Jump to content

Procedurally-Generated Random Maps Would Make Mwo Live Long And Prosper


98 replies to this topic

#21 Gorthaur

    Member

  • PipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 186 posts

Posted 12 April 2013 - 04:27 PM

I like the idea of completely random maps, but I don't see it as a feasible fix for the map issues. My idea is for them to create map making software for the community and have contests where players can submit maps and adopt them as official. I'd love for them to consider getting this game into the Steam Workshop.

#22 Harabeck

    Member

  • PipPipPipPipPipPipPip
  • 520 posts

Posted 12 April 2013 - 04:29 PM

I'd just like to point out that no game is, or has, used such a system for a match based competitive game. In fact, it's extremely rare that it's used at all unless: the game is a space game, or the game takes place in indoor facilities with lots of corridors and enclosed rooms.

This is not something PGI could just whip up. CCP wants to try it with Dust 514, but has not implemented it yet.

#23 Appogee

    Member

  • PipPipPipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 10,966 posts
  • LocationOn planet Tukayyid, celebrating victory

Posted 12 April 2013 - 11:30 PM

Simple procedural map creation doesn't have to be that hard.

It goes like this. You define a flat battle grid, pick random spots on that map where mountains and hills will be, and random heights for them, and then start building up the y (elevation) values around those points. Using a fractal algorithm makes the elevations non-uniform and natural looking. Then you basically ''decorate'' the terrain with textures, and objects, using the elevation data to decide what items are most appropriate where.

You can make it a lot more sophisticated than that - for example, a river that follows the path of the lowest elevation etc. But these are like additional modular procedures that run when the main map is created.

I'd like to briefly respond to the objections raised by others in the thread:

1. ''It's rare in multiplayer games.'' Maybe that's why we grow bored of most multiplayer games so quickly.

2. ''This is not something PGI could just whip up.'' Yes, every aspect of programming a game takes effort. This is no different. But the reward for the effort would be huge.

3. ''It would create imbalanced maps.'' Yes. Terrain is like that, War is like that. As I said in my OP, over the course of hundreds of games on hundreds of maps, the balance would turn out equal for all.

4. ''It would be too slow to push the each map to each player.'' No. As I said in my OP, the PGRM routine would run on each client PC from the same numeric seed, using the same tileset. So identical maps would be created by each PC.

5.''We should make suggestions (preferably in the suggestions forum) that are remotely within the realm of realistic reality. Not this.'' Your assertion that this isn't ''within the realm of realistic reality'' (as opposed to what, unrealistic reality...?) is just a bald assertion without supporting fact.

6. ''How would it not generate potential terrain glitches?'' By keeping it simple. As an additional safeguard, give the player a ''nudge'' key which elevates them by 10 pixels if they get stuck on a terrain item. It's even logical, given Mechs are walking machines, and could flex their legs in such situations.

7. ''PGRM are absolutely detested by anyone doing competitive gaming... you can't practice on them.'' Man up :) and learn as a team to battle on the fly rather than as a scripted battleplan. Anyway, you can have arena-style maps for your tournaments if you want. Many of the rest of us would prefer some variety.

8. ''I'm a game programmer and I say that complex map generation and balance aren't strong suits of procedural generation - particularly in this sort of game.'' As I said, doesn't need to be complex, and balance would equal itself out over time. As a game programmer, we look to you to advance games, and not be stuck in the rut of standardised and ''but this is how it is done'' thinking which has plagued FPSers since Doom 2.

9. ''Keep in mind who will be programming it.'' Heh. When Dust comes out, everyone from Homeless Bill to PGI will see how possible this is. It's eminently doable.

Edited by Appogee, 12 April 2013 - 11:35 PM.


#24 Teralitha

    Member

  • PipPipPipPipPipPipPipPipPip
  • Survivor
  • Survivor
  • 3,188 posts

Posted 12 April 2013 - 11:50 PM

While I support this idea... I know that PGI is not capable of doing it.

#25 xenoglyph

    Member

  • PipPipPipPipPipPipPipPip
  • 1,480 posts
  • LocationSan Diego

Posted 13 April 2013 - 12:15 AM

I think it would be awesome, especially for small "unimportant" CW battles. More important battles could still be held on established maps. Scouts wouldn't just be scouting for mechs, they'd be scouting the terrain as well.

Also, give us a low gravity moon map, like that one from MW:LL, it kicked ***.

#26 Chavette

    Member

  • PipPipPipPipPipPipPipPipPip
  • Little Helper
  • 2,864 posts

Posted 13 April 2013 - 12:25 AM

Now THIS is something that deserves a separate que, unlike 3rd person.

