For example, I wondered if I could do an 'outdoor' game so within a short while, I had a skybox up and running, with a terrain, some trees and a keyboard controlled camera.
I also had another idea of a 'bullet hell' style game, where instead of controlling a ship and dodging bullets, you would shoot bullets out and try to kill enemies. You would start off with a small amount of bullets and this would gradually increase until you were shooting hoards of bullets out.
I'd also like to have a go at writing a traditional horizontally scrolling shooter at some point.
My final thought was to create a 'retro' platform game, an 'homage' to the old ZX Spectrum games, but done in a modern style. To this end, I created a mapper and a scrolling, tiled 2d map and did some quick sprites on top.
I find this kind of prototyping really helps the development, it gives you a sense of what graphics and artwork you will need, helps with the control scheme and even highlights flaws in your game design - if something won't work or isn't fun then you can go back to the drawing board to fix it, without wasting a lot of time.
How do you guys go about deciding your next project? Am I alone in using this method? I'd love to know, drop some comments in below!