Status update + New video: Basic animations & cloth simulation in goal net
Let's break the silence with a status update from last month. There's also a new video, showing up some of the results from the last months development. The quality of the video is pretty poor, but as long as our videos are distributed through youtube, we'll just have to live with a fact that they compress all the videos. We'll try to publish these status updates from time to time. It helps all of you to get a clue what's going on in FBL09 development process and what's going to happen next.
During the last month we managed to integrate all the basic moves for the players. This significally improves the gameplay and the feel of the game. It also proves that our current content path works and we are able to import the motion capture data in to the game. The content path, in general, actually works a lot better than we even imagined. One important thing to notice is also, that we made it to work in such a short time, which allows us to concentrate the actual quality of the animations.
To get the animations to the level we want, we still need to spend a lot of hours to tweaking them. For that, we need to build some tools, which ofcourse somewhat slows down the actual development. We still believe, that good tools are worth it in a long term thinking.
Here's a short list of the currently integrated animations
The game is unneccesarily big environment for just testing and debugging the AI solutions during development, and for this reason we have a separate "simulator" for AI - it's a greatly simplified version of the game in 2D with circle characters going around doing their thing, and it graphically illustrates how the agents experience the world which obviously helps a great deal in testing and verifying the correctness of solutions.
Here is a screenshot of the AI simulator and some things the AI can already sense and do.
Techniques and methods made use of
Agents can currently sense, detect and estimate things like
Fieldplayer agents can currently succeed in - besides basic actions like passing the ball around, shooting, moving and breaking the ball control - i.e.
PTeam agents, coaches, send out coordinative instructions to their members mostly through playing zones
Things to do
Gameplay has also been developed during animation integration and AI development work.