Thread: What if surviors had to find their way out of a city...

When i first heard of L4D 1, i had imagined that players would start in the middle of a city and have to find their own way out... If such a system was implemented, would it renew your interest in the L4D series?


Heres my proposition:

Design a system to randomly generate city campaigns with unique layouts. Where survivors start in the center, and have to find their own route out, many paths would dead end/be obstructed. Lots of turning around and trying a different route... Not a perfect maze persay, there would not be a single correct path, but many; though there would be an equal amount of dead-ends.


Long story short, i need a project for my graduate class, this is one on my list of possibilities. It would be an implementation (and act as a tech-demo for) of a system that I have already written (though am in the ongoing process of completely revamping). In contrast, the actual difficulty would be in smoothing the output of the implementation, and designing and building enough content to populate a city without making it look copied and pasted. Which may actually be an potentially interesting sub project (see waay below for more on this sub-sub project). The true problem is determining WHEN campaigns are generated...





Over/through the Developmental Stages (early , rough, draft, final, etc, stages); in other words, over the programs development progress/progression-map (from start to finish, or rather the ascension from beta to live):

1. First versions would, really just be built of structures that players could not enter, so pretty much survivors fighting through city streets. Or rather towering blocks of doom (blocks representing where buildings should be).
   
   1. Easy to get to this point
   
   
2. Soon after, generated campaigns blocks would become actual buildings with varying styles of architecture.
   
   1. These buildings would come from a library, so you may recognize them, much like BC2 buildings (building templates, which instances vary slightly).
   2. At this point the campaigns would be different enough to mix things up, not to mention the project will stir serious interest at this point
   
   
3. Later buildings themselves would start being generated, with unique layouts.
   
   1. Slightly difficult to get to here, but not the true milestone, since generating eye-candy buildings of a given architectural style is fairly straight forward...
   
   
4. Then generated cities would have all the decals/details/debris lights/etc.
5. Eventually, buildings would start having player navigational interiors, small buildings, like the one on the corner of the first scene in the l4d 1 mercy campaign, then buildings like the one you spawn in on top in the first scene of mercy. Then super complex structures like business buildings,
6. Final version would desirably yield the above and alternative exits, ex airplane exit: i.e. players might fight through the city streets for the first scene or so, then come upon an airport and decide to enter and bam, playing through the airport like half the airport campaign. Actually, exactly like the airport campaign, only there would be no set linear path to follow. So you'd have to find the airport, or see signs/traffic-signs about it. Or a boat exit, or ...
   
   1. Though for these, I'd have to figure out if they are guaranteed exits, or if they have a possibility of being a dead end (which would suck to fight through the entire airport campaign to find out there's no airplane to escape with, though again, theres no linear element, so it's be easy to just continue on the campaign from the tarmak).
   
   

After enough popularity, i wouldn't be shocked to see Valve just add a new game type for it.


My concerns are:

1. that director is not very forgiving with survivors going in the wrong direction. IE survivors would rarely make it out alive.
2. for verses, the team to play zombie the first round would have an extreme advantage, ie they could just ignore the survivors for the first few minutes and take the time to scout a safe path, without the concern of zombies.

For 2), could be fixed by randomizing the obstructions on the map (kind of like cs_havana, in cs/css). However, the layout of the city would still remain the same, so there would be some advantage.... Maybe valve could add a "you are too far from survivors".... teleport? Which would limit how much exploration the infected could do...




Random Campaign would:

• negate the: "my way or the highway" methodology, thats currently consumed the series. IE the "such and such plan worked before, thus any other plan is irrelevant because the aforementioned plan worked, so its that plan or the highway". In other words, there's no trying anything new, really... because after all, they may be partially right, the maps do not change, so strategies for various bottlenecks of the maps naturally form.
• be bad ass, the way the series should have been (in regards to survivors in a city, finding there OWN route out),
• complete valves incomplete L4D re-playability cycle. That is to say, as it stands, the only thing that remains static is the linear path of the campaigns.
  
  • The game gets old fast when there is no improve, especially if you are on a team that is having trouble progressing.
  
  
• Damn near make L4D a RPG, valve could add a infinite survivor like mode, were players can run around the city.
  
  • Though completely unrelated and not a goal of the project, just a thought: Massive server with like, map seed of the month (new map every ___) with groups of survivors running about doing whatever, grouping up with other groups or ___thought incomplete__.
  
  

Before someone asks, such a system really could not be applied to games like TF2, since its maps are incoherent/adventitious, ie a lack of a coherent common theme to connect of all the tf2 maps.






I am NOT looking for developer opinions on the difficulty of this project, again this is just a implementation of a system I’ve already been developing for procedurally generating game worlds. Development wise, what I am actually worried about, is architecture, most architectural styles are the result of abstract thought. That though, is another discussion.


What I AM looking for: right now i need gamer (NOT DEVELOPER but GAMER) opinions, tester interest, ideas for further elaboration and extension of my proposal, possibilities of usages of the project (ex l4d rpg). However, if this is the project I roll with I'll need lots of dedicated testers of whom I can rely on for constructive options and testing. Really what I need now, should I choose this, is to finalize the project description, of which Ill use to develop the implementation of.



Also I realize Ive got a habit of posting such projects and not following through, so if it helps to say, most previous projects were being developed on my own time (which is very, very limited for personal projects) for my own intestest. However this is not only a school project (the implementation), but the actually development of the base system is something I have to get done regardless, whether the implementation becomes my semester project or not (which is not to say that if it did, support would end the day the semester finished, on the contrary I'd dedicated my full attention to it, because after all the base system is something I will see through regardless, and both will drive one-another).

Originally Posted by Bunni

Green Light, project's a go. Now accepting volunteers



Preliminary Findings :
I was doing a lot of preemptive testing last night, lots of good news. Via the map, we:

• Have control over survivor specific spawns (can separate survivors initial spawn)
• Maps do not have to be in a linear order (e.g. we can have safepoints, that are dead ends, and players have to remerge in the same area)

Bad news/Complications:

• Confirmed that survivors cannot exist in seperate instances (not that big of a deal)
• Navigational meshes will be a bitch for the first few stages, valve made a lot of changes to the navigation system in the l4d source engine version.
  
  • A lot of crucial mob related navigational nodes will have to be deliberately placed, e.g.: fences will have to have mob climbable mesh on them and be marked as climbable in nav.
  • This isnt really that big of a deal, as i have a thorough understanding of the underlying navigational technology of the source engine, Ill just have to fully investigate and identify all the relevant changes in l4d source engine. Mean while for the first few stages navigation will have to be done manually.
  
  

Bottom line, we can do damn near everything that was proposed by everyone up to this point (except the obvious daydreams, which came mostly from me), though there's some unforeseen complications that will lengthen the amount of preliminary coding (slight delays in the completion of a few stages) and cause a few bumps in the generated campaigns first few stages, but by no means qualify as obstacles.

Upcoming Days :


For the first few stages, while we nail down the procedural sequence which intelligently selects the area of land to build a city, and which generates the up-to the final city layout. Ill test both a procedural approach to generating the layout, and a random based algorithm which mimics the final product of the procedural approach.

I am in the process of finishing the development of a tool for testers which will most importantly allow testers to request the system to generate new city layouts directly from the server (you launch a program, have a few choices, you can dl and auto install the daily/weekly test map or can request a new random map be generated, dl'ed and installed). Since I am using the cuda architecture, the map's generate near instantly, however the map needs to render and this takes some time.

I had previously ported the compiling tools (VBSP, VVIS, VRAD) to CUDA for the most recent source SDK. However, the map format generated by most recent Source SDK compiling tools is not compatible with the L4D series. Which means we have to use the compiling utils provided by valve and there for compiling will take time (for the first few stages, compiling may take at most 5-10minutes, towards the very last stages, compiling will take upwards of hours-days to compile, but that's a far way off).

The good news is that I should be able to configure the compiling process to communicate the progress of the compile with the tool you use to request new maps and give you a general ETA time-frame window.

Furthermore, in regards to manually marking navigational zones in the first few stages, this will prevent any



First few weeks of development :


I think the best way to go about this will be to get a list of names (maybe even make a steam group if everyones cool with that? Will let me post events and those that are interested/available at the time and join up). For the first few stages (next few weeks, or a month a most), when I get to a point that i need opinions, criticism, ill generate a campaign, put it up on the aforementioned tester tool, and ill post an event on steam. Those available at the time can grab the map via the tester tool (which auto downloads and installs the map for you) and join up and well wonder around the map, check things out. Maybe play it through some with infected on, or disable the infected and zoom around with no-clip on (as well be focusing on the layout of the city and building assignments).

Some time in the future, ill start working on fully documenting all the l4d specific items/structures well need to have included-in and intelligently distributed throughout the the city for l4d to function correctly. These will be things like:

• structures/rooms:
  
  • the obvious, safe-rooms
  • maybe list of example infected "spawns", these are usually areas that are obscured from view or inaccessible to the survivors, like bathrooms or those dead end hallways that you cannot get into the end of on the l4d2 mall campaign, or fences the zombies can spawn behind and climb over, etc etc.
  • rescue closets for survivors
  • Finales (and escape vehicles)
  
  
• items distributed around the city:
  
  • med packs / adren / pills
  • weapon spawns/ weapon upgrades
  • gas cans
  
  
• etc

Any help with the above will be greatly appreciated.

Again, the Project should be at a far enough point to need testing within a few weeks.

Play Testing and Volunteering :

For now, the steam group for play testers is up, though private. If, after reading everything thus far, you are still interested in play testing the campaigns and giving constructive criticism, please let me know and I'll send an invite asap.

Also, I am in need of volunteers who are interested, able, and willing to dedicate time to aiding in the research process. That is, becomeing actively involved in the project and more than just a player who just tests and comments on the by-product campaigns from the lastest stage and version of The Creator. Helping with things like documenting the above L4D series specific items/structures; help organizing major test days; help processing user-input, and maybe updating the project proposal/summary (the OP) with accepted idea and changes, etc.

Either way, if you are interested in playtesting and/or volunteering to aid in research please let me know some how (reply to this post, forum pm, or yes, even steam-message ). By volunteering for either, you are not obligated to do any more than you want to and can opt out, rest, take a break at any time!



Thanks again!



Now lets get this party started,

Yes i realize the gravity of this proposal yields a cheesy tone,

Oh, another thought would be some way of marking where you have been. Because that will be one of the hard things about having random paths and dead ends. YOu don't want to be exploring an area you've already been in and wasting time and getting hurt more.

Like the survivors could carry around can of spray paint and be able to mark on buildings and streets that they have already explored. You know sort of like FEMA does during natural disasters like Katrina:



It could probabbly use the built in 'spray' function of the source engine, just modified a bit to allow multiple sprays to exist at once.

I'll be your guinea pig Bunni.

The concept bears many good ideas, it'll be nice to see an adaptive FPS with random terrain, and the possibility of making the game more than run n gun...

As for SV's rarely getting out alive... It's a zombie game - this should be the case

Originally Posted by Arreo

I love L4D but I just can't stand playing the same maps with the same routes over and over.

Off-topic, but part of the problem (in my experience) is, unlike in the first l4d, NO-ONE here plays custom maps anymore. Anytime you even mention one, it is usually met with "oh, its too long" "oh, its too short" "it looks dull" etc. etc. etc. A good map can get dismissed here in about 2 seconds because people are picky as fuck about every little god damn detail (one of the reasons I've given up on even trying to play customs and weed out the bad from the good).

Originally Posted by Arreo

So, to clarify. are you talking about a start point and one end point? Or are you saying that there would be one start point and multiple safe room end points.

Also, I think that you should focus mostly on coop play rather then versus since it increases the complexity by many magnitudes.

True though, i am more targeting co-op. As verses does not apply well to survivor exploration (which would be key to escaping the cities).

Single start point, with many, many, many end points. To give you some background, to get away from the current linear thinking of l4d campaigns: Image a city, with a grid with cells of size __ applied to the city. These individual grid cells, would be levels (in the sense of you have to go to the loading screen to go from one level/cell to another). The cells would be connected by windows, aka safe rooms. Now, while the city would actually appear like a city (ie a street on all sides of inter connected building blocks, like a grid), but in actuality, streets would be blocked, to yield a natrual, maze like system of PATHS. So players are limited to the paths they can take.

I cant find a good maze image right now. Think of a perfect maze:

source: Maze Generator in C# - Scott Beaudreau's Weblog - Site Home - MSDN Blogs

A perfect maze has only ONE path to the exit, any other turn off this path, ends in a dead end. What we do, is add more exits, and remove some of the walls that cause a path to dead end. Thats what cities will be like.


I dont think there is a limit to the number of scenes that a campaign may have, even more so, i believe they are only linearly connected via a configurable entity in the safe room that says "survivors hear load ___ map". Now that i think about it, safe rooms, do not necessarily have to act as a window to another level, they can put you back into the same level (but at the safe room door), so you can continue exploring the area, acting like a checkpoint if you die.



For verses though, just as easily as we can generate the outline of the city (really easy), we can just as easily specify the number of "paths", to really try and balance the scales for infected.




There'd be many safe rooms,

Originally Posted by Arreo

Oh, another thought would be some way of marking where you have been. Because that will be one of the hard things about having random paths and dead ends. YOu don't want to be exploring an area you've already been in and wasting time and getting hurt more.

Like the survivors could carry around can of spray paint and be able to mark on buildings and streets that they have already explored. You know sort of like FEMA does during natural disasters like Katrina:


It could probabbly use the built in 'spray' function of the source engine, just modified a bit to allow multiple sprays to exist at once.

Hmm good point, honestly, theres only been a few times that I've truely explored a city (that i was unfamilar with) on foot, but i think you are thinking more of a literal 2d maze with its walls raised:

source: Cool math games at Coolmath4kids.com - free math games for kids of all ages


Which would truely be dissorientating to navigate. But being a city, you have countless landmarks: buildings, skyscrapers, some buildings you can climb to get a nice nice elevated perspective of where you are in the big city, but if complete my goal buildings will be unique, so they will all be landmarks. The first handful of stages in the development though, will be very "are we going in circles" though, but just a phase...

Couldn't safe rooms just be used more so as a place to "take a break" so to speak. Basically it would have some weapons left over from past survivors, maybe health kits or something. Basically it'd be somewhere you could go to restock, take a break, maybe try to rethink plans, and so forth. If you really wanted to fuck with people, maybe some of them have a map or such to an evacuation point (i.e. like a CEDA point they were evacuated people), that could be used but leads just to another dead-end (or could be a chance thing that maybe they are still there, or maybe they aren't and you wasted time heading to a dead-end).