around those problem is to prevent saving after you start on a procedural generated planet and then make it so you can never return, but that doesn't seem like a lot of fun especially if something comes up and you just spend 30 min exploring and must leave.

[DuskWanderer]

Procedural generation wouldn't be a good idea for a very story-heavy game, simply because it could unbalance things. Also, since we as players unlock extra abilities through level up, that doesn't work for the types of enemies we'll face. 

[RoboticWater]

Not happening, especially not in an action game. By all means, have randomized armor/texture variants to give the enemies a bit more visual variety, but once we get into abilities, all bets are off. Active abilities are tough to balance because there's a lot more error involved in an action game. We'd end up with either a menagerie of enemies with wonky OP or UP power sets or a set of enemies generated by an algorithm with so many specific restrictions that BioWare would be better off just designing everything manually.
 
Even if we did have a GURPS style rule set where hot swapping traits is a matter of redistributing point values, that won't guarantee that any individual enemy, let alone an entire composition of enemies, will be fun and balanced to fight once the action starts. More importantly, enemy variety doesn't matter in shooters after a certain extent. Great shooters like Halo and DOOM do well because their comparatively small enemy variety was bolstered by level design.

 

While I'll marvel at the technical accomplishment a few decades down the line when procedural generation of this kind is feasible in the action genre, I'd consider it a greater artistic achievement now if BioWare made a game where every single NPC had a name and a place in their own environment. To put it plainly: games don't need more content, they need better content, and computers aren't ready to provide that yet.
 

Procedural generation has great merit to an exploration/shooting franchise

Which one? I can't think of a popular shooter franchise that uses procedural generation on this scale. There's a reason for it too: shooters are best designed to have only a limited number of unique enemy types with completely distinct silhouettes and attack patterns.

 

In fact, I can't think of very many games at all that have a system like the one you propose. The closest I can think of is a Bethesda game, but for the sake of balance, they can't swap out more than a couple of (mainly passive) abilities in each enemy, so most enemies in an archetype don't actually feel that different from each other. Heck, even the archetypes start to blend together.

 

The best implementation of procedural generated gameplay on the market currently can probably be found in a RogueLike game like Binding of Isaac, and that game doesn't randomly generate enemies. It creates small, pseudo-randomly laid out rooms with a relatively small cast of enemies.

[malloc]

I could see it working for some of the exploration elements but... what benefit would it add overall ?

[UpUpAway]

Not happening, especially not in an action game. By all means, have randomized armor/texture variants to give the enemies a bit more visual variety, but once we get into abilities, all bets are off. Active abilities are tough to balance because there's a lot more error involved in an action game. We'd end up with either a menagerie of enemies with wonky OP or UP power sets or a set of enemies generated by an algorithm with so many specific restrictions that BioWare would be better off just designing everything manually.
 
Even if we did have a GURPS style rule set where hot swapping traits is a matter of redistributing point values, that won't guarantee that any individual enemy, let alone an entire composition of enemies, will be fun and balanced to fight once the action starts. More importantly, enemy variety doesn't matter in shooters after a certain extent. Great shooters like Halo and DOOM do well because their comparatively small enemy variety was bolstered by level design.

 

While I'll marvel at the technical accomplishment a few decades down the line when procedural generation of this kind is feasible in the action genre, I'd consider it a greater artistic achievement now if BioWare made a game where every single NPC had a name and a place in their own environment. To put it plainly: games don't need more content, they need better content, and computers aren't ready to provide that yet.
 

Which one? I can't think of a popular shooter franchise that uses procedural generation on this scale. There's a reason for it too: shooters are best designed to have only a limited number of unique enemy types with completely distinct silhouettes and attack patterns.

 

In fact, I can't think of very many games at all that have a system like the one you propose. The closest I can think of is a Bethesda game, but for the sake of balance, they can't swap out more than a couple of (mainly passive) abilities in each enemy, so most enemies in an archetype don't actually feel that different from each other. Heck, even the archetypes start to blend together.

 

The best implementation of procedural generated gameplay on the market currently can probably be found in a RogueLike game like Binding of Isaac, and that game doesn't randomly generate enemies. It creates small, pseudo-randomly laid out rooms with a relatively small cast of enemies.

 

That's more where I was headed with this idea.  Use it in a limited to generate some variance in the battlefields to make the battles somewhat less precisely repetitive.  After a few playthroughs of the ME Trilogy, one can almost start shooting the enemies before they actually spawn... it's so predictable.

[RoboticWater]

That's more where I was headed with this idea.  Use it in a limited to generate some variance in the battlefields to make the battles somewhat less precisely repetitive.  After a few playthroughs of the ME Trilogy, one can almost start shooting the enemies before they actually spawn... it's so predictable.

The problem is that good pseudo-random still takes a lot of effort and the result will still probably feel random. I don't think I've played a game where I couldn't tell the levels were random, and I can guarantee you that I notice Skyrim's random enemies. Personally, I'd take unique and predictable over random and amorphous any day.

 

Besides, much of the repetition should be mitigated by better AI and dynamic objects. Despite a completely static construction, every level in a Halo game can be played through in any number of different ways because of the way vehicles, weapon drops, and enemies interact. The rest of the repetition should be offloaded to character progression. Six classes with multiple approaches to ability progression and various permutations of weapon loadouts should be more than enough to differentiate combat.

[goishen]

Well, the direction that I was going in with space battles, was that one, there would be space battles.  Two, we could board enemy ships.  Three, there would have to be and endless supply of ships.  Four, there would have to some type of procedural generation for those ships.  Otherwise we would be invading the same types of ships over and over and over again. 

 

It would be like ME1, just with say four different ships.

[Monk]

Well, the direction that I was going in with space battles, was that one, there would be space battles.  Two, we could board enemy ships.  Three, there would have to be and endless supply of ships.  Four, there would have to some type of procedural generation for those ships.  Otherwise we would be invading the same types of ships over and over and over again. 

 

It would be like ME1, just with say four different ships.

 

Yarr! You got the right idea, matey!

 

[LinksOcarina]

If it's side content...maybe...

 

But I highly doubt it will be utilized, unless the procedural generation is specific for each planet in some capacity. 

[Cheviot]

Procedural generation of enemies?  I don't see what advantage it would have over a developer-created new enemy type.  At least with the latter, you'd have specially-made models and a distinct art style.  And even if they did do procedural generation for this, the playerbase of a MP game wouldn't be my first choice for having the deciding vote; based on what I know of PvE players at large, they'd go for any enemy type that'd be easy to cheese and grind.

[spinachdiaper]

The few games I've played like Torchlight that do "random" dungeon generation become familiarly monotonous. I wouldn't want MEA to try and do something that ultimately ends up artificially different even though it's all the same.