|
Cheating at 3D
I've written some 3D code which I'm in the middle of testing. It now seems obvious that there's little point in doing the full 3D perspective transformation for any object that's not actually really close to the viewer, the model of a Eurofighter that I'm using to test with becomes only 15 pixels wide when the model is set to 500 metres away. Much further than that and it's not really recognisable and things actually start to break down as the polygons become a bit degenerate. At some point the object just becomes a single pixel, and that's not much fun to watch.
So, I'm thinking of cheating for certain objects that are far enough away and just calculating a center point and then drawing them at fixed scale(s) without the perspective transform. I may even precalculate and blit these objects in.
So, I'm looking for opinions here. Is it going to look bad? Does anyone know of any examples where similar things are done? Should I just use a simpler model for far away objects and forget trying to be clever?
|