After I wrote recently about procedural generation for dungeons, my brain kept chewing on the possibilities here. I decided to try the process outlined in the Dungeon Masters Guide (5e) Appendix A. This post will run through my experience doing that, as well as a look ahead at what could come next.
The first half of Appendix A consists of a procedure to generate dungeon layouts (passages, doors, chambers, etc.). Despite the fact that the first sentence of the appendix tells you that it will help you “quickly generate a dungeon,” my experience has always shown otherwise. Rolling on the tables, interpreting things in a consistent manner, and drawing them takes a while.
For this experiment, I used the Dungeon Unlimited tiles from Elven Tower. (Because I support Derek Ruiz on Patreon, I didn’t need to buy the set on Roll20, but I did need to upload them to my own library and organize them.) This turned out to have mixed results, mostly because the DMG has some chamber shapes that don’t fit well on a grid, like circles and trapezoids.
After about an hour of rolling dice, looking up the results in tables, placing tiles, etc., I still had not quite finished. The partial example hadn’t really generated something with any sense of consistency in the layout. That might not be a bad thing, but I didn’t really like it.
Honestly, I don’t think I’ll use these again. The larger community produces too much high-quality map content to need procedural generation here, and the cost in time to build it takes far too long. (I did like the tiles, though, and would use them in other situations, like when creating an improved version of a sketched map from a module.)
As Appendix A notes, map creation “is only half the fun”. Dungeons need to be filled with monsters, traps, treasures, and other fantastic features.
I decided to try the tables found here with a pre-drawn dungeon. Choosing more or less at random, I picked another product from Elven Tower. This time, I used an unlabeled version of the Bloody Petal Cult provided to Patreon supporters. The Roll20 version matched the grid exactly, making it easy to create the dynamic lighting layer.
I did not choose a “Dungeon Purpose” in this particular experiment, though maybe in a future example I will. This means that I didn’t really go through the various Chamber Purpose tables for each kind of dungeon, either. Instead, I rolled on the Chamber Contents table for each room. This basically consists of a somewhat more detailed version of the table from the BECMI Dungeon Masters Guide, but effectively the same in spirit (and easier to read). Each room can have a monster (dominant, allied, or random), a hazard, an obstacle, or a trap or trick, and potentially treasure. Subtables are provided for most of these.
In this example, I decided to work it up as a level 1 dungeon. That means I tuned it for a group of level 1 characters. To keep this fairly generic, the monsters mostly focus on an undead theme, with a couple of flying swords in one room. (All the monsters I used came from SRD content this time.) I also used the treasure tables from Chapter 7 of the DMG: individual treasure in most cases, but two treasure hoards can be found as well in a secret room and after the hardest fights.
The hazard and trap tables surprised me – I really liked them! They provide enough structure to build out the traps, but leave enough “blanks” to let me fill things in according to my desire. As an example, for one trap, I rolled the trigger as “5: Looked at (mural, arcane symbol)”, and the effect as “96: Symbol.” The symbol spell has 8 potential effects, and I rolled “Death”. Whoa – 10d10 necrotic damage? Even if the players make their save, they will average 28 necrotic damage, just for looking at a painting. Instead, I rolled for the severity, which was “2: setback”. According to the trap guidelines, that means 1d10 necrotic, which fit. And a “death symbol” fits the dungeon theme quite nicely, I’d say. This gives me room for DM interpretation: what is that symbol? Or is it a mural – even a painting? I decided to go with that last choice, actually.
This process felt a lot more satisfying. The tables don’t tell me anything about what the monsters themselves consist of, beyond the three very broad categorizations I listed above. Encounter building guidelines at this point have been tuned pretty well, and randomization via Kobold Fight Club helped a lot.
Of course, this ends up creating a fairly static experience. We need to jazz it up with a random encounter table. Off the top of my head, I’d roll a d6 every hour. On a 5 or 6, no encounter happens. Otherwise:
- 1d4 skeletons patrolling for intruders, weapons at the ready.
- Gray ooze drips through a crack in the ceiling onto a random character.
- A tiny homunculus is spotted, maintaining a watch on the adventurers.
- Avatar of Death materializes from within the floor. When destroyed, each adventurer feels a foreboding sense of doom wash over them.
The dungeons really lacks some “dressing”: what other interesting features does it have? Sights, sounds, smells, objects, etc? Appendix A provides tables for these, and I might use them (or something like them) later. That big empty room in the middle of the dungeon provides a great opportunity to embellish things. Maybe there are tracks leading off to the east, or corpses of adventurers who’ve come here before. Graffiti on the walls could provide clues as to the dungeon’s creator and master. And where does that exit in the northwest lead?
One thought on “Worked example of a procedural dungeon”
Take a dozen tiles of rooms. Toss them onto a table. if any land ontop of each other (at all, even just corners) they become a multiple story room. Add stairs/ladders or otherwise work out any wonkyness. Add corridors to connect them, ensuring loops. Then stock the place.