Prefab areas (the move away from procedural generation)

Racer1Racer1 Join Date: 2002-11-22 Member: 9615Members
After the Destructoid interview, its clear that procedural generation will go away, or at least be used much less often than had been anticipated. This means UWE will prefabricate all of the areas. I think replayability is still possible, and desirable (despite what was said in the interview), with only a small effort beyond designing the areas themselves...

I can see prefab areas working if:
1. The prefab areas are attached randomly (rather than always in the same place/orientation)
2. Each area has least a few random variations (either designed in or auto-generated)
- These would be not only aesthetic differences, but require you handle them somewhat differently depending how they show up (but are much less effort to add in then creating a whole new area).

The clincher (much more work to implement but would pay huge dividends):
3. Players can design new areas, to be submitted to, and rated by, the player community at-large (via an easy-to-use in-game tool). Each player could then choose the minimum rating they want to allow, and the game would automatically add random community areas to those players' games as needed.
- This could be done either with a design tool, or by saving an in-game area that the player has modified

Comments

  • SteveRockSteveRock Join Date: 2012-10-01 Member: 161215Members, NS2 Developer, Subnautica Developer
    edited June 2014
    I think attaching prefabs at random will lead to boring worlds. We actually did this for a while, with 16 meter prefabs, but everything was looking very flat (cuz the tiles have to line up in specific ways). We tried ways of adding more interest to this, inspired by Starcraft, but it was quickly becoming unwieldy for the level designers and very limiting for them. It's also just not very interesting once players figure out what it's doing, so it seems like it's not worth it.

    I do like the idea of letting players create worlds. This could get complicated in practice, but it's more of a known quantity.
  • Racer1Racer1 Join Date: 2002-11-22 Member: 9615Members
    I was thinking areas on the order of 200-400 meters across. This way, the whole area could be designed ahead of time, allowing for the detail you require (with aquatic life, danger areas, and even missions). I agree small tiles would be too difficult to stitch together believably.
  • SteveRockSteveRock Join Date: 2012-10-01 Member: 161215Members, NS2 Developer, Subnautica Developer
    Yeah, larger areas give us more control, but that still precludes us from doing huge features. For example, a Grand Canyon, or a huge lava tunnel that goes across the "sky". It's also weird to re-arrange these huge tiles independently of each other - are they going to be like completely independent islands? If not, they need to be blended together some how. We did do procedural cliffs for the PAX demo, but that requires a good amount of computation as well as limiting how the cliffs can look.
  • Racer1Racer1 Join Date: 2002-11-22 Member: 9615Members
    edited June 2014
    Good points.

    First, I don't see any reason certain areas couldn't be larger in size. It doesn't need to be a rigid grid.

    Second, I see two options for transition between areas:
    1. Use procedural generation to bridge the gap between the areas. Perhaps each "edge" of each area could give cues (either given by the dev, or just look at the terrain), and the procedural code would gradually change from one area to another.
    2. At the edges around each predefined area, require a buffer zone defined by the dev, which would still contain the "style" of that area (but possibly to a lesser extent). The game engine would have the freedom merge these areas together. This would still require a bit of magic (to match ground levels, etc), but not nearly as much as option one.
  • SteveRockSteveRock Join Date: 2012-10-01 Member: 161215Members, NS2 Developer, Subnautica Developer
    Yeah maybe that would work, but it's definitely more work on top of world creation tools we need to do either way. And is it really worth it just to re-arrange large areas? Not really convinced of that. It's much simpler for us to, for example, just randomize the player's start position each time.
  • Racer1Racer1 Join Date: 2002-11-22 Member: 9615Members
    I understand, time+manpower is always an issue.

    Well, at the very least, it might be nice to build them as separate areas, and glue them together beforehand. This way all of the pieces are in place -- so, if you decide you have the resources to later allow for random or player-submitted-areas (or if you provide that ability to the community), this can be done without much work.
  • SteveRockSteveRock Join Date: 2012-10-01 Member: 161215Members, NS2 Developer, Subnautica Developer
    yeah that is more or less the plan right now
  • GrimfangGrimfang Join Date: 2003-02-04 Member: 13086Members, Constellation, Reinforced - Shadow
    There are a number of boardgames that have a number of tiles, that are randomly placed for every game. While you might easily be able to identify the individual tiles, it makes for a new game every time. So if time permits, I would suggest leaving it open. Maybe the game where the kelp forrest resource tiles are next to the shark-pool and lava tiles makes for a more interesting and dangerous play than one, where they are further away, allowing for safer exploration of the kelp.
    Those tiles are just an example, but you probably get the idea.
Sign In or Register to comment.