This week ANGELINA and I are featured in the Tech section of UK science magazine The New Scientist, which is a huge honour and was a huge amount of fun to do. As part of the interview, we made a platform game for their site in the same vein as the Santa games you might have played back in December. I made a few tweaks to the system to improve the layout of powerups and general map design – I really recommend you go check the game out! The game may appear up here eventually, but if the New Scientist link continues to work I will probably leave it only on their site.
The New Scientist exposure had also led to a bunch of other mentions of ANGELINA all over the net. I’m really honoured! Check out the Press page for more. Meanwhile, here’s what’s going on with the project…
I had a meeting with Simon Colton, my PhD supervisor, last week. We discussed GARNET, which is the arcade game engine I’d been working on and the subject of a bunch of posts on the front page right now. I was becoming frustrated with GARNET, spending too much time fixing things like collision code and not spending enough time actually investigating the things the system was for. So for now, GARNET is being put on ice and I’m going to work on something a little faster and hopefully more focused.
It’s a short project (although I said that about the platform generation, and I’m still working on that three months after it ‘ended’) with a clear aim: getting ANGELINA to move from making single levels into more complete games. One frequent criticism of ANGELINA is that it’s really only capable of making singular levels right now, and I’d really like to fix that. I’ve got a clear idea of a small project that should be able to show this off, and hopefully it’ll lead to a system that can design full games, with a steady increase in difficulty and (possibly) tutorial-style levels to teach game mechanics. All done automatically – in theory!
Another topic that rose out of the meeting was bringing mood into games. I’ve talked before about how hard it is to get ANGELINA to choose meaningful content for her games. So far we’ve had zero-content games, like the arcade stuff, and human-selected content in the platformers. Before this PhD is done and dusted, I’d like to show that ANGELINA could, in some small way, influence the mood a game conveyed while it was being played. Simon had a great suggestion for a collaboration between his AI artist, The Painting Fool, and ANGELINA. I’ll probably start work on that a bit later in the year, perhaps once I’m back from EvoStar 2012 in April.
Thanks to everyone who is sending kind words, interesting questions or hilarious Skynet references. I maintain this blog primarily so I can get my research seen and discussed by people who play games every day – your feedback is so incredibly valuable. Don’t forget you can email me whenever to ask about ANGELINA: