Semicontinuity (A framework of analyzing videogame space)

I found myself thinking about the Dark Souls trilogy, again. Maybe one day I won’t have to think about them again… here is my attempt at exorcising them. Okay. I’ll refer to them as 1, 2, and 3 here.

So the other day I wondered some things. In particular:

  • I had meticulously (and wastefully) platinumed 3. Then why do I have little impression or sense of wonder when recalling it?
  • Why does 2 stand out as the most fascinating and exciting in my mind, even if I only played through it once?
  • Why despite 1’s first half being so strong, is it still not as memorable as 2?

I skimmed a speedrun of 3. It’s easy to see why it’s less memorable than 1 or 2. It has a fairly standard and safe art direction (undead decrepit stuff, dreary boring high fantasy). Environments out of that genre that don’t mystify me (sprawling castles, medieval villages in visually dense cliffs, swampy woods, European halls, catacombs…).

I can recall many of the spaces and levels in detail, but I don’t have the tinge of awe and jealousy I tend to get when being impressed.

Drawn in contrast to 2, some other reasons surface. 3 is far more world-design continuous (here on just continuous) than 2. 

See the Appendix at the bottom for definitions of the terms I’ll use in this post – hyper-, normal-, semi-, and discontinuous. I’ll assume you sort of understand what I mean by these terms going forward:

Okay, so Dark Souls 3 – or just 3 – the game is split into three chunks – Castle, Forest, and Snow Castle. There is a break between Castle and Forest (where you fly, conveying a sense of descent into the world), and a break between Forest and Snow Castle that’s articulated by a level (catacombs, conveying a sense horizontal-plane traversal into the world). These chunks are nearly hypercontinuous within themselves – while they’re not open world, there are handfuls of mutual vantage points, things are detailed and realistically sized. (3’s three chunks are circled here:)

Dark Souls 1 is roughly similar to Dark Souls 3 in continuity, but it takes place in a vertically oriented world. We could pick apart definitions as to what extent these two are continuous, but my focus isn’t on those games. Still, this glass model by Twitter use @rigmarole111 captures 1’s world well: Notice that the game has four strata: the top-level areas – Anor Londo, Duke’s Archive, mid-level areas (Sen’s, Parish, Burg, Garden), lower-level areas (Blighttown), and subterranean areas (not pictured). 

People love 1’s first half because you’re exploring the middle and lower-level areas, totally on your own, shrouded in mystery, and the game is good at helping you create connections between the world (as you play for the first time the structure is not at all obvious). The goal of the game is initially to reach Anor Londo, but you’ve got to explore all over before you get there, and many closed off/difficult paths will tempt you.

I would say 1 gestures at hyper-continuity at times, but actually some of its major mutual vantage points are faked (to good effect), and I would classify it as actually closer to normal-continuous. That is to say, none of its connections are confusing or baffling. If we are walking down, we’re going to the underground, and reach fire-y lava areas. If we go up, we go towards grand, royalty castle areas. It’s standard videogame world stuff (see any castlevania game), executed well in 3D.

Still… it’s not as inspiring as 2 to me. Why? By most popular accounts, 2 is a ‘mess’. The areas don’t make any sense, they connect randomly, there’s no ‘blog-worthy shortcuts!’.

Here’s a world map of Dark Souls 2, where I’ve highlighted the game’s paths.

A few notes:

  • Everything spirals out from Majula, a seaside village.
  • Game Structure:
    • Collect 4 Important Things at the end of 4 paths, located in: Black Gulch (deep dark area), Iron Keep (fire castle in mountains), Brightstone cove (church/caves near a settlement near the ocean), and Sinner’s Rise (moonlit fortress near the sea)
    • Visit Drangleic Castle, go deep deep down all the way to Undead Crypt
    • Go way way up to Dragon Shrine
    • End the game at Throne of Want (in the castle)
  • There’s basically no crossover between main paths and there are barely any mutual vantage points

2 varies from 1 because 2’s world is representing an entire continent. There are little to no mutual vantage points because as you walk from one area to another – even without loading screens – you might be implied to travel miles. The player is going on a world-spanning adventure, much like a JRPG, just without the game giving us the abstraction of a world map to make that feel more coherent. We walk through a mysterious tunnel from Majula to Heide’s tower, and emerge 10 miles away. A quick trek through some woods and a small cave system shoots us out at a shore all the way at the other end of the world. An elevator at the top of a poisonous windmill takes us to a fire-y lava keep. And so on and so forth.

