Saturday 23 May 2020

[BLOG] The Anatomy of a Dungeon Map

Over several years, in posts and forum comments, Yours Truly has ranted and raved about the general decline in mapping quality in tabletop roleplaying games, later switching to first person shooters, and then back again to RPGs. Wading through disappointing map after disappointing map, it is easy to get the impression making a good dungeon map – the sort that leaps off the page and encourages exploration, environmental puzzle-solving, and creative tactics – is a bit of a lost art. Many dungeons are in fact not dungeons at all; rather, they are illustrations depicting something like a dungeon map, but offering none of the dungeoneering experience due to their limitations. It becomes all the more important, then, to highlight the good stuff, maps with the right scope, complexity, and structure.

Good structure is especially tricky, since many promising maps conceal a rather banal layout under visual frills, as well as twists and turns which do not, in fact, do anything – they are visual noise masking linear corridors and the occasional, vestigial side branch you can visit before returning to the main one-way rollercoaster ride. Good structure is still more of an art than an exact science, but it is generally agreed that some structural features are better suited to “map flow” than others, by encouraging meaningful decisions, environmental interaction, and emergent gameplay:

  • non-linearity, aided by branching and looping elements;
  • three-dimensional environments with verticality, interesting interconnections between dungeon levels, and a variety of terrain (c.f. “jacquaying”);
  • relative openness, counterbalanced by occasional bottlenecks usually referred to as “pinch points” or “choke points”, and maintaining significant barriers to make navigation a challenge.

Not every dungeon has to have these features to be a good dungeon (and keying is the second half of the puzzle), but generally, they help. Furthermore, the principles apply to tabletop games and FPS games in different ways; thus, Ultima Underworld, classic Quake levels, or Thief’s Down in the Bonehoard embody these principles differently than Caverns of Thracia, Tegel Manor, or Tomb of Abysthor.

The Winter Tombs
The current post looks at good design through the example of The Winter Tombs, a free dungeon level by Dyson Logos. This will also be released as a dungeon dungeon by Jim Pinto, but for now, we will restrict ourselves to the map. This is a particularly good test case, since it is a map that has a pleasing complexity without obfuscating analysis, and its structural elements are easy to identify and discuss. Here, I will only reproduce a low-res specimen; for the larger map, go to Dyson’s site, download the map, and print it at home. For the sake of analysis, I rotated the map by 90 degrees, putting the entrances on the bottom (hereafter referred to as “South”), then I produced a line graph to showcase the map’s structure. So, what lies beyond the cross-hatching?

The Winter Tombs is a single-level dungeon map with a tomb/caverns theme. The closest analogy is Judges Guild’s classic Sunstone Caverns, a semi-keyed dungeon from their second campaign instalment. Like Sunstone Caverns, this one is densely (although a bit less densely) mapped to provide a large playing area in one map.

Dungeon graph
The first element that leaps before the eyes is the selection of ways in: right from the start, the explorers can choose among four tomb entrances and two cave mouths. One of the tombs is a dead-end: a simple, but pleasing trick. Others are connected to the dungeon in ways that integrate cavern and tomb elements, with signs of environmental degradation and blockage to complicate navigation. The way the characters choose to enter and exit through will have a meaningful effect on how expeditions develop. Nevertheless, there is no initial difficulty selection, at least no outwards sign that makes one or more entrances harder to discover or access. (As a counter-example, consider the lower level back door to In Search of the Unknown!) Notably, there are not many outright exits. They are in the distant reaches of the map (near the octagons and in the upper right corner). If this dungeon has lower levels, it will take a lot of initial effort to reach them – the overall permeability is quite low until secure routes are identified and firmly established!

Choke points and tricky bridges
Of course, many elements which appear complicated on a first look are essentially straight lines – a labyrinthine set of catacombs in the lower middle is a simple loop, impressive halls are essentially fancy corridors (bottom right), and many of the twisting cavern passages are simple “bridging” connections. However, the map uses these bridging pieces in a shrewd and disorienting manner, via over- and underpasses, slopes, and loops which reverse the direction of progression, diverting expeditions towards unplanned dungeon rooms. This is an underutilised navigation trick, and one which can be used well to draw the company into a danger zone. We can identify one of the dungeon’s main choke points in the bottom middle: a cave with four exits (one a dead end) is one of the main points linking multiple sub-sections. It is easy to reach from what looks like the “main” entrance, and it allows access to many further points of importance. He who rules the choke point rules the dungeon!

