Friday, April 29, 2011

Untested Collapsible Encounter Method

Here's a dungeon generator. It's fast-ish and requires less prep than a real dungeon but still requires some, but the main advantages are it's very reusable and will tend to produce dungeons that make some sense, with the danger and reward increasing as the PCs move further in...

You'll need to think of 8 monsters, put them in line from least scary to most scary (scariest is 8, wimpiest is 1), or in a rough hierarchical order from boss down through lieutenants, soldiers and finishing with vermin (at 1).

Optional step: Think up 8 traps and put them in order from least to most scary. If you can't think of any or don't have time, use the standard array (poison gas, needle trap, pit trap) and just arrange them so that Trap 8 does the most damage and Trap 1 does the least (you could go: d4, d6, blind, d8, d10, d12, d20, save or die, for example).

Optional step: Think of 8 treasures and do the same. If you don't have time to do this step, just go from like 10 gp to 800 gp or whatever is scaled appropriately.

Optional: Think of 8 locks and do the same. If you don't have time, assume lock 8 just takes the longest and makes the most noise and lock 1 is the easiest.

Now take a look at the graph (I have filled in some sample monsters and treasures on the right of the graph and have left traps and locks off).(click to enlarge all these)
This will all be explained, so hold on.

First: It works like this: the farther the PCs get from the entrance to the dungeon (in rooms or levels, but most of the time it'll be in rooms) the (usually) fiercer the opposition and (usually) greater the reward.

(This makes sense: this is the point of buildings, from a fortification standpoint--the boss hides away with his loot and puts layers of underlings between himself/herself and any entrances, or the fearsome beast keeps its pile of bloody bones in a cave far from hunters and other predators and the other animals lurk at a respectful distance to scavenge leftovers. Also, the further the PCs go from the entrance, the more likely they are to go where no looters have gone before.)

(Note also there are spikes in the graph making it so that the "front door" is likely to be defended as are the sort of mid-zone "living areas".)

When DMing any room, you look along the bottom for however many rooms you are from the entrance to the dungeon. (Let's say you're 3 rooms away. Keep in mind PCs may not know all the entrances to the dungeon.) Follow the vertical line up from there and you'll see that it intersects a green line representing monsters, a blue one representing locks, a red one representing treasure, and a pink one representing traps. When you intersect, look left (the numbers 1-8 down the right are just for reference later.)

For each room, roll 4d4--one die for each element. (For maximum speed use 4 different colors matching the graph colors but whatever.) (oh, hold on, ignore any line that's waaaay at the bottom below -3, that means you don't have to roll a d4 for that thing cause there's no chance of getting it) and add the modifier indicated for that element on the left. For example, in room 3, the trap modifier (pink) is minus one.

Then you get a number. If it's zero or below, that thing is not present in the room. If it's a positive number, then the room contains a thing of that type and level of awesomeness/toughness/scariness etc. Like if you rolled a 3 for traps and the mod was -1, then you'd have a 2 which means this room contains Trap 2, your second wimpiest trap.

Here's a bigger one for a bigger dungeon using a d10 instead of a d4--meaning you have to think up 20 monsters instead of just 8.
This method might be a tad cumbersome (4d4 for every room), like I said, I haven't tried it--however, I see advantages:

-It produces dungeons that make sense and does so relatively quickly.

-It is re-usable. If you're running a sandbox campaign, there is always the chance the PCs will run off to loot some theoretically-fortified locale or obscure ruin that you haven't written yet. If you can think of 8 types of inhabitants that might be in such a place, you've got a dungeon.

(obviously this method doesn't generate layout, I generally go by the "assume the room's square and has d4 exits" method if I have to pull a dungeon out of my ass)

-The only other way to produce results like this would be to write percentage charts for each room--i.e. room 8 has a 30% chance of monsters, a 60% chance of traps--PLUS have a second chart making it so you'd roll on a slightly different encounter chart the further you get from the entrance. i.e. in this method, the array of possible monsters near the ends is different than the array in the middle and it's all done on one chart.

-This does not have to produce linear dungeons--no matter which way the PCs go, as long as they're getting far away from their lines of escape they are going to get to something. If you do this long enough and re-write your encounter chart at intervals, you might end up with an interesting factional dungeon.

-This "graph" method can also be used for other kinds of encounters in a slightly different way.

Look at the graph again.: Assume the 4 colored lines all represent different kinds of wilderness encounters (green is a regular monster, pink is npcs, red is some inert, found locale, and blue is weather, f'rinstance) the numbers along the bottom represent "days from nearest settlement".

Now on each day you have a different chance of meeting different things and the further you get from civilization the greater the hidden rewards, the fiercer the beasts, etc. (Though you could alter the graph to make NPC encounters less common as you get further from civilization and make the weather curve less steep--after all, bad weather is often more likely near settlements since they are often closest to the water.)

Likewise imagine the colors represent different kinds of urban encounters (thieves, soldiers, animals, etc.) and the numbers along the bottom represent the median income of the neighborhood. Rather than a series of encounter charts where beggars are rife in Cheapside by unheard of in Thistlewood Lane (seriously ever notice it takes half an hour to figure out who you're encountering in the City State of the Invincible Overlord?), you can just write one table 1-20 and put beggars at the bottom of it.