Because 2 is missing a world-map-esque abstraction layer, the game world inevitably connects in some confusing ways. Walking 100 meters can actually imply travelling all sorts of distances. Thus 2 has inconsistencies to how you travel between spaces, and it approaches being discontinuous, and thus becomes a semi-continuous game at times.

Back to Dark Souls 1: while playing, you are oriented within the world. Even as you uncover new places, you’re still aware, roughly, where you are vertically within the game’s world, and if you’re not aware (perhaps as you go to blighttown for the first time), a mutual vantage point (Great Swamp to Undead Burg) comes along and situates you within the world. Or if you wandered into Demon Ruins early and are disoriented, you’ll still become oriented once you see Demon Ruins from Tomb of the Giants.

As a player explores a game’s world, tension slowly builds as they sense they’ve drifted far from home, far from the familiar. And within this tension is where lack of continuity can be used to achieve various effects. Connect the player back to a familiar place and let them feel grounded. Or lead them further and further from the familiar, connecting areas in surprising ways.

2 doesn’t do anything to situate you within the island unless you go and look it up. It’s just one area after another, and the sense of endlessly going deeper into some fever dream of a continent. You just pick a random direction from Majula and keep going and going and going, and it feels like you may never reach any end point. You just know you’re either “sort of far” from Majula or “really far” from Majula.

Eventually, you do hit the end of a path, and the magic wears off, and you get one of the 4 magic items and go on to find the next.

This kind of semicontinuity is fascinating to me. It happens when a game establishes a consistent physical logic but then chooses to break its rules at certain times, or to at least not be consistent in one method of arranging the game’s world. Dark Souls 2 creates fairly continuous areas such as Forest of Fallen Giants, but at the same time does absurd things like having a well take you to a rat-infested catacomb, which THEN takes you to a poorly-lit cave with shambles of wooden housing, and finally some eerie, green cave. Nothing in the game really explains how this connects spatially to the rest of the game (unlike 1’s Lost Izalith, which you can see from elsewhere in the game). Instead, 2 is content to shove you from one area to the next. 2 feels like the randomness of some old videogames’ worlds, but expressed through a detailed 3D world with no loading screens. It’s full of interesting and inconsistent ideas, and like some Twitter mutuals put it – feels like playing a romhack of Dark Souls.

2 is the most memorable for these reasons. You take an elevator down from the top of a castle, and for some reason it shoots you out into a sprawling, blue, underground lake:

And if that wasn’t enough, there’s another elevator which takes you to an underground crypt.

In another section, a mansion sits off the side of a road: you then take the world’s tallest elevator up to a floating set of islands and dragon temples. Now, these strange transitions extremely memorable just because of how bizarre they are, peppered in between fairly reasonable transitions (e.g. walking through misty woods, to forested pathways, to some shady ruins). These moments are strange because the game has lots of non-strange transitions (like the interior of Drangleic Castle), and the game plays them both off as normal.

What semicontinuity gets at for me is the exciting ways we can represent our game worlds as designers. There’s no need to adhere to any one formula or idea: the ways areas are connected can be carefully thought through and become expressive.

It’s something we thought about while making Anodyne 2: Return to Dust (Spoilers ahead.. if you quit Anodyne 2 after 1 hour this is a good time to go finish it!!).

