31 replies to this topic

[Pelanthoris]

Ok, so this is a bit more of a question to people with more knowledge in game design, rather than a mature suggestion.

The idea is generation of semi-random maps. There are some examples out there, I'm not sure about recent games, but let's throw an oldie as an model: terror city-maps on the UFO: Enemy unknown (1994). These maps are created from premade templates of city blocks, warehouses, stores, etc. The a random set of these templates are glued together. It think most people know what I'm talking about. If you know a better and more recent example, please post it.

 

Would it be possible/reasonable to create maps in this fashion in Hawken or in fps games in general? It would change the game dramatically as map knowledge is one of the critical skills in arena shooters. I know it couldn't be totally random, because the maps should be balanced in some way, but even if there were some balance issues, they wouldn't matter as much because as no one would know of them in advance.

 

IIRC, it's not implemented in xcom, but it shouldn't be much more difficult to implement random stuff inside the templates either (blocked roads, collapsed floors or walls, extra crates for cover, etc.). This would reduce the benefit from knowing the templates extremely well.

 

There could even be certain (random) mutators to the maps, like this: create random area in a city; there was a battle (wrecks of mechs generated); city was bombarded from south (more collapsed walls on south sides of buildings). So a good player could except certain things even in the beginning and use it to his/her advantage.

 

I've used a city map as an example, but I guess you could do something else like this too.

 

Spoiler

Would this open a possibility for new game modes? I'm thinking about something like this: asymmetric teams, like 3vs9. Big team has to defend a large (5-6 maybe) number of targets, which the small team tries to destroy. Big team gets points by killing members of the small team.

 

I know everything about the state of this game, but one can dream, right? Ideally I don't want post stating that it would be a waste of developer resources, Hawken is ded, not possible at this point, etc. I know. And I'm bad at reading irony/sarcasm, so please, just post how you feel about this idea and would you like to have it in your ideal world of Hawken.

[6ixxer]

I think this is a cool idea.

We have a laser skirmish place a few suburbs over where they change the layout... and they have to physically push the damn walls.

[MomOw]

I think that random map have to be avoided in TDM, MA, Siege to avoid balance issues.

 

In DM, coop bot or coop TDM this could be fun.

[Zaxik]

Not sure how much dynamic this could be, but I know that Global Agenda game (same engine) used to have multiple variants of the same map with only very minor differences. Like crates/obstacles/scenery being on a different place.

 

But it was never random, I guess technically it was just multiple maps, even though they had the same name and general layout.

 

Not sure how well this could work in UE3 if we're talking fully dynamic/random.

[Kopra]

Procedural maps in Hawken would be cool. You could have symmetric maps for MA/Siege and asymmetric for TDM/DM. It could work very well for abstract kind of maps filled with Gigastructure and you could have plenty of different types of sections consisting of different pieces.

[asipo]

i play street of sim city when kid

you can make your own map (using sim city as the map editor) and then drive inside it

 

edit: there is also another game like mech commander

Edited by asipo, 10 April 2016 - 04:54 AM.

[Neraste]

This could be cool, but I have some concerns.

 

"Official" maps have been created with level design in mind, i.e. with places like choke points, lobbies and more quiet areas wisely displaced (ok, sometimes not: Bunker). I've discussed once with a board game creator: his game was centered around a map he spent a lot of time to generate, even using math optimization algorithms. Testing was also a large part of the process. I think it is algorithmically possible to generate maps, but will they be optimized and interesting? I also have discussed with someone really into game development, who told me procedural map generation is great to "fill the gaps" between human made "interesting" areas, like a forest between two towns in a RPG. "Interesting" places created by procedural generation (like dungeons in Daggerfall, if anyone remind them) are known to become repetitive and boring (but this is more a RPG point of view).

 

Also, the generator has to be smart enough to not create à la Minecraft generation errors, like collision between chunks and unusable areas. It has to be climber proof as well: even human created maps can be exploited, generated maps risk to be real Swiss cheese...

