236 replies to this topic

[Kristov Kerensky]

KraftySOT, on 04 August 2015 - 12:40 PM, said:

http://www.no-mans-sky.com/

Your argument is invalid.

Ah, yes, because a single player game is the perfect example of what you can do with a multiplayer game! Oh, wait, no it's not, not even remotely the same, huh, go figure right?

It's a great system for single players games where balance doesn't matter, and with the cartoon graphics in NMS, clipping issues probably aren't really noticable, so it's not a big deal.

Multiplayer games, it's not a good system, it's why it isn't used very often, too many issues to justify using it.

LOTS of things sound great when you ain't got a clue how coding actually works, or how to balance maps for multiplayer use, that doesn't mean they ARE great however.

I've done some work using procedurals, it's fun, but the multiplayer issue really are just too much of a PITA to make them worthwhile overall. NMS looks interesting for people who want a game that's not multiplayer and has literally unlimited exploration ability, and who don't mind cartoon graphics, which really kills it for me, sad to say.

[bad arcade kitty]

pbiggz, on 04 August 2015 - 02:32 PM, said:

1 team having an advantage based on terrain doesn't matter in a procedural (i.e. totally random) map, because in order for either team to win they have to move and scout and locate the other team, which also doesn't have any idea where anything or anyone is on the map. That's important. Both teams have no intel other than the basics. Temperature, gravity, time of day, etc. Both teams will find themselves moving in and out of defensible areas because thats what happens when you don't know the map.

Think about every time a new map hits the random queue, its a mess, everyone goes everywhere. often those games are the most interesting because of it, even if this condition only lasts a day or two, and everyone settles into a pattern. This wouldn't happen with random maps, because nobody would have time to settle into those patterns.

it does matter, scouting will not affect it much, for instance in skirmish one team spawned near a highground which is easy to hold, they get it and have a great advantage over the another team, especially if that highground possibly have only 1 route by which it can be stormed, or in conquest one team spawns near several points and some hills barring them, they get those points and easily guard them, etc

Edited by bad arcade kitty, 04 August 2015 - 02:48 PM.

[Johnny Z]

This isnt doable in a good way any time soon. Also I dont expect what amounts to noobs to even try and do this properly.

Tough to explain but all ai needs what is called a nav mesh generated. Nav mesh auto generation is impresice at best and buggy at worste. Nav mesh is auto genrated once a map is fully complete then gone over by hand and cleaned up to remove overlaps, height maps, cover areas(AI uses cover to avoid fire) so ai can figure out what everything is... Thats a good system and I have seen worse. If it isnt cleaned up by hand it will create all sorts of bugs for the ai(they will basically get "stuck") and even result in bloated memory usage to a small degree for each map.

Mechwarrior Online will have npc assets and the nav mesh basically needs to be done by hand eventually for it to work properly.


Second problem, auto generating one "tile" and the joining it to another tile with nav mesh already done is in my experience also buggy resulting in many overlaps and or "holes". A total mess if not gone over by hand.

Can these problems be worked out? It looks like X-COM has done it very recently in its turn based game thats coming out soon but thats an entirely different ball game and even they have had problems with it and likely still will at launch to some degree. They used the straight roads in between sections of maps, to make the joins which are entirely a staight line then generated some minor assets on the roads to make it look passable visually. X-COM is using a simple kind of ai since it isnt real time but turn based.

Other procedural games are using very simple AI, like very simple.(zombies and yes thats one of the reason zombies are so popular in games these days)

Mechwarrior Online will be using AI assets at some point so procedural isnt easily possible. Not to mention the maps wouldnt look very good. Again it is possible some time in the future I assume, but the quality and performance issues and other issues make it a tough sell. At least with what is out there for map making tools at this point in time.

Edited by Johnny Z, 04 August 2015 - 04:02 PM.

[Mister Blastman]

Kyocera, on 04 August 2015 - 02:34 PM, said:

But how can you do that whilst creating the same experience for both teams? Yeah we all know that our terrific matchmaker already makes things ridiculously one-sided but adding another (very wide) variable would be chaotic. Can you imagine if the lower ELO team got the bad map generation? More to the point, how could you guarantee that the lower ELO team got the better map? What criteria and software analysis could determine which side of the map is better for which team?

Would the generated maps be symmetrical to make it fair? Would there be some variable added to the higher ELO team to make their life more difficult?

As much as I would so love procedurally generated maps, in this game we just can't have it. Until we can invent an AI with the ability of Skynet to carefully craft the right maps, it's just never going to happen.

Both teams share the same map. Both teams have free reign to move about it. Procedural maps have can multiple sub-seeds with geometric modifiers so you can have weightings on quadrants with raw scores and totals to ensure there is a fair distribution--or you can have none at all.