We initially establish a videogame-y logic to shrinking into people located in the 3D world, putting the game in the vicinity of normal-continuous. But as the game goes on, it’s revealed that the linkage of “3D character” and “2D world” is not always paired. Iwasaki Antimon takes you to a mailbox of a person Iwasaki obsesses over. The fur of a dog-creature ends up being a village you reside in for weeks. The desertnpc takes you to something completely unexpected. Pastel Horizon, Minorma’s Orb and New Theeland connect through an alternate Anodyne 1’s Nexus. The notions of space and how they should connect shift from a straightforward videogame-y premise to something disorienting and discontinuous. Anodyne 2’s continuity shifts as you get further… and it uses this to suggest a certain location in 2D as further away.

These sorts of formalized techniques can be used hand-in-hand with the story themes to make certain ideas resonate stronger with the player. For example, putting Dustbound Village inside a nano area makes you wonder about the rest of the 2D worlds you’ve visited – and gets you to rethink what you’ve learned about your role as a Nano Cleaner in Anodyne 2’s world. The strange large areas in the Outer Sands break down an attempt at a ‘logical’ understanding of Anodyne 2’s world’s fabric and suggests something that exists on a more fantastical layer. Thinking about and using continuity in different ways can create different headspaces for a player to inhabit as they read what you’ve written or prepared for them to play through.

Anyways, all degrees of continuity can result in interesting games. On the hypercontinuous end, we could have an exploration of a single bedroom, not necessarily sprawling and repetitive corpgames. Or an exploration of a single town (like in Attack of the Friday Monsters). There’s no right approach to making games or picking the level of continuity. But if you’re designing games it’s good to find some approach that you care about.

Appendix on defining continuous/etc

As always, game design theory is… sort of made up, right? These ideas may bear fruit when analyzing certain games, they may be worthless for others.

I’ll define continuity as a game’s tendency to have unbroken, realistically-sized spaces that you travel through without many cuts. On one end of the spectrum, games can be hypercontinuous (painful accuracy to real life – exteriors of buildings always match the interiors, the camera never cuts, there is no fast travel). I don’t know of many large games that are hypercontinuous, but the majority of recent open-world corporate games (corpgames) are close to this (Breath of the Wild, God of War, Grand Theft Auto).

Or games can be discontinuous – levels connect with seemingly no underlying logic (Yume Nikki, Goblet Grotto, Chameleon Kid).

But most games tend to lay in the middle – normal-continuous. This is when a game employs the standard set of breaks in its spatial traversal that we expect: a JRPG brings you from a world map (abstract) to various towns and dungeons. We traverse the world both concretely (through dungeons and towns) but also abstractly, through a world map. However, to the player, it’s always understood where you are in this world and the game generally intends its world to be remembered as a continuous space, even if it may change the level of abstraction from time to time. Normal-continuous represents what today’s commonly accepted practices are in terms of how a game articulates its transitions from one area to another.


  • Characterized by a game whose spaces have an experience of moving about them that is very close to real life.
  • Likely to not have restricted levels (e.g. you can walk anywhere)
  • Exists at big and small game scales: From “See that mountain? You can visit it!” to “spend a day sitting in an apartment.”
  • Usually 3D, just because 2D more or less is always an abstraction from “realistic” representations of 3D space


  • Most games. E.g. JRPGs with world maps, visual novels with different areas
  • Spaces may not be ‘realistic’ in representation or connection
  • a player can make sense of where they are in the world most of the time, because areas are connected with consistent logic and the game makes an effort to let the player situate themselves (through a world map, etc.)
  • travelling between areas is consistently represented and has consistent outcomes. E.g. when a game has 3D houses you can enter, if you enter the door and then always enter an interior area, the game is following its own rule that “Going into houses leads to an interior space”. A continuous game is unlikely to violate this rule. Other rules could be – Doors take you to new areas, doors take you to the same area if you re-use them, etc. A discontinious approach would be that sometimes a house’s door takes you to a beach or something


  • Either a mixture of discontinous and normal/hyper-continuous approaches, intentionally (or not ) breaking ‘good practices’ and breaking the sense of logic to its game’s world


If a game never situates its player in a familiar place, but shows one surprising locale after another, it enters the realm of discontinuous, like dream-exploration Yume Nikki, labyrinthian platformer Chameleon Kid, semi-procedurally-generated 0n0w (  or  spiraling 2D/3D maze 10 Beautiful Postcards ( , or various early PC games that didn’t seem to ‘care’ about player’s navigation or “good practices”. Games that use this approach can explore unique feelings, sometimes terrifying. Digital space within discontinuous games can be implied to be infinite and unknowable, and that’s a very interesting situation to place a player into.