[coldform]

well, warframe used randomly generated maps by creating a conglomeration of pre-designed rooms  that have hallways following a set standard.

the server randomly determines a set of rooms, determines how they are connected, sets down starting points and finish points, then dumps the whole thing to the connected players.

 

think roguelike.

 

we would need to have both fully designed and finished maps alongside the roguelike maps, As I'm sure that roguelike maps during competitive matches would lead to cases of environmental bias in favor of one team.

[Amidatelion]

The big problem is that Hawken has a LOT of geometry problems with the way its actual physical infrastructure is drawn. As others have said above, on TDM and DM it could be cool, especially lore-wise, but without a human to go over and smooth edges that catch mechs on them as they dodge/boost, I think a lot of these generated maps would be unplayable.

 

That said, 6ixxer brings up a potentially point - have a normal map, but with a configuration that changes. Bridges deploy, walls move, pits open up, towers rise, all at random intervals. That'd have a similar effect, and be really cool in game.

[CraftyDus]

This is an idea by casuals for casuals, and would be casual fun.

Like Coldform said, warframe does it to the delight of people who enjoy warframe.

But for a true contest of skill, comp sorta needs maps to be static.

[Badtings]

If anyone out there knows the software to make NEW maps, I want to work w/you. I will create artwork and help carry the load. Willing to learn, but need help. 

[coldform]

The big problem is that Hawken has a LOT of geometry problems with the way its actual physical infrastructure is drawn. 

 

This.  the visual geometry and the collision geometry - the one that you don't see, but the one that the game uses to determine what is a wall, a floor, a ramp, etc. - are different.  it seems like ADH used some sort of collision mesh generator that used visual geometry as a basis, then simplified it to improve performance, but it looks like the process was less than perfect.  this leads to such issues as a simple bump in a floor that one would assume would be trivial becoming something that completely stops the forward motion of the mech, and, of course, map exploits.  Although there also appears to be effort placed in sorting out some of these imperfections,  the sheer length of time that we have played on these maps without these problems being addressed has amplified our awareness of their presence.

 

the real solution to the collision/visual geometry cohesion is to actually have map designers design the collision mesh by hand, in the same manner as they design the visual aspect of the map.  This is why I suggested the use of the roguelike map generation suggestion.

 

again, this should only be used for co-op missions, not for the actual PvP stuff.  A HUGE BIG OVERWHELMINGLY ENORMOUS aspect of a player becoming better at hawken is learning the maps, and how to use that knowledge.

Edited by (TDM)coldform, 10 April 2016 - 09:15 AM.

[JackVandal]

I like this idea, once the setup is done, it could be a lot of fun for some of the non pvp stuff.

[I2DI]

It can work for example if you have huge map and set semi-random borders in it. That way particular areas of the map is still known by you + less random map shape overall. But yeah, for any serious pvp it's not gonna work because balance and yata-yata-yata.

[Kopra]

The problem with match scheduling isn't tpg, its hawken players.
I can count on one hand how many matches I saw played at a time/day other than on match night at 9pm east in 10 years of tpg.
Then came hawken.

This. the visual geometry and the collision geometry - the one that you don't see, but the one that the game uses to determine what is a wall, a floor, a ramp, etc. - are different. it seems like ADH used some sort of collision mesh generator that used visual geometry as a basis, then simplified it to improve performance, but it looks like the process was less than perfect. this leads to such issues as a simple bump in a floor that one would assume would be trivial becoming something that completely stops the forward motion of the mech, and, of course, map exploits. Although there also appears to be effort placed in sorting out some of these imperfections, the sheer length of time that we have played on these maps without these problems being addressed has amplified our awareness of their presence.

the real solution to the collision/visual geometry cohesion is to actually have map designers design the collision mesh by hand, in the same manner as they design the visual aspect of the map. This is why I suggested the use of the roguelike map generation suggestion.

again, this should only be used for co-op missions, not for the actual PvP stuff. A HUGE BIG OVERWHELMINGLY ENORMOUS aspect of a player becoming better at hawken is learning the maps, and how to use that knowledge.