Two main structuring elements are also easy to see. These are the level’s waterways and a very large set of caverns. These play a different role. Water blocks or impedes movement, and conceals invisible monsters from the deeps. Hence, the rivers and the lake are barriers. The E and SW river branches separate the map into its southern and northern sections, while the NW branch further subdivides the northern part into two sections, almost as if they were separate sub-levels (I would certainly be tempted to design the key this way). There are multiple places where the rivers are crossed by bridges, and more where it would be fordable to an ambitious group. We can call these environmental challenges limited barriers. The lake is something that would probably be impassable at first, but become a potentially good way to access the rest of the level once the characters return with a canoe, build a raft, and neutralise whatever threat might inhabit the lake. Ironically, the easiest access to the lake (right from the “main entrance”) is bound to be completely useless on the first visit, aside from offering a tantalising glimpse of things to come!

A case of two octagons
The caverns are not barriers: they represent a nexus point. Although similar to choke points, nexus points are relatively open structural elements, which usually offer multiple ways of traversal, and collect multiple routes departing in various directions. You can see another one on the opposite side: the larger octagonal room with its five main exits (the NE one does not realy count as a full one). This is also a piece of dramatic architecture which stands out from the lower-level dungeon texture: it is YUGE, regular, and perfect for a complex set-piece encounter. Of course, sometimes, appearances are misleading: the other octagon just off to the NW does not actually do anything in the context of the map – it is a linear route to a lower level. Also, nexus points may start off as choke points, ruled by a nastier monster or puzzle before being cleared and used to the explorers’ benefit.

Dungeon highways
There is one more interesting feature concerning this map, which may not be noticeable first, but which has a strong bearing on its flow. This is the presence of long corridors linking distant corners of the dungeon, and once you look for them, you will find a bunch. We might call them accelerators (or fast lanes?), since clearing them allows fast travel through the dungeon. The clearest accelerators are found in the distant reaches of the level, including one which just “caps” the whole thing with a sequence connecting everything to everything. Once you get there, you can choose a way back just as freely as you could at the beginning, a dark “mirror image” of the dungeon’s multiple entrances! Others include the corridors bisecting the octagons, and the corridor to the right, going from the entrance areas all the way to the cavern. As the company finds the accelerators, they will prove very valuable in subsequent expeditions, getting them past the entrance areas and into the depths of it!

So, this is what a good dungeon level looks like. It walks the right balance between openness and navigation challenges, it has a good sense of progression, it is structured in fun ways that suggest both exploration puzzles and exploration solutions, and overall, it has a pleasing complexity that takes effort to figure out, but does not descend into unpleasant pixel-bitching and the exploration of dull, featureless mazes.


  1. Topology is the concept you are scrabbling about at great length.

    I despise these easy-to-draw dungeons, endlessly patching together doodles on the same plane, as if architects were children with crayons commanding drudges to make real their drawings taped together.

    It is easy to prove this is not 'strange' or 'alien' architecture but instead ignorant doodles on a blank page because the pretended 'natural' features follow the same ignorant design. The natural 'caverns' and 'tunnels' don't demonstrate even a basic understanding of geology and the effect of water under gravity over time across varying kinds of stone.

    Natural features obey natural laws, and given the effort required to carve out space underground every cubic metre will have purpose.

    Not one of you has a clue.

    1. You're absolutely right, everyone making maps is in utter error. That said... there is a more grievous mistake that may or may not have slipped right under your nose.

      Taking a game that puts physics and natural laws 2nd seat to the utter ridiculousness mechanics that are hit points and armor class, ...that is the height of ignorance.

      I fully endorse that ignorance and you should too.

      People playing these fantasy games should just get stupid, the more intellectually impaired the better... the only caveat, remain self-aware and make sure you are in on the joke.

      Nothing in fantasy RPG's should be taken seriously, EVER. THAT is the CLUE you possibly should be looking for.

  2. Kent wrote: "Topology is the concept you are scrabbling about at great length.

    I despise these easy-to-draw dungeons, endlessly patching together doodles on the same plane, as if architects were children with crayons commanding drudges to make real their drawings taped together.

    It is easy to prove this is not 'strange' or 'alien' architecture but instead ignorant doodles on a blank page because the pretended 'natural' features follow the same ignorant design. The natural 'caverns' and 'tunnels' don't demonstrate even a basic understanding of geology and the effect of water under gravity over time across varying kinds of stone.

    Natural features obey natural laws, and given the effort required to carve out space underground every cubic metre will have purpose.

    Not one of you has a clue."

    These are the scornful warblings of a man who no longer loves Dungeons & Dragons, and deserve no lengthy response. I will just note: don't you dare lecture me on speleology, because I spent much of my childhood crawling around in the caves my dad visited with his spelunker buddies; and got pneumonia in Beke and a bloody head in Kossuth in addition to the scrapes and bruises. I knew what a cave map looked like before I got my hands on the Lord of the Rings, you herculean numpty.