So, all you need to make your own is 4 pens and a piece of graph paper. That's good, right?

Or, if you're a computer programmer: there you go. This is probably how some computer games work already.

Now I personally have a list of groups or types of baddies in my campaign, with notes like this:

Villain(s): Hex king
Lieutenants: Eyes of fear & flame
Grunts: Caryatids, skeleton
Around: gloomwing moths
Trap regime: Creepy black fairy-tale like a music box that prevents magic so long as it plays
Treasure: Old magic and art
Schemes: Some kind of pact with Nephilidia

Villain(s): Insect gods (En-Gorath, Hammurabi, etc.) Chasme
Grunts: Githyanki, Jackalmen
Tough guys: Giant bugs, CIFALgangers
Trap regime: Ancient Egyptian-style curses & mechanical traps
Treasure: Gold, insect magic, blood gems
Schemes: Destroy Slaads

I have 20-some of these: groups of monsters that are on the same "team". If the PCs unexpectedly burst into a stronghold I can randomly pick a faction (or maybe I know from context whose it is), plug the notes into the graph and hey, that's enough dungeon until the PCs have left or the session ends and I have time to figure out what's really going on down there.


yourboogieman said...

I have always been a big fan of "Eyes of Fear and Flame." All this Fiend Folio love lately has gotten me mighty nostalgic.

John Evans said...

You've packed a lot of information into the chart there. Pretty slick!

Welcome to Dungeon! said...

A lot of promise here. It reads like it has pretty good usability already, but I'm all for seeing successive refinements of this process.

-C said...

I may have to stop reading your blog, being I'm trying to come up with ways to do similar things for the re-release of my dungeon stuffs and don't want to be accused of outright theft.

Excellent post

Blair said...

Oh Jeez, it's Thursday's Polynominal Function Transformations exam comin' back to haunt me...

I love your example faction compositions, with the heavy FF component but esp. how they're so random ("the insect god has githyanki and jackalman followers?!") but I can totally see it in a Sandman or Swamp Thing comic; I should have said that Vornheim was Fairy Tales filtered through nineties DC Vertigo comics.

Adam Thornton said...



"Or, if you're a computer programmer: there you go. This is probably how some computer games work already."

If I have a few hours this weekend....(no promises, got a birthday party and some Dominion playing and an oyster-and-cocktail bar to go to, and a new foster dog to keep calm, but we'll see)

Dan said...

This is related something that's always bubbled in the back of my mind but I've never got around to expressing.

Seeing how important numbers and their patterns are to RPGs like D&D, it seems crazy that so little progress has been made in the way we formulate, interpret and display those numbers.

In the age of excel, when spreadsheets can do so much amazing stuff and most school graduates ought to be able to whip up a graph... I just hope that you've started something here that spreads a lot further.

richard said...

regarding your heuristic for "depth" from the entrance equalling power, you've just recreated Hillier & Hanson's space syntax. The big inversion of this is care institutions (asylums, prisons, hospitals) where the most powerful people are between the entrance and the least powerful ones. I have a post brewing on dungeons as building types for exactly this. Most loosely-designed dungeons seem to be apartment buildings to me.

Blair said...

Ananlysis of my first attempt results with a 23 room dungeon:

- A disproportionate butt-load of #6 monsters (Lamias in my case)
- No treasures over #4 (haha! that's a problem for the players not me ;P )
- Lots of Traps!

I'll try with the d10/20 version next.

Adam Thornton said...

Do you just draw the graphs arbitrarily but generally sloping from lower left to upper right, with spikes where there "should" be something (like the entrance guards, the choke-point-to-the-inner-part-of-the-fort, or whatever), or is there some technique for determining the graph height as a function of from-entrance depth?

Zak S said...


just drew the graph. it's about the principle not the numbers. the numbers can be adjusted.

Roger the GS said...

I guess once you put this to the test you'll see if the properties you code into the irregular function earn their keep - as opposed to a simpler escalating modifier to each roll.

The real nugget here is the mini-encounter tables for different groups in the campaign, which has me thinking in a number of different directions at once ...

Zak S said...


if you're running a more "cinematic" and less "the world is the world"-type campaign I could see the escalating modifier method working fine, or a model where the numbers along the bottom represent units of real-time spent playing.

Also: for intelligent, coordinated monsters alerted to the PC's presence, the escalating modifier thing for monsters sounds perfect.

Christian Kolbe said...


Hizzah! Another great, easy generator. As a DM working my way through my own dungeons this gives me a great way to randomly generate the possibility of an encounter while still being sure that there is some environmental/player choice probability that comes into it.

Like a few of your other things, I am gonna take this and run with it.

scott said...

this whole thing screams "swords and wizardry". i could cook up a whole evening of old school fun with something like this at my disposal.

what kinds of monsters are in the bandit lair? check. traps and treasure, check.

grab my favorite random room desriber and then rock on.

you could also get a lot out of the "gaming paper" dungeon geomorphs i think.