I heard a rumour from a cousin's dog's namesake that the collision mesh tool used for Hawken maps is wonky because it's optimized for human sized actors (100 unreal units?) while the actual actor size is much bigger, causing problems.

Edited by (KDR) Kopra, 10 April 2016 - 10:39 AM.

[coldform]

I heard a rumour from a cousin's dog's namesake that the collision mesh tool used for Hawken maps is wonky because it's optimized for human sized actors (100 unreal units?) while the actual actor size is much bigger, causing problems.

 

hmm... an outcropping that would slow down a human sized actor/pawn/character is registered as a short wall, but would be something a mech would be able to step over easily... if that's the case, then a simple mistake of scale is the problem.  I plausible theory.

[6ixxer]

A HUGE BIG OVERWHELMINGLY ENORMOUS aspect of a player becoming better at hawken is learning the maps, and how to use that knowledge.

^This

...is why I want maps that aren't static.

 

I think you should be good at playing hawken and adapting to your surrounds, instead of relying on the environment being the same and knowing them intimately.

 

I want to see initial games on a new map to see what fuzzy bunny goes on when veterans don't know where the hell everything is.

[ticklemyiguana]

 

But for a true contest of skill, comp sorta needs maps to be static.

 

I'll play devil's advocate here.

 

 

You are defining skill as including map knowledge, whereas there is a converse skill that is being skilled at adapting to new and unfamiliar environments, and that is a skill that is not tested with static maps.

 

On Mirage, I have a smoke that blocks off the window in mid. I throw it from T spawn. I land it every single time because I know where to put my feet and I know where to aim, and I even have a jumpthrow script to ensure that I'm putting it exactly where it needs to be.

 

Is that skill? No. Just a little bit of dedication - but it can completely fuzzy bunny up a CT side that isn't expecting it. That sort of imbalance is possible in static maps.

[CraftyDus]

Equating knowledge of the one smoke you've learned on mirage with map imbalance is a poor example.

And if you are playing people that don't expect sniper window to be smoked in the mid push, you have a responsibility to taunt them in chat.

[CraftyDus]

Also if you need a jumpscript to throw nades, I feel sorry for you man.

[CounterlogicMan]

You can make a random map generator for any game. The reason why it typically isn't used in fps games is stated above. Balancing a map in a player versus player fps has a pretty heavy reliance on where you place things and the distances between things. The maps in fps games are hand crafted to provide a lot of depth for players that really take the time to learn and explore every facet of the map. The flow of a map is also very important in fpses, where are the hot spots? Traversal areas? Sniper lanes? How long does it take to get from A to B? etc. There also needs to be a part of the map for each kind of weapon/ability/item, in this case mechs, to be powerful. How can you achieve this with an algorithm and keep the gameplay compelling and not frustrating or unfair for players?

 

Games like warframe, diablo, xcom(to an extent), etc. can get away with random generation of maps because they are largely player versus environment. Which is made much more compelling when the experience is mixed up every time you play. You can only make the AI so smart, players will eventually get bored of killing the same AI in the same spots, in the same shooting gallery, or cave over and over again. When you make the game player versus player you have to limit the amount of changing factors in the environment because gameplay is much more dynamic and difficult when all of the players are given the opportunity to learn and practice in a consistent and fair environment. So you want consistency in the environment to give both players a chance to outplay the other based on their skill and map knowledge. Think of it like sports, there is a playing field that two teams compete against each other on. In video games each side of the playing field isn't always a mirror image of the other side like in sports but each side is designed to be fair and provide its own advantage the other side doesn't have.

 

When you leave it up to a computer to randomly generate a map, you throw the likelyhood of fairness and consistency out the window. Fairness and consistency don't matter in pve because the game will always be unbalanced slightly in the player's favor. In pvp games you want to strive for balance and consistency so the focus is on the player's skill not the environment randomly favoring one player/team over the other one.

