Zelda Temples and "Realistic" Dungeon Design

I had a thought recently about dungeons that was revived today by a Zelda Water Temple discussion on the NSR discord:

I believe everyone agrees that The Legend of Zelda series is the gold standard for dungeon design in video games. Even titles that leave a bit to desire in that aspect (I’m looking at you Breath of the Wild) are still above average, while the best dungeons in the franchise have no equal in any other game.
In my opinion part of that greatness comes from the fact that those dungeons feel real.

A lot of dungeon/puzzle design in videogames (and TTRPGs) is that they don’t make sense outside of the fact that they are part of a game. They aren’t logic as a security system, neither as a challenge to a possible wielder of whatever thing that place houses. They are only throw there to pose a interesting challenge to the player.
In Legend of Zelda those places just make sense, the challenges of a temple feel like a trial to forge the hero that will save the world, the traps and puzzles in ruins make sense as a security system that keep outsiders out, but open up easily to those who know the secrets.

Now, to the part that matters more to us, how to do that in our games? We delve a lot of dungeons at the table, and more than often they have the same problem of just feeling like an artificial barrier between us and the loot.

What are good examples of dungeons that feel realistic, and how to achieve that result?


Mmm interesting premise from the start. Thanks for making the thread.

I believe everyone agrees that The Legend of Zelda series is the gold standard for dungeon design in video games.

For a VERY specific game-type, sure. Like uhhh The Minish Cap dungeons are great puzzleboxes with fun combat and a ton of really interesting unfolding concepts that are taught to the player in good timing. That’s an excellent dungeon design. But also Immersive Sims want dungeons that are so much less “solved puzzles”. The Stone Tower Temple from OoT is phenomenal, but giving that to someone who put all their points into sneak and poisons, then asking them to solve a whole dungeon with Proficiency (Ocarina) and a level of illusion spellcasting wouldn’t feel as satisfying as it does in Link’s grand adventure. Not to say that your premise is wrong! They’re baller dungeons, for that specific type of play. The reason Breath of the Wild’s shrines all work as good dungeon set pieces because your Link and my Link are essentially the same character. But in a game with broader character roles, players will find points of bounce.

Now, to the part that matters more to us, how to do that in our games?

My boy Dan leads a top ten over here. Level design is an art, not a science unfortunately. I think there’s a lot of really good actionable advice, but I’m not sure it’s a solved problem.

Overall, though, I wouldn’t count out just how much of the Dungeon/Puzzle feeling at home in the game is you buying into the Magic Circle of that play experience. And that’s where I’d always start: What’s your player buy-in, what’s your themes and story. It’s like how the idea of Pokemon gyms being all one type is ridiculous because that makes them intensely vulnerable to someone just rocking up with the opposite element (“Hi Misty. Here is my team of 6 Bellsprouts”), but the themes of pokemon create the Magic Circle of progression that means that as a player you’re just trying to ride that stress curve of winning without being overleveled to boredom. It’s why Portal can just be a series of puzzle rooms: because Glados sells the narrative so well that testing is all there is. It’s why Control is allowed to have the Ashtray Maze - People, Place, Plot, Purpose.

This is also why small conceptual dungeons are good starts. Ya boi the deku treebeard is a good first dungeon because structurally it’s a tree. It’s the easiest to enter that magic circle with. By the time you’re flipping gravity or floating on invisible platforms, you’re so invested that you’ve bought in, you’ve crossed within the magic circle.


I just got @QuestingBeast’s “Alchemist’s Repose”, which has a wonderful reprogrammable automaton guardians mechanic. Go check it out!

Also, pulling in @deldon into this convo, he might have some good ideas? (no pressure :laughing:)


I don’t ever aim for “realistic” - actual real ruins and underground edifices rarely have the complexity and maze like aspects useful for dungeon design. Or at least procedural, exploration based design.

This for example is the Interior of the Great Pyramid:

It contains what I’d count as about 4 keyed areas, and all are on scales that don’t really work well for RPGs. The King’s Chamber there is about 30’ x 20’ and it requires a massive set of granite beams and supports (the tower looking thing above it) to avoid collapse. Likewise, the protections of the Great Pyramid (which didn’t even work for a few decades - it’s documented as having been cracked open by ancient historians) primarily consisted of walls of rock, and blocks that were dropped into place closing off passages. Needless to say, it doesn’t make for a very good RPG dungeon…

Alternatively one can look at more contemporary underground constructions - World War II had its share of them, including this underground airplane factory. It’s a complete RPG scenario, built by an evil empire with slave labor to create doomsday weapons on a massive scale - it is surely cursed and haunted. It’s scale is enormous - each cross tunnel is around 500 feet long.


The main issue though is that the map is extremely symmetrical (it’s meant to function as a factory, not confound adventurers).

Honestly your best bet for real world dungeons are the Derinkuyu underground city or maybe the Wieliczka Salt Mine.

The point I’m trying to make is that the game dungeon, be it Zelda’s excellent computer action rpg design, or something like the Caves of Chaos, is a fantastical space. It’s meant to make sense only within the context of the game and its mechanics. It’s “built” with the goal of making navigating it interesting and fun within the context of the rules it’s designed to be run with. Feeling Real is a concern and a goal to strive for, and “Gygaxian Naturalism” does work - for an exploration based game your dungeon should have an internal logic that players can unravel. The Throne Room should be near the Guard Post or Receiving Hall, not behind the Waste Pit and dungeon inhabitants should find themselves defensible lairs with access to the things they need (water? food sources? exits?), but the realism always takes a back seat to functionality.

This of course can mean that if you aren’t designing for a game with exploration mechanics (supplies, turn-keeping, random encounters, encumbrance etc) your dungeon doesn’t need to look the same - a flowchart or a few rooms … heck the Great Pyramid - despite its difficult to translate into play verticality might work for something more scene based.


I agree with what @SidIcarus has said, that Zelda dungeons are the gold standard of Zelda-like dungeons and that doesn’t necessarily apply to other games or subgenres. I also agree with what @GusL has already said re: Realism vs. “Feels Realism”; Zelda dungeons don’t usually contain rooms included only to sell the player on the reality of the game world.
You point out two very good dungeon design themes, and I agree that Zelda dungeons do a lot of things really well, so what I’m trying to get at is what exactly are you looking for when you ask for “good examples of dungeons that feel realistic”?

Regardless, I do have some thoughts based on what I believe you’re asking for.

Zelda dungeons carve out a unique niche in dungeon design that I do feel is largely unexplored by TTRPGs. There are a handful of extant examples (@directsun’s Seer’s Sanctum comes to mind, though I can’t exactly remember many more) but for the most part the focus is either on the classic dungeon-as-mythic-underworld or dungeon-as-resource-management-challenge. Most video game dungeons don’t really fit into this mold since they are usually designed as levels intended to be beaten, with a beginning, middle, and end much like a story (Most WoW dungeons fall into the category). TTRPG dungeons can certainly be designed around telling similar sub-stories (in fact I would argue that the intent of WotC’s D&D’s dungeon design advice has been to popularize this as the default) but for the most part dungeons are designed as interesting places containing isolated challenges and infinite potential challenges.

The “secret sauce” that keeps gamers pining for more Zelda-like dungeon design in their video games is that while most dungeons are designed as levels to be beaten, nearly all Zelda dungeons are designed as puzzles to be solved. Getting to the end of the dungeon is the reward, and beating the Boss Monster is more of a cherry on top.

If we want to design more dungeons that fit into this niche, then we need to embrace some new principles of level design specifically for this niche.

