EPC2018 | Wave Function Collapse in Bad North | Oskar Stalberg

176K views · 8 years ago · talk · intermediate · evergreen

Oskar Stalberg's 2018 talk at the EPC game conference on using the Wave Function Collapse algorithm for procedural generation in his game Bad North. He explains how tile-based constraint solving builds varied, believable island levels.

HN treats it as the clearest explanation from the creator himself of how wave function collapse works in a shipped game, and cites it alongside his Townscaper work.

“The game Bad North used it to good effect, so depending on the game it may be a very useful tool in the toolbelt.”

10 comments from Hacker News

C
@Cognitron2023 · in “CityDreamer: Compositional Generative Model of Unbounded 3D Cities”
Oskar Stålberg has some great talks on YouTube about how the systems in Bad North and Townscaper work. ▶ this video link link link
RecommendsView on HN
s
@symmetricsaurus2022 · in “The “Wavefunction Collapse” generation algorithm explained clearly (2018)”
Oskar Stålberg has famously used WFC in Bad North and Townscaper. Here's a nice presentation from him explaining it: ▶ this video
RecommendsView on HN
b
@btown2021 · in “WaveFunctionCollapse: Generates bitmaps that are locally similar to the input”
The creator Oskar Stålberg has a great overview talk of Wave Function Collapse and how he applied it to his previous game Bad North - it’s absolutely worth a watch! ▶ this video
RecommendsView on HN
P
@PythagoRascal2024 · in “Tiny Glade 'built' its way to >600k sold in a month”
As far as I know they are using a customised variant of the "wave-function collapse" technique, used and popularised by Oskar Stalberg in his games "Bad North" and "Townscaper". The technique boils down to hand-crafting tons of tiles with adjacency rules about which tiles can slot together. When the user adds/removes a tile the algorithm iteratively tries to find fitting tiles and, if needed, changes neighbouring tiles for ones with the best transitions. He gave a talk where he goes into detail about this[1]. You can also find more if you google his name and "wave-function collapse". [1]: ▶ this video
CitesView on HN
n
@nico2024 · in “Tiny bitfield based text renderer (2022)”
Yes, it is probably its main usage The author of the GitHub repo is a pretty successful game author and has used it many times for his games Here’s a talk by him talking about procedural generation using wave function collapse: ▶ this video
b
@btown2021 · in “Townscaper Running in the Browser”
And here's another, slightly more technical talk from him about the "wave function collapse" algorithm he uses for procedural generation in his games: ▶ this video
h
@hesdeadjim2021 · in “WaveFunctionCollapse: Generates bitmaps that are locally similar to the input”
I played around a lot with this library: link Fun stuff, but I struggled to get a lot of value out of using it for level gen. You get cool patterns, but levels need structure and intent to be interesting. Adding constraints to the algorithm becomes a big-oh nightmare and you end up with frequently unsolvable paths as the algorithm recurses. The game Bad North used it to good effect, so depending on the game it may be a very useful tool in the toolbelt. ▶ this video
CitesView on HN
d
@detaro2021 · in “How to create a game like Townscaper?”
Its creator has a bunch of details on their Twitter (link although digging through long Twitter history is of course a pain) and given some talks: ▶ this video link
b
@bschne2020 · in “Townscaper – A building game inspired by the design of Scandinavian cities”
Not sure what of this was actually used in townscaper, but someone in another discussion of the game (can't remember where unfortunately) referred to these two sources: - Oskar Stålberg's 2018 Talk "Wave Function Collapse in Bad North" (▶ this video) - Müller et al, 2006 - "Procedural Modeling of Buildings" (link)
p
@pdoub2020 · in “Brick Block (2016)”
Please note that this is an old demo that has since grown into a full-blown game: link (Says Windows-only, but has a working Beta for Mac) Here's a talk by Oskar on how generating such worlds work: ▶ this video