#27 Volthorne

    Member

  • PipPipPipPipPipPipPipPip
  • Elite Founder
  • Elite Founder
  • 1,929 posts
  • LocationCalgary, Canadia

Posted 13 April 2013 - 12:32 AM

View PostAppogee, on 12 April 2013 - 11:30 PM, said:

9. ''Keep in mind who will be programming it.'' Heh. When Dust comes out, everyone from Homeless Bill to PGI will see how possible this is. It's eminently doable.

You mean IF Dust comes out. It's been, what, 2/3 years since I last heard anything about it? It's the next DNF.

Procedurally generated maps are an absolute *****. Look at Minecraft - it can take up to a few minutes to ank out the first 8x8 section of chunks, and that's an example of SIMPLE procedurally generated terrain.

Edited by Volthorne, 13 April 2013 - 12:33 AM.


#28 xenoglyph

    Member

  • PipPipPipPipPipPipPipPip
  • 1,480 posts
  • LocationSan Diego

Posted 13 April 2013 - 12:41 AM

View PostVolthorne, on 13 April 2013 - 12:32 AM, said:

You mean IF Dust comes out. It's been, what, 2/3 years since I last heard anything about it? It's the next DNF.

Procedurally generated maps are an absolute *****. Look at Minecraft - it can take up to a few minutes to ank out the first 8x8 section of chunks, and that's an example of SIMPLE procedurally generated terrain.


It's not an incredibly difficult problem to solve. Our battlefields aren't even interactive in any way. It's just some terrain. They could even generate all the maps on their end, include them for download in weekly patches. Vote at the end of the match if you liked the map. Crappy maps are weeded out, the best are kept in rotation.

As already been said, obviously the maps will pale in comparison to the nice hand crafted maps we currently have, but I still like the idea.

Edited by xenoglyph, 13 April 2013 - 12:45 AM.


#29 Appogee

    Member

  • PipPipPipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 10,966 posts
  • LocationOn planet Tukayyid, celebrating victory

Posted 13 April 2013 - 12:46 AM

View PostVolthorne, on 13 April 2013 - 12:32 AM, said:

Look at Minecraft - it can take up to a few minutes to ank out the first 8x8 section of chunks
Minecraft has interactive terrain and in hundreds of stacked layers. MWO needs one layer and not an interactive one. While the textures and objects in MWO are more complex, the actual single layer map would be much simpler.

#30 Ghogiel

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • CS 2021 Gold Champ
  • CS 2021 Gold Champ
  • 6,852 posts

Posted 13 April 2013 - 12:50 AM

I wish they would do it just to show the people who advocate it how it would suck.

#31 FunkyFritter

    Member

  • PipPipPipPipPipPip
  • 459 posts

Posted 13 April 2013 - 12:57 AM

Would be cool in the context of a fully functional community warfare, but we're a long ways off from that. There's a long list of features I would rather see implemented before the idea of trying to create a decent map generator is worth considering.

#32 Moriquendi86

    Member

  • PipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 97 posts
  • LocationWarsaw

Posted 13 April 2013 - 12:58 AM

I'm against this idea. First of all making tools to create such maps is difficult and time consuming process, think like 1 year from today to first public prototype even if they will assign huge team to this.
Also randomly generated maps doesn't mean better maps, personally I prefer when someone deliberately builds terrain and add objects to create specific gameplay opportunities and additional skill factor of knowing maps. Random maps only adds variety.

#33 Appogee

    Member

  • PipPipPipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 10,966 posts
  • LocationOn planet Tukayyid, celebrating victory

Posted 13 April 2013 - 05:18 AM

View PostMoriquendi86, on 13 April 2013 - 12:58 AM, said:

personally I prefer when someone deliberately builds terrain and add objects to create specific gameplay opportunities and additional skill factor of knowing maps.
Even if that means you only get a choice of 5 to 10 maps, in endless rotation, forever...?

Don't you get bored with knowing in advance how each battle will go, where the brawls will be, where the snipers will hide, etc?

To me that's like Groundhog Day.

#34 Jetfire

    Member

  • PipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 1,746 posts
  • LocationMinneapolis, MN

Posted 13 April 2013 - 05:30 AM

Not a fan, there are reasons games like SC2 and TF2 do not use them either. They are very hard to implement and very hard to balance and can only give you so much.

It would be much faster and easier to just add more map variants. Forest Colony is a good example, Ship/No Ship, Path by Cave/ No Path by Cave. I would add things like occasionally caves are caved in and unpassable, more buildings and obstables that are at multiple possible locations. Again, not randomly generated, but a random selection of several presets ala SC2 map variants.

Also further variation of the weather could make the maps more diverse. I really think we need a full moon variant, a pitch dark variant, a thunderstorm variant and bright sunny day variant to almost all the maps. The cold planets can get a snow storm variant, a clear variant and high winds variant (only white out closer to the surface).