    1. But Melan, haven't you heard how obsessed with geology simultation Gary Gygax was? And what about those experts on sedimentary, igneous, and volcanic rocks - Jaquays, Moldvay, Kuntz, etc?

      In my day, you weren't allowed to play DnD if you hadn't got a CertHE in Geology. The best DMs were those who had themselves been ancient processes of water moulding the rock in the night below, prior to their emergence as elemental avatars who ran Basement Nights at Golden Games on 5th Street.

      You people know nothing about AD&D.

    2. Why are your city maps not as remedial as these pre-teen dungeon patch-works? Is it because you have learned something about cities since you were twelve years old.

      Gygax's dungeon maps are brilliant, D1-3, G2, excellent. They are NOT indifferent unplanned random scribbles. They are designed. Your notion of good design is crappy, no consequence, computer design not real world architecture.

    3. Dear Kent, a genius of your magnitude should have realized by now that most people are just not interested in real world architecture or geology while playing games. If they were, they'd drawing up nifty designs on the drawing board or poring over rocks instead. Your wisdom is wasted on the dullards and heathens here.

    4. Why are your city maps not as remedial as these pre-teen dungeon patch-works? Is it because you have learned something about cities since you were twelve years old.

  3. Re: Naturalism. I feel it's deeply irrelevant. All that matters, is the vibe communicated to the players; that disbelief can be suspended. Designing realistic cave systems, for example, would only please the ego of the DM.

    If you think about it for a moment, everything about the environment explored, is only understood by players on a personal scale: the immediate impressions by the flickering light of a torch or lantern. Even the best made maps are unavoidably filled with inaccuracies due to the scale upon which features are appreciated.

    That's not to say 'don't craft caverns that are geologically accurate'. Please yourself, by all means. But expecting scientific legitimacy to have fundamental meaning beyond what one assigns to it, I think is absurd.

    1. I like caves, I grew up around caves. Most caves suck as dungeons. They are either linear corridors (underground rivers), tiny two-to-five-room affairs, or large but unmappable fractal mazes. Insisting on geological realism in this case is missing the point.

    2. It would also be fair to say that many of the "natural" corridors and rooms were not formed by geological processes. They look more like the work of some subterraneum tunnelling creature, like rats of enormous size.

      Given the proliferation of underground excavations and dungeon complexes, I would hazard that the simple physics of excavation, whether by man or by animal, is different somehow in the D&D world, as compared to ours.

      That said, there are a few real world cave systems that look useful for dungeoneering, e.g. James Brown in West Virginia. But yes, Robber Baron (Texas) is bordering on fractal maze, as is Mark Twain (Missouri).

  4. A good coverage of the topic. Perhaps with the addition that symmetrical maps merit the death penalty because they quickly remove the thrill of discovery from the dungeon crawling process?

    "(As a counter-example, consider the lower level back door to In Search of the Unknown!)"

    Ah lass, the famed exit permitted ever egress, never entrance. I am wracking my brain for a good middle ground but could not delve further then the teleporter in B6, by no means a conventional second entrance but perhaps of some interest to scholars of topological persuasion.

    I have yet to find an excellent treatment of NPC use of geography in dungeons because A) suitably complex dungeons have been out of vogue for a long time and B) Enemy strategy is almost always left to the GM, and rightfully so, with immediate reactions and nearby reinforcements noted but never extending beyond that. A treatment more akin to actual play would have the Orcs (or whatever) set ambushes, attempt to re-occupy choke-points, attempt to draw in pernicious adventurers before flanking them etc, but its difficult to map possibilities or provide guidance that extends beyond the general.

    As a final addition I would ask the hosts's opinion on the use of secret chambers that may be inferred by careful and precise mapping. Pleasingly versimilitudious or enabling of Sonic-obsessed bean-counters and compulsive equipment-trackers?

    1. Symmetry gets not only dull, but has a strong disorienting effect. FPS mapping and 3d navigation always make large, symmetrical places into devious mazes.

      Good point on B1; should have looked that up!
      Good strategic notes are fairly rare. You can always improvise, but a battle plan combining ease-of-use and smart basic defensive/offensive strategies would be an enjoyable pace of change. I mostly play my NPCs as limited in their capabilities by lack of information, habit, or stupidity, but the occasional foe who is a master of his environment is a very scary opposition! Of all classic modules, thespian-suspect Ravenloft actually has decent tactics outlined for Count Strahd! Rappan Athuk and Tomb of Abysthor had some decent stuff, as well.