  • Dungeons should include some paths that are blocked until certain parameters are met elsewhere in the dungeon.
  • Multiple paths to those parameters should be included, but at least some sections of the dungeon should act as a choking point, so that the parameters cannot be bypassed.
  • Dungeons should have a clearly defined “end state” that signals to the players when they have overcome its primary challenge, and the players should have a good idea of where this is before they are able to reach it.
  • Navigation and orientation within the dungeon should be one of the challenges.

Like I wrote above there is a lot of good design that contributes to Zelda dungeons’ level of polish, but I think that these points help restructure how to think about level design in order to achieve a similar effect.


@deldon touches on some very good points in getting that Zelda feel.

The biggest problems I see with the majority of ttpg adventure puzzles is what you’ve said @Xenio, they just feel thrown in there and don’t make sense in context.

The best puzzles, imo, are ones that aren’t looking for an arbitrary answer to solve. Instead, their solution comes from understanding the logic of the dungeon or world itself.

Mark Brown from GMT has some fantastic videos where he breaks down puzzle techniques for video games.

One of the tings he talks about that helps in designing rewarding puzzles, is teaching the player. Let them understand mechanics of the world piece by piece.

In our case, slow feed the players information so that they can work out for themselves the logic of the dungeon. Let them solve puzzles using that logic and then build on it adding slightly more complex puzzles or challenge the players in a way that then challenges their assumptions and forces them to rethink and expand their reasoning about the logic of the dungeon and the world.

In Puzzle Dungeon: The Seers Sanctum, most puzzles involve specifically looking at something. The players can begin experimenting with this concept in some of the rooms.

Questions and answers they ask and learn as they explore and experiment and move on to more challenging puzzles:

  1. Does everyone have to be looking at the thing for an affect?
  2. Does not looking at something have an affect?
  3. Do other things count as ‘looking’?

These questions naturally get asked as the players learn more. The players then attempt to test their hypothesis and once they think they know how it works, attempt to figure out how to get the mechanics of the dungeon working for them.

These are my favorite puzzles. The ones that encourage experimentation and understanding of the mechanics around you. It’s one of the biggest things that lend to that “Zelda dungeon feel.”


To the question of how we make these dungeons feel realistic I couldn’t really say. Typically, I like to tie in a specific narrative thing that tells the players “hey, this is why this looks this way”. For instance a wizard who obsessed with chess having puzzles in their dungeon based around specific types of movement depending who you are. This doesn’t really make much sense without the context of what the wizard is into. Or, a goblin warren that is being tunneled into by a massive burrowing worm, causing parts of it to collapse and new tunnels to open up. The party can discover the worm is searching for something, like a stolen item, causing its tunnels to lead toward this treasure. Without this giant creature this doesn’t make any sense, but the fact that its after something creates a flow to the dungeon.

A thing about Zelda dungeons that I love and think should be explored more in the TTRPG space is the “Hub and Spoke” system. The idea that you have a central entry chamber with branching paths, some of which are blocked off, but by exploring others you can find keys or switches that remove those blocks. It seems like this would work better in an RPG than a video game since having to spend the time to backtrack to the central location in a video game can be tedious, whereas in a TTRPG the players can just…say they go back (assuming there aren’t any traps or enemy’s in the way). Now all the time and effort is spent where it should be spent, in solving the puzzle.

Finally, check out this wonderful video I found that breaks down a lot of the core mechanics in Zelda dungeon design: How Zelda's Puzzle Box Dungeons Work - YouTube


This is a really good topic. There is a kind of modularity to some Zelda dungeons that’s interesting, in the sense that Cadence of Hyrule can procedurally generate maps that a.) each have fairly unique layout and b.) still capture the feel of dungeons from 16-bit Link to the Past.

I’m also reminded of the shrine dungeons in Breath of the Wild, where each iteration re-uses elements from the previous shrines, while adding a new piece or theme to keep things fresh. And they’re all more or less bite sized … which makes me think that one could develop a Zelda-style megadungeon, where you unlock mini-shrines that are all connected to one another … :wink: