Procedural Dungeon Generation Algorithm

Here’s a really cool article by A Adonaac that describes, step by step, a neat algorithm for building dungeons. It’s based on a technique used by the developers of TinyKeep, using a physics-sim to separate a bunch of randomly placed cells that are then linked up based on a minimum spanning tree of the delaunay triangle graph of the detected rooms. 

Notice how Voronoi tessellation and Delaunay triangulation can come in handy when you’re building a generator. Another interesting thing to note is the use of a specific normal distribution rather than just picking a linear random value for cell size. This is a good example of how controlled chaos is the key to improving your generators.

A Adonaac goes into a ton of detail about how the procedural generator works and discusses some tweaks and potential pitfalls along the way.

http://gamasutra.com/blogs/AAdonaac/20150903/252889/Procedural_Dungeon_Generation_Algorithm.php

The 2015 ProcJam is only a couple of months away and in anticipation the official art pack by Marsh Davies has been released! Get started working on your next big procedurally generated ideas.

The art pack is licensed under Creative Commons BY-NC, and you don’t have to wait until ProcJam to start using it. Mark your calendars for November 7th!

EDIT: I knew I forgot something! Here’s the link to the announcement post and the download:
http://www.procjam.com/2015/09/procjam-art-pack-now-available/

Ravel Landscapes (live excerpts)

Built in vvvv by Quayola & Natan Sinigaglia, this realtime generative project translates the music of the live piano into a visible landscape. (The musician is Vanessa Wagner, and music was composed by Maurice Ravel.)

Random Github Project Name Generator

You’re all ready to start on your next great Javascript app, sure to bring you fame and fortune, but there’s one problem: you can’t think of a name!

Enter the random Github project name generator, written by Glenn Conner.   It generates a new name based on a Markov chain analysis of all of the popular Javascript libraries on Github. No doubt the next job apply to will want five years experience in requenceJS, brework, and Motifox. 

While generating random names is amusing, there’s also a practical use for it: generating a random name when you start a project lets you concentrate your creativity on building the thing rather than trying to come up with a clever name for it. Rather than name your next art project “Untitled_01″ try rolling some dice and name your next blank canvas something like “Raspberry Respad”.

http://mrsharpoblunto.github.io/foswig.js/

I got a alligator for a pet! 

A novel by @pentametron

Pentametron is a twitter bot by Ranjit Bhatnagar which searches twitter for tweets that are inadvertently written in iambic pentameter and retweets them as rhyming couplets. For NaNoGenMo, I got an alligator for a pet takes this one step further and outputs a book of sonnets.

The formal qualities of the rigid poetic structure act as a filter to the found-text source material, turning what were intended as ordinary mundanity into lyrical expression.

http://pentametron.com/alligator.pdf

Noctis IV

Explore a procedurally generated galaxy full of unknown planets.

Somewhere around 1996, a developer who goes by Alex released the earliest version of Noctis. Noctis IV, the last official version has been supplemented with community-developed mods.

The game is pure exploration, with 70 billion stars to explore, catalog, and share notes with other players. Some planets are lifeless hunks of rock, while others are full of life, like that rabbit-kangaroo thing I took a snapshot of. The planet I found seemed to have some flying lifeforms as well, but I never got close enough to get a good look at it.

The controls are a bit weird, and the DOS-based nature of the game is showing its age, but Noctis remains one of the landmarks of procedurally generated exploration games.

Temple of the Roguelike is blog-thing covering all varieties of roguelikes. While the blog itself hasn’t been super active lately, the archive stretch back to 2007 with lots of coverage of diverse roguelikes. Add in an active forum, and a searchable database of all of the roguelike announce at Roguebasin, and you’ll probably find enough roguelike-flavored things to keep you occupied for quite some time.

http://www.roguetemple.com/

Partitura 001

Sound visualisation is another application for procedural generation (not just Audiosurf). Particularly if you want to have a visualisation that works for arbitrary music, a generative process is basically required.

Created by Quayola and Abstract Birds, Partitura is custom software created with the vvvv, which is a toolkit for live programming. Its generation is based on a system of relationships, translating the inputs of music structure, gestures, and audio analysis into the coherent linear visualisation. Thus the structured input gives the output recognizable but unpredictable patterns, the ordered chaos that generative art thrives upon.

Then I Will

Created for NaNoGenMo 2013 by Rodrigo Lanas, Then I Will is a novel constructed out of phrases sampled from Twitter.

Using tweets as a source means that the raw material used as an input to the generator make at least a little bit of sense on their own. Using them in the generator recontextualizes them into a continuous first-person narrative of obsessive planning for the future, suggesting an elaborate Rube-Goldberg-esque plan that seems to start from writing a NaNoGenMo novel generator and escalate into a complex plan of revenge.

http://contraculto.com/then-i-will/

https://github.com/Contraculto/then-i-will

Fractals are an important tool for procedural generation.

A fractal is a pattern that has a repeating pattern over multiple scales. Since they create amplification–complex patterns from simple inputs–they can be an effective way to add detail. Additionally, since many natural things exhibit self-similar fractal patterns–mountain ranges, coastlines, trees, ferns, clouds, lightning–fractals are useful for approximating them. There 

The pictures above are successive frames from a deep zoom into the Mandelbrot set. As you zoom deeper, you can see that the larger structures are repeated at smaller and smaller scales, in different configurations, neverending. Zoom deep enough in the right place and you’ll find distorted versions of the whole Mandelbrot set that themselves contain distorted versions of the entire set.

One of the critical things that enabled Benoit Mandelbrot to explore fractals in depth was the then-new computer-visualisation of fractals, which allowed the previously abstract mathematical functions to be seen rather than just imagined or roughly approximated. Some of the earliest terrain generation came from these computer visualization experiments with fractals.