Case Study – Introversion

As part of my research into procedural and generative processes in games I have been carrying out a number of ‘case studies’ of prominent games and their designers. Usually this takes the form of an overview/analysis of the methods used, coupled with an associated interview. In this case I was lucky enough to grab hold of Chris Delay from Introversion Software after the Channel4 World of Love conference earlier this year.

Introversion are a small independent game developmet company based in the UK. After the success of their initial game Uplink the company became financially stable enough to produce further games. Throughout the development of their projects Introversion have strived to remain independent and frequently use procedural techniques to bypass various content development costs and also to generate interesting aesthetic and environmental elements for their games.

The following article was produced after an unstructured interview I carried out on the 25th June 2010. There is also follow-up email transcript at the base of the page.

Chris Delay is the lead creative programmer at Introversion, throughout his career he has used procedural and generative techniques in the production of the companys games. His usage of these techniques stems mainly from a need to solve various content creation issues, but there is also a clear sense in which the aesthetic results of such processes tie into the artistic vision of introversions games. Outsourcing content production or hiring additional design staff can potentially derail the creative direction and finiancial security of a small developer. Using mathmatical solutions to overcome these concerns was the primary motivation behind Delays experiments.

Delay sees this approach strictly as production method, the core gameplay of his projects resulting from more traditional rulesets. He expresses no real concern that the player is unaware of the generative processes behind the games produciton and tightly controls the results of these processes to directly serve the intended gameplay experience.

Delay makes reference to historical implementations of procedural generation in game design, citing Elite as a classic precendent of algorithmically built gamespaces. Although memory and data storage are less of an issue with todays hardware Delay feels there is still a real need to tackle exponential content detail and bandwidth issues, especially for independent developers. One of the key games in Introversions catalogue is the retro-futuristic RTS Darwinia.

Darwinia level map and view

Delay explains how the island terrain of Darwinia was created through a customised midpoint displacement algorithmn (diamons-square or plasma style). This routine produced a much wider range of structures than were applicable for the RTS style game design and through a combination of parametric adjustment and aesthetic selection the final island forms were selected. Similarly the tree designs were achieved through use of bespoke l-system functions and express a wide range of forms, from more naturalistic to obviously fractal and grid aligned models.

Fractal tree example

Darwinia demonstrates an obvious relationship between generation methods and the visual/narrative theme of the game. Its neon vector landscapes an flat lit polygons echo the feel of early 3D computer graphics and the graphic design of Tron. Using procedural methods to achieve these visual effects is an interesting example of how the appearance of computer-math forms has a strong nostalgic component and specific cultural reference. Delay deliberately included ‘loading screens’ in the launch portion of the game that also play on the digital remeniscence of tape based storage and demoscene cracks. In one of these ‘intros’ Delay incorporated a version of Conways life algorithm, humourously replacing the cells with tiny images of the games protagonists. This refers wryly to the larger story of the darwinian world as a mathmatical laboratory straining to achieve some sort of trancendental emergence.

Procedural city generation

In Subversion Delay has implemented a detailed city creation tool that allows the production of many different city types. These will utlimately form the background to the projects ‘mission imposible’ style gameplay. Again Delay insists that the range of possible city forms must be guided by the usability of them from a gameplay perspective. He even suggests that each city could be compressed to the initial seed and those seeds delivered to players as levels which will then be deterministically ‘grown’ to identical versions on their own machines. The buildings within the city are likewise procedurally designed based on specific stochastic nodegraphs, ensuring that pathfinding solutions are possible for every constructed space.

Office block construction processes

Delay expresses great respect for the detailed emergent simulations of Dwarf Fortress, but like many others finds the coarse design of both the interface and graphics a significant barrier to the player. Equally he admires the emergent complexity of Spore in terms of content generation but feels the gameplay is far less satisfying than it could have been.

Like most game designers he is reticent to consider his work in a digital art context, and although Introversion have engaged with several academic R&D excercises the player experience remains the main barometer with which any generative processes are tempered. Delay would consider producing alternative, less game-like versions of his software for other contexts but is concious of preserving the mystique of his production processes as another value added aspect of the main commercial game. In all their projects Introversion have maintained a consistent art style that reflects their interest in both the mathematical processes of gameplay and the visual representation of this experience in popular culture. Uplink pays hommage to the hacker aesthetic of the early 80s, Darwinia reflects the Tron era of visualising digital space and Defcon mimics the strategic military computer simulations of War Games and the cold war. Introversions stylistic approach is perhaps to reflect on the cultural reception of technologies rather than employ technologies for their own autonomous sake.

When considering generative techniques in a wider cultural or philosphical framework Delay expresses that the loops of technological progress (mobile devices retracing the hardware development of PCs from 10 years ago etc) inevitably lead to ideas being recyled in waves of popularity based both on the practicalities of production and the aesthetics of deisgn.

Email Interview 08 July 2010

Do you feel that the mathematical basis of much of Darwinia/Subversion
world-building directly influencs the visual aesthetic? (and is it at
all important for you that players realise the ‘science’ behind these
game design processes)

Yeah it definitely does. Procedural generation use in world building often leads to very mathematical looking results – if you compare a hand made landscape with a fractal generated landscape, you can tell a huge difference. People aren’t capable of reproducing the effect themselves, but you can tell a fractal landscape has been generated by an algorithm or taken from real data.

If the advancement of storage, memory and bandwidth has eliminated some of the
historical reasons for using procedural technology what do you see as
its key advantages

Although we now have basically unlimited memory and storage in our games (on pc certainly), a new challenge has come about : context is now extremely expensive to produce. I can take weeks of work to build levels for games now, and every year that passes the cost to produce content increases. This is not really sustainable, and mixing in procedural generation (even if it occurs at the developer end, rather than the gamer end) can speed that process up hugely. Procedural generation scales much better than human designers : a 10km city takes just as long as a 1km city for a generator, but for a team of artists that’s a massive difference.

Do you think generative /procedural systems have a place in ‘active’
gameplay as well as in content generation.

Yes, although it’s harder. Typically procedural generation techniques can be very cpu intensive while they are running, so you tend to run them in advance and cache the results – eg the levels for Darwinia are generated and stored in vertex arrays for the live game. Something like Valve’s director in Left 4 Dead is a procedural game system, in that it dynamically alters the game based on the status of the players, resulting in a lot of replayability.

When using procedural functions how do you isolate the desired result
(ie. do you start from a wide chaotic range and refine or determine
the outcome more tightly from the beginnning)?

Trial and error I guess!

How frequently does the result of your procedural content suprise, shock
or disappoint you?

Disappointment happens a fair amount, because it’s hard to totally avoid problems. My city generator for Subversion still sometimes generates road networks that run through buildings. It may seem obvious but in a 10Km city, its very hard to determine accurately “there are no faults in this city”.

But the real benefit is it always suprises me. The cities always look great, way better than I could design myself, and they always look different.

“Rather than using procedural techniques to invent new mathematical
constructions Introversions visuals tend to reflect the cultural
representation of these technologies in film and tv.”
Would you agre with this statement? If so what do you feel about this
relationship between complex software technologies and their

Yeah films and TV are very good at taking something and making it cool for the audience. So we use the blueprint style of rendering all throughout Subversion, because that’s how movies display the high-tech blueprints their protagonists are looking at.