      WRT secret chambers, it really depends on context. I try to build them in a way that makes them possible to be found by paying attention to the environment, without undue pixel bitching - which is the worst. Obviously, man-made buildings make this easier to accomplish than, say, caverns.

    2. I developed a fondness for Doom Mod Playthroughs and Romero was without question a DnD player. The difference between early FPS maps and the linear drivel that came to infest the genre is heartwrenching.

      Thank you for the recommendation. It seems I am doomed to cover all of DnD. I shall add it to the Book of Infinite Reviews.

      Speaking of which, I don't want to enable Kent since he will berate but refuses to illustrate, ever the weird kind in the corner at a prom. 'I could dance much better then that if I wanted too' mutters he, clasping his bottle of blackbush like a shield.

      That being said, Kent brings up an intriguing point. A dungeon must make some sort of internal sense so another layer is added. Storerooms, Barracks, Cisterns, Temple etc. etc. The question becomes, can you have a good, jaqueyed large map without it violating this versimilitude principle, or are they utterly unrelated?

    3. I think these are unrelated concerns. In fact, a dungeon can also be both "realistic" (in the context of a fantastic world of intrigue and adventure) and "fantastic" (even in the context of a fantastic world of intrigue and adventure) at the same time, since these concepts are only opposites for the imbecilic.

      And that shall be my Kentian wisdom of the day.

  5. I think that an important feature of dungeons--I mean ones actually used in play--should be their susceptibility to oral explanation and description to the group of players at hand.

    Most Referees can't clearly explain scenery and its dimensions and most PCs are not surveyors. Much of the players' experience is where their minds fill in the gaps between what the Referee says and the choices they must make.

    If any player wants to make a map, then the Referee, who adopted a certain set of odd dimensions and irregularity in the dungeon at hand, should be able to describe those spaces clearly enough. Most of the lovely and free maps that I find online are not suited for spontaneous description for those who can't see it. The Referee's explanatory skills, and the limits on them, should be a factor in dungeon design, unless players don't mind being utterly lost. If they are lost, then many of the clever considerations in dungeon design, catering to "skillful" players, are not useful. This includes mastery of "realistic" architecture and cave formation. If the Referee can't describe it so that the players know how to make choices, then it doesn't matter how cool the map looks behind the DM's screen.

    I'm reminded of the Terrain Features post by Detect Magic.

    If your players don't now the difference between a gorge, a ravine, and a gully, it doesn't matter if the Referee knows the difference or uses those words. The same goes for different kinds of cave systems and architecture.

    Melan, your prescriptions for dungeon design make sense for story purposes. If they are followed, then a dungeon should not originate as a pastime with graph paper and wandering pencil, but should begin with subway-map flow design intended for a satisfying experience of exploration and risk. If that is so, then I wonder why anybody needs the beautiful maps DMs crave today, unless for inspiration. A subway sort of map will suffice. Such maps do suffice for my own humble home games, and they may likewise for published games, too.

    The maps we use today are still on the model of those designed by wargamers who wanted miniatures on the board. They are often not suitable for the kinds of adventure stories that the hobby has evolved to tell.

    I think your schematic flow designs are more effective than the beautiful maps by the charting savants of these internet years.

    1. Once again we find different people with incentives that run at cross-purposes, or are at least orthogonal. Nice professional-looking dungeon maps with cross-hatching and everything get views, clicks & sales. Better-looking art gets more attention.

      I like a nice map myself, but nobody needs to see the contours of the fucking cobblestones while trying to run a game.

    2. I would not divorce aesthetics from function, even if this post is about the latter. Much of what we do is for aesthetic purposes; two statistically identical monsters can be described so differently that they will seem to be complete opposites, and two functionally identical dungeon sections can result in a very different experience through description.

      That said, maps which can be clearly described have the advantage of encouraging and facilitating player mapping, which complicated ones (including some of mine) don't. This is an important gameplay choice with tremendous impact! (There is also the option of VTT with a fog of war option, which can be surprisingly effective in simulating in-game vision - but that's another subject.)

      Finally, let us not mistake the model for the thing being modelled. Models deliberately simplify reality in order to make a point. Yet, as every first-year geography student is taught, "the map is not the territory!"