I think the Thunderstorm variant alone does a ton for you as NV should become saturated due to lightning, limiting its use. Thermals should have a hard time seeing through the cold rain and picking up cold rain soaked mechs. So you would rely on lightning to get good visuals.

#35 Quinton99

    Member

  • PipPip
  • Overlord
  • Overlord
  • 38 posts

Posted 13 April 2013 - 05:39 AM

Some valid points brought up by the "No" crowd, but I'd like to point out to those that are in the no camp because they think that it's too hard to do, or that it would delay the development of the game, that PG of maps is already a feature of CryEngine 3. In essence, that huge bulk of time creating the tools to do it well has already been spent, and PGI is already using them. Google "CryEngine 3 Sandbox" for more details on that.

Don't get me wrong, I like the fact that someone is putting a lot of effort into the maps that we have, but the problem is that labour of love takes a disproportionate amount of time to the value added. To echo the sentiments of someone earlier in this thread: Have a look at the heat maps of Alpine Peaks and Tourmaline Desert. How long did we wait for those maps, only to have massive portions of them go virtually completely unused? Not only that, but Alpine is definitely unbalanced. Tourmaline is better, but again everyone pretty much just heads for the dropship in the middle because if you're in an assault mech and you head for the outsides you're not even going to see anyone for 15 minutes. We'll see how 12v12 changes that, but I'm not optimistic.

3500 worlds in the inner sphere indeed. Right now we have 6 maps (night and snow variants don't count). Even if they magically increase that number by an order of magnitude by September, which let's be clear is NOT going to happen, this game is going to be massively underweight at launch. I cannot overstate how bad that would be for the longevity of the game.

#36 Hypercation

    Member

  • Pip
  • Big Brother
  • 14 posts
  • LocationSlovakia

Posted 13 April 2013 - 06:07 AM

+1
Also, you can make PG maps 100% balanced (if balanced means symmetric) by generating one half and then mirroring it.

Edited by Hyperion110, 13 April 2013 - 06:09 AM.


#37 Noobzorz

    Member

  • PipPipPipPipPipPipPip
  • 929 posts
  • LocationToronto, ON

Posted 13 April 2013 - 06:39 AM

The other day I got the foot of my mech wedged in what seemed like flat ground in Forest Colony. I stood there and traded alphas with a PPC stalker when cover was literally 2 seconds away from a dead stop because I couldn't move.

And now you're telling me this wont happen on procedurally generated maps?

No thank you.

Edited by Noobzorz, 13 April 2013 - 06:40 AM.


#38 Thedrelle

    Member

  • PipPipPipPipPip
  • Bad Company
  • 160 posts

Posted 13 April 2013 - 06:49 AM

View Postp00k, on 12 April 2013 - 02:34 PM, said:

i like procedurally generated maps. no annoying terrain "features" to get glitched on

Untrue. I'm sure the Procedurally generated maps would have as many if not more places to get caught up on.

That being said, I agree with OP. We need a LOT more maps. It should seem like we are dropping on hundreds of locations on various different planets.

and do you want to reuse locations? save the seed!

One more thing. they could use hand made maps for specific key locations, where more points in the meta are awarded.

Edited by Thedrelle, 13 April 2013 - 06:51 AM.


#39 Moromillas

    Member

  • PipPipPipPipPipPipPip
  • 943 posts
  • LocationSecret **** moon base

Posted 13 April 2013 - 06:55 AM

Quality maps with an algorithm? Is that even possible? Where would you even start making such a program.

#40 Appogee

    Member

  • PipPipPipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 10,966 posts
  • LocationOn planet Tukayyid, celebrating victory

Posted 13 April 2013 - 07:12 AM

View PostMoromillas, on 13 April 2013 - 06:55 AM, said:

Quality maps with an algorithm? Is that even possible?
Yes. Never as good as a hand-crafted and tweaked map. But once you get the engine and algorithms right, you can have an infinite variety of ''good enough'' maps on the fly. It honestly isn't rocket science.

View PostMoromillas, on 13 April 2013 - 06:55 AM, said:

Where would you even start making such a program.
See my post above.

Or ask Notch.

Or just, you know, apply coding techniques, starting with an array and tile set, like we used to back when we were actually creating new computer games, instead of just reskinning pre-made game engines over and over :P

View PostNoobzorz, on 13 April 2013 - 06:39 AM, said:

The other day I got the foot of my mech wedged in what seemed like flat ground in Forest Colony. I stood there and traded alphas with a PPC stalker when cover was literally 2 seconds away from a dead stop because I couldn't move. And now you're telling me this wont happen on procedurally generated maps?
No, you're telling me that it's happening on the very small number of hand-crafted maps that currently exist.

Talk about inverse logic.

Edited by Appogee, 13 April 2013 - 07:11 AM.






23 user(s) are reading this topic

0 members, 23 guests, 0 anonymous users