D&D 5E DM resource: on maze generation

[FormerlyHemlock]

I found this useful and want to recommend it: http://www.jamisbuck.org/presentations/rubyconf2011/#title-page

It has some thought-provoking observations on mazes which are obviously relevant to any game designer or DM creating a dungeon or other maze in his game.

Naturally, as a DM, you don't have to construct all of the "walls" out of stone. A monster can be a "wall" which seals off a portion of the maze, too, and so can a clue. And as a DM you don't necessarily need to restrict yourself to perfect mazes--in fact you might deliberately seek out an imperfect maze so that the players have multiple routes to victory.

It's worth a read.

 

[Quickleaf]

I found this useful and want to recommend it: http://www.jamisbuck.org/presentations/rubyconf2011/#edge

It has some thought-provoking observations on mazes which are obviously relevant to any game designer or DM creating a dungeon or other maze in his game.

Naturally, as a DM, you don't have to construct all of the "walls" out of stone. A monster can be a "wall" which seals off a portion of the maze, too, and so can a clue. And as a DM you don't necessarily need to restrict yourself to perfect mazes--in fact you might deliberately seek out an imperfect maze so that the players have multiple routes to victory.

It's worth a read.

Interesting. As a non-mathematician, what struck me was that the "Hunt and Kill" variant of Eller's algorithm created less dead-ends (better for D&D maps I'd argue), and that the Recursive Backtrack algorithm first began by dividing the maze into zones (useful for D&D dungeon/mega-dungeon design).

Kruskal's Weave was pretty cool to get you thinking about 3-dimensionality.

I didn't really see what was so great about the Growing Tree, however, which he says is his favorite algorithm.

Not sure there's much more to it for DMs to extract for gaming, however. What did you get out of it?

 

[Hussar]

Did I do something wrong? I clicked the link above and it took me to a picture of two dots and a line connecting them. No article.

 

[FormerlyHemlock]

Did I do something wrong? I clicked the link above and it took me to a picture of two dots and a line connecting them. No article.

Sorry, my fault--I didn't realize the URL I had was an interior link. Correction: http://www.jamisbuck.org/presentations/rubyconf2011/#title-page

It's a slide show. You should be able to navigate between them with the arrow keys.

 

[LordEntrails]

Use arrow keys to navigate, or go to the start of the presentations: http://www.jamisbuck.org/presentations/rubyconf2011/

 

[FormerlyHemlock]

Not sure there's much more to it for DMs to extract for gaming, however. What did you get out of it?

One thing I got was that you can view adventures in general as a kind of graph, which means you can use maze-generation techniques to design e.g. randomly-generated mysteries, complete with as many red herrings as you want, as long as you have ways to generate labels for rooms. For this purpose, the "biased" algorithms like Sidewinder that are too easy to solve may actually be better than complex mazes--it limits the amount of work you have to do as DM.

Here's one example of a murder mystery that kind of peters out at the end (needs more work) but illustrates how I'd use the maze generator as a forcing function for creativity. Short corridors indicate quick resolution (little table time invested before getting a result). Branches indicate decision points presented to the players. Obviously they are free to go off-script, but the maze gets me thinking about possible ways for them to go off-script and how I could prepare fun stuff on multiple branches that still eventually leads to a solution.



Another thing I got was that you could make a regular dungeon crawl using e.g. Hunt-and-Kill, and each separate segment that you generate winds up belonging to a different kind of monster, which gets its own exit to the outside world. So maybe the gnolls get segment one, and bulettes inhabit segment two. You'll wind up with fairly organic-looking monster territories, each with its own distinct type of spoor.

I agree that the weave variants are also very inspiring.

I'm still kicking around in my head ideas for how to apply these ideas, but so far that's what I've got.

 

[Imaculata]

I coded a random maze generator for a Zelda clone a while back. What it did, was:

-Pick a random maze-size in rooms horizontal and vertical, within constraints
-Choose a random room as the starting room, but always a room away from the sides of the maze. This is contrary to normal dungeons in Zelda games, where the start is always at the border of the maze.
-Create a random path from the starting room, with a random length. The algorithm makes random left, right and straight ahead turns, but if it gets stuck, or reached the end of the length that I randomly set, it places the exit there.
-It then creates a boss room, just before the exit. Basically it just takes one step back, and then checks how big it can make the boss room.
-It then selects random rooms that it visited along the way (except for the exit and boss room), and if they are next to an empty space, it expands the maze in that direction with another random path.
-The algorithm labels each path, and also from what room along the main path it originated. This way it can decide where to place locked doors, and where to place the keys for those doors, so that the player can reach them.
-It places treasures and special items in dead ends of the other paths.
-It creates a random number of secret rooms from the left over empty rooms, if they are adjacent to at least 2 rooms from the maze. It then decides randomly how many secret doors lead into each secret room. I didn't want the placement of secret rooms to be entirely random. By making a secret room always border on two or more existing rooms, they will always be placed in a corner between two rooms, or in an empty eye. So they are always in suspicious locations.
-Go through each room, and select random prefab templates for the rooms, to give them unique looks. Some of the contents in the room are also randomized.

 

[hejtmane]

This is a great one for a dungeon builder

5e Random Dungeon Generator

https://donjon.bin.sh/5e/dungeon/

 

[Gradine]

Do many folks still run their players through literal mazes? I always imagined they became passe in much the same fashion as they did in computer games and interactive fiction ("twisty little passages" indeed).

Where I do see the value in this is exactly as [MENTION=6787650]Hemlock[/MENTION] demonstrated, through adventure/scenario design. Though there's danger there too in constricting yourself to the branches you've created, since you can never really account for the range of possible player actions. It's something that needs something supplemental, a list of quickly improvisable scenes and encounters if the PCs get off track. This "extras" can be designed to provide additional clues to steer the party back in the direction you want them to go, or they can perhaps help you follow whatever tangent the PCs are on about and have that loop back around to the original adventure.

There's a great story I remember reading recently (I wish I could find it, my google fu is failing me) about a DM who was running a murder mystery, and that the way they had written it their "Miss Scarlet" analogue was the killer. However, the players ended up creating an elaborate reason for all of the clues they had found to instead pointed to "Colonel Mustard". The thing is, their reasoning made a ton of sense to the DM, who had never even considered the players' theory but was actually rather intrigued both by their theory and by the implications of the players actually being right. He ended up deciding to change the plot mid-session so that the players were actually correct; he decided that was the solution that provided the best possible play experience for everyone involved. I'm sure there are some DMs who would balk at such a suggestion, but it sounded like it worked great for that group.

I'm pretty far off-topic, I realize. But I do like the idea of using these branches to help design more open-ended adventure scenarios (what I think the Alexandrian would term Node-based scenario design).

 

[Imaculata]

Do many folks still run their players through literal mazes? I always imagined they became passe in much the same fashion as they did in computer games and interactive fiction ("twisty little passages" indeed).

In the campaign I'm currently running the players are coming up to an island that is a literal maze of coral, created by a dragon.

But I have some ideas on how to tackle this maze a little differently. I want to give my players the experience of navigating a maze, without going through endless boring empty corridors. So what I want to do, is show the maze as a relatively small abstract series of tiles. Each tile represents a section of the maze, with many corridors, and a challenge. So rather than focusing on navigating each and every corridor, I want to focus on the interesting things that the players encounter inside those sections.

I suspect there will be a lot of role playing and random encounters, instead of a boring crawl through countless dead ends.