 

There are ways to tweak random generator algorithms to make sure that certain parameters are always met. Things like making sure there is always a treasure room and an exit in a dungeon crawler with randomly generated maps. Diablo III is a great example of random map generation that uses a rotation of certain set pieces, random filler tiles, and quality of play conditions to keep the experience fun. I can't imagine how insanely complex one of these algorithms would have to be to make sure the maps are compelling, fair, caters to all the things available to players, and laid out in a way that allows good flow of gameplay each match for an pvp fps game. If a game out there has this already, I want to play it!

TL:DR yes it is possible, but in all likely hood would not be fun or balanced.

 

 

Here is an example of one of the many ways it can be done for dungeon crawler games.

Edited by CounterlogicMan, 10 April 2016 - 04:15 PM.

[wolfrock]

This is an idea by casuals for casuals, and would be casual fun.

Like Coldform said, warframe does it to the delight of people who enjoy warframe.

But for a true contest of skill, comp sorta needs maps to be static.

It would likely be a mess if attempted but if a single player or co-op challenge mode generator actually did work, it would be casual fun and likely much more engaging than the current bot destruction mode, with daily challenges, etc.

 

Gotta have something to spice it up for the other 99%. They might even play longer and possibly spend some money on MC. Toxikk (remember that game? frag like it's when? too busy with grils to play vidya back then) is putting out some sort of single player mode to try to get more than < 50 concurrent players a day to try it. Not saying it'll work but at least they're trying.

Edited by wolfrock, 10 April 2016 - 06:27 PM.

[ticklemyiguana]

Equating knowledge of the one smoke you've learned on mirage with map imbalance is a poor example.

I didn't do that.

 

 

Also if you need a jumpscript to throw nades, I feel sorry for you man.

Better 100% consistent than 99%.

[Pelanthoris]

I'm really glad about all the constructive feedback this suggestion has received. Nice insights on the actual technical stuff on likely problems if Reloaded actually tried this. Again, I'm no expert, but most pointed problem is the visual/collision geometry disparity and all related problems. I'd guess that procedural maps would need a totally different workflow to that used on current maps and I think the current problems arise from the current workflows. Even if current tools were used on generation of the templates, they would be just as buggy as current maps, although there would be a new set of challenges on template borders.

 

As for the comp question, I agree with 6ixxer and Tickle. Why comp is the measure of the same performance again and again? I haven't been interested in comp since the original CS and seeing people play de_dust in the same way over and over again. I sometimes get a laugh when I see people play CS:GO and feel like nothing has changed in 15 years. IMHO comp would get a whole lot more interesting (although slower too) if every match was on a new setting that no one was familiar with. But I really don't want to push my opinion on others, so never mind.

 

Still, I can clearly see the likely balance issues caused by randomization of environment so I would never make all maps random. IMHO the random maps could be a distinct set of game modes, so you wouldn't find premade and random maps on the same server... But as many have stated, randomized maps probably would be best on coop-stuff.

 

Again, thanks for all the good feedback, love to see a good discussion on stuff people don't agree on.

[DallasCreeper]

I don't like CS:GO. It makes me way too salty. 

[CraftyDus]

On Mirage, I have a smoke that blocks off the window in mid. I throw it from T spawn. ....... but it can completely fuzzy bunny up a CT side that isn't expecting it. That sort of imbalance is possible in static maps.

 

 

 

 

Equating knowledge of the one smoke you've learned on mirage with map imbalance is a poor example.

 

 

 

I didn't do that.

 

[ticklemyiguana]

Yeeeeeeah. That's not map imbalance.

Edited by ticklemyiguana, 11 April 2016 - 05:33 AM.

[Hyginos]

Player generated maps would be better than procedurally generated maps.

[TheButtSatisfier]

Can anyone think of a PVP FPS where the maps are randomly generated (to some degree)? I'm not asking as a sideways criticism of OP's idea, I'm genuinely curious.

[CraftyDus]

warframe has those map modules

[Kopra]

Can anyone think of a PVP FPS where the maps are randomly generated (to some degree)? I'm not asking as a sideways criticism of OP's idea, I'm genuinely curious.

[StubbornPuppet]

Would rather have a built in map editor/creator.  I hear that's not really friendly with Unreal though.