On Friday 16 November 2001 12:57 am, you wrote: > The idea is interesting. One question obviously is - is it easier to set > up something like that, or just expand the map by hand/script? It's debatable, I'll try to write a map generater with some extra time I have this weekend. I don't know enough about the engine to modify it, but I can right out a map format. > The main problem is the resulting map that may get generated and how it > mixes in with quests and the like. Some quests say 'find the caves near > the south edge of the main mountain range' or the like - if that mountain > range is now someplace else, or does not really exist, such messages become > pretty useless. > You also need make sure the map generated fits in - you probably don't > want a case where the world now ends up as a bit of islands and no easy way > to get from one to another. I would guess that the map builders would have to worry about this with the new map anyway. Plus a number of different numbers could be plugged into the generator until it generates a map the similar to the current map. > Note that the current world map uses about 700k (and I note that probably > 10% of that is overlap spaces). So even if the new map is 100 bigger, that > amounts to 70 MB - yes, a bit of space, but on most any modern systems, I > can't see that being much a bigger. I hear a lot of talk of increasing map detail so that outdoor maps are the same scale as the cities. If this turns out to be the case, the map could easily be 100 times bigger in each dimmention. In this case, the map will take 7GB if stored on disk. Yes, I have that much free, but it's still a lot to use up for one map. > Well, maybe. The world map is obviously going to have entrances to > dungeons and the like - these dungeons need to lead back out to the same > place - they need to use some coordinate to do that. IMO, a bit of the > effort in expanding the world map is not increasing the scale, but updating > all of those exits to the new paths. Not increasing the scale, increasing the size by allowing players to pass beyond the border to whatever may be there. > Debatable. Presumably, no matter what the scale/generation, people know > that they want to go to certain places (eg, the dungeon that is located > there, or the south east corner, or whatever). So while the terrain the > player is travelling through may end up being different, the important > pieces will remain in the same places (eg, dungeons, etc). And unless the > map is infinitely large, in any case, places will probably get explored. > And while that is an interesting idea, unless there are interesting things > on all areas of this infinite map, you really want care. Interesting if you care about looking at scenery. Or fighting generated monster populations. > > 5. The map would use a universal coordinate scheme. Thus monsters could > > track players through map block without worrying about map transitions > > and portals. > > With the new map tiling code, this currently works. Cool > Other thoughts: > > The map as a whole has to be consistent. If you have a player starting in > the far east, and another in the far west, the system must be such that as > they move (and new map gets generated), if they meet, the transition makes > sense. > > Second, it would seem most likely that each server would generate this > world map (or areas) once, and not again. Otherwise, it is hard to explain > why that mountain range that used to be there is not there anymore. Perline Noise functions are deterministic. I can ask the terrain function for one x, y location on the world map, and it would return the same exact result every time I call the function. I suggest the following site for Perlin Noise info. http://freespace.virgin.net/hugo.elias/models/m_perlin.htm > Third, computer generated maps are pretty much never as good as human > ones. Presuming you do as you say and some features (eg, rivers and the > like) are put in the over ride maps, you know have the problem that a river > may run through a mountain range because the generator put a range where a > river was running. See above. The map designers would know what the map looked like as they made changes and I hope would not run a river straight through a mountain, and could fix any mistakes the terrain engine made. > > Multi > > server capability(Different maps run on different server, One server is > > authority on player stats) > I have yet to see a compelling reason to do this, and lots of reasons not > to (complication, syncrhonization, more points of failure) OK maybe on doing this on such a scale is not a good idea right now. But maybe taking the image server out of the main server, and putting them on authoritative image servers around the internet would have some benefits. Later Adam