Game Design, Programming and running a one-man games business…

Procedurally generated blandness

There was a time when the two buzzwords guaranteed to generate hype and news coverage were the words ‘procedural’ and ‘generation’. They were most popular as ‘procedural generation’, less exciting when describing people as the ‘procedural generation’…anyway…

I’m not sure it really lived up to the hype. There was a time when we really needed this stuff. Elite couldn’t have generated an entire universe within 16k without it. And when you are doing a small indie game on a budget but want a large world, it can make sense. the problem is, you hand over control over design not to designers, but to mathematicians. Sure, some of the best developers come up with hybrid systems, where the designers are still in charge, but I do worry that we have gone too far down the road of ‘look lots of randomly generated stuff!’ and not enough down the ‘this is a wonderful hand crafted world’.

I love big open-world games, but I hate it when I start to recognize the maths behind it. Yup, another little fishing village I haven’t been to before but…isn’t this just the last fishing village with the houses at different angles and positions? is that *really* all we can do these days?

I find myself thinking about this because of Democracy 3. If you have played the game you might recognize the ministers screen. It has randomly generated minister portraits like these:


Before that, in Democracy 2 they were individually drawn like this:


I think D2’s look way better than D3’s. I think the random generation thing went too far. The problem is, with D2, you kept seeing the same faces again and again. I couldn’t afford the variety.  It wasn’t exactly game-wrecking, but even so, it was annoying. For Democracy 3: Africa we are going with a hybrid. The artist created all the assets and we are selecting a big bunch of individuals:


I think thats a good compromise. D2 had 13 ministers of each gender. Democracy 3 Africa already has 70 each, and will likely have more, and I think they will still look better than the base game truly random ones. Am I right?



7 thoughts on Procedurally generated blandness

  1. I think that the beneficial part of procedural generation for indies is that it allows for a shift from creating content to creating systems that create content. It lets games by small teams be much larger than would be possible with hand-crafting.

    And yeah, I agree that it’s often done badly, or done for the sake of doing it, or done in the wrong places or with the wrong content… but that’s just bad design. I think the “true art” of procedural generation is making it invisible, just the same way that good game art hides its seams.

  2. Postures, angles, lighting: they’re all the same on your procedural portraits. Compared to a google search like , that’s even worse than your fishing village example.

    That said, maybe if you distinguished your portraits by background hue and texture, they would be more easily distinguishable from each other.

    1. indeed that is a good point. They are all made up from matching components, so posture changes and lighting changes are difficult, although more color variant is definitely something I need to do.

  3. Perhaps the way is to add some more variation in the basic posture? Obviously, this means an increase in the number of assets, but it might be worth it for the variety.

    I’m currently working with an artist on a similar system myself (the reason why I’ve been looking at yours with some interest), and that is one thing that I’m thinking of trying. I’m also hoping to introduce some variation in the expression. Looking forward to seeing how that works.

    As it is, I don’t think procedural generation like this has any problems that can’t be solved by spending enough money; problem is, of course, that we (indies) usually go for this to save money. So it becomes a game of quality vs quantity.

  4. Looks like a slight shadowing from the hair overlay would have helped D3. Although shadowing was used more in D2 but presumably dropped in D3 to make composition easier. (or was it a style decision)

  5. Talking about Elite, people are really not ready for the upcoming ultra-bland No Man’s Sky….I watched their GDC lecture about how they generated the art (given by the art director, so more artistic than technical talk) and you can basically expect 2^64 variation of the same “cool” looking planets with weird animals, again and again, for ever and ever…etc…etc…

Comments are currently closed.