Discussion

Since there was so much we could add to the game: different features, different kinds of models for the game objects, etc, we tried many kinds of approaches in making the game good. In the end though, the game didn't turn out to be that different from what we expected it to be when we first started this project.
The fact that we chose to improve a simple idea for a game, made it easier for us to start making changes to the game, since we already had a framework that could be improved further.
There were a lot of things that turned out very well, the models for instance. At the beginning of the project we kept the simple model for the ship and used spheres to represent the asteroids. During the development of the project we became more and more eager to try out new models. The problem was that obj-files were hard to come by and this resulted in conversion from collada to obj to pwn. To convert collada files to obj, we used the program Meshlab, which is a good program for converting between different 3-d formats, as well as making simple changes to the models.
This allowed for us to make use of some awesome models that we found online. There was however some problems, obviously the troublesome conversion procedure but also the fact that some models had to be transformed or altered to work properly. Some did not even work at all, which was infuriating when it was good-looking models.
The background sprite turned out pretty well, even though it increased the load times drastically.
It's quite hard to think of where the game could be improved. Performance wise could be to use a more optimized texture compession algorithm, to reduce the memory needed for the sprite, which is basically a texture. Other improvements would in our case have to be more nice features, like the player ship firing two shots at a time instead of one or adding more levels to the game.
Another thing that is featured in our game is the collision detecting. This is only used for the enemies so far, but is one of the things that isn't really as optimized as it could be. In theory the procedure of detecting a collsion is done correctly in our game, however there are certain aspects of the models that can complicate these calculations. Where the origin of the model is located f.eg. or the shape of it. Currently it creates a sphere round the center of origin of the model and checks if the shot passes through the inside of the sphere, in that case the instance disappears.
There's is finally one more thing that we were a tad irritated about, namely the rand function used to obtain a random state for each asteroid/enemyship as they were created. It didn't seem very random that the asteroids were created at the same locations every single time we ran the game, but since it seemed to work okay anyway we decided to keep it.