In choosing not to embody a place and be hypercontinuous, a discontinuous game begins to embody an abstract sense of feelings depending on the game, but also the game opens up more space for a player to relate *their* spatial memories with that of the games.

Anodyne 1 is somewhat discontinuous, by way of Yume Nikki’s influence. And I think it’s a big reason why many people remember its world and take away different things – in a discontinuous game world, a player is left to form many of their own associations based on what images and text the designer provides.

As a game changes its digital space from more realistically connected to more discontinous, the world may feel unknowable, a fever dream of unrelated parts and images that seem to go on endlessly, a fearsome nightmare or dream, all contained within a few bits on a computer’s hard drive…

(If you liked this post… I’m on Twitter! . I also run a discord for my game studio Analgesic Productions at

Or… join our studio’s newsletter and check out our games at!

Oh no! Sparseness in 3D games

Lately just thinking about – the transition of 2D to 3D. What happens with that.



Something I’ve thought about as I’m a few weeks from being done with Anodyne 2 is that it’s easier to create an explorable space that can capture your interest 3D, there’s just a sparseness problem to it all.

That 2D Harvest Moon image above looks quite empty, but it still feels like a visual whole. The noisiness of the tilework creates a coherence, a cozy sense of ‘farm’. something about the 3D in the upcoming friends of mineral town remake – looks bizarrely empty. Now I’ve seen the trailer and some shots of it (like the towns) still look nice. But just when it comes to one-off areas in 3D games, they’re so so expensive.I think 3D presents a unique challenge when adapting 2D. Because 2D games really “work” by reducing a lot into a flat plane, and when you unflatten things, you get all this bizarre empty space you have to account for. like see the harvest moon. you can just pixel art a mailbox and wood dividers and stuff, but for 3D boom you have to model and texture all that crap. and most of it is just colliders.

Some games don’t account for the emptyness and you get weird just like… empty places.

How do you account for empty space? Well, to me it seems like:

  • throw money at it, which basically means throwing PEOPLE at it, which means uh oh! now you have ridiculous commercial expectations so this doens’t seem to be a great way to go about it
  • throw time at it (“my 20-year love letter to shadow of the colossus! oops now the universe has heat deathed”)
  • ignore the problem and hope it goes away

REAL FARM (thanks gamefly)


TIME AND ETERNITY (thanksgamestop)


now i’m not saying that an empty feeling game is a problem, but moreso that i think that if a game feels really empty, the ‘spatial feel’ is a dimension of a 3D game you can use to your advantage, so it’s better to figure a way out than just well, not doing anything. Not to mention that having a huge 3d space to fill is anxiety inducing and spending time just throwing assets at it doesn’t really make the game better… idk.. 10 art assets in a 3x3x3 meter room can be more expressive than 10 art assets in a football field, is what i’m saying

so the last option is

* do something about it.

don’t really feel like saying much, but i think shadow of the colossus MUST have seen a similar problem. they went ‘Oh shit, there is so much fucking space to fill in 3D! fuck. we are fucked. wait what if empty space was the point. haha’

As much as banjo kazooie is equivalent to intentionally forgetting your wallet at home after driving to the grocery store, you have to give credit to these early 3D games for being good about space management. Things are kept relatively small and dense. say what you want about good old BK but those levels did have a sense of place and personality, even if it was well, banjo kazooie.

