Thank you for indulging me gentlemen

I have searched Steam extensively but it's not to say that there isn't something I missed, I'll keep looking

I have seen many games use procedurally generated open world environments like minecraft, but none that are action/skill-emphasized and set in a contemporary realistic-ish environment.

I understand why you might think procedural generation would be ill-suited for generating complex balanced and detailed maps similar to those you find in gta games. But, my idea is that user generated/customized maps could be uploaded, downloaded, and shared between players allowing countless unique and potentially (albeit not necessarily) brilliant maps (like you see with minecraft).

Does anybody think this would be a good/viable idea as I do?

Am I crazy to think this would be awesome? Or am I missing something obvious that breaks the whole game?

Of course any number of pre-made maps could be released with the game. This is just a matter of level design which would obviously be a lot of work.

Procedural Generation would be the only thing making this project achievable for a small team. So I agree with your point about enabling small developers, that's what I was banking on.

I understand the fear of trying to do too much at once, but the "SimCity" part of it is not the emphasis, it's the gta-type game play. I think the scope is small enough to be doable, like you say, just difficult.

Have you ever played Sim Chopper? It's an awesome shitty rotary aviation game that is capable of importing Sim City 2000 maps and converting them into 3-D playable areas. I always thought it was a fascinating concept. As I wrote this I realized I may be channeling that game somewhat but as interesting as it is it doesn't come close to providing the game play experience I had in mind.

Thanks again yo

Procedural maps are easy. Yes, lots of old games used them.

You've got a huge number of old arcade games that were basically the precursor of the endless runner. Right turn or left turn randomly selected, forever.

Then you've got a simple "random track". There were several old games I remember in the 1990s back when people bought force-feedback steering wheels and yokes. You can make driveable paths, but people quickly discovered that they lack story.

Still today there are many tools in Unity for it. I've seen several games that use Easy Roads or use the Racing Game Kit. Yes, you can drive around a track, travel through waypoints, and even pick up powerups, but that is not what GTA is about. That is almost more like what Need For Speed is, but even that is not a random course.

GTA is well-designed, carefully considered, planned world. Real humans start with the basic algorithms and then clean them up, then either use the results to fit the story or make modifications to make the story fit the models. "Random" is usually a poor choice for stories.

Yeah user customization or something equivalent provided by the design team before release would be necessary to optimize the maps. Also think gta3 not gta5 with an emphasis on things like vigilantly missions paramedic missions. I'm not super concerned with story missions even though I feel they could be incorporated with the risk of some missions becoming un-playable due to some particularities of the given map. There would have to be work arounds incorporated into core features of the game.

By this I mean mechanisms similar to the ropes and bombs in spelunky. In spelunky some levels are unbeatable without a couple bombs or ropes. I haven't completely figured these workarounds out, outside of the fact that players could have some degree of control over the terrain and the buildings/roads/buildings generated by the game.

so in short you are agreeing that it all sounds feasible?

thanks for the direction btw frob

Thanks a lot for the time

define feasible...

Doable? Yes.

technologically challenging? Most probably not, as long as ther result is not expected to be perfect... or good.

Easy to design? No (Depends on your expierience, for a beginner a clear no).

Easy to program? No (See above).

Doable in a year, working alone? No.

Doable in a year, with a good sized team off expierienced developers? Maybe.

Is a Procedrual GTA a good idea? -> Has been discussed by other above, it is highly subjective, so I give it a clear Maybe

Now, either start working on this game already, and maybe, in 10 years, you will be able to finish something that comes close to your current aspirations....

Or just start simpler, and see how far you get with that (and learn the needed skills for more complex stuff on the way)

There are myriads of things you can do to either scale down your current idea to a more achievable goal for a single person or small team, or start with smaller games to get your feet wet. This has been discussed to death in other threads, so I will not repeat that information here. Just have a look at the FAQ Thread here.

Getting believable, driveable cities would be a challenge. But not impossible go for it.

Also, the only reason to ever be concerned about whether a game already exists, is so that you can go shamelessly steal all their good ideas and learn from all their mistakes. If you're only hope for success is by being the only game of a certain type, then you're pretty much doomed to failure already.

"procedurally generated GTA"
...
This may result in "bad" or "stupid" maps. So you also want to incorporate a mechanism that allows players to modify cities and [terrain].
All these things combined would allow players to personally create and [customize] game maps using a set of prefabricated and/or custom [designed] buildings, ramps, [terrain,] objects, cars, or whatever else makes sense ...

The procedurally generated part says "random" to me (so that each time I play, the city is laid out differently). I think there are reasons why that isn't going to be popular.

The user-created part sounds more reasonable to me: sort of a Sim City that you can drive around in.

btw, I moved this to Game Design since you're designing a game, not asking technical questions like how to procedurally generate a city. (That would want to be asked in a technical forum.)

This idea has seemed like a no-brainer to me for a while. I'm surprised it hasn't been made already. Is there a good reason why it hasn't?

Well, I'd say the big draw of the GTA games is the richness of the cities; all the different neighborhoods and styles and cultures, and the interesting details that make it seem like a world. It's relatively easy to procedurally set up buildings and streets, hard to support the illusion that it's a real, living place. And the more familiar your audience is with something, the harder it is to generate that illusion algorithmically, especially if you're trying to capture it in detail. (Hence fantasy dungeons, alien planets, wilderness: things the audience only has foggy ideas about.)

There was an interview at RPS with Introversion about their procedural-city-crime-simulator Subversion, and why they cancelled it: http://www.rockpapershotgun.com/2011/11/08/introversion-explain-subversions-fate/

The Idea could be summerized as "procedurally generated GTA"

...

Has anyone seen anything like it?

Since this is a game design related question: yes, it already exists, its called GTA

The average player most likely does not play through a game (maybe getting bored after 75% ), so from this player perspective it is just GTA with content, regardless if some designer made this content or if a computer generated it. Eventually it is just a clone, because there's no real new innovative gameplay, just an other way to generate the content.

Thanks for all the comments and suggestions guys

That article was exactly what I was looking for Valrus, you are the man.

That subversion game sounds awesome, what a shame. Perhaps the generated city is tougher than i imagined but 95% sounds pretty good to me. This becomes especially true when you use tactics like what he spoke about regarding the banks. (generating a bunch of banks and then using a finite but large number of them that were deemed "acceptable" to more or less achieve the desired "randomness"). Could such a method not be used in basically every facet of the city generation, until it produces an acceptable map? Is that just too much work? or is it unworkable for some other reason that's beyond me? It seems like they touched on a lot of my points, like the user control of the environment. I hope someone pulls something like this off.

This project is obviously way way beyond my skill level. (haha even that's an understatement) but I really wanted to know what you big shots thought about my baby-brained idea.

I will post more stupid crackpot BS ideas like this if anyone has interest.

I got a shitload more specific ideas about this game but I think those would be regarded as a waste of time. If anyone wants to hear them feel free to pm me.

-Steady out

The closest thing I can think of is a game similar to Simcopter (same engine in fact) called Streets of SimCity.

http://media.moddb.com/images/games/1/13/12724/StreetsOfSimCity-4.png

It was basically a driving game but the random generation of cities could be done in SimCity 2000 and then played instantly.

Depending on the game mode, the cars had weapons (similar to Interstate '76 /'82) but much of the game was about driving around collecting and delivering packages. You also couldnt get out of the car unlike GTA :)