Cognitive Discodance is a one man enterprise buoyed by commissioned artists and friends. And when he gets busy buying a new house and getting married to an amazing, lovely lady, turns out gamedev pauses. So here we are, an 11 month delay after the last update. But that was all well worth it.
I really expected to balance life, work and gamedev, but that was a big ask. Luckily that’s all normalized (until we have kids, eek!) So I’ve been back at the gamedev grindstone.
I’ve been filling page after page of my notebook with ideas, theory and paper prototypes of The Day After. The core of the game is playing through stories generated by an AI director. The game is broken up into scenes and the Director chooses scenes based on how you’re going and what makes a good story. While I’ve had the theory for the scenes in a good shape for a long time now, I’ve only just nailed down a good model for the Director’s decision-making process. It’s a flexible structure, but falls neatly into some other computer science theory. So it might be new for games, but it’s not new for science. The best combination!
Since the honeymoon I’ve been working pretty consistently on Unreal Engine 4 and learning Blender. Apart from learning the software, I’ve been mocking up user interfaces and testing some really, really basic prototypes.
Currently I can populate a game screen with some fake scene data and then make it bounce between scenes on a button press, tracking a minimal bit of information – how good the game is going. The scenes are being read from a file (data driven programming!) instead of being hard-coded.
Not tricky at all, but I’m learning UE4’s quirks and experimenting how to do things correctly. In particular this code would transition to multiplayer without much issue. Which is not even in the vague realm of relevance at the moment, but nice to know.
I’m enjoying Blueprints, even though I could easily replicate the functionality in C++. I have a Pavlovian aversion to writing C++ since last time I was writing my own engine, the linking and compiling was doing my head in. I’ll need to write code for the AI Director, but the vast majority of gameplay code might be in Blueprints because it’s so easy to put together, tinker with and refactor.
Blender is starting to stick into my brain a bit better. I’m getting more adept at making simple inorganic objects and things like trees. I’m learning good techniques for UV mapping and good polygonal flow. I’m not cocky enough to try something complicated like a human character, but it’d be nice to build my skills.
I’ve been researching art styles and shader techniques. In particular I want to be able to render things like cartoons, which has been done before (toon/cel shaders via Sobel filters, for example). But looking at what other people have done, there’s quite a spectrum of quality available. I’d like high-quality toon rendering, so I’ve been reading the odd Siggraph, IEEE or Disney research paper. I’ve also been re-reading the old slides on Team Fortress 2’s art style and thinking about what I’d like. I think replicating Alice Carroll’s art style is a good initial goal. As a small art target, I’d like to replicate one of the commissioned pieces in 3D. Most of the props are easy to model (apart from the characters) and the lighting is interesting but not too complicated.
On the writing side, I’ve been filling out the character bible a little more. Story components are coming as easily as they ever have, but wordsmithing is hard since I do mostly technical reading and writing nowadays. Plucking a delicious phrasing from the brambles of thought is rewarding but hard.
When running headlong at The Day After is too hard, I’ve been tinkering with two smaller projects – a Twitterbot game and a tiny VR game. These are interesting projects in their own right, but side distractions to the main event that is The Day After.
So we’re back, but there’s so much more to do!