but back to anodyne 2, because I enjoy talking about ourselves. so if you pick up the game next month, there are these interior areas. i call these ‘mistakes’ JUST KIDDING. they’re wonderful but they were EXPENSIVE to make, in terms of time. when you get to Cenote city, that place was.. expensive. marina had to fill it with too many buildings. we then found that SURPRISE outdoors naturey areas that are sparse and hilly are easier to make! hm. should have done that more often. but at the same time, nearing the end of development, I am lagging behind so maybe that giant city bought me time to do all the random programming stuff I needed to do. but that’s a different issue – coding uniqueness and time… sigh, we are still learning so much about how to be careful about time…

anyways, in anodyne 2by modeling a few rocks and ground textures and using a general environmental shape concept, we made pretty neat 3D areas that are big-feeling but still feel like that sparseness is intentional rather than a big ‘oops haha empty’. sometimes it’s a matter of slight terrain variance to break up flat ground perceptions. othertimes visually ‘messy/dense’ textures (like anodyne 2 has a lot) help to create denseness. idk. there are a lot of tricks. Some of the outdoors nature 3D areas in Anodyne 2 could have gone faster if we had better tools, but then we’d need a tools programmer..


Here’s an area we took out of the game. Well, not really – we used many of these textures in the game in a similar area, but this picture specifically does not show up in the game. There are a few ground textures, grass texture, then a single rock 3d model copy pasted a bunch. The structure on the right was expensive to make because we didn’t even use it in the final game (except the red huts). But it was a learning experience for 3D asset making I guess.

also the other way we dealt with it was moving gameplay heavy stuff into 2D, so you don’t spend all the time in 3D, but that also has various production issues that blah don’t feel like talkign about.

I made the mistake a few times over Anodyne 2 dev of creating these small, one-off areas in 3D that are just entrypoints into 2D levels. I think they were wastes of time. Memorable, maybe. But I’ve been playing Yoshi’s Crafted World, and every single level is this disconnected singularity of 3D art assets that has no connection to later levels. Like a visual explosion.

At least in Anodyne 2’s case, all our areas are connected so we have the coherence benefit of those one-offs still fitting into a big picture. The next time we make a game we can be more careful about these things. In the case of AAA extravagance like Yoshi, you just have a gigantic grab bag of these little party poppers of levels… which look traditionally BEAUTIFUL but there game is just boom boom boom visual explosions so I can’t remember much actually. plus it’s a simple 2D platformer so i don’t really remember the spaces much


It’s funny, another game I’ve been playing is the opposite of Yoshi – 10 beautiful postcards by thecatamites. In some ways there’s a parallel I see between Yoshi and this game in that from a visual standpoint, both are games where you are constantly whiplashed between visually distinct and dense areas. In Postcards you’re travelling in an almost endless maze of colorful areas, but they have thematic overlap. In Yoshi you’re just going into one random mechanic-fest after another. it shows just how much money large Yoshis are wasting on these set pieces when there isn’t a drop of meaning to be found in terms of the bigger picture. On the other hand…

Postcards is interesting because it presents an example of what would happen if small-sized (or in this case solo) developers did a huge amount of unique visual content. Of course when you’re a small team you can only polish so much, so Postcards has very traditionally ‘unpolished’ visual art. which for me doesn’t matter at all outside of various commerrcial implications – But it goes to show that there is an interesting effect in scale, meaning arising as multiple distinct spaces start to connect to each other, like reading one word after another…

Anyways the point of this is for the love of god don’t make an 3d game with big open spaces unless you are okay spending a lot of time.

alright, not really going to edit this. good bye, time to go finish anodyne 2… ho ho


(EXCERPT) Tokyo Lo-fi: Shin Megami Tensei Nocturne

Hey all, so recently I published an essay on Shin Megami Tensei Nocturne in the Heterotopias journal – here’s an excerpt of it which I originally posted as a preview on Gamasutra.

This is an excerpt from Heterotopias 003, the third issue of the game and architecture zine, which is now available to purchase for $6. To read the rest of the article, which further analyzes the techniques used in Shin Megami Tensei III: Nocturne, go grab your copy over on

In Joni Kittaka and I’s eight-plus hour game Even the Ocean, I spent days fixing the animations of a mole that walks along tunnels in a single area, rather than replacing the mole with something less animation-heavy. This level of detail applied to every aspect of the game—such as cutscenes, art style revisions, and level design—contributed to the development time of three-and-a-half years.