The point is--the maps will be unknown. Do you know you have a better position than the other team? Do they? Nobody knows where they will spawn. The spawn locations will be random, too (within a min/max distance of each other... or maybe not).

Scouting would be critical. As random as they may be, the outcome will still favor the better player, every single time.

[Milocinia]

It's a fair point but I just don't see it being fairly implemented.

Oh look I've been banned again for flaming PUGs in game... Oh wait my forum profile is active but I can't log into the game

Edited by Kyocera, 04 August 2015 - 03:22 PM.

[Johnny Z]

Another game coming out soon that has "some" procedural maps in it is Fallout 4 although it isnt obvious.

The pre designated base building areas for the player will have what amounts to procedural map generation because the player can build what he or she wants real time and the ai wont bug out("presumably"). This is using some fairly detailed nav mesh generation on each piece the player adds I assume and is successfully joining it to any other nav meshes.

They are using predesignated areas for base building for performance reasons presumably.

All these other examples in this topic are using very simple games and less than attractive as examples.

Edited by Johnny Z, 04 August 2015 - 03:34 PM.

[Lyoto Machida]

bad arcade kitty, on 04 August 2015 - 02:27 PM, said:

well, the problem with such maps it's that they will be horribly unbalanced, so a team which got a better spawn point will usually win, it will be like a lottery, they also probably will not be fun having too much open space or poor routing from one point to another etc; to make procedurally generated maps fun and balanced... a hard task

Who cares if it's unbalanced? It's only for one match...so you hotdropped into a bad position? Suck it up and fight your way out of it...or don't. Move on to the next game...

[Mystere]

Kyocera, on 04 August 2015 - 01:36 PM, said:

I mean don't get me wrong I would love to see something like this but in the context of the extremely whiny MWO player base, I really don't ever see it being possible.

FTFY.

[Mystere]

bad arcade kitty, on 04 August 2015 - 02:48 PM, said:

it does matter, scouting will not affect it much, for instance in skirmish one team spawned near a highground which is easy to hold, they get it and have a great advantage over the another team, especially if that highground possibly have only 1 route by which it can be stormed, or in conquest one team spawns near several points and some hills barring them, they get those points and easily guard them, etc

A competent opponent will just not decide to storm that hill but instead create another plan. So what's the problem again?

[Funkin Disher]

I wouldn't mind them randomly generating a couple of maps to use each month. You only have a limited time to get used to them before they are swapped out, and you don't have to wait for them to get generated for each drop.

The favourites could even get compiled and turned into comp maps, keep those designers at work.

[Soy]

well, yes. without it this game is kinda... well, shallow as fk. which it is [obvsly]

[Mister Blastman]

I love the feeling of getting lost on a map and never knowing if an enemy is around the corner. That "new map feeling." Ahhhhhhh.

[lol lol lol lol lol lol lol lol]

Well to be completely honest...

They took the time to put in place every planet, didn't they? I think it'd be a silly waste for them to only be populated with the same set of maps over and over and over again. You could have saved the poor souls an assload of time by letting them copy/paste "mechbayplanet, iceplanet, lavaplanet, rockyplanet" instead of naming and positioning each and every one of them if the planets aren't important enough to put different stuff on each of them.

Edited by 00ohDstruct, 04 August 2015 - 07:08 PM.

[Xetelian]

Maps need cover. As long as you can guarantee you get bog maybe canyon or caustic and alpine in equal shares, then I'll totally pay a founders fee for brand new coding of a system that could do that.

However I suspect to really harness the power of random generation we'd need larger maps.

[Johnny Z]

https://www.youtube....e&v=4WUG55EC4v0

This shows how nav mesh is done for competent ai so it doesnt get stuck on invisible walls, or walk through walls in other situations, or terrain and can use cover and other intelligent behavior.

This doesnt show the nav mesh auto generation or height map generation or cover generation etc... This guy skipped auto generation because it would have been tougher to correct the mess than start from scratch by hand.

Like I said above many games use zombies because they will just head directly at the player and thats it and can use fairly simple nav meshes and ai. See a zombie get stuck on walls? Thats a nav gen problem.

Other games may create npc or animals in an open area and they wont travel far and that would work in procedural map generation.

Some systems are more difficult than this to use. This is an extremely good system.

Fallout 4 has made improvements to the above system to allow for procedural map generation on a small scale in predisgnated areas for base building and they are extremely proud of that. So a player can place a wall and an npc wont walk into it not knowing it is there for instance.

The Crytek engine Mechwarrior Online is using will be similar to this and I know nothing about it but I think it is good.