    3. Melan, it seems we agree on everything. Just as statistically identical monsters can feel different in play, thanks to the narration of the referee, so two twenty-by-twenty rooms will be completely different, for the same reason. The map is not territory, and the players' experience of territory is in the communication between referee and players, and where the players' minds fill the gaps. I like pretty maps as much as anybody else, of course, but when it comes to designing "dungeons" that facilitate player exploration and choice, I think starting from a flowchart like what you describe rather than randomly drawn (or randomly rolled) maps is more likely to result in a "good dungeon" from the players' point of view. In other words, I agree with you, and go further, perhaps, in saying that a "good dungeon" is more likely to result from flow design, not doodling rooms and passages.

  6. Fortunately for you, all with Echoes and Xyntillan, this article won't forevermore cement you as "the guy who wrote that dungeon layout article" :P

    1. I also wrote that rant about the tyranny of fun, which people are apparently still salty about 14 years later. That has to count.

  7. In general, the article explains well how a well-designed dungeon works.

    Unfortunately, I never had luck with actually running huge dungeons like dyson's above. What seems to be the sweet spot for me is about 15-20 keyed areas, preferably one level, but with multiple entrances.
    I have about 2-3 hours to prepare for each game night, so drawing dungeons is usually an activity I have to skip. Most of the time I use flowcharts, or even "d12 interesting loactions ins an otherwise unmappable locale".

    1. Dave's Mapper is surprisingly good at the floorplan stage. And you can stock the things in a rapid-fire way. That said, I respect individual tastes, even if I believe that some advantages of dungeons really emerge over a certain size, a bit like how a city is not just a very large village, but something functionally different.

    2. It's not really individual taste. While I very rarely use big dungeons, or 30+ page adventures, I admit they are usually more rewarding than the scenarios I can come up with.
      The real question is: should I make something up in 2-3 hours for next weeks game, or should I start digesting some bigger premade material and have perhaps a monthly game?

      Just as a general inquiry, as I know you both run and design more complex dungeons than me: how much time do you spend preparing for each session? Can you fit into 2 or 3 hours per week?
      I am really puzzled how others do it. My experience is that when I buy a premade adventure/dungeon, reading, understanding, and fixing (!) it usually takes more time than what I have. Sometimes I do it though, not because it saves some time for me (it doesn't), but because it improves my game with ideas. Provided it has cool ideas in the first place.

      (there are exceptions though. A few bigger modules which I used often in the past, and I can run it without preparation. Death on the Reik and Sinister secret... comes to mind)

    3. It is more than 2-3 hours, although usually not *much* more than that. My only secrets are practice, and picking a time when I am particularly inspired. However, there is some extra background work. I tend to carry around a notepad to jot down basic ideas like:
      "Doriano the Wizard - does business in the name of the cosmic Balance"
      "Filberto the last thief - he was not hanged due to the coming dooms"
      "The Cult of the Bat, idol."
      "Lumps the Innkeeper, private room of card-playing veterans."
      "Sir Rapax, stone knight and Niabex, Faerie of the Mountains"
      "The hungry pit wants food in exchange for dungeon key."
      "The Dark Rangers, robber spies, walk the forest paths."
      "Flandevole, head of the winesellers - guarded cellars."

      When I do the actual preparation, I build fairly quickly from these idea fragments by tying them together into a common framework, whether it is a situation or a map. There are some pieces which get filtered out (as it happened to some things from the ideas above), some which fit neatly into a concept, some which are changed, and some which are left half-developed, and stay odd during play.

    4. Galu: my experience with premade dungeons/adventures of medium or big size is that the initial preparation takes a lot of time. Once you are confident with it and kick the campaign off, it needs much less preparation between sessions (although it will inspire you to expand it and build upon it, so you'll put in a lot of work anyways :D).

  8. Brilliant post Melan. I will be re-reading it many times. I love the "accelerators" and "choke points" notions.


  9. Melan is analysing a twelve year old gamer's notion of a dungeon. Most so called AD&D gamers, IME, tend to resent the natural growth of knowledge and maturity of normal folk when brought to the game table. They want to re-experience ad nauseam that first twelve year old high.

    Gygax's and Jaquays' maps are well designed and mock Melan's faux-mathematical 'insight'.

    (I do not deny that Melan knows far more than all of you other morons)

    1. Kent, what's the point of your nauseating blabber here? Believe it or not, everybody got the oh-so mature and sophisticated message (i.e. all the world's a moron, except you)a long-long time ago, so you might as well STFU. Nothing of value would be lost.

  10. I thought this post was really well done so I added a link to it in my BEST READS OF THE WEEK! You can check it out at this link if you want:

  11. Thanks for your useful blog post on Dungeon maps, Melan.

  12. I discovered this in Knock #2 and was impressed with how easily you conveyed advanced dungeon design concepts to an amateur such as myself. Using Dyson's map as a prop really helped get your points across. Thanks!