For a long time I believed the only way to reduce the costs of content-heavy games with high-fidelity art was to make them smaller. But, recently, I’ve found another and perhaps better solution to reduce the traditional quality of the art. What stopped me from this discovery before was the prevalent attitude towards what are deemed ‘lo-fi graphics’—a widely-held view is that they are  inherently bad. But this shouldn’t be the case. They are not always a rush job or a mistake. They have, in recent years, even emerged as an art style in their own right. We can and should do more than dismiss lo-fi graphics as a historical stepping stone, inferior to photorealistic or HD art styles.

Consider games before 2010 with total or partial lo-fi art styles. Some of these older games aspired towards photorealism while lacking the hardware to do so effectively. Shin Megami Tensei III: Nocturne (hereon Nocturne) is a visually striking game, despite its ‘old’ graphics. First released in Japan in early 2003, it is a monster-collecting JRPG set in post-apocalyptic Tokyo. It had a team of about 15 artists, as opposed to the roughly 500 (counting contractors) artists that worked on Final Fantasy XV.

Nocturne  is full of ‘low-quality’ textures, objects, and spaces, yet its art style conjures an atmosphere more thematically dense and memorable than many contemporary games. Interrogating the art style shows Nocturne achieves this atmosphere through abstraction-based techniques for visual and spatial design. What is exciting about these techniques is that some of them do not require years of technical training in programming or visual arts, and thus, these techniques are easy for game developers to use to create new forms of visual expression in their games.

Nocturne’s setting

In Nocturne, Tokyo has just gone through a nuclear apocalypse, and so it has the player travel the city in order to decide how to rebuild it. Nocturne makes no attempt to create a realistic Tokyo, instead, it focuses on abstractions of public and commercial spaces like subways and malls, occasionally with more directly referenced real-world spaces, such as Shibuya Crossing.

While the artists aimed to dress the game’s spaces in realistic textures, the art direction is ethereal—this is accomplished through the eerie lack of human presence, a refusal to make accurate recreations of Tokyo spaces, and the use of unrealistic lighting. The latter is especially captivating, as it sees ambient lighting illuminate empty rooms in bizarre shades of purple or blue, lights cast in impractical ways, and shadows appearing without visible light sources.

Post-apocalypse Shinjuku Medical Center


Shibuya Crossing


Blurriness, sharpness, and immersion

Without actively looking, it’s harder to notice the lack of photorealism in the floor tiles of Nocturne’s subways than it is to notice it in a poster on the wall. Advertisements, printed materials, and commercial items like clothing appear throughout Nocturne, all of which invite players to parse them for written or visual information. But these objects aren’t suited to that type of inspection and only cause the fidelity of the game to fracture. Posters and ads appear too blurry or too sharp; a stack of newspapers resembles a cube rather than individual sheets, making it clear these were objects placed as atmospheric shorthands rather than sources of information.

These types of objects act strangely in games. Look at the “Yushima Station: Safety Message” picture. If the object is blurry, like the station exit diagram on the left, it immediately separates the player’s reality from the game’s should they attempt to read it. There also exists a problem when the object is clear and readable, like the warning sign on the right, as this isn’t a game about obeying train signs, and its sharpness makes it weirdly readable when almost everything else around it is hazy. The artist who placed this sign probably wanted it to stand as a nod to the pre-apocalyptic history of the station platform, as a way to ground it in a reality closer to the player’s, but its inconsistency with the textures around it pull it more towards achieving the opposite.

One solution to this would be to remove all the posters and signs, but that may make Tokyo strangely bare. Whereas rendering the posters blurry makes them too obviously fake, having them sharp enough to read causes them to stand out too much in this indistinct world. It is better to view these ‘problems’ as a technique that can, like the train station signs, be utilized to carefully place references to reality into a less realistic space.

Yushima Station: Safety Message and Station Exit Map


 To read the rest of this article purchase Heterotopias 003 over on

Sean Han Tani is a developer of Anodyne (2013), Even the Ocean (2016), and currently, All Our Asias. He also lectures about game design and game music production at the School of the Art Institute of Chicago.