Edited by Johnny Z, 04 August 2015 - 10:59 PM.

[Rebas Kradd]

Mister Blastman, on 04 August 2015 - 05:22 PM, said:

I love the feeling of getting lost on a map and never knowing if an enemy is around the corner. That "new map feeling." Ahhhhhhh.

You're talking about a hardcore feature. Not to be condescending, but those who push hardcore features often fall into the trap of thinking that everyone wants the same thing they do. The fact that we voted Repair and Rearm out of the game should be an object lesson here.

The reality is, being taken out of their comfort zone is the LAST thing players want, and the last thing smart game designers try to do. Games that do it are generally hailed by gameplay watchdogs but don't necessarily sell all that well. The entire game industry is built around map familiarity. Players want it.

I see where you're going when you say that the advantage of an unbalanced map would be undone by both teams being unfamiliar with it. But that's not going to muffle the screams of the losing team, and the threats of asking for refunds, when they lose on a horribly unbalanced map. It'll be a player satisfaction issue same as 6PPC Stalkers and gamemode voting were. Brings all the cockroaches out of the woodwork. PGI has to think of this basic human nature.

If you want some variety in maps, tell PGI to loosen their deathgrip on their map design philosophy. They put a ridiculous amount of work into the maps that just doesn't have to be done. Well, some of it does - they rely on dense terrain in order to enable brawling. But some of it is just for the sake of needless beautification. Ask them to start cranking out simpler, terrain-based maps like the non-base side of Boreal, but much bigger, like the MWLL scale. Instant need for scouting. Far, far faster to make. Much less limiting because not everything is a blasted lane anymore. Everyone wins.

[Aresye]

While I think 100% on the fly, "every map is unique" procedural generation is definitely not ever going to happen, it doesn't explain why procedural generation can't be used.

- But what happens if the map generated is unbalanced towards a particular side?
Well then...uh...the MAP TEAM can then TWEAK the procedurally generated map to BALANCE it!

- But what happens if the collision map doesn't line up?
Well then...that SAME MAP TEAM can then TWEAK the physics mesh to FIX it!

Doesn't procedurally generated maps just give you the general mesh anyway? It's not like the map team CAN'T edit a map after it's generated, right?

Say I generate a map in Maya using a high-poly plane with a displacement map driven by a fractal node. I don't HAVE to settle with what the fractal gives me. I can bake the displacement to the plane (via converting to polygons) and then tweak the geometry to fix areas I don't like. Oh, what's that? The Red team's spawn point will end up on the top of a cliff face? Here, let me move that spawn someplace else and/or fix the map itself by getting rid of the cliff! Huzzah!

I know there's definitely more to it, but in general NO, having procedurally generated maps doesn't mean they can't be tweaked to avoid balance and gameplay issues.

[Kissamies]

Funkin Disher, on 04 August 2015 - 03:41 PM, said:

I wouldn't mind them randomly generating a couple of maps to use each month. You only have a limited time to get used to them before they are swapped out, and you don't have to wait for them to get generated for each drop.

The favourites could even get compiled and turned into comp maps, keep those designers at work.

When I was thinking of this, I was dreaming of them releasing a number of generated and vetted maps each month or two months, and at the end of the each period there is a mini-event in which the players get to vote for one of those maps that will be kept and the others discarded for new batch of generated maps. The winned map would then probably be enchanced and futher balanced by the mappers.

[Nightshade24]

Mister Blastman, on 04 August 2015 - 11:30 AM, said:

Well, I'm sorry to burst your bubble but those worlds are generated realtime at sixty frames per second without substantial delay. On average, I might have to wait ten seconds for the terrain to render in an area a hundred times larger than the River City remake. That's as long as it takes to load into the game. I am free to move about, explore or do whatever as it renders the smallest details and once they are, whee! Off I go. I can explore the whole planet! No pop-up. It's just... there.

What's the big deal?

There's no downloading. There's no uploading. The only thing that is shared is the random seed and the parameters--small packets worth of information. This is trivial. This is now.

and yes and how does this fair when say someone is on the low end of there internet capabilities or hardware?Just a question because I do not want to play with 5 fps while teleporting, warping around for hte first 4 minutes and the rest of the time shooting at enemies behind invisible buildings that are not rendered yet on my side but he obviously knows they are there and is laughing at my misfortune in the poor trade off for heat and ammo...

[TheCharlatan]

While i like the idea, making (or buying and tweaking) a procedural map generator for MWO would probably cost A LOT OF $$$, and be a huge risk.
What if it dosen't work right?
What if players don't like it?

We all know that this game needs more maps. There are many ways to do that, but this one sounds pretty dangerous.